SCAPS manual

Marc Burgelman
Koen Decock, Alex Niemegeers, Johan Verschraegen, Stefaan Degrave

University of Gent
Department of Electronics and Information Systems (ELIS)
Campus Ardoyen, Technologiepark 914, Grote Steenweg Noord
9052 Gent-Zwijnaarde
‘Belgium’

Version: 8-4-2021
 Table of contents

Table of contents .............................................................................................................................................. i

Chapter 1 : About SCAPS ................................................................................................................................. 1
Chapter 2 : Getting started ............................................................................................................................... 1
2.1 The basics ...................................................................................................................................... 1
2.2 Run SCAPS: ..................................................................................................................................... 2
2.3 Define the problem: ....................................................................................................................... 2
2.4 Define the working point................................................................................................................ 2
2.5 Select the measurement(s) to simulate ........................................................................................... 2
2.6 Start the calculation(s): .................................................................................................................. 2
2.7 Display the simulated curves, ... ..................................................................................................... 3
2.8 …e.g. the I-V curves ...................................................................................................................... 3
2.9 Editing the problem ........................................................................................................................ 4
2.10 Speeding up: Batch calculations .................................................................................................... 4
2.11 Speeding up: Recorder .................................................................................................................. 4
Chapter 3 : Solar cell definition ....................................................................................................................... 5
3.1 Editing a solar cell structure ........................................................................................................... 5
3.2 Reference conventions for voltage and current .............................................................................. 5
3.3 Contacts ......................................................................................................................................... 8
3.4 Layer thickness ............................................................................................................................ 10
3.5 (Graded) semiconductor layers .................................................................................................... 10
3.5.1 Adding, duplicating, splitting, removing layers from the cell structure ........................................ 10
3.5.2 Temperature dependence of parameters ....................................................................................... 12
3.5.3 Grading: general approach ............................................................................................................ 12
3.5.4 Composition grading .................................................................................................................... 14
3.5.5 Parameter grading ......................................................................................................................... 15
3.5.6 The optical absorption constant () or (h) of a layer .............................................................. 19
3.5.7 The actual position dependent grading results .............................................................................. 24
3.5.8 A materials approach .................................................................................................................... 25
3.5.9 A frequently asked question (FAQ) about grading ....................................................................... 27
3.6 Defects and recombination ........................................................................................................... 28
3.6.1 Adding defects .............................................................................................................................. 29
3.6.2 Multivalent defects ....................................................................................................................... 29
3.6.3 Energetic distribution of defect levels .......................................................................................... 31
3.6.4 Impurity photovoltaic effect (IPV) ............................................................................................... 32
3.6.5 Short overview of defect values .................................................................................................... 32
3.6.6 Radiative and Auger recombination ............................................................................................. 33
3.7 Metastable defect transitions ........................................................................................................ 33
3.7.1 Principles ...................................................................................................................................... 33
3.7.2 Introduction of a metastable transition.......................................................................................... 34
3.7.3 Help, the buttons to introduce/edit metastable properties are not available! ................................. 35
3.7.4 Numerical settings ........................................................................................................................ 36
3.8 Interfaces ...................................................................................................................................... 36
Table of Contents ______________________________________________________________________ ii

3.9 Tunnelling .................................................................................................................................... 36

3.9.2 Numerical tunnel settings ............................................................................................................. 39
3.10 The blue button ............................................................................................................................ 40
3.11 Saving and loading problem definitions ....................................................................................... 40
Chapter 4 : Working point definition ............................................................................................................. 43
4.1 General ......................................................................................................................................... 43
4.2 Illumination conditions ................................................................................................................ 43
4.2.1 Internal SCAPS calculation ............................................................................................................ 44
4.2.2 Generation from file ..................................................................................................................... 45
4.3 Generation and spectrum models (SCAPS 3.3.05) ......................................................................... 45
4.3.1 The internal optical model of SCAPS ............................................................................................. 45
4.3.2 Analytical models for the generation and for the spectrum: motivation........................................ 47
4.3.3 New facilities in SCAPS 3.3.05 ...................................................................................................... 47
4.3.4 Analytical models for the generation G(x) .................................................................................... 48
4.3.5 The SCAPS Model Panel ................................................................................................................ 49
4.3.6 The generation models .................................................................................................................. 51
4.3.7 The spectrum models .................................................................................................................... 51
4.4 The initial working point .............................................................................................................. 52
4.5 Shunt conductance and series resistance ...................................................................................... 52
Chapter 5 : Single shot calculations............................................................................................................... 55
5.1 Calculation roadmap .................................................................................................................... 55
5.1.1 Meshing ........................................................................................................................................ 55
5.1.2 The pathway to a solution ............................................................................................................. 57
5.1.3 Small signal analysis .................................................................................................................... 58
5.2 Setting up a single shot simulation ............................................................................................... 58
5.3 Numerical parameters .................................................................................................................. 59
5.4 Numerical limitations ................................................................................................................... 60
Chapter 6 : Result analysis ............................................................................................................................ 61
6.1 Navigating to the analysis ............................................................................................................ 61
6.2 Zooming and scaling .................................................................................................................... 61
6.3 Curve info and legend .................................................................................................................. 62
6.4 Measurement specific options ...................................................................................................... 62
6.4.1 The energy band panel .................................................................................................................. 62
6.4.2 The generation-recombination panel ............................................................................................ 63
6.4.3 The IV-panel ................................................................................................................................. 63
6.4.4 The ac-panel ................................................................................................................................. 64
6.4.5 The CV-panel ............................................................................................................................... 64
6.4.6 The Cf-panel ................................................................................................................................. 66
6.4.7 The QE-panel ............................................................................................................................... 67
6.5 Managing measurement data ........................................................................................................ 67
6.5.1 The Manage measurements panel ................................................................................................. 68
6.5.2 Structure of a measurement file .................................................................................................... 68
6.6 Saving results ............................................................................................................................... 70
Chapter 7 : Batch calculations ....................................................................................................................... 75
7.1 The batch set-up panel.................................................................................................................. 75
7.1.2 Custom defined values .................................................................................................................. 76
7.2 Varying entire definition files ...................................................................................................... 77
7.3 Varying parameters of the initial state work point ........................................................................ 77
7.4 During calculation… .................................................................................................................... 77
Chapter 8 : Recorder calculations .................................................................................................................. 79
8.1 Setting a recorder ......................................................................................................................... 79
8.2 Recorder calculations ................................................................................................................... 80
8.3 Analysing the recorder results ...................................................................................................... 80
Chapter 9 : Curve fitting ................................................................................................................................ 83
9.1 General principles ........................................................................................................................ 83
9.2 Setting up the curve fitter ............................................................................................................. 83
 ____________________________________________________________________ Table of contents

9.3 Defining groups of simultaneous curve fitting parameters ........................................................... 86

9.4 Analysing the curve fitting results ................................................................................................ 87
9.5 Curve fitting résumé ..................................................................................................................... 89
Chapter 10 : Scripting ................................................................................................................................... 91
10.1 Running SCAPS externally ............................................................................................................ 91
10.2 Running a script ........................................................................................................................... 91
10.2.1 Automate mouse clicking ........................................................................................................... 92
10.2.2 Run external programs within SCAPS .......................................................................................... 92
10.2.3 Running dynamically linked libraries ......................................................................................... 92
10.3 The script editor ........................................................................................................................... 92
10.4 The SCAPS script language ........................................................................................................... 93
10.4.1 General ....................................................................................................................................... 93
10.4.2 Load commands.......................................................................................................................... 94
10.4.3 Save commands .......................................................................................................................... 95
10.4.4 Action commands ....................................................................................................................... 98
10.4.5 Clear commands ....................................................................................................................... 102
10.4.6 Set commands........................................................................................................................... 103
10.4.7 Get commands .......................................................................................................................... 111
10.4.8 The extract command ............................................................................................................... 119
10.4.9 Math commands ....................................................................................................................... 120
10.4.10 Loop commands ..................................................................................................................... 126
10.4.11 Show command ...................................................................................................................... 127
10.4.12 Plot commands ....................................................................................................................... 128
10.4.13 Calculate commands ............................................................................................................... 128
10.4.14 The run commands ................................................................................................................. 129
Chapter 11 : Script support for simulating multi-junction (tandem) cells .................................................... 132
11.1 Introduction ................................................................................................................................ 132
11.2 Problems with the direct simulation of tandem cells in SCAPS ................................................. 132
11.3 Basic strategy: calculate top and bottom cell separately, place them in series afterwards .......... 132
11.4 Splitting the tandem cell into top and bottom cells ..................................................................... 133
11.5 Strategy 1: ‘adapt generation’ .................................................................................................... 133
11.6 Strategy 2: ‘adapt the optical filters at the contacts’ ................................................................... 134
11.7 Strategy 3: ‘replace parts of the cell with electronically inactive layers’ .................................... 136
11.8 Connecting the top and bottom cells in series to a 2-terminal tandem cell ................................. 138
11.9 Accessing the internal optical properties of a cell ...................................................................... 139
11.10 Application examples ................................................................................................................. 140
11.10.1 Introduction ............................................................................................................................ 140
11.10.2 The illumination, absorption and generation conditions ......................................................... 141
11.10.3 Results of the tandem simulation with ‘adapt filters’ .............................................................. 142
11.10.4 Comparison of the 3 script strategies for tandem cells ............................................................ 144
11.11 Conclusion ................................................................................................................................. 146
References ................................................................................................................................................... 147
 Chapter 1: About SCAPS

SCAPS is a one dimensional solar cell simulation program developed at the department of Electronics and
Information Systems (ELIS) of the University of Gent, Belgium. Several researchers have contributed to it's
development: Alex Niemegeers, Marc Burgelman, Koen Decock, Johan Verschraegen, Stefaan Degrave. A
description of the program, and the algorithms it uses, is found in the literature [1-6].
The program is freely available to the PV research community (universities and research institutes). It runs
on PC under Windows 95, 98, NT, 2000, XP, Vista, Windows 7, and occupies about 50 MB of disk space.
The program can be freely downloaded (but...: don't sell, don't distribute further, refer when you publish
results obtained with SCAPS). Please report to Marc Burgelman ([email protected]) when you
have downloaded a SCAPS version (your name and your institution name and address, and the promotor's
name for doctorate's students).
Up to now, there was no consistent manual for the program but there was (and is) a collection of add-on
manuals describing the novelties in every new version. This manual is based on those previous documents.
Also, there are two short and recommendable documents: a text Getting Started 2011.pdf, and a presentation
SCAPS introduction.pdf. These documents are doing exactly what they promise.
SCAPS is originally developed for cell structures of the CuInSe2 and the CdTe family. Several extensions
however have improved its capabilities so that it is also applicable to crystalline solar cells (Si and GaAs
family) and amorphous cells (a-Si and micromorphous Si). An overview of its main features is given below:
 up to 7 semiconductor layers
 almost all parameters can be graded (i.e. dependent on the ocal composition or on the depth in the cell):
Eg, χ, ε, NC, NV, vthn, vthp, µn, µp, NA, ND, all traps (defects) Nt
 recombination mechanisms: band-to-band (direct), Auger, SRH-type
 defect levels: in bulk or at interface; their charge state and recombination is accounted for
 defect levels, charge type: no charge (idealisation), monovalent (single donor, acceptor), divalent (double
donor, double acceptor, amphoteric), multivalent (user defined)
 defect levels, energetic distributions: single level, uniform, Gauss, tail, or combinations
 defect levels, optical property: direct excitation with light possible (impurity photovoltaic effect, IPV)
 defect levels, metastable transitions between defects
 contacts: work function or flat-band; optical property (reflection of transmission filter) filter
 tunneling: intra-band tunneling (within a conduction band or within a valence band); tunneling to and
from interface states
 generation: either from internal calculation or from user supplied g(x) file
 illumination: a variety of standard and other spectra included (AM0, AM1.5D, AM1.5G,
AM1.5Gedition2, monochromatic, white,...)
2 Chapter 1: About scaps

 illumination: from either the p-side or the n-side; spectrum cut-off and attenuation
 working point for calculations: voltage, frequency, temperature
 the program calculates energy bands, concentrations and currents at a given working point, J-V
characteristics, ac characteristics (C and G as function of V and/or f ), spectral response (also with bias
light or voltage)
 batch calculations possible; presentation of results and settings as a function of batch parameters
 loading and saving of all settings; start-up of SCAPS in a personalised configuration; a script language
including a free user function
 very intuitive user interface
 a script language facility to run SCAPS from a ‘script file’; all internal variables can be accessed and
plotted via the script.
 a built-in curve fitting facility
 a panel for the interpretation of admittance measurements
 Chapter 2: Getting started

2.1 The basics

SCAPS is a Windows-oriented program, developed with LabWindows/CVI of National Instruments. We use
here the LW/CVI terminology of a ‘Panel’ (names used in other softwares are: a window, a page, a pop-
up…). SCAPS opens with the ‘Action Panel’.

3.

4.

2.

5. 6.

Figure 2.1 The SCAPS start-up panel: the Action panel or main panel. The meaning of the blocks numbered 1 to 6 is
explained in the text.

There are dedicated panels for the basic actions:

1. Run SCAPS .
2. Define the problem, thus the geometry, the materials, all properties of your solar cell
3. Indicate the circumstances in which you want to do the simulation, i.e. specify the working point
2 Chapter 2: Getting started

4. Indicate what you will calculate, i.e. which measurement you will simulate.
5. Start the calculation(s)
6. Display the simulated curves, ... (see section 6)
This is further explained below.

2.2 Run SCAPS:

Click the above pictogram on the Desktop, or double-click the file scaps3200.exe in the file manager
(or any other SCAPS version). SCAPS opens with the Action Panel.

2.3 Define the problem:

Click the button set problem in the action panel, and chose load in the lower right corner of the panel
that opens. Select and open e.g. the file NUMOS CIGS baseline.def: that is the example problem file
of the practicum session at the NUMOS workshop, Gent, 30 march
2007. This file is supposed to be in the folder /scaps/def, where
/scaps/ stands for the directory where you installed SCAPS, and
where the SCAPS .exe file resides. If necessary, browse to find this
file. In a later stage, you can alter all properties of the cell by clicking set problem in the action panel.

2.4 Define the working point

The working point specifies the parameters which are not varied in a measurement simulation, and which
are relevant to that measurement. Thus:
 the temperature T: relevant for all measurements. Note: in SCAPS,
only NC(T), NV(T), the thermal velocities, the thermal voltage kT
and all their derivatives are the only variables which have an
explicit temperature dependence; you must input for each T the
corresponding materials parameters yourself.
 the voltage V: is discarded in I-V and C-V simulation. It is the dc-bias voltage in C-f simulation and in
QE() simulation. SCAPS always starts at 0 V, and proceeds at the working point voltage in a number of
steps that you also should specify.
 the frequency f: is discarded in I-V, QE() and C-f simulation. It is the frequency at which the C-V
measurement is simulated.
 the illumination: is used for all measurements. For the QE() measurement, it determines the bias light
conditions. The basis settings are: dark or light, choice of the illuminated side, choice of the spectrum. A
one sun ( = 1000 W/m2) illumination with the ‘air mass 1.5, global’ spectrum is the default, but you have
a large choice of monochromatic light and spectra for your specialized simulations. If you have an
optical simulator at your disposal you can immediately load a generation profile as well in stead of using
a spectrum.

2.5 Select the measurement(s) to simulate

In the action-part of the Action Panel, you can select one or more of the following measurements to simulate:
I-V, C-V, C-f and QE(). Adjust if necessary the start and end values of the argument, and the number of
steps. Initially, do one simulation at a time, and use rather coarse steps: your computer and/or the SCAPS
program might be less fast than you hope, or your problem could be really tough... A hint: in a C-V
simulation, the I-V curve is calculated as well, no need then to specify it separately.

2.6 Start the calculation(s):

2.7 Display the simulated curves, ... 3

Click the button calculate: single shot in the action panel. The Energy Bands Panel opens,
and the calculations start. At the bottom of the Panel, you
see a status line, e.g. “iv from 0.000 to 0.800 Volt: V =
0.550 Volt”, showing you how the simulation proceeds.
Meanwhile, SCAPS stands you a free movie how the
conduction and valence bands, the Fermi levels and the
whole caboodle are evolving. When you see the hated
divergence message, you’re entitled to get into a bad
mood, but don’t exaggerate. Anyway, you did not loose
the I-V points already calculated.

2.7 Display the simulated curves, ...

After the calculation(s), SCAPS switches to the
Energy band panel (or the AC-band panel). You Save Results
can now look at your ease to the band diagrams,
carrier densities, current densities,… at the last
bias point calculated (stop your calculations
earlier, or use the pause button on the Action Panel
if you want to look at an intermediate state at Go to other panels
ease). You can output the results (buttons print,
save graphs, show (then the numbers are
shown on screen; cut & paste to e.g. Excel is
possible), or save (then the numbers are saved to
a file). You can switch to one of the specialized output Panels (if you have already simulated at least one
corresponding measurement). We only show the example of the IV Panel.

2.8 …e.g. the I-V curves

The meaning of the plot, show or save buttons is as for the Energy Bands Panel. Again, you can switch
to the other output panels (energy bands, ac, C-V, C-f and QE, if already calculated), and to the Action Panel
to do a new calculation, or to stop (important: you can only leave SCAPS from the Action Panel!). Several
small remarks:
The color of the last calculated curve is indicated (tip: when the graph gets too crowded, go to the Action
Panel and click clear all
simulations to clear all graphs). The 2.
recombination curves are only shown for
the last simulation. The color of the legend 5.
5.
corresponds to the color of the curve 1bis.
(indicated as 1bis).
If Curve Info is switched ON and you
click the cursor on a curve in a graph, a
pop-up panel will appear which gives
3.
information about the graph, curve and the
point which you clicked.
1.
Here you can display a measurement file
4.
(only one measurement at a time!). Select
e.g. the file Numos Ex 1 light.iv or Numos Ex 1 dark.iv which you should find under
/SCAPS210/measurements.
4 Chapter 2: Getting started

Hint: when you are doing many simulations, be friendly and helpful to yourself, and write some comments in
the comment box before printing the Panel: you’ll be glad to have done so when the time of writing (an
article, your doctorate) comes…
You can change the range and scaling of the axes with the Scale button. If you press the CTRL-button and
select a rectangular area in a graph, the graph will zoom-in to the selected area. Pressing the CTRL-button and
clicking the right mouse button results in zooming out.

2.9 Editing the problem

Go to the Action Panel, click set problem. You are now in the Solar Cell Definition Panel. Click on a
layer name, and you enter the Layer Properties Panel where you can change all parameters of that layer. Use
your intuition and/or read the rest of this manual.

2.10 Speeding up: Batch calculations

When you want to explore the influence of one or a few parameters to the solar cell characteristics, you can
take profit of the batch option. When you click Batch set-up, a panel opens where you can choose
which parameter to vary, over which range, and in which mode (Lin, Log or custom). You can also define
more than one parameter, and vary all of them (in a nested way or ‘simultaneous’), but be modest to start. A
batch calculation is launched when calculate: batch is clicked.

2.11 Speeding up: Recorder

In a regular single shot or batch calculation, the detailed panels are only available for the last measurement
point. To be able to see them as a function of the batch parameters you can launch a record calculation. You
should first select the properties which you want to keep track of by clicking Record set-up. Browse
through the property-lists, and don’t forget to press one of the insert buttons to add a property to the
recorder list. By clicking calculate: recorder, a recorder calculation is launched. Cell parameters are
varied according to the Batch set-up, and all simulations are performed which are needed to determine
the asked properties. This means the selected measurements on the action panel are ignored!
 Chapter 3: Solar cell definition

The recommended way to introduce your solar cell structure into SCAPS is to use the graphical user interface.
This way you can interactively set all parameters while SCAPS watches over you, so that you don’t define
impossible or unrealistic situations. This chapter explains which situations can be modelled and how to
introduce them in SCAPS.

3.1 Editing a solar cell structure

When clicking the ‘Set Problem’-button on the action panel, the ‘Solar cell definition’-panel is displayed.
This panel allows to create/edit solar cell structures and to save those to or load from definition files. These
definition files are standard ASCII-files with extension ‘*.def’ which can be read with e.g. notepad. Even
though the format of these files seems self-explaining it is however strongly advised not to alter them
manually.
Layer-, contact-, and interface properties can be edited by clicking on the appropriate box as shown in
Figure 3.1. In a similar way, layers can be added by clicking ‘add layer’

Figure 3.1 Defining a solar cell structure

3.2 Reference conventions for voltage and current

The user can input own reference conventions for the applied voltage V and the current J in the external
contacts. When setting a new problem, or editing an existing problem that does not contain any reference
data (e.g. an older .def file), the new options in the solar cell definition panel (Figure 3.2 right) are invisible,
and the default reference conventions are set. Upon checking the option in the More Numerical Settings
Panel (Figure 3.2 left), these options are visible and can be operated right away. When a newer problem is
loaded that contains reference information, the checkbox ‘allow change of…’ is set automatically, and the
6 Chapter 3: Solar cell definition

three options of Figure 3.2 right are enabled. (As of 2-1-2014, this More Numerical Settings Panel is not yet
available to the user; the option “allow change of … references” is always enabled).

Figure 3.2 Setting user reference conventions for voltage and current. Left: checkbox in the More Numerical
Settings Panel. Right: new facilities in the Solar Cell Definition Panel.

The three new facilities are:

1. ‘apply voltage V to’: when ‘left’ is set, then the right contact is the reference contact, and the
voltage V is applied to the left contact; this is the default, and the only possible option in
SCAPS<3.2.0. When ‘right’ is set, the left contact is the reference contact, and the voltage V is
applied to the right contact; in an JV curve, this correspond to a reversal of voltage axis
compared to the traditional JV curves in SCAPS.
2. ‘current reference as a’: when ‘consumer’ is set, then the current reference arrow is set such that
P = JV is the power consumed by the cell, and thus - JV the power generated by the cell.
When ‘generator’ is set, then the current reference arrow is set such that P = JV is the power
generated by the cell, and thus - JV the power consumed by the cell. Setting of the current
reference arrow thus depends both on the selected voltage reference and on the
consumer/generator selection.
3. ‘Invert the structure’: the solar cell structure is mirrored along the x axis: the leftmost layer
becomes the rightmost layer, and so on. This inversion of structure also swaps the interfaces,
and all grading information in the layers and the defects. Clicking two times the inversion
button brings the original cell back. This inversion only concerns the structure: the illumination
side, the voltage and current reference settings all remain unchanged.
With these 3 settings, one can define 8 different problems, resulting in 4 different aspects of the JV curves.
We illustrate this in Figure 3.3 and Figure 3.4, to make the user more familiar with the concepts of voltage,
current and power references.
3.2 Reference conventions for voltage and current 7

pn
V to left contact
referred as
consumer

pn
V to right contact
referred as
consumer

pn
V to right contact
referred as
generator

pn
V to left contact
referred as
generator

Figure 3.3 Possible references of V and J for pn structures. The calculations are for the problem file simple pn.def,
and the illumination is always from the right.
8 Chapter 3: Solar cell definition

np
V to left contact
referred as consumer

np
V to right contact
referred as
consumer

np
V to right contact
referred as
generator

np
V to left contact
referred as
generator

Figure 3.4 Possible references of V and J for np structures. The calculations are for the problem file simple pn.def,
and the structure is inverted with the button ‘Invert the structure’; the illumination is always from the right.

Internally in SCAPS, only the default reference is used (voltage applied at the left contact, current reference
arrow from left to right, resulting in a reference as a consumer. In all output (graphs, show/save tables), the
result is shown consistent with the user’s choice of reference. Note that the electric field in the SCAPS output
is not subject to the user-set V and J references: it is always referred to the positive x-axis, thus from left to
right.

3.3 Contacts
The contact properties can be set by either clicking the front or back contact button on the cell definition
panel, which opens the ‘contact properties panel’, Figure 3.5.
3.3 Contacts 9

Figure 3.5 Contact properties panel.

The metal work function Φm (for majority carriers) can be input by the user. However, the user can also
choose the option “flat bands”. In this case, SCAPS calculates for every temperature the metal work function
m in such a way that flat-band conditions prevail. In SCAPS versions before 1-1-2014, a simplified algorithm
described below was used. When the layer adjacent to the contact is n-type Eq.(1) is used, when it is p-type
Eq.(2) is used, when it can be considered to be intrinsic Eq.(3) is used.
 NC 
 m    k BT ln   (1)
 ND  N A 

 NC 
 m    E g  k BT ln   (2)
 N A  ND 

N 
 m    k BT ln  C  (3)
 ni 
As can be seen, only the shallow doping density is taken into account in order to calculate the flat band metal
work function. When there is a considerable amount of charge in defects present near the contact however, it
is thus possible that the flat band option will not lead to flat bands .
In recent SCAPS versions (after 1-1-2014), also charge in deep defects is considered; an algorithm
involving the solution of a non-linear algebraic equation (expressing that total charge = zero) then replaces
the simple equations (1) to (3).
When the layer next to the contact is either n- or p-type (NOT INTRINSIC) a recalculation of the barrier
height with respect to the Fermi level and conductance/valence band are calculated and displayed in the
contact properties panel. These values however only serve as an indication to the user, they are not used in
the simulation.
At the contacts a (wavelength dependent) reflection/transmission R() or T() can be set, see §4.2.1.
These can be set either as a constant value (wavelength independent) or as a filter file. These filter-files are
standard ASCII-files with the extension ‘*.ftr’. Several files are provided with the SCAPS installation,
however, the user can easily make more files. If a line in this file can be interpreted as starting with at least
10 Chapter 3: Solar cell definition

two numeric values, the first value is interpreted as the wavelength (in nm) and the second as the reflection
(in %). All other lines are ignored and treated as comment. Often a SCAPS simulation will need R() and T()
at a wavelength outside the range specified in the filter file: this is e.g. the case if an R() file was specified
between  = 300 nm and  = 1100 nm, and a simulation was asked under illumination with the default
spectrum AM1_5G 1 sun.spe that is specified between 305 nm and 4045 nm. Then extrapolation will be
used for the spectrum wavelengths 1100 nm <  < 4045 nm. For the SCAPS extrapolation rules, see Section
3.5.5.3.

3.4 Layer thickness

Until SCAPS 3.3.00, January 2014, the input values of layer thickness was always in units of m
(micrometer), the thickness range was from 1 nm = 0.001 m to 10 cm = 105 m, and thickness was always
displayed with three decimal digits shown. This was annoying when one had input d = 0.0025 m (thus 2.5
nm): SCAPS was calculating with 2.5 nm, but the thickness display showed 0.002 m (thus 2 nm); and input
of a thickness lower than 1 nm was automatically set to 0.001 m = 1 nm.
Since SCAPS 3.3.00, August 2014, the allowed input range of thickness is extended to 0.01 nm = 10 -5 m at
the thin side to 1 m = 106 m at the thick side. The number of decimal digits is still limited to 3, but the user
can select several units for the thickness display: Å (1 Ångström = 0.1 nm), nm, m, mm and cm. When the
unit selected is not the traditional micrometer, the value and unit are displayed in magenta colour. When the
display unit was set e.g. to m, and one had input 0.00245, the display would show the value rounded off to
3 digits: 0.002 m; but when one selects nm as display unit, 2.45 nm will be shown (see Figure 3.6). When a
new problem is loaded from file, thickness will be displayed in nm if d < 10 nm; in m if 10 nm  d  1000
m; and in mm if d > 1 mm.

Figure 3.6 Display of layer thickness, when d = 0.00245 m was input: once with m as display unit (in black, left),
and once with nm as display unit (coloured, right).

3.5 (Graded) semiconductor layers

All parameters of a semiconductor layer can be edited by clicking ‘add layer’ or on the appropriate layer
button in the ‘Solar cell definition panel’. All properties can be graded, as will be discussed below. However,
some remarks not related to grading should be made first.

3.5.1 Adding, duplicating, splitting, removing layers from the cell structure
By right-clicking a layer button in the ‘Solar cell definition panel’, a panel opens where you can remove this
layer, or duplicate it, or split it, see Figure 3.7.
3.5 (Graded) semiconductor layers 11

Figure 3.7 Panel to remove a layer, or to duplicate or split it.

With ‘duplicate’ an identical layer is inserted after (= to the right of) the layer you right-clicked; in
particular, the inserted layer has the same thickness as the original layer. The split option is there from SCAPS
3.3.01, 4-3-2015 on. With split, the thickness and the grading properties (see below) of the original layer are
changed. The original layer is called the ‘left split layer’, and the inserted layer is called the ‘right split
layer’. The splitting action conserves the thickness of the original layer. Upon clicking ‘split’ in the panel of
Figure 3.7, the Split Layer Panel of Figure 3.8 opens. There, you can set the thickness of the left split layer or
of the right split layer, as absolute thickness in m or nm, or as relative thickness (a fraction of the original
thickness). The name of the layer and its duplicate are changed, e.g. in the example of Figure 3.8 with
original layer name ‘p-layer’, the left split layer will be named ‘p-layer left’ and the right split layer ‘p-layer
right’.

Figure 3.8 The Split Layer Panel allows to split a layer whilst conserving the total thickness and the grading
properties of the original layer.

Also, the grading properties of the split layers are adapted so that the overall grading of all properties is
conserved (advancing to the terminology explained in 3.5.3). This works perfect for most grading types. For
‘exponential grading’, a set of transcendental equations has to be solved, we hope that this will work fine for
all values you would try, but there is no guarantee. For ‘Beta-function grading’, it is mathematically not
possible to find Beta-function solutions for the grading in both split layers. SCAPS will set the Beta-function a
and b of the split layers in a very rough approximation, that will be unsatisfactory in many cases, especially
when a > 1 and b > 1. It is up to the user to check the results, and judge if they are acceptable or not.
For ‘grading from file’, two new grading files are created for the split layers, and their names are the
original grading filename preceded by ‘leftsplit_’ or ‘rightsplit_’. Exception: (SCAPS 3.3.03, february 2016)
When the grading file has the range: shared over adjacent layers keyword, the grading file is used without
modification in the two split layers; no two files with different names are created.
12 Chapter 3: Solar cell definition

When there is no space to duplicate or split a layer, i.e. when the maximum number of layers is already in
use (actually 7), the duplicate and split buttons are disabled (and have an appropriate label). Also, you cannot
remove a unique layer from a cell structure; when you try, the remove button will be disabled and have an
appropriate label.

3.5.2 Temperature dependence of parameters

The density of states in the conduction/valence band are temperature dependent according to (4). T0 is the
default temperature (at which the parameter value should be defined in SCAPS and equals 300 K). Likewise
the thermal velocity is temperature dependent (5). All other parameters are assumed to be temperature
independent. The diffusion coefficient D=μkT/q which is used in the calculations is temperature dependent
(contrary to what was mentioned in the very first SCAPS manual (version 2.0))
1.5
T 
N C T   N C T0   
 T0 
(4)
1.5
T 
NV  T   NV T0   
 T0 
0.5
T 
vth T   vth T0    (5)
 T0 
Should you want to give a temperature dependence to other parameters, e.g. Eg(T), n(T), p(T), …, it is
entirely your task. The way to do so is:
In the Batch Set-up, define the temperature T as a batch parameter. Define for example Eg as a next,
‘simultaneous’ batch parameter, and set its value from an Eg(T) list (should the temperatures T in your list
not form a regular series, you should set the T values also from a list). You should take care to save both lists
and maintain them together.

3.5.3 Grading: general approach

All layer-parameters can be graded. The principles of the algorithms used to simulate graded solar cell
structures have been presented in [3]. To give a suitable and materials oriented description of the grading of
the various materials parameters, SCAPS derives all parameters consistently from the composition grading of
a layer. Each layer is assumed to have composition A1-yBy. The user defines the properties of the pure
compounds A (e.g. A = CuInSe2) and B (e.g. B = CuInS2), and the composition grading y(x) over the
thickness of the layer: thus defining the composition values y at the left and right side of the layer, and by
specifying some grading law in between. All materials properties P are then derived from the local
composition parameter y(x), that is, P[y(x)] is evaluated. Several grading laws are implemented in SCAPS and
offered by the user interface: uniform, linear, logarithmic, parabolic (two laws), power law, exponential,
effective medium, from file and a Beta function. These grading laws can be used to set the composition
grading y(x) over a layer, as well as to set the composition dependence P(y) of a property. These grading
laws and their parameters can be set on the Grading panel, see Figure 3.9. The grading lays used are
summarized in Table 3.1.

Table 3.1 Available basic grading laws. SCAPS takes care of possible numerical problem, that could occur for very
small or very large arguments; hence these laws are not always strictly followed but are more complicated under
specific circumstances.
Name P  y  remarks

Uniform PA  PB
3.5 (Graded) semiconductor layers 13

Linear PA  yB  y   PB  y  y A 
yB  y A
Parabolic PA  y B  y   PB  y  y A   y  y  y  y A  b: bowing factor
b B
yB  y A  yB  y A 2
Parabolic2 2 There are two parabolas,
 y  y0 
y  y0 : P0   PA  P0   
one to each side of the point
 y A  y0  [y0,P0] which can be given
in the user interface.
2
 y  y0 
y  y0 : P0   PB  P0  
For extrapolation a fourth

 y B  y0 
order equation is used

Logarithmic yB  y y  yA
yB  y A yB  y A
PA  PB
Exponential  y  y  y  yA  P0: background value
sinh  B  sinh   LA,B: characteristic lengths
P0   PA  P0   LA    P  P   LB 
B 0
 yB  y A   y  yA 
sinh   sinh  B 
 LA   LB 
Beta function  y  yA  βa,b(x) is the incomplete
PA   PA  PB  a ,b   beta-function (Ix(a,b) in [7]
 yB  y A  on p.226)
Power law m m: power
 1/ m yB  y y  yA 
 PA   PB1/ m  
 yB  y A yB  y A 
Effective P P P P Not available for
medium  yB  y  A   y  yA  B  0 , thus: composition.
PA  2 P PB  2 P
It is the Bruggeman
equation.
b  b2  8PAPB
P with
4
( y  y A )  2 PB  PA   ( y B  y )  2 PA  PB 
b
yB  y A
Two diffusion-type grading laws were implemented in SCAPS 3.3.06, december 2017. They are only
available for graded densities, dependent on position x (thus not: dependent on composition y), thus
for shallow doping densities ND(x), NA(x) and for defect densities Nt(x). In both laws, the input
parameter xdiff represents a diffusion depth, and is given by xdiff  2 D Tdiff  tdiff .


 xx
 
Diffusion: 2 For dopants or impurities
exp   
N tot
N  x 
Gauß law proj diffusing in a layer from a
  xdiff 2  limited source, that contains
xdiff   a total amount of impurities
2 Ntot (cm-2). The parameter
xproj is the ‘projected range’.
It describes the penetration
of the (maximum position
of) the ion beam in an ion
implantation process. For
the diffusion depth xdiff,
there is a sign convention:
 if diffusion is from the
14 Chapter 3: Solar cell definition

left side of the layer

into the right side, a
negative value for the
diffusion depth xdiff
should be input.
 if diffusion is from the
right side of the layer
into the left side, a
positive value for the
diffusion depth xdiff
should be input.
Diffusion:  x  When impurities diffuse
complementary N  x   N s erfc   from a source that keeps the
error function  xdiff  surface density constant at
law Ns (cm-3; either Ns = Nleft or
Ns = Nright). The same
convention for the sign of
xdiff holds.

Figure 3.9 The Grading panel, in this example an exponential composition grading is set.

3.5.4 Composition grading

Composition grading y(x) is the basic grading of the layer and has extra possible grading laws: the definition
of uniform is somewhat more complicated than for parameter grading and a the grading can be loaded from a
file. The composition grading can be set by clicking the ‘Layer composition grading type’, see Figure 3.10,
which displays the ‘Grading Panel’
3.5 (Graded) semiconductor layers 15

Figure 3.10 Setting the composition grading y(x). Select a grading law for the composition (in this example
exponential). The values yleft=0.2 and yright=0.5 are just indications here. You can only set them on the grading panel
which appears when selecting a grading law.

3.5.4.2 Uniform composition grading

There are three possible definitions of ‘uniform’:
 ‘uniform pure A (y = 0)’. The composition in this layer is y = 0 for all positions x. You see only the
column of the materials properties of the pure material A (y = 0), with no button available to set a
grading of these parameters. All parameters p get the value p (y = 0). Position grading of the doping and
defect density is still possible, e.g. NA(x),…
 ‘uniform pure B (y = 1)’. The composition in this layer is y = 1 for all positions x. You see only the
column of the materials properties of the pure material B (y = 1), with no button available to set a
grading of these parameters. All parameters p get the value p (y = 1). Position grading of the doping and
defect density is still possible, e.g. NA(x),…
 ‘uniform y, 0 < y < 1’. The composition in this layer is y = constant for all positions x, and you can set
this constant composition in the grading panel. You see both columns of the materials properties of the
pure material A (y = 0) and B (y = 1), and you can set a grading of each of these parameters, to give them
the uniform value p (y).
Even though it is possible to set a (position dependent) grading of doping and defect densities when the
composition grading of the layer is either uniform A or uniform B, it is strongly advised to use the uniform y-
option.

3.5.4.3 Composition grading from file

The composition grading profile can be loaded from a file. The rules and conventions used are the same as
for specifying parameter grading ‘from file’, and are described further in section 3.5.5.2.

3.5.5 Parameter grading

3.5.5.1 Position dependent parameter grading

Materials physics impose that we should implement also position dependent grading (and not only
composition dependent grading) as an option for the shallow doping densities ND and NA, and for the defect
densities Nt.
16 Chapter 3: Solar cell definition

Figure 3.11 Here, the shallow donor density is given a grading as a function of composition y: ND (y). The two values
displayed are the values of the pure A material (1014 cm-3) and the pure B material (1016 cm-3). The shallow acceptor
density NA(x) is given a grading as a function of position x: NA(x). The two values displayed are the values at the left
side (1015 cm-3) and at the right side (1014 cm-3). Notice the use of the colour code (red for position, blue for
composition grading).

3.5.5.2 Grading ‘from file’

All graded profiles (composition grading y(x) and parameter grading P(y) or P(x) ) can be specified ‘from
file’. The (ASCII-text) files containing these profiles should be saved in the folder ‘grading’, and the default
extension is .grd. The file which contains the grading profile should consist of two columns of numerical
data, the first number representing the position in the layer in μm, the second representing the composition
(usually a number between 0 and 1). All lines which cannot be interpreted as two numerical data are treated
as comment. Make sure that the data are in the first column are in ascending order!
When the grading file is used to specify grading of a parameter as a function of composition y, thus P(y),
the first column is interpreted as the composition. Normally, only the values between y = 0 and y = 1 are
used (see remark in the next section). If necessary, SCAPS will extrapolate the file data to obtain the property
for the pure A material (y = 0) and for the pure B material (y = 1).
When the grading file is used to specify grading of a parameter as a function of position x, thus P(x), the
first column is by default interpreted as the position x in µm. Note that this position-dependent grading is
only possible for the composition y(x), for the doping densities NA(x) and ND(x) and for the defect densities
Nt(x). If the layer thickness exceeds the range of the first column, extrapolation is used. However, you can
also force SCAPS to interpret the first column as the relative position x/d, that is the position relative to the
layer thickness. In that case, only the data between 0 and 1 are used. To specify whether the first column is
the absolute or the relative position, scaled to the thickness, insert a line in the grading file, starting with:
position: absolute (the default), or:
position: relative
To get the property of the x or y values required by the program, interpolation between the file input
values is used. This interpolation can be linear, or ‘logarithmic’ (that is, the property is first ‘plotted’ in a
logarithmic plot, and is then linearly interpolated). To specify which interpolation mode to use, insert a line
in the grading file, starting with:
interpolation: linear (this is the default for y, Eg, , , , meff), or:
interpolation: logarithmic (this is the default for NC, NV, ND, NA, Nt, vth, Crad, CAuger)
Addendum SCAPS 3.3.03, february 2016. There is an option to apply one and the same grading file over
several layers. You can specify this by inserting a line in the grading file, starting with:
range: this layer only (the default), or:
range: shared over adjacent layers
The grading file is checked for the occurrence of ‘range: this’ or ‘range: shared’ only. When
the shared option is set, the grading file applies over all layers of a group of adjacent layers:
o it can only be applied to properties that are graded as a function of position x, not as a
function of composition y. These are: y(x), NA((x) and ND(x). As of now, not Nt(x).
o the layers in the group should be adjacent to each other
o ‘grading from file’ must be set in all layers of the group
3.5 (Graded) semiconductor layers 17

o the same grading file (exactly the same name, thus the same file) must be set in all layers of
the group
o this grading from file must be set for the same property: all y(x), or all NA(x), …
The rules for extrapolation, should it be necessary, are described in the next paragraph. They apply to all
interpolation of file input data in SCAPS, in particular also to absorption () files.

3.5.5.3 Extrapolation conventions in SCAPS

Extrapolations can be necessary in several SCAPS input files:
o Files specyfying the wavelength dependence of an input property: absorption files ();
spectrum files Spec(); filter files for reflection R() or transmission T(); optical capture
cross sections for the IPV effect, n,opt() and p,opt(). When these data are needed outside
the  range specified in the file, extrapolation is used: linear extrapolation for Spec(), and
filter files R() or T(). And logarithmic extrapolation for (), n,opt() and p,opt() (that
is, first plot  or  on a logarithmic scale, and then extrapolate linearly).
o Files specifying the position dependence of an input property: generation files G(x),
composion y(x), general grading files Property(x) or Property(y) (where property might be
Eg, , NC, …, NA, ND, Nt, …). When these data are needed outside the position x range (or
composition range y) specified in the file, extrapolation is used: linear extrapolation for y,
Eg, , , n, p. And logarithmic extrapolation for G, NC, vth,n, vth,p, Cr (direct recombination),
Cn and Cp (Auger recombination), NV, NA, ND, Nt.
To prevent that extrapolated data shoot out to infinity, SCAPS applies a few precautions when logarithmic
extrapolation is asked. Therefore, the first two points and the last two points in a file table are examined. The
basic rule is that an extrapolated property is not allowed to increase when going away from either side of the
specified interval. Additionally, when an end point is zero, all values before or after this end point are set to
zero. These rules have been adapted and refined somewhat in the evolution of the SCAPS versions. Therefore,
slightly different results can be obtained with different versions; a main cause is a different extrapolation of
() files for long wavelengths. Figure 3.12 below illustrates these extrapolation conventions. Tip: you can
make sure that values outside the specified range are extrapolated to zero by adding manually a zero end
point to your data file.
Notice that a property can get a negative value by extrapolation when linear interpolation is used. We
allow this because we want to support users who run simulations with a negative composition y < 0. For
example, a user could define Ga0.75Al0.25As as the ‘pure A’ material, and Ga0.25Al0.75As as the ‘pure B’
material. Then y = -0.5 corresponds to GaAs and y = 1.5 corresponds to AlAs. There is little chance that a
user would like to do so with the well documented Ga-Al-As system, but e.g. with the still largely unknown
CZTS system, we cannot foresee for which compositions () data would be available to a user, and for
which compositions she or he would like to simulate…
18 Chapter 3: Solar cell definition

2 1.E+03
P(y) or P(y) or
1.5 P(x) 1.E+02 P(x)

1 1.E+01
0.5
1.E+00
0 y or x or x/d
-0.5 0 0.5 1 1.5 1.E-01 y or x or x/d
-0.5 -0.5 0 0.5 1 1.5

Figure 3.12 Conventions for extrapolation of file input data in SCAPS. The definition interval is between the two
vertical thick lines. The data specified in the file are shown with open symbols; the SCAPS extrapolation with solid lines
of the same colour. Left: ‘Linear interpolation’. Extrapolation is also linear, and notice that the extrapolated values can
be negative. Right: ‘Logarithmic interpolation’. Extrapolation is also logarithmic, but with the restriction that P cannot
rise. For the lowest curve, P = 0 at the edges y = 0 and y = 1 of the specified abscissa range, and the extrapolation yields
P = 0 outside the specified abscissa range; these values are shown here as P = 0.1, just to make them appear in the
figure, but internally they are set rigourously to P = 0.

Note from SCAPS 3.3.07 of december 2019:

Up to december 2019 there was a SCAPS requirement that the input files were strictly ordered to increasing
values of the independent variable (the wavelength , or the position x, or the composition y); and that no
duplicate values (two or more values with exactly the same  or x or y) occurred. From december 2019,
SCAPS eliminates double occurrences (by retaining the first occurrence, and discarding all next occurrences),
and then orders the data to increasing  or x or y (internally, the input files are not changed). This will make
SCAPS to run with your input file, which is a clear advantage to earlier SCAPS versions. However, the user
should realise that double occurrences or ordering mistakes should warn a user against a possibly incorrect or
unintended input file. Also, what SCAPS is now doing with these deficiencies (automatic sorting, and
removing all but the first occurrences of duplicate points) might not be what the user intended. Therefore, a
warning is given to the user (see below).
Also, all these input files (with the exception of grading files, see the above remarks about y < 0 or y > 1)
are now given a value check: SCAPS forces the values to be  0 before extrapolation; and the extrapolated
values of the R() and T() files are forced in the range 0  R,T  1. Again, while SCAPS will now run with
every R() or T() file, and sends a warning to the user. We advise the user to add one or more extra points
at the short and long  end of these files, to take over the control of the extrapolation, instead of leaving it to
SCAPS.
These new checks and warnings (thus: duplicates, sorting errors, range errors) are performed:
o not immediately while loading the .spe, .gen, .abs, .opt and .ftr files from the command line,
or from a .def or .scaps file: after this load action, you are still in the Action Panel, and have
the occasion to change all these files to your taste.
o but after pressing the OK button in the Action Panel: then your choice of input files is
definitive
o the range checking for the filter files (R and or T) however is done at each calculation: only
then, the wavelength range is known, and SCAPS can know whether or not extrapolation will
lead to extrapolated values outside the range [0, 1]. In cases you will be given more than one
warning, as one click to the calculate button can launch several calculations, e.g.: the work-
point calculation under illumination (then the  range is that of the spectrum file, in a typical
case) and a QE simulation (then the  range of the QE setting is merged with the  range of
the work point (if under illumination), and can thus differ from the previous  range).
3.5 (Graded) semiconductor layers 19

o these warnings are by default given to the screen, and the user should acknowledge the
warning (thus, press OK) to proceed. The warnings can also be directed to a log file, without
interruption for the user (this is the default setting while running a script). See Figure 7.5.
o if you are annoyed with these range-warnings, there is only one good solution: edit the filter
file, and add extra points, so that extrapolation in whatever  range you will consider will
not shoot out of [0, 100 %].
o and there is a worse, but faster solution: you can silence all inter/extrapolation warnings in
the Numerical Panel (Figure 3.13). This setting is saved in the .scaps files (but not in the .def
files!), and will be loaded from these. Older .def files will not have this setting saved, and
the default setting will be used: ‘do give warnings’.

Figure 3.13 SCAPS 3.3.07, december 2019: new option at the bottom left of the Numerical Panel. The default setting is
shown. The other setting is “Do not warn and proceed”.

3.5.6 The optical absorption constant () or (h) of a layer

3.5.6.1 The optical absorption constant α: ‘from file’

SCAPS absorption files are ASCII-files (simple text files) with the extension ‘*.abs’, and they reside by
default in the [SCAPS]\absorption subdirectory. If a line in this file can be interpreted as starting with at least
two numeric values, the first value is interpreted as the wavelength (in nm) and the second as α (in 1/m).
Note: as in all SCAPS input files, SI units are required, with the exception of nm for wavelength  and m for
position x. All other lines of an .abs input file are ignored and treated as comment. The interpolation and
extrapolation rules of section 3.5.5.3 apply, and -interpolation is always ‘logarithmic’, not ‘linear’.
A limited library of absorption files which are present in the SCAPS distribution; you will notice that we use
the comment lines to document the source of the () data, thus they contain a reference. Next to absorption
data of real materials, we also offer a set of .abs files that can be used in modeling work more oriented to
theory: there is a set of ‘Gray xEx.abs’-absorption files which implement a wavelength independent α. Since
SCAPS  3.3.07, it is (much!) more comfortable to use one of the new SCAPS absorption models to this
purpose (see section below).
With the instructions above you can easily make your personal absorption files. Just do not forget toput  in
1/m (not 1/cm), and we cannot enough advise to use the comment lines to document the () data file (for
your own later comfort, not for SCAPS…).

3.5.6.2 The optical absorption constant α: traditional model (SCAPS  3.3.06, december 2016)
The optical absorption constant can be set from either from a model or from a file, see Figure 3.14. When it
is set from a model, α(λ) is given by (6).
20 Chapter 3: Solar cell definition

Figure 3.14 Setting the absorption constant from file (left) or from model (right). The grading can be shown or set by
clicking the ‘Show/set alpha’-button.

 B
      A   h  Eg (6)
 h 
Here Eg is the actual band gap of the material, and A (in cm-1eV-1/2) and B (in cm-1eV+1/2) are the model
parameters.

3.5.6.3 The optical absorption constant α: new models (SCAPS  3.3.07, january 2018)
The new SCAPS models for optical absorption constant are discussed and illustrated in detail in a dedicated
application note: SCAPS Application Note Absorption Models.pdf, downloadable from our
site. Here we give a brief description, refer to the application note for more information.
Components of the new SCAPS model for optical absorption
SCAPS  3.3.07 offers 6 sub-model for optical absorption that each can be present or not. These sub-models
are described by:
  h    bg back ground constant 
 
 c  u h  E g  E g -step c if h  E g , 0 if h  E g
 E g  h
  0   0  1 E g -sqrt and 0 if h  E g
 h  E g
n1
 E g1   h  
  
 1 1     1  power1 and 0 if h  E g1 (7)
 h   E g1 
n2
 E g 2   h 
  2   2    1

power2 and 0 if h  E g 2
 h   E g 2 
 E g  h 
  exp  -  sub-bandgap "glued to other mechanisms"
 E0 

Here (+) means that the sub-model can be omitted or can be added by the user to the total (h) in the SCAPS
user interface. The names ‘back ground’, ‘E g-step’… are the sub-model names that are used by SCAPS. A few
remarks should be made:
 There should always be at least one absorption sub-model active. When the user tries to unclick all sub-
models in the SCAPS user interface (see below), the back ground model will be forced to be present.
 The model ‘Eg-step’ is there mainly to please theoreticians. It delivers a constant absorption constant c
‘above the band gap’, thus h > Eg or  < g, and zero  ‘below the band gap’. The band gap is always
3.5 (Graded) semiconductor layers 21

the band gap Eg that was input as one of the electronic properties of the material/layer. Hence it will also
be varied when Eg is varied as a batch parameter, or in the script.
 The ‘Eg-sqrt’ model is the only model that was implemented in traditional SCAPS  3.3.06. In the new
SCAPS  3.3.07, this model can be active or not, as the user decides. The user interface shows the model
parameters 0 and 0 also in their traditional form A and B. The band gap is always the band gap Eg that
was input as one of the electronic properties of the material/layer. Hence it will also be varied when Eg is
varied as a batch parameter, or in the script.
 The two power models ‘power1’ and ‘power2’ allow much versatility in modelling the (h) or ()
behaviour of a layer/material. There are 4 parameters per ‘power model’: 1, 1, Eg1 and n1 (and alike for
power 2,with index 2). In the user interface panel, g1 is given as an alternative parameter for Eg1: change
one of the two, and the other will be adapted in the user interface.
 Important note on the meaning of the band gap parameters Eg1 and Eg2:
 When Eg1 (or Eg2) are positive, they should be understood as a fixed parameter: they are not
related to the actual band gap Eg of the layer/material. In particular, this means that also
when Eg is varied (batch or script), the parameter Eg1 (or Eg2) keeps its fixed value.
 When Eg1 (or Eg2) are input as a negative number, they should be understood as a ‘relative
band gap’, or a multiple of the actual Eg: their value is substituted internally with |Eg1|Eg.
When Eg is varied then (batch or script), the internal parameters Eg1 (or Eg2) are varied with
it.
 Example: take a layer/material with band gap Eg = 1.2 eV. When setting Eg1 = 1.5, the
internal Eg1 parameter will always be 1.5 eV, regardless of the changes made to Eg. But when
the input was Eg1 = -1.5, the internal Eg1 value is set to 1.51.2 eV = 1.8 eV; and when one
would vary Eg in a batch calculation from 1 eV to 2 eV, the internal Eg1 would vary from 1.5
eV to 3.0 eV.
 The sub-bandgap tail mechanism is (of course) only available when the user has checked at least one of
the 4 ‘band gap mechanisms’ Eg-step, Eg-sqrt, power1 and power2.
 The proportionality constant in the last term of Eq. (7) is chosen in such a way that the sub-band gap tail
is ‘smoothly glued’ to the (h) or () curve of all other mechanisms.
 The Eg-sqrt mechanism of the new SCAPS works with the model parameters 0 and  0, not with A and B.
However, traditional SCAPS definition files will be read and the (A, B)  (0, 0) conversion will be
done automatically.
 When the Eg-sqrt model is the only ‘band gap model’ present, the new SCAPS  3.3.07 will also output
the (A, B) parameters, and the new definition files will be read and treated correctly by traditional SCAPS
versions.
 Also when other band gap models (than the Eg-sqrt model) are present, the new SCAPS versions will
output some ‘best’ (A, B) values. Traditional SCAPS versions will run with these new definition files, but
of course will not give exactly the same result as the new SCAPS versions, since they do not have the Eg-
step, power1 or power2 models.
The SCAPS user interface for absorption models
The optical absorption block
The optical absorption block in the Layer Properties Panel has been changed in the new SCAPS 
3.3.07, see Figure 3.15. When the optical absorption of a material/layer is set to ‘from model’, the
‘set absorption model’ button is active; upon clicking, the Absorption Model Panel opens (Figure
3.16). This panel is operating much like the grading panel and the (spectrum or generation) model
panel, that are perhaps familiar to the SCAPS user.
22 Chapter 3: Solar cell definition

Figure 3.15 The optical absorption block in the Layer Properties Panel in the new SCAPS versions ( 3.3.07). For the
‘pure A material’, absorption ‘from model’ is set, and a summary of the absorption models present is shown. For the
pure B material’, absorption ‘from file’ is set, and the full path of the absorption file is shown.

Figure 3.16 The Absorption Model Panel of SCAPS  3.3.07.

Description of the Absorption Model Panel:

 The title displays the layer name, here ‘n-layer’, and the actual value of the band gap Eg, here 1.200 eV.
 The 6 SCAPS absorption models ‘back ground’… to ‘sub-Eg tail’ are displayed in a line below the graph.
Each sub-model can be clicked on or off. The button with the sub-model name is only active when the
sub-model is checked; upon clicking such button (here the ‘power2’ button was clicked), this button is
highlighted in light blue, and the sub-model parameters (2,  2, Eg2, n2) are displayed and can be edited.
Some parameters have an ‘alternative form’; this can be merely another unit (2 and  2, in 1/cm or 1/m),
or another informative form of the parameter (Eg2 or g2). (These are ‘alternative facts’ that are true  ).
3.5 (Graded) semiconductor layers 23

 Upon changing a parameter value (or value of an alternative parameter), a new () or (h) curve is
calculated and added to the graph. When it starts to look too messy, you can clear all graphs or all but the
last calculated one.
 You can select two abscissa (horizontal axis) variables (wavelength  or photon energy h) and two
ordinate (vertical axis) values (just , either in 1/cm or in 1/m). When change the abscissa or ordinate
selection, all graphs but the last graph are lost.
 You can also show the () or (h) data contained in a SCAPS absorption file; once a file selected, you
can show or hide the file data with a toggle button. Remark: showing file data does not change the
absorption mode from ‘from model’ to ‘from file’! This mode remains ‘from model’, but at least you can
compare your model try-outs with real data.
 Tip: this is also a great way to look to an absorption file when setting up a SCAPS model: immediately
you see what it is like!
 The number of mesh points ( values or h values) in the top right part of the panel of Figure 3.16, is
only relevant for this panel. In calculations, SCAPS decides the -values at which to evaluate (), the
generation G(x, ),… based on the calculation settings ordered by the user: e.g.  values in the spectrum
file,  values at which QE() are ordered,…
 Of course all other familiar SCAPS facilities are there: linear/logarithmic view with only one click, saving
the graph in some graphical format (great for presentations and reports – usually the graphical quality is
not good enough for decent publications), showing the data (and copy/paste them e.g. in Excel).

3.5.6.4 The SCAPS grading algorithm for optical absorption constant 

The grading of the optical absorption (no matter whether is has been defined from a model or from a file)
needs a dedicated interpolation algorithm [3] to determine a grading dependent α(λ,y(x)). To interpolate the
optical absorption constant α(λ,y) for some composition between the pure material A with composition y = 0
and absorption αA(λ), and the pure material B with composition y = 1 and absorption αB(λ), SCAPS uses the
following algorithm. First, determine the cut-off wavelengths λgA and λgB of the pure materials, and a
characteristic wavelength λ0A and λ0B in the near UV wavelength range. Usually, the α(λ) curves have a
maximum (peak) in the near UV, if not one can take an arbitrary value for λ0A and λ0B. Then determine the
cut-off wavelength λg of the compound with composition y: λg = 1240nm.eV/Eg(y), and the UV peak
wavelength λ0 from linear interpolation between λ0A and λ0B. A first estimation for α(λ) is then obtained by
evaluation αA at a wavelength λA given by

A 

 gA     0    0 A  g    (8)
 g  0

A second estimation is found by evaluation αB at a wavelength λB found in a way similar to Eq. (8). Then
take a weighted logarithmic average between the two estimations:
log   1  y  log  A   A   y log B   B  (9)

The merits of this interpolation algorithm are discussed in [3].

24 Chapter 3: Solar cell definition

Figure 3.17 The optical absorption constant (, y) of a compound material with composition y, thus A1-yBy,
calculated by SCAPS with a dedicated interpolation algorithm. Here, the pure A material is CuInSe2, and the pure B
material CuInS2. The data can be obtained in tabular form by clicking Show Alpha Data (the green button)

3.5.7 The actual position dependent grading results

The graded properties get their position dependence directly (ND (x), NA(x), Nt(x) when these options are
selected), or indirectly via the x-dependence of the composition y: P[y(x)], all other cases. The x-dependent
values that are finally used in the calculations can be obtained from the Cell Definition Panel, but only after
at least a calculation is done (a working point calculation is sufficient, unclick I-V, C-V, C-f, and QE). Hereto
click one of the green save, show or graph view buttons on the Cell definition panel. The Graph View button
was added in SCAPS3.3.1, March 2015, finally . .
Upon clicking of the View Grading button, the View Grading Panel of Figure 3.18 is opened. In SCAPS 
3.3.07, also an ‘optical absorption button’ is added it. By clicking it, a view of the absorption constants ()
or (h) of (the middle of) all layers together is shown in one graph. The  values apply to the middle of
each layer (important if the layers were graded, otherwise not). Also, we added a Show Data button at the
bottom right of this panel. Is works as the ‘show buttons’ in all other SCAPS panels. In particular, you can
copy/paste from this SCAPS output window into your favourite program for further processing of the data
(Excel, Origin, whatever).
3.5 (Graded) semiconductor layers 25

Figure 3.18 The View Grading Panel allows a quick overview of all parameters that can be graded. Parameters that
one would possibly like to view in one graph are grouped together.

When using the green save button (not the show-button) you also get the values of all α(λ,x)-values, where
the λ-values are the wavelengths present in the spectrum you used to perform the simulation and the x-values
the mesh-points. Combining this facility with a uniform y-graded layer allows you to construct personalized
absorption files as well .
If you want to have a graphical view of the graded data, the best way to go is (was?) to use the recorder
function which is introduced in Chapter 8. Then you have access to more variables and saving and showing
options as well.

3.5.8 A materials approach

All parameters of a layer can be defined separately, but they can also be loaded from and saved to a
‘.material’ file, using the appropriate buttons on the layerpanel.
A material consists of all layer parameters excepting the layer name, the layer thickness, the doping
density and the defect properties.

3.5.8.1 Saving materials

Materials can be saved. This results in an ASCII-file with extension ‘*.material’. There are four different
options to save a material: both materials, pure A, pure B and pure Y. The options pure Y and both materials
are not available when the composition grading is set to pure A or pure B. In that case only one option
(pure A or pure B) is available.
26 Chapter 3: Solar cell definition

Figure 3.19 Available options for saving materials

Saving a material using the pure A/B option only saves the parameter values of the A/B-material of the
layer. All parameter grading is set to uniform.
Saving a material using the both materials option saves both parameters of the A/B-material of the layer
together with the appropriate parameter grading parameters.
When saving a material using the pure Y option you should give a fraction y which defines the material to
be saved as A1-yBy. The appropriate parameter values for this material are then calculated according to the
values for the A/B-material of the layer and saved as a uniform material (similar to pure A/B). The
absorption for this material is also calculated as an interpolation of the absorption for A and B and is saved in
an absorption file. This file is automatically added to your absorption folder and gets the same name as your
material-file extended with ‘.abs’.

3.5.8.2 Loading materials

Materials which have been saved, can of course be loaded as well. Three options are available: pure A,
pure B and both materials. When the composition grading of the layer was saved to pure A/B however, only
the pure A/B option remains available for loading.
Loading a material using the both materials option sets all material parameters of the A- and B-side of the
layer, including the parameter grading. When loading a material which has been saved as a pure A/B/Y
material, the parameters of the A- and B-side of the layer will be identical and all parameter grading will be
set to uniform.
Loading a material using the pure A/B option only sets the parameters of the A-/B-side material of the
layer. If the parameter grading was uniform it will be set to linear.
If the materials file contains information on thickness, on the shallow doping densities or on defects, you
can load this information or not: select or unselect the appropriate check box in the Load Material Panel
(Figure 3.20). As a materials file can contain file names (optical absorption files, grading files), you should
make sure that these files are indeed present in the appropriate SCAPS directory.
Material files can also be varied in a batch calculation (see Chapter 7).
3.5 (Graded) semiconductor layers 27

Figure 3.20 Available options for loading materials

3.5.9 A frequently asked question (FAQ) about grading

Sometimes we think of a desired or supposed band diagram, with a sloped (non horizontal) conduction band
and/or valence band. The model input parameters are the grading of the electron affinity (x) and of the band
gap Eg(x). There is sometimes misunderstanding about the relation between the (x) and Eg(x) input, and the
band bending EC(x) and EV(x) they cause: the result depends on the doping density and on the voltage and
illumination conditions. The rules of thumb are given below and illustrated in Figure 3.21 and Figure 3.22
below:
 In equilibrium, in a neutral, p-type region (e.g. the absorber bulk in CIGS), EF = EFp = horizontal
(equilibrium), EV is a fixed amount kTln(NV/NA) under it (depending on the doping density NA;
supposing that both NA and NV are uniform and not graded): thus the valence band will be horizontal,
regardless the band gap and/or electron affinity grading. The conduction band EC is then placed at a
distance Eg above EV, thus: EC(x) = EV + Eg(x), and will thus be sloped when Eg(x) is graded. You will not
see any effect on electron affinity grading.
 in equilibrium, in an n-type region, EF = EFn = horizontal. The conduction band EC is placed at a fixed
distance kTln(NC/ND) above it (this is a constant when both ND and NC are uniform and not graded).
Thus the conduction band will be horizontal, regardless the electron affinity grading. The valence band
will then be placed a distance Eg under EC, thus: EV(x) = EC - Eg(x) and will be sloped when Eg(x) is
graded.
In all other circumstances (applied V, illumination, in a depletion layer, grading of the doping densities ND
(x), NA(x) or of the densities of states NC(x), NV(x) ) there are no simple rules of thumb, it gets too
complicated. Trust e.g. SCAPS for it. But always check SCAPS (that is, check what you have input) by always
looking to the equilibrium band diagram and applying the rules of thumb above where possible (that is, in the
flat-band regions).
28 Chapter 3: Solar cell definition

electron energy

left = 4 eV vacuum level

EC right = 5eV
Eg left = 1 eV
EV Eg right = 1 eV

 grading, noEg grading: will this yield both

sloped EC(x) and sloped EV(x) ?

EC

EF

EV

Figure 3.21 Developing ‘intuition’ about grading. Grading schemes of electron affinity (x) and of band gap Eg(x):
(x) is graded, but Eg is not. What is suggested by the naïve sketch (top) and what really happens in a pn junction
(SCAPS calculation, bottom).

electron energy
electron energy
left = 4 eV vacuum level
left = 4 eV vacuum level
right = 4eV
EC EC right = 5eV
Eg left = 1 eV
EV Eg right = 2 eV Eg left = 2 eV
EEgg right = 11 eV
right = eV
EV
no  grading,Eg grading: will this yield
constant EC and sloped EV(x) ?  andEg opposite (will this yield constant EV
and sloped EC(x)?)
EC
EC
EF
EV EF

EV

Figure 3.22 [more…] Developing ‘intuition’ about grading. Grading schemes of electron affinity (x) and of band
gap Eg(x): [left] no  grading but an Eg(x) grading and [right] both (x) and Eg(x) are graded, but their sum (x) + Eg(x)
= 6 eV = constant. What is suggested by the naïve sketch (top) and what really happens in a pn junction (SCAPS
calculation, bottom).

3.6 Defects and recombination

In a diode, current is converted from hole current at the p-contact to electron current at the n-contact. This
means that somewhere in the diode recombination MUST take place, even in the most ideal device. So the
3.6 Defects and recombination 29

user MUST specify recombination somewhere, at least at one place (in a layer, at a contact or at an
interface). If (s)he does not do so, a convergence failure will result in non-equilibrium conditions (non zero
voltage, and/or illumination).
In the bulk of a semiconductor layer, three different kinds of recombination processes can be introduced:
through defects, radiative and Auger.

3.6.1 Adding defects

Up to seven defects can be introduced in a semiconductor layer. The parameters governing each defect can
be edited by clicking the appropriate Add/Edit-button, Figure 3.23, which opens the ‘defect properties
panel’, Figure 3.24. Also, you can right-click on one of the ‘defect summary text boxes’; a panel then opens
where you can remove, duplicate or add a defect, much the same as it was with removing, duplicating and
adding a layer (Section 3.5.1 and Figure 3.7, page 10); of course, ‘splitting a defect’ is not available as it has
no meaning.

Figure 3.23 Adding, editing and removing defects.

Figure 3.24 The defect properties panel

3.6.2 Multivalent defects

The most common defects in semiconductors are either donor or acceptor defects having two possible charge
states. However, there exist defects with more than two different charge states (multivalent defects). In
SCAPS you can add up to four defect levels (= five charge states) for each defect. The charge present on a
defect can be varied from -3 to +3 elementary charges. Because the most common multivalent defects are
double donors (charge states 0, +1 and +2), double acceptors (charge states 0, -1 and -2) and amphoteric
30 Chapter 3: Solar cell definition

defects (charge states 0, +1 and -1), special facilities for them are provided in SCAPS. The charge states
connected to a defect can be set by the ‘defect type button’
If the defect is not multivalent (donor-acceptor-neutral), all defect parameters can be set on the defect
density panel, otherwise clicking the defect type button will result in opening the ‘Multiple level defects
properties panel’, Figure 3.25.
The algorithms used by SCAPS to calculate the recombination through defects are explained in [2].

Figure 3.25 ‘Multiple level defects properties panel’

3.6.2.2 A neutral defect

The “neutral” defect is an idealization of a defect which contributes to the Shockley-Read-Hall
recombination but does not contribute to the space charge. In the case of a “neutral” defect, only the product
of  and Nt affects the dc and ac solutions, through the carrier lifetimes; n e.g. is given by 1/(n.Nt.vth) : in
this case, the defect centers contribute to recombination but not to the space charge. This type of defect can
be chosen if one wants to specify electron and hole lifetimes without specifying a defect density which
affects the space charge.
In other words: a neutral defect does not exist in reality, it is an idealization to help you to create a model
step by step.

3.6.2.3 More than two charge states

When choosing the defect type as double donor, double acceptor, amphoteric or custom defined multilevel
the user is redirected to the multiple level defects properties panel (see Figure 3.25). Some of the properties
which were already available on the defect properties panel are available here as well, together with the
properties which are specific for a multivalent defect. Starting from this panel the user can edit the defect and
its levels.
Only when the defect type custom defined multilevel has been selected the user can add up to four
different levels, otherwise the number of levels is restricted to two and the charge types are predefined. The
‘insert’ buttons are used to add a specific level. In order to remove a level the user has to click on a level
with the right button of the mouse.
3.6 Defects and recombination 31

The charge type and the energy of a level can immediately be accessed on the multiple level defects
properties panel. The charge types MUST be introduced in an ascending way (level 1 the most positive
charge), or all levels must be neutral. In most cases the most positive level will be situated the closest to the
conduction band, a warning is given with the option for the user to define the defect differently, but the user
is not obliged to comply with the warning.
The other defect properties of a level can be edited by double-clicking on the text-box belonging to each
level. This opens a panel similar to the defect density panel, however the properties which are not level
specific but general to the entire defect are dimmed and can’t be edited. The only properties which can be
edited are the capture cross sections (for electrons and holes), the energy level and the IPV-properties
(§3.6.4).
Several features of the energetic distribution can also be accessed on the multiple level defects properties
panel. The energetic distribution, the reference for defect energy levels and the characteristic energy are the
same for all levels of one defect. More info about these properties can be found in§3.6.3. If the checkbox Use
Correlation Energy is checked, the energy difference between level 2 and level 1 is displayed in the window
next to the checkbox. When the user edits this correlation energy the energy level of level 2 will be changed
accordingly with respect to level 1.
The degeneracy factors (see [2]) take as a default value Eq. (10), with H the total number of charge states.
H!
g s  CH
s
 (10)
s ! H  s !
However, their values can be set to one as well by clicking the appropriate checkbox.
When the user has finished editing the properties of the levels the multiple level defects properties panel
can be quitted by the return to the defect definition panel-button. Then the defect properties panel is again
displayed, but the level specific properties are no longer visible. It is a kind of survey of the defect. Now the
changed to the defect can be saved using the add/accept-button or ignored using the cancel-button.

3.6.3 Energetic distribution of defect levels

Next to a discrete defect energy level, also defect whose levels display a distribution in the band gap can be
modeled in SCAPS. There are five different options, listed in Table 3.2.

Table 3.2 Overview of the energetic distributions of defects available. Et is the energy level (‘trap level’), Ec is the
characteristic energy (not the conduction band edge energy here!). The defect density Nt(E) is given in cm-3/eV (except
for single, where it is given in cm-3). Npeak is the energetic density at the peak of the distribution, in cm-3/eV. Ntot is the
total density, integrated over all energies in (cm-3). The parameters Et, EC, and Ntot or Npeak, and the shape of the
energetic distribution (‘single’, or ‘uniform’…) are set in the Defect Properties Panel for each defect. The width
parameters wG and wt are set for all defects together in the Numerical Panel.

Range N t  E   ... 
N tot N peak  remarks

Single  Et ; Et  N tot  Et

 Ec Ec 
N tot  EC N peak
 Et  2 ; Et  2 
Uniform N peak

  E  E 2 
 wG wG 
Gauss  Et  2 Ec ; Et  2 Ec  N peak  exp    t
  N tot  EC N peak (1)
  Ec  
 
 E  Et 
CB tail  Et  wt Ec ; Et  N peak  exp   N tot  EC N peak (2)
 Ec 
32 Chapter 3: Solar cell definition

 E  Et 
VB tail  Et ; Et  wt Ec  N peak  exp    N tot  EC N peak  (3)
 Ec 

Remarks:
(1) wG can be defined in the numerical panel; default value: wG = 6.0; see Figure 3.26; G stands for Gauß.
(2) wt can be defined on the numerical panel; default value: wt = 7.0; see Figure 3.26; t stands for tail.

Whenever the defect energy distribution is not single, this distribution is discretized as being a specified
number of single defect levels. This number can be set on the numerical panel, its default value is 7.

Figure 3.26 Defect settings on the numerical panel with their default values. Top: the number of discretization levels,
bottom: wt and wG.

The energy level has to be defined with respect to a reference energy level. This can be set on the defect
density panel or on the multiple level defects properties panel. The available choices are: above EV (above
the valence band level); below EC (below the conduction band level) and above Ei (above the intrinsic level
Eg k BT  NV 
Ei  EV   ln   ).
2 2  NC 

3.6.4 Impurity photovoltaic effect (IPV)

SCAPS is able to simulate the IPV-effect. Its parameters can be set on the defect density panel. The
algorithms and an example is given in [8, 9]. More examples are found in articles referring to this article, e.g.
[10]. An example ‘*.def’-file is provided with the SCAPS installation with ample comments.

3.6.5 Short overview of defect values

When all defect-parameters are set, an overview is given in the layer definition panel, Figure 3.27.

Figure 3.27 Overview of the defect parameters

Next to the charge type, the defect density (and its spatial and energetic distribution) also an overview of
the carrier lifetime and diffusion length is given. These are calculated (Eqs. (11) and (12)) only at the left and
right side of the layer for every energy level. Only its maximum and minimum value are listed.
1
 (11)
vth N t
3.7 Metastable defect transitions 33

L  D (12)
IT SHOULD BE STRONGLY EMPHASIZED THAT THESE VALUES ARE ONLY CALCULATED TO GIVE THE USER
A ROUGH IDEA ABOUT THE RECOMBINATION OF THIS DEFECT, NONE OF THESE VALUES ARE USED IN THE
ACTUAL SIMULATIONS! MORE ADVANCED ALGORITHMS ARE USED THERE, see e.g. [2].

3.6.6 Radiative and Auger recombination

It is possible to introduce radiative and Auger (band-to-band) recombination in SCAPS, according to Eqs. (13)
and (14).


U radiative  K np  ni2  (13)

U Auger   cnAn  c pA p np  ni2  (14)

The parameters K, cnA and cpA can be set on the layer panel. SCAPS does not give a hint for a plausible K
value ; for Si you can take KSi = 1.810-15 cm-3s-1, for GaAs KGaAs = 7.210-10 cm-3s-1. In CIGS the
recombination is most probably dominated by defects, you can take K = 0, unless you want to simulate
electroluminescence. In Si appropriate values for the Auger constants are: cnA  c pA  3  1031 cm-6s-1 .

Figure 3.28 Band to band recombination parameters. From top to bottom: K, cnA and cpA.

The contributions of the different recombination processes to the current can be assessed on the IV-panel
(see §6.4.3) and in the recorder facility (see Chapter 8). When you are sure that all direct band-to-band
recombination is radiative, and that all emitted photons can get out of the cell to your measurement set-up,
you can interpret Jradiative as the electroluminescence signal. However it is not labeled as such in SCAPS.

3.7 Metastable defect transitions

In the chalcopyrite material system often metastable behaviour of the samples is observed. This is possibly
due to the presence of defects which undergo transitions which are accompanied by large lattice relaxations
and thermal activation over energy barriers. SCAPS is able to simulate this behaviour, the algorithms which
are implemented and an example can be found in [11]. An introduction to metastable defects in SCAPS was
given in [11’].

3.7.1 Principles
Metastable effects take place on a long timescale (which can vary a lot, depending on the actual conditions).
SCAPS does not solve transient problems. Hence, the following philosophy has been followed.
A metastable defect is considered to exist of two different configurations, a donor and an acceptor
configuration, which are separated by energy barriers associated with lattice relaxations. Each separate
configuration behaves as a conventional defect, which can have a multivalent character. The transition
between the different configurations requires a simultaneous capture or emission of two free carriers together
with the thermal activation of an energy barrier. Hence, at lower temperatures, this transition will not occur.
Similarly the transition will not be observed in an admittance measurement when the change of the quasi-
Fermi levels occurs at higher frequencies.
34 Chapter 3: Solar cell definition

In a typical experiment, the studied sample is brought to a well defined metastable state, i.e. a distribution
over the donor and acceptor configuration, by applying certain initial voltage/illumination condition at
elevated temperatures for a sufficiently long time. Afterwards the sample is cooled down in order to inhibit
further changes in the metastable state of the sample and measurements are performed.
In order to simulate this, first the distribution over the acceptor and donor configuration of the defect is
calculated under initial voltage/illumination conditions, see §4.4. This distribution, together with the total
defect density is used to calculate the defect density belonging to the different configurations of the
metastable defect. These defect densities are kept constant and are used to perform further simulations under
measurement voltage/illumination conditions.

3.7.2 Introduction of a metastable transition

A metastable defect is assumed to have two different configurations (acceptor and donor). Each of these
configurations has to be introduced as if it were a regular (multivalent) defect. Metastable properties can only
be set for defects with single energy levels, not for defects that occupy a band of energies. If at least two
defects are present in a layer which have compatible charge states, they can be set to be the two
configurations of a metastable defect by checking one of the ‘metastable transition’-boxes on the layer panel,
Figure 3.29. By clicking the ‘edit’-button, the ‘Metastable Defect Definition Panel’ (Figure 3.30) opens,
where all parameters governing the metastable transition between the two configurations can be set/edited.

Figure 3.29 Allowing a metastable transition.

Figure 3.30 The Metastable Defect Definition Panel

3.7 Metastable defect transitions 35

Its main components are discussed below:

Charge states: This is a list which gives all possible combinations of charge states available which can be
transformed into one another through a double capture/emission process. If none are available, you will not
be able to access the Metastable Defect Definition Panel. If more than one option is available, you can select
the one you want.
Activation Energies: The transition between the two configurations of the metastable defect can proceed
through four processes: electron capture (EC), hole emission (HE), electron emission (EE) and hole capture
(HC). Each of these processes has its own activation energy which can be set. The Fermi level position
where the two configurations are equally stable under thermal equilibrium conditions is called the transition
energy (TR) and can also be set. These four activation energies together with the transition energy are not
independent as they have to obey to the principle of detailed balance. Only three of them can be chosen
independently, the two remaining are then calculated by the program. You can decide which energies you
want to provide in the ‘Activation Energies to be set’-list. When a band gap grading is present, the calculated
energies can vary throughout the layer. The values which are displayed in the user interface are then those
which are valid for the right side of the layer.The transition energy can be set with respect to the valence
band, conduction band or intrinsic energy level in a similar way as a regular defect level.
Metastability type and electron/hole processes: The double capture and emission processes can be
obtained in different ways. The electron capture process e.g. can consist out of two simultaneous single
electron capture processes (Double electron capture) or a single electron capture together with a
simultaneous hole emission process (Electron capture + Hole emission). This has an influence on the details
of the detailed balance calculation and needs to be set. These properties can also automatically be set to the
standard value for the common VSe-complex and InCu-complex in the CIGS material by selecting
respectively ‘Se Vacancy’ or ‘DX centre’ in the Metastability type-list.
Capture cross sections: You cannot change the four capture cross sections in the Metastable Defect
Definition Panel, they are shown there for your information only. Instead, you should set these in the
Multilevel Defect Properties Panel of Figure 3.25. To know which of the n and p values of the defects
(possibly multivalent) constituting the metastable defect, you should (alas ) understand some of the details
of the metastable transition mechanisms. To help the user a bit, this information is also displayed on the
Metastable Defect Definition Panel from SCAPS 3.3.07 of november 2018 onwards, see Figure 3.30.
Defect density: The defect density of the metastable defect can be set in a similar way as for a regular
defect. Pay attention! When a defect is set to be one of the configurations of a metastable defect density, the
defect density which you can set on the defect properties panel will be ignored. Instead at every meshpoint
the relevant defect density of this configuration will be calculated from the total defect density of the
metastable defect.
Initial donor occupation guess: In order to start the iteration process an initial guess has to be made for
the fraction of the metastable states that are in the donor configuration. This fraction is the same for all
meshpoints.

3.7.3 Help, the buttons to introduce/edit metastable properties are not available!
One of the aims of SCAPS is to keep the user interface very intuitive and easy to learn. Since the first release,
a multitude of extra facilities have been added to the program. These extensions increase the number of
situations which can be simulated, but also complicate the learning process of a user making his/her first
simulations.
The simulation of metastable defects is considered to be ‘advanced’ and should not be one of the worries
of a rookie in the bright world of numerical simulations with SCAPS. As a result, most of the buttons
governing metastabilities are only accessible/visible when the user really expresses her/his intention to use
them:
36 Chapter 3: Solar cell definition

 The ‘Metastable Defect Definition Panel’ can only be accessed if at least two defects are present in the
layer AND at least one of the charge states of one defect can be transformed into one of the charge states
of a second defect through a double capture process.
 The ‘Initial State Workingpoint Panel’ can only be accessed if at least one metastable defect is present in
the definition file which is currently loaded.

3.7.4 Numerical settings

The calculation of the occupation of the different metastable configurations is done in an iterative way. The
maximum number of iterations and the minimum remaining error can be set on the numerical panel, see
Figure 3.31.

Figure 3.31 Numerical settings for metastable simulations

When the calculation is numerically unstable it can be stabilized by using a clamping factor. This process
is described in [11]. This factor should be chosen between 0 and 1, 1 corresponds to no clamping. The
smaller this factor the more stable, but also the slower the calculation gets. Values below 0.001 are usually
very slow.

3.8 Interfaces
Between any two semiconductor layers an interface can be defined. The main algorithms used in SCAPS are
discussed in [12].
The model which is implemented for interface transport in SCAPS is thermionic emission. The thermal
velocity of the interface transport equals the smallest thermal velocity of the two neighboring layers. The use
of this model implies that there will always be a (small) step in the quasi-Fermi level energy values at an
interface, even when there are no band offsets.
Just as in a bulk layer recombination centers can be present at an interface. The definition of interface
defects is very similar to bulk defects. However, there are only three possible defects and there charge type
cannot be multivalent. Recombination in interface states is modeled by the Pauwels-Vanhoutte theory [13],
which is an extension of the Shockley-Read-Hall theory.
Two tunneling processes are implemented which involve interfaces: intraband tunneling and tunneling to
interface defects, see §3.9.

3.9 Tunnelling
SCAPS treats (some) tunnelling mechanisms in a semi-classical way: band to band tunnelling, intraband
tunnelling, tunnelling to interface defects and tunnelling to contacts. The main algorithms are discussed in
[4], a more elaborate version is available in [14] (in Dutch, on request). PAY ATTENTION: Tunnelling is only
taken into account in the solution of the dc-problem, not in the ac-problem. Hence only indirect tunnel
influences on the admittance (through the setting of the dc-state of the sample) are simulated!
Band-to-band tunnelling is the tunnelling between conduction and valence band states, see Figure 3.32.
Intraband tunnelling is the tunnelling between states both in the same (conduction/valence) band, see Figure
3.33. The valence and conduction band consists in (almost) every structure out of bulk layers and interfaces.
3.9 Tunnelling 37

Band-to-band tunnelling and intraband tunnelling can be allowed/forbidden for every layer/interface
separately by checking the appropriate checkbox on the layer/interface panel.
Next to band-to-band and intraband tunnelling, also tunnelling between the bands and an interface defect,
or the bands and a metal contact is possible, see Figure 3.34. These processes can be set for each interface
defect separately on the interface defect properties panel and for each contact on the contact definition panel.
Whether or not these tunnelling processes are allowed does not depend on the layer/interface settings. E.g. if
tunnelling to the left contact is allowed on the contact properties panel, tunnelling between this contact and
all layer/interfaces will be allowed even if the ‘allow tunnelling’-checkbox is unchecked in one or more
layers/interfaces.

EC
EC
EFp
EV EV EFn
EFp
EFn

Band-to-band tunneling at reverse bias Band-to-band tunneling at forward bias

Relevant part of the Layer Properties

Panel
Figure 3.32 Band-to-band tunnel mechanisms at reverse and at forward bias. Note that band-to-band tunneling at
forward bias (Esaki tunnel diode) is only possible with very heavily doped, degenerated semiconductors, where the
Fermi level is inside the conduction and/or valence band; these heavy doping phenomena (Fermi statistics, band gap
narrowing…) are not implemented in SCAPS. At the bottom, the check box and meff* values to be enabled in the Layers
Properties Panel are shown.
38 Chapter 3: Solar cell definition

EC
EF
EV
EC

intra-band tunneling in the conduction band

EC

EV
EF
EV

intra-band tunneling in the valence band

Relevant part of the Interface Panel

Figure 3.33 Intra-band tunneling in the conduction and in the valence band. At the bottom, the check box and meff*
values to be enabled in the Interface Panel are shown.
3.9 Tunnelling 39

EC
EC

EF EF

contact EV EV an interface state

contact tunneling tunneling between a band and an interface state

Relevant part of the Contact Properties Panel Relevant part of the

Interface Defect Properties Panel

Figure 3.34 Contact tunneling and tunneling to interface states: implemented in SCAPS. At the bottom, the check box
and meff* values to be enabled in the Contact Properties Panel and the Interface Defect Properties Panel are shown.

3.9.2 Numerical tunnel settings

The tunnel probability is determined by the effective mass of the two states which are involved in the
process. These masses can be different, and can be set differently in SCAPS. You can select on the numerical
panel, which of the two masses to use. The default choice is to use the minimum effective mass.
SCAPS uses the WKB-approximation in order to calculate tunnel probabilities. This approximation is only
strictly valid for large enough barriers. When no precautions are taken, the tunnel rate for very thin and
shallow barriers is hugely overestimated. In order to avoid this you can set a minimum height of the tunnel
barrier (default = 2kBT), for such small barriers the current is determined by drift and diffusion rather than
tunnelling anyway.
Up to now four different tunnel processes are present in SCAPS. Each of them can easily be forbidden/
allowed by (un)checking the appropriate box on the numerical settings panel.
As it is often inconvenient to check all allow tunneling boxes on the different layer and interface panels,
this can automatically be performed by clicking the check tunneling in all layers button on the numerical
panel; it is still advised to check afterwards which tunnel mechanisms are enabled/disabled in the various
panels.
 40 Chapter 3: Solar cell definition

Figure 3.35 Tunnelling settings on the numerical panel

3.10The blue button

In SCAPS 2.8.4 and earlier, it happened that a panel requiring input was hidden under other panels, and
difficult to find; it could also happen that such panel disappeared, which caused a hang-up of the program.
These problems have been fixed now. There are some limitations to the number and kind of panels that can
be opened or operated simultaneously. Above all, the F1 key acts as an emergency button that brings the cell
definition panel on top; on that panel, you find a new blue button that displays a nicely
structured list of all input panels currently open, and that allows you to navigate between them (or to quit all
input at once). Pressing the F2 key redirects you to the current layer or interface panel.

3.11Saving and loading problem definitions

Almost every setting in SCAPS can be saved to a file and loaded again. These files are standard ASCII-files
which can be read and edited with e.g. Notepad. However, editing is at own risk, and we highly advise
against it. An overview of the main file-types with their default extension and directory are given in Table
3.3.

Table 3.3 Main file-types with their default extension and directory
SCAPS filetypes default extension default directory
problem definition .def scaps\def
material properties .materials scaps\materials
action list .act scaps\def
initial workingpoint .wp2 SCAPS\def
batch settings .sbf scaps\bdf
recorder settings .srf scaps\bdf
scaling .scl scaps\scaling
SCAPS scripts .script scaps\script
all SCAPS settings in one .scaps scaps\def

Other files can not be edited within SCAPS but are needed as input files. These files are text files (ASCII
files) that you can construct and edit with a simple text processor, e.g. Notepad, see Table 3.4. Each file has a
header that can be used to explain what the data represent and where they are coming from. The files
distributed with SCAPS can serve as an example to construct ones own input files. Take in mind that SCAPS
expects meter-based SI units in input files, except for position x (m) and for wavelength  (nm).

Table 3.4 File-types of SCAPS input data files

SCAPS filetypes units default extension default directory
spectrum () nm, W/m2 .spe scaps\spectrum
generation G(x) µm, 1/m3.s .gen scaps\generation
absorption () nm, 1/m .abs scaps\absorption
filter T() or R() nm, % .ftr scaps\filter
optical capture cross section n(), p() nm, m2 .opt scaps\optcapt
grading y(x), Eg(y), NA(x), … x in m, properties in m- .grd scaps\grading
based SI units
3.11 Saving and loading problem definitions 41

When definition files are exchanged between users (or between your lab computer and your home
computer), it can happen that a .def file contains an input file name (.abs, .ftr, .opt, .grd, …) that is not
present on the computer. When leaving the solar cell definition panel with the OK button, SCAPS checks if all
files are found, and if not brings you to a panel where you can either browse for an existing filename, or set
the property to the default internal model instead of ‘from file input’.

Figure 3.36 SCAPS checks if all necessary files can be found: after loading an action list or loading all SCAPS settings,
and after leaving the Solar Cell Definition Panel with ‘OK’ (see arrows in screen shots at the left side). Then, a panel
opens that displays a list of all files not found in the problem definition (see screen shot at the right side). You can either
browse for an existing file, or set the property ‘to model’.

Please note that such check is not carried out when the problem has been input or changed by the SCAPS
the batch set-up: you still have the possibility to cause a program crash by inputting non-existing files. Also,
when SCAPS is in non-interactive mode (see 7.4 and Figure 7.5), the Check Files Panel is not opened, but
everywhere a file is not found, the corresponding property is set ‘to model’ (this is equivalent with clicking
all the ‘Set to model’ buttons, at the right side of Figure 3.36); a message of what was unfound and has been
set to model is written to the SCAPSErrorLogFile.log file.
You can leave the Check files Panel without fixing all unfound files (that is, either browsing for an
existing file, or setting the property to model). The calculate button is then disabled, and you see a red
warning at the bottom, right of the action panel (Figure 3.37).

Figure 3.37 When not all unfound files are fixed, a red warning is displayed at the bottom/right of the action panel.
 Chapter 4: Working point definition

The workingpoint specifies the parameters which are not varied in a measurement simulation, and which are
relevant to that measurement.

4.1 General
These parameters are (immediately) available on the action panel (Figure 4.1).

Figure 4.1 Setting the working point conditions, available on the action panel.

 The temperature T: relevant for all measurements. Note: in SCAPS, only NC(T), NV(T), the thermal
velocities, the thermal voltage kBT and all their derivatives are the only variables which have an explicit
temperature dependence; you must input for each T the corresponding materials parameters yourself.
 The voltage V: is discarded in I-V and C-V simulation. It is the dc-bias voltage in C-f simulation and in
QE(λ) simulation. SCAPS always starts at 0 V, and proceeds to the working point voltage in a number of
steps that you also should specify, see §5.1.2.
 The frequency f: is discarded in I-V, QE(λ) and C-f simulation. It is the frequency at which the C-V
measurement is simulated.
 The work point voltage V and illumination intensity (e.g. in sun) are approached from the equilibrium
state in Number of points = n steps: the voltage from 0 to V linearly in n steps, and the illumination
intensity Pin logarithmically from 10-10Pin to 1Pin.
 The illumination: is used for all measurements and is discussed below in §4.2.

4.2 Illumination conditions

When performing simulations under illumination, you can further specify the illumination conditions. The
basis settings are: dark or light, choice of the illuminated side, choice of the spectrum. If you have an optical
simulator at your disposal you can immediately load a generation profile as well in stead of using a spectrum.
SCAPS 3.3.05, december 2016: both the incoming spectrum and the generation can be specified either by
an input file, or by an analytical model. This is treated in Section 4.3.
44 Chapter 4: Working point definition

4.2.1 Internal SCAPS calculation

SCAPS is able to perform simple optical simulations in order to calculate the generation in your structure. The
internal optical SCAPS model is described in Section 4.3.1 below. Hereto the illumination conditions have to
be specified (Figure 4.2).

Figure 4.2 Specifying the illumination conditions.

 The illumination side can be changed (left/right)

 A light spectrum file can be selected. Several spectrum files are available in the SCAPS distribution. The
default is a one sun (= 1000 W/m2 = 100mW/cm2) illumination with the ‘air mass 1.5, global’ spectrum.
A lot of monochromatic spectra are available as well. They come in two versions. They either contain a
total illumination power of 1000 W/m2 (1 sun) in the wavelength band (e.g. ‘600 nm.spe’), or they
contain a number of photons that would yield a light current JL = 20 mA/cm2 if QE were unity (e.g.
‘600nm fixed JL.spe’). All spectrum files which come with the SCAPS distribution are documented with
comments. All spectrum files are ASCII-files and can hence be user defined. The first line contains the
number of wavelengths that appear in the spectrum file. This line is immediately followed by two
columns. The first column specifies the wavelength λ (nm). Suppose the spectral density (in Watt/m2.nm)
is given by P: the incident power in the wavelength interval [λ - dλ/2 , λ + dλ/2] is then given by Pdλ (in
Watt/m2). The second column in the spectrum file contains this incident power Pdλ (in Watt/m2) in a
wavelength interval dλ around λ. As such, the total incident power is given by the sum of the values in
the second column. The wavelength interval dλ considered around a wavelength λi is: dλ = (λi+1 - λi-1)/2.
For the first wavelength λ1 it is dλ = λ2 - λ1, and for the last wavelength λN it is dλ = λN - λN-1. The
maximum number of wavelengths appearing in the spectrum file equals 2500. The number of incident
photons, per m2 and per s, of wavelength between λ-dλ/2 and λ+dλ/2 is then set equal to 5.0351015 × λ ×
Pdλ , λ in nm, Pdλ in Watt/m2.The numerical factor equals 10-9/hc where Planck's constant h and the light
velocity c are expressed in SI units. For convenience, the spectrum file may be preceded by comment
lines, which begin with the character “>” or “/”. These lines are not read by SCAPS. They can be used to
identify the source or the purpose of a given spectrum file.
 The incident power of the incoming light can be weakened or amplified by applying a neutral density
filter. The value ND of this filter can immediately be set, or the transmission of the filter T = 100% 
10-ND. You can to set ND < 0 or T > 100 %, to do simulations under concentrated light. The incident
power can also be be attenuated by an internal/reflection at the front contact which should be specified
on the cell definition panel.
 No reflections at the interfaces are implemented. Only at the contacts you can specify a (wavelength
dependent) reflection/transmission (§3.2).
 A short/long wavelength cut-off filter can be applied.
4.3 Generation and spectrum models (scaps 3.3.05) 45

4.2.2 Generation from file

There is also a possibility to input the generation in the structure immediately. This can be useful if you have
a better optical simulator than SCAPS, or if you want to mimic e.g. EBIC measurements.

Figure 4.3 Settings when using a generation file.

The advised way to create a g(x) file is as follows: set up a problem in SCAPS; perform a calculation with
the internally calculated g(x). Save the results in the EB-panel (energy bands panel), or, alternatively, press
show in the action panel and copy/paste the desired data in a user file (e.g. Origin or Excel). Then you have
the x-coordinates of the nodes used by SCAPS. Make a generation file with the g(x) data from your own
model, calculation or measurement… Mind that SCAPS input files want data in SI units (only position x is
wanted in m, not m). As a consequence, a SCAPS generetion file expects the g(x) data in two columns, first
column x in m, second column g in m-3s-1. For an example of a valid format, see the examples coming with
SCAPS (file extension.gen).
You can always save the generation g(x) that was set by the internal SCAPS illumination model in a valid
SCAPS generation input file using the script command save results.SCAPSgenerationfile, see
section 10.4.3. And then you can replace the g data in the second column by your own data.
To use your own generation file, click ‘Generation of eh pairs determined from file’ in the action panel,
and select your file. When the g(x) data in your file do not use the x values used by SCAPS, the program will
have to interpolate between your x values, and this might cause inaccuracy. To judge the interpolation, you
can visualize the g(x) data in the Generation-Recombination profiles panel, see §6.4.2. The g(x) profile you
will see then is, however, only updated after a calculation has been performed with the calculated or user
specified g(x) .
Additionally you can still use an attenuation filter, which would mimic the effect of a neutral density filter.
Again, setting an attenuation > 100 % corresponds to a measurement under concentrated light. The ideal
light current in the file (Figure 4.3) is the total number of eh pairs in the generation file, thus integral of all g
values over the cell thickness, and divided through the elementary charge q. The idelal light current in the
cell is the ideal light current in the file, multiplied with the transmission of the attenuation filter.
A generation file is a table of x, g values (x in m, g in m-3s-1). It contains information about the number of
eh pairs generated, but not about the energy of the photons that created the eh pairs. As a consequence, no
solar cell efficiency can be calculated. However, a ‘collection efficiency’, that is the ratio of the short circuit
current to the ideal light current in the file, can be calculated, and is displayed in the IV panel. Also, a
relevant spectral response QE cannot be calculated, since the generation file does not contain any
information on wavelength.

4.3 Generation and spectrum models (SCAPS 3.3.05)

4.3.1 The internal optical model of SCAPS

In its default mode, SCAPS calculates the generation of eh-pairs G(x) from the incident photon flux
Nphot0() from a very coarse optical model. This model contains only reflection/transmission at the two
46 Chapter 4: Working point definition

contacts, and absorption in the semiconductor layers. Thus: no interference, no scattering, no intermediate
reflectors… For a single layer of thickness d, with uniform optical absorption constant () (uniform here
means that  does not depend on position within the layer, thus (, x) = ()), and for light incident from
the left with photon flux Nphot0(), the photon flux Nphot(, x) at each position in the layer is given by
1  Rback    exp  2  d  x      
N phot  , x   N phot0     Tfront     exp   x      (15)
1  Rback    Rint exp  2d     

where
Tfront() is the (possibly wavelength dependent) transmission of the front contact; the front contact is
where the light is incident, here the left contact. Tfront is in SCAPS a property of the contact, and thus is input
in the Contact Properties Panel of the front contact.
Rback() is the (possibly wavelength dependent) reflection at the back contact; the back contact is at
the opposite side from where the light is incident, here the right contact. Rback is in SCAPS a property of the
contact, and thus is input in the Contact Properties Panel of the back contact.
Rint is a (not wavelength dependent!) internal reflection at the front contact. In SCAPS, the value
of Rint or its complement Tint = 1 - Rint, should be input in the Solar Cell Definition Panel, see Figure 4.4.

or
Figure 4.4 Input fields for the internal reflection Rint, or its complement the internal transmission Tint = 1 - Rint, in the
Solar Cell Definition Panel. The position of the input fields depend on the side where the light is incident. These input
fields are only visible if a reflection at the back contact Rback is set.

Eq. (15) is simple to interpret: If there is no reflection at the back contact (Rback = 0), there is only one pass
of light through the cell, and the light is extinguished with the familiar exponential function exp(-x). If Rback
 0, but the internal reflection at the front contact is zero, Rint = 0, there is a second pass of the light reflected
at the back contact, and an additional extinguishing function exp(+x) appears in (15) for the reflected light.
If there is internal reflection Rint  0, there is an infinite number of passes of the light, that are summed up in
(15) yielding the denominator.
In SCAPS, this model is extended to more than one layer, to layers with non-uniform (= position
dependent) absorption (, x), and to the possibility of light to be incident on the other contact. But there is
no more physics than that already described in Eq. (15).
From the photon flux Nphot(, x) at each position and for each wavelength, the generation of eh pairs is
then calculated as
G  , x     , x   N phot  , x  (16)

This is then integrated over all wavelengths of the incoming photon flux:
 max  max
G  x   G  , x  d    a  , x   N phot  , x  d  (17)
 min  min

where min and max are the short and long wavelength limits of the incoming photon spectrum Nphot0(), as
it is specified in the spectrum input file.
4.3 Generation and spectrum models (scaps 3.3.05) 47

Users that are not satisfied with the optical model of SCAPS, can directly specify the eh generation G(x) ate
each position. The optical model of Eqs. (15)-(17) is then not used. Since the impurity photovoltaic effect
(IPV) relies on this internal optical model (with extensions), assigning IPV properties to defects is
incompatible with a user specified generation G(x). Precautions in SCAPS prevent the user to define both in
one problem.

4.3.2 Analytical models for the generation and for the spectrum: motivation
Direct specification of G(x) through an input file with extension .gen is intended for users who want to use
SCAPS for the electrical calculations, but prefer to do the optical calculations themselves. A number of users
have applied SCAPS with a file input for G(x) to simulate electron beam induced current (EBIC)
measurements. Typically, they would define intense generation in a small sheet of the cell, and calculate the
current response when the central position of the beam (the position of the ‘intense generation’) is swept or
scanned over the cell thickness. For each position of the e-beam, a separate generation file should be made,
and this proves to be a rather complicated and time consuming procedure. There was a little improvement for
those brave users with SCAPS 3.3.02 of july 2015, where script commands where introduced that called one
file from a list of file names, but still it was labour intensive work.
Another problem is the mesh, thus the set of x values. SCAPS has an internal mesh generator that defines
the positions x where the semiconductor equations are solved. But a user often has his own mesh generator
for his own optical model. When the two meshes do not coincide, interpolation is used by SCAPS, but that can
be a cause of numerical inaccuracy. For example, a narrow EBIC profile G(x) specified by a user could fall
completely between two adjacent SCAPS mesh points in the middle of a thick layer, so that SCAPS wouldn’t
even ‘see’ the e-beam…
That is why we introduced in SCAPS 3.3.05 of December 2016 the possibility of defining G(x) from an
analytical model, where all parameters of the model (e.g. ‘central beam position’) can be manipulated in the
SCAPS batch, recorder and script facilities. And in the same move, analytical models for the spectrum were
implemented.

4.3.3 New facilities in SCAPS 3.3.05

The main SCAPS panel, the Action Panel was modified somewhat.

Figure 4.5 Using the internal optical model of SCAPS: define a spectrum, let SCAPS calculate G(x) with Eqs. (15) -
(17). The selection switch marked on top of this figure is only active when the light is switched on.

The default setting is shown in Figure 4.5: a spectrum is specified from a file, and the internal optical
model of SCAPS is used. External optical filters can be applied: two spectrum-cut-off filters and a neutral
density filter. When the model/file switch in the spectrum part of Figure 4.5 is set to ‘Analytical model for
spectrum’ (instead of ‘Spectrum from file’), the new facilities to calculate a spectrum from a (simple) model
are enabled, see Figure 4.5 The available spectrum models are discussed below.
When in Figure 4.5 the option ‘Directly specify G(x)’ is set, one can set the generation G(x) from an input
file or from analytical model, see Figure 4.7. The available generation models are discussed below.
48 Chapter 4: Working point definition

Figure 4.6 Setting an analytical model for the spectrum, here a black body spectrum is selected.

Figure 4.7 Directly specifying the generation G(x). Left: take G(x) from an input file with extension .gen (as in
previous SCAPS versions). Right: calculate G(x) from an analytical model (new in SCAPS 3.3.05). The selection switch
marked in red is only active when the light is switched on.

4.3.4 Analytical models for the generation G(x)

Figure 4.8 The available generation models

As of now, the user can select one from seven models as shown in Figure 4.8. When selecting one of
them, e.g. here a Gaussian generation profile G(x), the new SCAPS Model Panel opens, see Figure 4.9. The
panel is operated in much the same way as the SCAPS Grading Panel, that is maybe familiar to many users.
4.3 Generation and spectrum models (scaps 3.3.05) 49

Figure 4.9 The SCAPS Model Panel, showing a modelled gauss curve G(x), and also the generation profile from an
input file, that can be used for comparison with the G(x) model. The meaning parameters is (should be, I hope) self-
understood.

Figure 4.10 Linear and logarithmic view of the generation model ‘rectangular with Gaussian tails’, with parameters:
central position 1 m, total width of rectangle 0.5 m, standard deviation of Gaussian tails 0.3 m. Ideal light current:
10 mA/cm2.

4.3.5 The SCAPS Model Panel

The number of x points is set at the top right of the Model Panel (Figure 4.9, n = 41 is set). But modelled
points with x < 0 are discarded. The remaining number of points is displayed for information. In the example
of Figure 4.10, 101 x-points were asked, but only 73 have x > 0 and are retained. x points beyond the cell
thickness d, thus points with x > d, are not discarded, although they are not used in the SCAPS calculations of
course. But when one would increase the thickness of a layer, points beyond the original thickness could fall
inside the cell, and they will be accounted for. The x-mesh of the generation profile G(x) is independent from
50 Chapter 4: Working point definition

the x-mesh of the SCAPS calculations, and usually contains much less points. One can or cannot include the
G(x) mesh points into the SCAPS mesh by clicking this option in the Numerical Panel, see Figure 4.11.

Figure 4.11 In the Numerical Panel, one can ask (or not) to include the mesh points of G(x) into the SCAPS mesh.

Also in the right part of the Model Panel of Figure 4.9, one can select the abscissa (x-axis) and the ordinate
(y-axis) of the plot, see Figure 4.12.

Figure 4.12 Possible choices for the abscissa (x-axis) and ordinate (y-axis) to be displayed in the Model Panel for
generation models.

The x-axis options are trivial (micrometer or nanometer units), but the y-axis options are more interesting.
The definitions are:
x x
Gcumul, L R  x    G  x dx and J ideal, L R  x   q   G  x dx
0 0
(18)
d d
Gcumul, R  L  x    G  x dx and J ideal, R  L  x   q   G  x dx
x x

These integrals are calculated numerically. The total generation, expressed as a light current, thus
d
J ideal  J ideal, LR  d   J ideal, R  L  0  q  Gtot  q   G  x  dx (19)
0

is displayed in the Action Panel, see Figure 4.7 right (the value 9.9997 mA/cm2 indeed slightly deviates
from the parameter value 10.0 mA/cm2 that was set). If e.g. a substantial fraction of the G(x) profile
modelled would have x < 0 and thus lay outside the cell, the displayed Jideal value would substantially be
lower than the Jideal parameter value set.
The name of the model parameters is self-evident, we hope. For most parameters, you have two alternative
forms of the same parameter; input one and the other is adapted automatically. Examples are: positions in
m or in nm; total (integrated over cell thickness) Gtot values as a photon flux (# photons.s-1cm-2) or as an
ideal current (mAcm -2). And finally, as is customary in SCAPS, there is only one mouse click between you
and your logarithmic thoughts .
You can display also the G(x) data from an input file, but that is only for information: it allows to compare
a modelled G(x) with G(x) data from a file. To actually set G(x) from an input file, proceed as in Figure 4.7
left. (And don’t forget, SCAPS is expecting the input files as a table of x (in µm) Geh (in #/m3.s), separated
by white space (blanc or tab)).
4.3 Generation and spectrum models (scaps 3.3.05) 51

4.3.6 The generation models

4.3.6.1 Constant generation G

There is only one parameter, Gtot, expressed either as a photon flux or as an ideal current. This generation is
smeared out uniformly over the cell thickness. Thus, if the cell thickness is changed subsequently, the
generation G(x) is changed accordingly, but Gtot remains the same.

4.3.6.2 Exponentially decaying G(x)

The first parameter is Gtot or Jideal, as in the previous model. The other parameter is the characteristic length
of the exponential decay. The G(x) laws are

Gtot  x  Gtot  dx 

G  x  exp    or G  x   exp    (20)
Ldecay, L  Ldecay, L  Ldecay, R  Ldecay, R
   
In these equations, d is the actual cell thickness.

4.3.6.3 Gaussian generation profile G(x)

The parameters are: Gtot or Jideal, the central position xcentral and the standard deviation . Part of the profile
could fall left of the cell (thus x < 0), and then will be discarded. When the cell is too thin for the defined
profile, part of the Gauss curve with x > d will not be accounted for.

4.3.6.4 Rectangular generation profile G(x)

The first parameter is Gtot or Jideal, as in the previous models. The other parameters are the central position of
the rectangle, and the full width. When you specify e.g. xcentral = 1 m and wfull = 0.3 m, the generation will
be a constant between x = 0.7 m and 1.3 m and 0 outside. When the cell thickness was only 1 m, or you
set a total thickness of 1 m afterwards, only the left half of the generation profile will lay within the cell.

4.3.6.5 Rectangular generation profile G(x) with tails

This works like a rectangular profile, but with a tail left and a tail right; both tails are symmetric. They can be
exponential (then define the decay length), or (half) Gaussian (then define the standard deviation).

For all G(x) profiles with exponentials or Gauss curves, it might be worth while to look at the logarithmic
graph!

4.3.7 The spectrum models

The available spectrum models are shown in Figure 4.13.

Figure 4.13 The available spectrum models

Again the SCAPS Model Panel opens upon selecting a spectrum model. The options for the x-axis and y-
axis for spectra are shown in Figure 4.14.
52 Chapter 4: Working point definition

Figure 4.14 Possible choices for the abscissa (x-axis) and ordinate (y-axis) to be displayed in the Model Panel for
spectrum models.

The abscissa can be switched between wavelength and photon energy. The ordinate can be based on the
photon flux (the number of photons incident at each wavelength) or the photon power (the photon power
incident at each wavelength). Both can be shown as such, or integrated from the short wavelength limit of
the spectrum (‘cumulative spectrum’).
There are two versions of white spectra: one has a constant photon flux over the specified wavelength
range (expressed as a total photon flux or ideal current), the other has a constant photon power over the
wavelength range (expressed as total incoming power, in suns or in mW/cm 2; 1 sun = 100 mW/cm2 = 1000
W/m2). Similarly, there are two versions of monochromatic spectra, one with specified total photon flux, and
one with specified total photon power. The black body spectrum is there to please our colleagues more
oriented towards fundamental theory. Of course, the black body temperature is the main parameter (input
Tblackbody or the wavelength of the maximum, according to Wien’s displacement law). As you will observe
when playing with this model, black body radiation extends very far in the infrared wavelength range. To
avoid spending many -mesh points on not-useful wavelengths, one can limit the  range of interest.

4.4 The initial working point

In order to perform simulations for metastable defects (§3.7) two (different) working points are needed. The
working point which is present since the birth of SCAPS still corresponds to the working point which is
applied during the ‘measurement’. A second working point ‘the initial working point’ corresponds to the
conditions which are applied to the structure in order to bring it in the desired metastable state. For example
if you want to simulate a red-on-bias experiment, this working point will consist of a negative voltage bias
and red illumination.
Pay attention! The temperature at this initial working point is usually high as the metastable transitions are
thermally activated. Many metastabilities can not be induced at low temperature because the transition is too
slow and you never get to an equilibrium situation in a reasonable time span. SCAPS always calculates
equilibrium conditions, and waits in fact until an infinite amount of time has passed. You should take this
into account when setting a low initial working point temperature .
The ‘Initial State Workpoint Panel’, where you can set the initial working point can be accessed from the
action panel only if at least one metastable defect is present in the problem definition.

Figure 4.15 Accessing the initial workpoint panel: press this button on the action panel.

You can save and load all settings of the initial state workpoint into and from a file. These files have
extension .wp2 (from ‘second work point’), and reside in the \def folder.

4.5 Shunt conductance and series resistance

It is possible to introduce an external shunt conductance and series resistance to the structure on the action
panel, see Figure 4.16.
4.5 Shunt conductance and series resistance 53

Figure 4.16 Introducing external shunt conductance and series resistance

Both resistances can be switched on/off and for a shunt conductance you can either define its resistance or
conductance.
These external resistances only affect I-V, C-V and C-f simulations (no QE simulations) The voltage which
you apply to the sample (either the working point voltage or the voltage in the iv or C-V simulation) is
interpreted as the internal voltage over the sample, without external resistance. This voltage gets then
recalculated after simulation in order to get the external applied voltage in I-V and C-V simulations. In small
signal simulations, the value of G and C are also recalculated after simulations.
The values of the external resistances are saved in the definition file of the solar cell structure.
 Chapter 5: Single shot calculations

The main functionality of SCAPS is to solve the one-dimensional semiconductor equations. In the bulk of the
layers these equations are given by (21) together with the constitutive equations (22).
       def 
  0   q  p  n  N D  N A  
 x  x   q 
 J
n  U  G  n
 n (21)
 x t
 J p p
 U p  G 
 x t

  n n E Fn
 J n   q x

 (22)
 J    p p E Fp
 p q x
Together with appropriate boundary conditions at the interfaces and contacts, this results in a system of
coupled differential equations in (Ψ, n, p) or (Ψ, EFn, EFp). SCAPS numerically calculates a steady state and a
small signal solution of this system. Hereto, the structure is first discretized (creating a mesh). A steady state
workingpoint situation (see §Chapter 4) is calculated and when required a small signal analysis is performed.

5.1 Calculation roadmap

5.1.1 Meshing
The first step in every calculation is to discretize the structure. The meshing algorithm of SCAPS provides:
 Coarse meshing in the middle of a layer
 Finer meshing near the interfaces and contacts
 Two discretization points (with identical spatial coordinate) for each interface
 The mesh can be optimized during calculation.
The result of the discretization algorithm is illustrated in Figure 5.1.
56 Chapter 5: Single shot calculations

300

250

meshpoint #
200

150

100

50

0
0 2 4 6 8 10

x(µm)

Figure 5.1 Result (right) of the discretization of the structure on the left. (Red: p-type, blue: n-type, green: intrinsic)

The basic algorithm is designed to provide a lot of points in regions where properties experience large
variations (close to interfaces/contacts) and fewer points where the properties are expected to remain fairly
constant (in the bulk). However, when performing simulations with strong gradings, with multivalent defects
or with the IPV-effect this meshing procedure might be unsufficient. Hence, it is possible to optimize the
mesh at every iteration step. This option can be set on the numerical panel as shown in Figure 5.2.

Figure 5.2 Recalculate mesh settings on the numerical panel

When recalculate mesh is switched on at each iteration the ratio between the following properties at two
adjacent meshpoints are assessed: exp(qΦ/kT), exp(EFn/kT), exp(EFp/kT) (all dimensionless), the
recombination rate R and the generation rate G (both R and G in /m3). When that ratio is larger than the
maximum ratio fmax set on the numerical panel, extra meshpoints will be inserted to enable a better
simulation. If the ratio is smaller than the minimum ratio fmin set on the numerical panel, the meshpoint is
removed. Precautions are taken not to remove too many meshpoints in a row. The ratio of the recombination
and generation rate are only taken into account if q|R| < Rlimit/L or qG < Glimit/L respectively. Here Rlimit and
Glimit are parameters set on the numerical panel (atypically for SCAPS, these parameters are displayed and
saved in units of A/cm2), and L is the total cell thickness. At room temperature, the default value fmax = 1.60
corresponds to a difference of about 12 mV or 12 meV of , EFn or EFp between two adjacent mesh points,
whereas the default value fmin = 1.05 corresponds to a difference of 1.2 mV or meV.
The recalculate mesh settings are all saved in the definition file.
The effect of the recalculation of the mesh is illustrated in Figure 5.3, where the defect occupation of an
amphoteric defect is calculated.
5.1 Calculation roadmap 57

Figure 5.3 Occupation of an amphoteric defect calculated without (left) and with (right) mesh recalculation

5.1.2 The pathway to a solution

When clicking the ‘Calculate: single shot’-button, the measurements you selected are calculated for the given
structure. Hereto the system of (21) is solved numerically, using a Gummel iteration scheme with Newton-
Raphson sub-steps [12, 15]. The parameters of this scheme are available on the numerical panel as
convergence settings. As a rule of thumb, the best way to improve convergence is first to increase the
number of iteration steps and then perhaps increase the termination criterion thresholds.
In order to get to a solution, an initial guess has to be made. Care should however be taken, because when
this initial guess differs too much from the solution of the system, the iteration procedure will fail to
converge (after a reasonable number of iterations steps). SCAPS does not ask an initial guess from the user,
but calculates several situations in order to get to a solution under the working point circumstances and at the
first measurement point. The strategy which is followed is illustrated in Figure 5.4.
Startpoint

Equilibrium Situation (dark; 0V)

LIGHT

DARK
Short circuit (light; 0V)

Workingpoint Situation

1st Calculation point

closest to workingpoint

Figure 5.4 Getting to the working point and first calculation point…

Every calculation starts at the startpoint. This is a very simple situation, assuming the quasi-Fermi levels
to be zero throughout the structure and no potential drop to be present over the structure. This is used as an
initial guess to get to the equilibrium situation (no illumination, no voltage applied). When the working point
conditions are under darkness, this equilibrium condition is used as an initial guess to calculate the solution
under working point conditions. When illumination is switched on however, the short circuit situation is
calculated in an intermediate step to serve as the next initial guess.
When the difference between two situations is too large, the iteration procedure will diverge. Therefore,
SCAPS implicitly introduces intermediate situations between the equilibrium/working point; equilibrium/short
circuit and the short circuit/working point situation. Hereby gradually increasing the applied bias voltage or
58 Chapter 5: Single shot calculations

the illumination strength. The number of intermediate steps to be taken can be controlled on the action panel,
see Figure 4.1. When divergence occurs, increasing this number can sometimes lead to convergence,
however, this slows down your calculation.
When the working point situation is calculated SCAPS immediately calculates the first calculation point of
the measurements to be simulated (without any sub-steps). For C-f and QE simulations, this first point is
identical to the working point conditions. For I-V and C-V simulations a voltage range has been given,
independently from the working point voltage. SCAPS will then jump to the edge of this voltage range which
is the closest to the working point voltage. IT IS THUS STRONGLY ADVISED TO CHOOSE THE WORKING POINT
VOLTAGE IDENTICAL TO THE STARTING POINT OF THE VOLTAGE RANGE YOU WANT TO SIMULATE.
When performing batch (or recorder) simulations, every calculation is started at the startpoint again. This
is often unwanted, e.g. when gradually increasing the working point voltage it is more logical to use the
previous batch calculation as an input for the next calculation. Varying, voltage, frequency, series/shunt
resistance in a batch calculation, this latter approach can be taken by checking the appropriate checkbox on
the numerical panel.

5.1.3 Small signal analysis

Capacitance and conductance simulations are calculated by performing a small signal analysis at the working
point conditions (of course with the voltage (C-V) and frequency (C-f) get changed). This analysis leads to an
expression for the small signal current, which is a complex number. The cell structure is then analysed as if it
were a parallel connection of a (frequency dependent) capacitance and a (frequency dependent) conductance.
J
Y     jC    G   (23)
uext

5.2 Setting up a single shot simulation

SCAPS is able to simulate four different measurements: I-V; C-V; C-f and QE. The settings of these
measurements are immediately available on the action panel, see Figure 5.5. Of course, many more user
defined measurements can be simulated by exploiting the recorder facility .

Figure 5.5 Measurement simulation settings

 Each of the measurement simulations will only be performed when its appropriate checkbox is checked.
 A voltage/frequency/wavelength range can be set up.
 The number of points (limited to 201) can be set up, or a voltage/wavelength increment or the number of
points per frequency decade can be given.
 When a C-V simulation is performed you get the I-V simulation for free (no need to specify separately).
When performing I-V simulations under illuminated conditions, one is often only interested in the voltage
range up to the open circuit voltage. However, before starting the simulation Voc is still unknown. Hence, one
can ask SCAPS to stop the I-V simulation under illumination as soon as the current becomes positive by
checking the ‘stop after Voc’-option.
5.3 Numerical parameters 59

5.3 Numerical parameters

All parameters contained in the Numerical Panel are listed in Table 5.1.

Table 5.1 The numerical parameters in SCAPS, with their units as displayed in the Numerical Panel, and their
default values.
numerical parameter name default value unit
Convergence settings
maximum number of iterations 1000 -
clamping of changes between iterations
of electrostatic potential  0.5 kT/q
of electron Fermi level EFn 0.5 kT
of hole Fermilevel EFp 0.5 kT
stop criterion: stop if all updates < stop criterion
of electrostatic potential  0.001 kT/q
of electron Fermi level EFn 0.001 kT
of hole Fermilevel EFp 0.001 kT
Mesh generator settings, see section 5.1.1
recalculate mesh between iterations yes -
maximum number of mesh adaptions at each voltage 10 -
maximum ratio between neighbouring mesh points (fmax) 1.60 -
maximum ratio between neighbouring mesh points (fmin) 1.05 -
generation limit (Glimit), expressed as current density 1.0 A/cm2
recombination limit (Rlimit), expressed as current density 0.001 A/cm2
Calculation of quantum efficiency QE, see section 6.4.7
calculation mode constant Nphot -
monochromatic photon particle flux Nphot in QE calculation 31018 cm-2s-1
monochromatic photon power flux Pphot in QE calculation 0.1 mW/cm2
Defect settings, see section 3.6.3
number of discretization levels for distributed defects 7 -
width of tail-like energy distribution in multiples of Echar 7.0 Echar
width of Gaussian energy distribution in multiples of Echar 6.0 Echar
Tunnel settings, see section 3.9.2
minimum height of bulk tunnel barrier 2.0 kT
choice of tunnel mass min of -
adjacent
layers
allow band-to-band tunneling no -
allow intra-band tunneling no -
allow tunneling to interface defects no -
allow tunneling to contacts no -
Update of metastable occupation, see 3.7.4 -
use clamping in iteration of occupation fmetastable no -
maximum number of iterations 250 -
maximum relative error 0.001 -
Batch and Recorder calculations, see sections 7.4 and 8.2
to calculate work point, start from previous batch result yes -
maximum V for the recording of solar cell characteristics 2.0 Volt
minimum V for the recording of solar cell characteristics 0.02 Volt

Since SCAPS 3.3.02, version august 2015, the numerical parameters of Table 5.1 can be set back to their
default values by clicking a button in the Numerical Panel. When saving a problem as an all SCAPS file
60 Chapter 5: Single shot calculations

(extension .scaps), or as a .def file from within a script, all numerical parameters are saved. When saving as a
.def file in interactive mode, the numerical parameters are only saved when their value differs from the
default.

5.4 Numerical limitations

Even though many people believe computers to be perfect when performing calculations, they are not!!
Every computer has its machine accuracy, which is defined as the smallest (in magnitude) number which,
when added to 1.0, produces a floating-point result different from 1.0. This accuracy usually is a small
number (< 10-15), so that in a pragmatic view one can consider it “accurate enough and close to perfection”.
Even though this view is commonly adopted, it is branded as “naïve” and plain “fiction” in [7]. For the
simulation of semiconductors, this accuracy is indeed often insufficient, and care should be taken to avoid
numerically unstable operations. SCAPS does take care, but sometimes things can still go wrong. Below you
can find a list of phenomena which should set of an alarm in your head indicating a numerical problem, and
tips and tricks to avoid them and to avoid convergence failures:
 A negative capacitance or a capacitance increasing with increasing frequency is often not physical.
Though exceptions exist, most of the time this behaviour is a result of a numerical error when calculating
the capacitance from the small signal current. This occurs when the imaginary part of the small signal
current is (in magnitude) much smaller/larger than the real part. This might be the case when large
voltage biases or strong illumination conditions are applied.
 The current through the sample (visible in the lower left corner of the EB-panel, in green) should be
constant through the sample. SCAPS is optimized to calculate pn-junctions (p to the left, n to the right)
and is much less stable for np-junctions. IT IS HENCE STRONGLY RECOMMENDED TO KEEP THE P-SIDE OF
YOUR STRUCTURE TO THE LEFT AND THE N-SIDE TO THE RIGHT. np-junctions tend to lead to non-uniform
currents through the cell, which is unphysical.
 Keep the variation between different calculation steps limited to avoid divergence. The reason for this
should be obvious when you have read §5.1.2.
 Keep it realistic! SCAPS is developed and tested to simulate realistic situations, hence things can go
wrong when simulating unphysical situations.
 Don’t overdo! Do you really want to know the current at T = 2 K and V = 3 kV? Charge carrier densities
scale exponentially with 1/kBT. Hence, the stability of the algorithms exponentially decreases with
decreasing temperature, which initially leads to a higher number of iterations needed to get to
convergence and at even colder temperatures to an inescapable divergence.
 Chapter 6: Result analysis

SCAPS performs a lot of calculations any time you click the ‘calculate’-button. Your job as a user is to
analyse those results. Fortunately there is a lot of help. The analysis-panels can easily be accesed from the
action panel or any other analysis panel, see Figure 6.1.

6.1 Navigating to the analysis

The analysis-panels can easily be accesed from the action panel or any other analysis panel, see Figure 6.1.

Figure 6.1 Navigating to the resuls from the action panel (left) or any other panel (right)

Several options are available on every panel: saving data, showing data, saving graphs and plotting the
panel (sending to a printer). There are options available for scaling and zooming of graphs and to show more
info about the plotted curves. Other options are panel specific.
At the bottom of every panel there are two comment windows. The left window is auto-generated and
gives the definition file used with its last saving and the when the simulation was performed. The right
window can be used to write personal comments.

6.2 Zooming and scaling

Most of the graphs are scalable. For most axes a logarithmic or linear scale can be chosen and/or the absolute
value of the property can be plotted. The axis-range can be set by clicking a ‘scale’-button, which opens the
following pop-up menu.
62 Chapter 6: Result analysis

Figure 6.2 The Scaling panel

 The scaling of both axis can be set automatically (default) or manually

 A minimum/maximum value can be given
 The settings on this panel can be saved/loaded (generating an ASCII-txt-file)
 Clicking the minimal positive button sets the y-axis scaling to manual and sets ymin to the smallest
positive datapoint which needs to be plotted. This might be very useful when there are negative
datapoints in a semi-logarithmic diagram.
(Almost) all diagrams have moreover zooming facilities. In order to zoom in you should hold the CTRL-
button pressed whilst drawing a rectangle with the mouse on a graph. For zooming out you should hold the
CTRL-button pressed whilst right-clicking with the mouse. Resetting the initial zooming can often be done
as well by changing the log/lin property of the axis. This only works if the axis scaling is set to auto.

6.3 Curve info and legend

Performing several simulations, the graphs can get quite crowded (unless you click ‘clear all simulations’
before every simulation). Hence, facilities are provided to get more info about the simulated curves. When
the curve info option (top right corner of any panel) is switched on, any click on a graph will render a pop-up
menu with information about the graph, curve and point you clicked on. The point/curve closest to the
position where you clicked will be selected.
It is also possible to interactively hide and show curves on graphs. This option is available by right-

clicking on the -button (left-clicking leads to plotting the entire panel). A pop-up window is
opened which gives a list of all available curves (with a small legend). (Un)checking items in the list leads to
hiding/showing the curve on the panel.

6.4 Measurement specific options

6.4.1 The energy band panel

This panel shows the band diagram, the carrier densities, the current densities and the occupation of defects
of the last calculated work point. When no measurement is selected or a QE or C-f measurement is selected it
displays thus the working point conditions as set on the action panel (or in the batch). If a C-V or I-V
simulation is performed it displays the conditions of the last calculated voltage point. These conditions are
mentioned on the bottom left part of the panel.
6.4 Measurement specific options 63

The panel is displayed during calculations, as such it provides a small movie of how the bands evolve
during simulation. When your computer is too fast, you might want to slow down this movie. This is possible
by checking the ‘pause at each step’-option on the action panel, see Figure 5.5. This way the program will
pause at every calculation step and you can move forward by clicking the continue button (on the EB-panel
or the action panel) or abort by clicking the stop-button (on the action panel).
The scaling of the x-axis on all graphs of the EB-panel is the same. Hence, unless using the zooming
option, all graphs display the same part of the structure you simulated.
The defect level energy of the defects present and the transition energy of metastable defects are indicated
on the band diagram graph. When the defect has an energy distribution (e.g. Gauβ or CB tail), the energy Et
as set in the problem definition is displayed. This energy can of course be graded. The colour legend used for
displaying the defect levels can be chosen either grouped according to the defect type (e.g. single acceptor,
amphoteric,…) or grouped according to the charge state of the defect level (e.g. 0/-, 2+,+,…).
On the occupation probability graph the defect level occupation is displayed. One can choose whether to
display the occupation with electrons (default) or holes. For a metastable defect transition the ‘occupation
with electrons’ represents the fraction of defects in the acceptor configuration and the ‘occupation with
holes’ the fraction of defects in the donor configuration. Just as for the band diagram graph a colour legend
can be selected according to defect type or defect charge states. Additionally, a colour code according to the
charge of the states can be chosen. In this case the fraction of defects in a specific charge state are displayed.
When this option is selected, neutral defects and metastable transitions are not displayed however.
When the defect has an energy distribution (uniform, Gauß, tail) the occupation of several (typically 7)
sublevels in this distribution is shown, and the occupation of outermost levels of the distribution, considered
in the calculation, is drawn in a thicker line. You will observe that this occupation diagram can get quite
complicated and not straightforward to interpret. The use of the Curve Info feature (Section 6.3) is highly
recommended in interpreting the occupation graphs: click on a graph, and Curve Info shows you what this
graph represents.

6.4.2 The generation-recombination panel

A more detailed view of the defect occupation and of the recombination and generation can be found on the
generation-recombination panel. On this panel several recombination-generation processes can be show by
checking the appropriate checkboxes. An overview of the available defect(level)s is given in a tree structure.
(Un)checking them will make the corresponding curves (un)visible. A graphical view of the simulated
structure or a more detailed overview of the simulation conditions can be shown by switching the

-button.

6.4.3 The IV-panel

The results of the current-voltage simulations are shown on the IV-panel. The left graph displays all I-V
simulations. The right graph gives detailed information about the recombination currents in the last
simulation. This allows to see the main recombination mechanism in the structure for varying voltages.
If the simulation is performed under illumination, the solar cell parameters are calculated and shown, see
Figure 6.3. If SCAPS needed to perform an extrapolation to determine these parameters, e.g. because the
simulation range was too narrow, the warning LED below the parameter turns red. As most users set the
voltage range from Vstart = 0 to some Vstop, the need for extrapolation will first show up for Voc and thus FF
(e.g. if Vstop < Voc, when the references for voltage are such that both are positive numbers). When the voltage
is more restricted, Vstop < Vmpp (mpp is the maximum power point), then also extrapolation is needed to
determine the efficiency . When Vstart was set to 0 (the default value), Jsc will be determined precisely (no
interpolation or extrapolation needed). When extrapolation was needed, and when SCAPS deems the result
not physical, both the extrapolated (but unreliable) parameter and the warning LED are not shown. This is
64 Chapter 6: Result analysis

e.g. the case if SCAPS would find Voc < 0 or Voc > 10 Volt (we do not believe in supermegagigantic good solar
cells…). So, if some of your solar cells parameters do not show up (e.g. you see Jsc and  but not Voc and
FF), just extend your voltage range to include Voc, and do not immediately start to harass us with e-mails
about ‘malfunctions’ in SCAPS1.
As already mentioned, when the generation was ‘from file’, no efficiency can be calculated, but the
collection efficiency is given instead.

Figure 6.3 Some examples of visualization of the solar cell parameters Voc, Jsc, FF and . The three upper lines are
for a simulation of simple pn.def with Vstart = 0 and Vstop = 0.2 V, 0.6 V and 0.8 V. The bottom line is for a
measurement.

The I-V graph is by default directly given as an I(V) graph, but the current density J can also be scaled or
subtracted automatically with the short circuit current by changing the current mode on top of the graph
(Figure 6.4). This enables you e.g to perform analyses as discussed in [16] or to study an illuminated I-V -
curve with a logarithmic scaling (this is also possible by clicking the ‘absolute’ checkboxes next to the V and
I axes).

Figure 6.4 Options to display the I-V curves: full custom scaling; fast buttons for linear/logarithmic and absolute
value/algebraic value; options to scale (with Jsc) or shift (over Jsc) all J-V curves.

6.4.4 The ac-panel

This panel gives an overview of the small signal current (left side of the panel) and of the small signal
variations in potential and quasi-Fermi levels (right side of the panel). These properties are complex
numbers. The top graphs give the amplitude, the bottom graphs the phase.

6.4.5 The CV-panel

This panel displays the capacitance and conductance as a function of applied voltage. Moreover, a Mott-
Schottky diagram and the apparent doping density are calculated. In order to calculate this apparent doping
density profile, a numerical differentiation of the data is needed together with a choice of the relative
permittivity. These options can be set by clicking the ‘Analysis method’-button, which opens the
‘Admittance (capacitance) analysis panel’ which is discussed in §6.4.5.1.

1
In the first months of 2019 a real avalanche of such help-cries: “help! Voc and FF do not show up: what is wrong with
SCAPS?” arrived at M.B.’s desk. All of them were by your researchers simulating perovskite solar cells with Voc > 1.0 V,
with the standard voltage range setting Vstart = 0 and Vstop = 0.8 V. All of these help requests could have been avoided by
simple inspection of the IV curve and a little bit of common sense — or by reading this recent addition to the SCAPS
Manual (Marc B., 25-3-2019).
6.4 Measurement specific options 65

6.4.5.1 Admittance (capacitance) analysis panel

This panel allows to set the dielectric permittivity and the parameters of the numerical differentiation used in
the calculation of the apparent doping density profile and the AS-analysis (§6.4.6). It also allows an
automatic calculation of the attempt-to-escape frequency on the AS-panel. The graph of the apparent doping
density profile is not immediately adapted to a change of these parameters. The new parameters are only
valid for new simulations. The parameters can however be immediately updated in the AS-analysis as there
is a ‘recalculate’-button on the AS-panel.

Figure 6.5 The Admittance (capacitance) analysis panel

Only relative dielectric permittivity values on the left/right side of the layers can be chosen by clicking
any of the select buttons on the panel.
The layer which is chosen here determines automatically the choice of the layer-parameters (density of
states and thermal velocity) that are used to calculate an attempt-to-escape frequency ν0. This frequency is
calculated as (24) when the switch is set to electrons or as (25) when it is set to holes. The capture cross
section is determined by choosing a defect(level).

 T2
0  0  n vth,n NC (24)
2

0T 2
0    p vth, p NV (25)
2
The numerical derivative is calculated by fitting either a best straight line or a best parabola through a point
and its neighbours. The number of neighbours which have to be taken into account and the order of the
polynomial (straight/parabola) in the middle and at the edges of the simulation/measurement range can be
chosen.
66 Chapter 6: Result analysis

6.4.6 The Cf-panel

This panel displays the capacitance and conductance as a function of simulation frequency. Moreover, a
Nyquist plot and a plot of G/ω is given. A facility has been added to analyse the results according to the
admittance spectroscopy (AS) method outlined in [17] and extended in [18]. By clicking the ‘Admittance
spectroscopy’- button the AS-panel is displayed, see Figure 6.6.

Figure 6.6 The AS panel

The terminology and parameters used in this panel are those defined in [17, 18]. You can select several
properties as abscissa and ordinate

Figure 6.7 AS panel: abscissa (left) and ordinate (right) choices

The depletion width used in the formulas is immediately determined from the capacitance value at a
specified frequency. The dielectric permittivity used to calculate this depletion width and the parameters
governing the numerical differentiation can be specified by clicking the ‘Analysis method’-button, which
opens the ‘Admittance (capacitance) analysis panel’, see §6.4.5.1. When clicking the ‘Data from cell
definition’-button the attempt-to-escape frequency is calculated from the data set on the ‘Admittance
(capacitance) analysis panel’ as well. Its value can however afterwards be changed again. The built-in
voltage which is needed for the calculation of Nt can be chosen temperature dependent according to (26).
N N 
Vbi  Eg  k BT ln  A D  (26)
 NC NV 
6.5 Managing measurement data 67

6.4.7 The QE-panel

This panel allows to analyse the QE simulations. Specifically, the meaning of QE in SCAPS is external
quantum efficiency QE: the number of electrons leaving the cell (ass current) divided by the total number of
photons incident on the cell.
On the horizontal axis one can display either the wavelength or the photon energy of the monochrome light.
On the vertical axis one has a wider choice:

Figure 6.8 Ordinate choices on the QE-panel

The fourth option (spectral response SR) is a recent addition (august 2017) to serve users that want to
simulate photodetectors. SR is defined as the short circuit current divided by the total incident light power,
and thus is in A/W (this is also mA/cm2 divided by mW/cm2, should you insist). Note that SR = QE/h when
SR is in A/W, QE dimensionless and h in eV. The four last options are sometimes used to derive a band gap
value from QE measurements or simulations.
The QE is calculated by comparing the current at the workingpoint conditions and the current when
adding an additional amount of monochromatic photons. This number of photons can be set on the numerical
panel, Figure 6.9. When both currents are large compared to their difference, it is possible that your
computer makes a numerical error in the subtraction leading to a non-physical QE which is negative or larger
than 100 %.

Figure 6.9 Setting the number of photons for the QE simulation in the Numerical Panel.

In SCAPS 3.3.02, version august 2015, the defaults values displayed in standard units of the SCAPS panels
(thus: cm, mA, mW-based), and were changed to Pphot = 0.1 mW/cm2, corresponding to 0.001 sun, and to
31018 /cm2s, that is for ( = 620 nm, h = 2 eV) photons, also roughly corresponding to 0.001 sun. These
default values could be representative for actual QE measurements with a monochromator set-up.

6.5 Managing measurement data

SCAPS provides facilities to compare simulation results with measurement data. Measurement data can be

loaded by clicking the -button, which redirects to the ‘Manage measurements panel’, Figure
6.10.
68 Chapter 6: Result analysis

6.5.1 The Manage measurements panel

This panel shows a list of the measurement files which are currently loaded. Additional measurements can
be added, measurements can be removed or replaced and the order of the list can be changed. By
(un)checking items in the list you can control which measurements have to be displayed in the graphs. When
selecting an item in the list, a summary of its properties are displayed on top of the panel. This allows to
check whether you have loaded the measurement you intended and whether the working point conditions of
the file have been read correctly.

Figure 6.10 Manage measurements panel

6.5.2 Structure of a measurement file

Any ASCII-text file can be read as a measurement file. In particular, any of the simulation results which
have been saved by SCAPS can be read as a measurement file. So, if you feel unsure whether your file will be
read correctly, you can try to start from a SCAPS-results file.
The file extension indicates which kind of measurement it contains. The allowed extensions are ‘.iv’, ‘.cv’,
‘.cf’ and ‘.qe’.
A measurement file is read until the first line which can be interpreted as measurement data is found. This
is a line starting with at least two numeric values. All lines above are considered to be part of the header
(which might be mere comments or contain information about the working point conditions) All lines below
the first data line are considered to contain measurement data. As a result, any line which can not be
interpreted as data which occurs below the first data line will be discarded and can thus not be interpreted as
working point conditions. Hence, when more than one measurement is contained in a file only the working
point conditions of the first measurement will be read, even though the data points of all measurements will
be loaded!

6.5.2.1 The workingpoint conditions

SCAPS will try to interpret the information in the header of the file as the working point conditions of the file.
When any of these words is found at the beginning of a line, the numerical value which follows is saved.
None of these words is case-sensitive. An overview of the code-words used is listed below.

Table 6.1 Workingpoint code words

code-word(s) unit Default value remarks

area; cell area; cellarea; cm2 1.0 See below
6.5 Managing measurement data 69

temperature; temp K 300

bias V 0.0 Only for C-f and QE
frequency; freq; frekw Hz 106 Only for C-V
Incident power; Pin; P in; P_in mW/cm2 100 Only for I-V and QE
Incident light; Light power
dark # light Whenever this word is found, the
incident power is set to zero
Only for I-V and QE
Units See §6.5.2.2

 The area is used to scale the current, capacitance and conductance values to an area of 1 cm2.
 Alternative units for the area are supported. Adding any of the words [“cm2”, “cm^2”, “cm2”]; [“mm2”,
“mm^2”, “mm2”] or [“m2”, “m^2”, “m2”] to the area statement will change the unit of the area read to
cm2; mm2 or m2. If none of these words are found the unit of the area is cm2.
 In a similar way as for the area, different units for the incident power are supported: mW/cm2;
mW/cm^2; mW/cm2; W/m2; W/m^2; W/m2; mW/m2; mW/m^2; mW/m2; W/cm2; W/cm^2; W/cm2.
 The temperature and bias voltage are only used to calculate admittance spectra
 The incident power is used to calculate the efficiency.
 When the incident power is zero (or when the code-word dark has been found) the I-V parameters (thus:
, Jsc, Voc and FF) will not be calculated.

6.5.2.2 The measurement data

When a first data line is found, all subsequent lines will be interpreted as data lines. Up to five numeric
values will be read. The meaning of these numbers varies according to the measurement type as listed below.

Table 6.2 Data in the columns of a measurement file

measurement #1 #2 #3 #4 #5
I-V V (V) I (mA)*
C-V V (V) C (nF)* G (S)* W (μm) Napparent (cm-3)
C-f f (Hz) C (nF)* G (S)*
QE λ (nm) QE (%)

 The properties marked with a * will be scaled with the area, when no area has been read, an area of
1 cm2 is assumed.
 The depletion width W and the apparent doping density Napparent can be read from the measurement-file,
but they can also be calculated from C and V by SCAPS. This choice is governed by a switch button on
the capacitance analysis panel.

The default units for the measurement data are given in the table. It is however possible to change them. In
order to do this, a line starting with the code-word “units” has to be added in the header. Next to this code-
word a list of properties with their unit can be specified.
Examples: units: I: A V: mV
Units: C: pF V: V G: mS
A list of property-names which can be changed for every measurement type are given in the Table below:
70 Chapter 6: Result analysis

I-V V: I: J:
C-V V: C: G: W: N:
C-f f: C: G:
QE lambda: QE:

Pay attention! The semicolon ‘:’ is obligatory.

6.6 Saving results

When clicking the save or show button, the following properties are saved in an ASCII file which can be
read using e.g. notepad or MS Excel. More options to keep track of and save properties are available in the
recorder facility, see Chapter 8.

Table 6.3 Overview of the properties which can be saved on the EB panel

property remark
x(μm)
Ec(eV) Conduction band energy
Fn(eV) Quasi-Fermi level for electrons
Fp(eV) Quasi-Fermi level for holes
Ev(eV) Valence band energy
n(/cm3) Free electron density
p(/cm3) Free hole density
rho(defect) (/cm3) Charge density in defects
net doping (/cm3) ND - NA
rho(/cm3) p -n + ND - NA + charge in defects
E(V/cm) Electric field
jn(mA/cm2) Electron current density
jp(mA/cm2) Hole current density
jn_tunnel(mA/cm2) Electron tunnel current density
jp_tunnel(mA/cm2) Hole tunnel current density
jtot(mA/cm2) Total current density
generation(#/cm3.s) Generation rate
recombination(#/cm3.s) Recombination rate
cumulative generation Cumulative generation rate (summation from the side where the
(mA/cm2) light is incident, to the opposite side)
cumulative recombination Cumulative recombination rate (once with summation from left to
(mA/cm2) right, and once summed from right to left). Also interface and
contact recombination taken into account
jn1(mA/cm2) Interface recombination current for electrons (left side of the
interface)
jn2(mA/cm2) Interface recombination current for electrons (right side of the
interface)
jp1(mA/cm2) Interface recombination current for holes (left side of the interface)
jp2(mA/cm2) Interface recombination current for holes (right side of the interface)
jn1[with tunnel](mA/cm2) Same as jn1 but now with tunneling
jn2[with tunnel](mA/cm2) Same as jn2 but now with tunnelling
jp1[with tunnel](mA/cm2) Same as jp1 but now with tunnelling
jp2[with tunnel](mA/cm2) Same as jp2 but now with tunneling
js(mA/cm2) Total surface recombination current density
pi(C/cm2) Surface charge density

A note on “cumulative” generation and recombination

6.6 Saving results 71

The cumulative generation is always calculated starting from the side where the light is incident. Thus:
x
Gcumulative  x    G  x dx illumination from left, thus:
0
Gcumulative  0   0 (27)
Gcumulative  d   Gtotal

or
d
Gcumulative  x    G  x dx illumination from right, thus:
x
Gcumulative  0   Gtotal (28)
Gcumulative  d   0

The cumulative recombination is offered in two directions of summation (integration): from left to right, and
from right to left:
x d
Rcumulative,L R  x    R  x dx Rcumulative,R  L  x    R  x dx (29)
0 x

The notation 0- means that the left contact is included, and d+ that the right contact is included. At an
interface, Rcumulative takes a discontinuity (jump) equal to the interface recombination. With these definitions,
Rcumulative,LR  0  Rleft contact Rcumulative,LR  d   Rtotal
(30)
Rcumulative,R L  0  Rtotal Rcumulative,R L  d   Rright contact

Table 6.4 Overview of the properties which can be saved on the Generation recombination panel when saving the
generation data. All properties are split up per layer, per defect in this layer and per charge state in this defect. (first
index = layer, second index = mechanism, third index = defectlevel (0 is the most positively charged and hence the
closest to the valence band))

property remark
x (µm)
Total recombination Total recombination rate
(#/cm3.s)
SRH recombination Shockley-Read-Hall recombination rate (at defects)
(#/cm3.s)
Direct band-to-band excitation of electron-hole pairs by light (this
Geh (#/cm3.s) equals zero in dark)
Radiative recombination Radiative recombination rate
(#/cm3.s)
Auger recombination Auger recombination rate
(#/cm3.s)
Gross IPV electron effect: direct optical excitation from defect to
IPV Gn conduction band
Net electron generation: IPV Gn + thermal generation - thermal
net Gn recombination
Thermal electron generation (thermal emission of electron from
Gnth defect to conduction band)
Thermal electron recombination (capture by defect of conduction
Rnth band electron)
72 Chapter 6: Result analysis

Gross IPV hole effect: direct optical excitation from defect to

IPV Gp valence band
Net hole generation: IPV Gp + thermal generation - thermal
net Gp recombination
Thermal hole generation (thermal emission of hole from defect to
Gpth valence band)
Thermal hole recombination (capture by defect of valence band
Rpth hole)

Table 6.5 Overview of the properties which can be saved on the IV-panel.

property remark
v(V) voltage
jtot(mA/cm2) Total current density
jbulk(mA/cm2) Bulk recombination current density
jifr(mA/cm2) Interface recombination current density
js_n(mA/cm2) Back contact recombination current density
js_p(mA/cm2) Front contact recombination current density
j_SRH(mA/cm2) Defect recombination current density
j_Radiative(mA/cm2) Radiative recombination current density
j_Auger(mA/cm2) Auger recombination current density
Voc (V) Open circuit voltage (only under illumination)
Jsc (mA/cm2) Short circuit current density (only under illumination)
FF (%) Fill factor (only under illumination)
eta (%) Efficiency (only under illumination)
V_MPP (V) Voltage at the maximum power point (only under illumination)
Current density at the maximum power point (only under
J_MPP (mA/cm2) illumination)

Table 6.6 Overview of the properties which can be saved on the AC-panel.

property remark
x (µm)
jn.re(mA/cm2) Real part of the small signal electron current density
jn.im(mA/cm2) Imaginary part of the small signal electron current density
jp.re(mA/cm2) Real part of the small signal hole current density
jp.im(mA/cm2) Imaginary part of the small signal hole current density
jdispl.re(mA/cm2) Real part of the small signal displacement current density
jdispl.im(mA/cm2) Imaginary part of the small signal displacement current density
j.re (mA/cm2) Real part of the small signal current density
j.im (mA/cm2) Imaginary part of the small signal current density
psi.re(#kT/q) Real part of the small signal potential
psi.im (#kT/q) Imaginary part of the small signal potential
Fn.re (#kT) Real part of the small signal electron quasi-Fermi level
Fn.im (#kT) Imaginary part of the small signal electron quasi-Fermi level
Fp.re (#kT) Real part of the small signal hole quasi-Fermi level
Fp.im (#kT) Imaginary part of the small signal hole quasi-Fermi level

Table 6.7 Overview of the properties which can be saved on the CV-panel.

property remark
6.6 Saving results 73

v(V) voltage
C(nF/cm2) Capacitance
G(S/cm2) Conductance
W(µm) Depletion width
Napp(/cm3) Apparent doping density
jtot(mA/cm2) Total current density
jbulk(mA/cm2) Bulk recombination current density
jifr(mA/cm2) Interface recombination current density
jopp_n(mA/cm2) Back contact recombination current density
jopp_p(mA/cm2) Front contact recombination current density

Table 6.8 Overview of the properties which can be saved on the Cf-panel.

property remark
f(Hz) Frequency
C(nF/cm2) Capacitance
G(S/cm2) Conductance
Z.re(ohm.cm2) Real part of the impedance
Z.im(ohm.cm2) Imaginary part of the impedance
Z.magn(ohm.cm2) Magnitude of the impedance
Z.phase(°) Phase angle of the impedance

Table 6.9 Overview of the properties which can be saved on the AS-panel.

property remark
f(Hz) Frequency
C(nF/cm2) Capacitance
Et(eV) Activation energy
-w*dC/dw(nF/cm²) Scaled derivative (see [18])
Nt(cm^-3/eV) Calculated defect density

Table 6.10 Overview of the properties which can be saved on the QE-panel.

property remark
lambda(nm) Wavelength
QE(%) Quantum efficiency
photon energy (eV) Photon energy
 Chapter 7: Batch calculations

When you want to explore the influence of one or a few parameters to the solar cell characteristics, you
can take profit of the batch option. When you click ‘Batch set-up’, a panel opens where you can choose
which parameter to vary, over which range, and in which mode (Lin, Log or custom). You can also define
more than one parameter, and vary all of them (in a nested way or ‘simultaneous’). Now, up to nine batch
parameters can be defined, but be modest to start. A batch calculation is launched when ‘calculate: batch’ is
clicked. After a batch simulation all parameters on the panels are reset as they were before the calculation.

7.1 The batch set-up panel

The batch set-up panel allows you to vary up to nine different parameters, when you want to vary more you
can also vary entire definition files (§7.2) or exploit the script facilities (Chapter 10). An example of the
batch set-up panel is shown in

Figure 7.1 The batch set-up panel, illustrating most of the options

 All (most of the) parameters present in the currently defined structure can be varied. Also working point
conditions can be varied.
 When checking ‘simultaneous’, this parameter will be varied together with the parameter above (and
thus in the same number of steps). If ‘simultaneous’ is not checked, the parameters are varied in a nested
way.
 Parameters can be varied in a linear, a logarithmic or a custom defined way.
 Parameters which are files can be varied by entering a list of file names: generation files, spectrum files,
filter files, grading files, optical capture corss-section files, initial state workpoint files.
 Some parameters can only take two values (on/off), e.g. illumination.
 When changing any of the illumination parameters (ND filter, spectrum file…) as a batch parameter, the
illumination is automatically switched on.
 When changing one of the effective mass parameters, the accompanying tunnelling mechanism is
automatically switched on.
76 Chapter 7: Batch calculations

 It is possible to save/load the parameters on the batch panel in a ‘*.sbf’-file which is a standard ASCII-
file.
 It is possible to print the batch panel in order to remember your settings.
 [SCAPS 3.3.04, october 2016] It is possible to suppress (or not) the updating of the screen (drawing of all
curves in the Energy Bands panel) during batch execution: check the appropriate box in the Batch Set-up
Panel. This is speeding up the batch execution somewhat, though less than we hoped for . This option
can also be set in a script, with a set batch_display_mode command (see Section 10.4.6).

7.1.2 Custom defined values

When checking the custom list option, a Set-button appears which when clicking it opens a panel which
allows you to give a enter a list of parameter values.

Figure 7.2 Setting custom batch values

 Parameters can be typed or copy/pasted in the right list interactively. Afterwards you should press
‘Update’ to allow SCAPS to interpret your typing work. Data which can not be interpreted as a number
will be ignored.
 The values in the left list are the values SCAPS will use. These can be sorted.
 Parameter value lists can be saved and loaded. The resulting files are standard ASCII files with extension
‘*.bdf’. This allows you to make custom lists with any other program (MS Excel, Origin, Matlab…) and
load them in SCAPS. The layout of this file is rather tolerant. You just add the parameter values one
below the other. All lines which can not be interpreted as only one number are ignored as being
7.2 Varying entire definition files 77

comment. We strongly recommend the user to add some comments to the file, so that she/he remembers
what data are present in the file, even after not using it for a while.
Some parameters are files rather than numbers, e.g. a spectrum file. Then you have no choice but varying
the parameter in a custom way. Clicking the Set-button will open a panel where you can set a list of files.
Files can be added/removed/replaced. The order of the file-list can be changed. The list of files can be
saved/loaded as well. This is again a standard ASCII files with extension ‘*.bdf’. In the bdf-file you then
have to list the filenames of these files preceded by “file:”. All lines in the bdf-file which are not preceded by
this will be treated as comment. The listed files should be placed in the appropriate SCAPS folder: For
example, the spectra and the generationfiles which you list should be placed in the “spectrum” and
“generation”-file of your SCAPS-folder respectively.

7.2 Varying entire definition files

SCAPS offers the opportunity to vary entire definition-files in a batch calculation. This option is activated by
checking the ‘vary definition files’-checkbox. A list of files can then be set by clicking the ‘edit/load list of
definition files’-button. This list of files is treated in a similar way as any other parameter which has a file-
nature rather than a numeric-nature, see §7.1.2.
When definition files are varied in a batch, only those parameters which are present in ALL of these
definition files can be varied further as a batch parameter.

7.3 Varying parameters of the initial state work point

There are two ways to vary the initial workpoint variables in the batch set-up: all together, by specifying a
list of .wp2 files (Figure 7.3), or each parameter separately (Figure 7.4).

Figure 7.3 Selecting all parameters of the initial state workpoint as a batch parameters; al list of .wp2 files has to be
specified, where each .wp2 file was made by saving the initial state workpoint in a file.

Figure 7.4 Selecting a single parameter of the initial workingpoint as a batch parameter.

7.4 During calculation…

It might happen that you launched a (big) batch calculation, and all of a sudden you realize you made an
error in the set-up. Instead of waiting until all calculations are performed or instead of aborting SCAPS, you
can interrupt the batch calculation by keeping the SHIFT-button pressed. The calculation will then stop (but
only after it has finished the calculation of the current parameter value)
78 Chapter 7: Batch calculations

You might want to launch a big batch calculation overnight, but when you return next morning, you see
that the calculation stopped at step 3 due to a convergence error  which gives a pop-up window that should
be acknowledged. Inventive users have used pieces of wood to keep the ENTER-button pressed (which
solved the problem ), but there exists a better way out. On the numerical panel you can ask SCAPS to write
error messages to a log-file rather than to inform on the screen.

Figure 7.5 Error handling on the numerical panel

The batch facility is a powerful tool, but it also enables you to let things go wrong. Be aware that you are
more able to set irrealistic parameter values. For example you can set a layer width equal to zero (which will
abort SCAPS ). When varying files, take care that all files exist in the correct folders.
Every new calculation in a batch starts a the starting point, as explained in §5.1.2, this can however be
changed to speed up the calculations. Do take care however to add the voltage/frequency/resistance as the
last (bottom-most) parameter in the batch then.
 Chapter 8: Recorder calculations

In a regular single shot or batch calculation, the detailed panels are only available for the last measurement
point. To be able to see them as a function of the batch parameters you can launch a record calculation. You
should first select the properties which you want to keep track of by clicking ‘Record set-up’. By clicking
‘calculate: recorder’, a recorder calculation is launched. Cell parameters are varied according to the Batch
set-up, and all simulations (and only those) are performed which are needed to determine the asked
properties. This means the selected measurements on the action panel are ignored!
If this option is used with a little bit of imagination, you can device all kind of measurement simulations
with SCAPS.

8.1 Setting a recorder

With the Recorder facility, you can do a batch calculation and register or record selected cell properties as a
function of the batch parameter(s). For example, you can record cell efficiency as a function of some doping
density, (NA), or whatever, there are really many possibilities.
The list of properties to be recorded is made in the Record Set-up, see Figure 8.1. With the help of the five
choice-menus (type-property-layer-defect-level) the user can chose a property and add it to the list on the left
side of the panel using an Insert or Replace button. There are eight types of properties to be recorded: I-V
characteristics, General properties, Cell definition, Interfaces, Energy band panel, Generation panel,
Occupations and AC panel.

Figure 8.1 Setting up a recorder.

Pleas note carefully that it is also possible to record the cell definition properties! This allows you to see
how SCAPS interpreted the parameters you set (in the cell definition or in the batch). This might be very
helpful when dealing with graded variables. However, in recent SCAPS versions, this is easier done with the
80 Chapter 8: Recorder calculations

green Graph View button in the Cell Definition Panel, once at least one calculation has been done; an
equilibrium calculation, that is at V and in dark, is sufficient, see Section 3.5.7.
You are able to record properties which are not available in the simulations, e.g. the occupation of defect 3
when there is only one defect. Of course SCAPS can not record what is not available. However, this will not
lead to an interruption of the calculations.
Recorder settings can be loaded and saved in a similar way as batch settings, only the extension of the files
are now ‘*.srf’.

8.2 Recorder calculations

Clicking the ‘Calculate: recorder’-button leads to a batch recording, which is only available when a batch is
set. In this case the action list as it is set on the action panel will be ignored. SCAPS will determine which
calculations need to be done in order to record the asked properties. For example when a property of the AC-
panel is to be recorded SCAPS will perform a C-f simulation at the frequency determined by the working
point. When a cell characteristic is to be recorded an I-V simulation will be performed starting from 0 V
stopping at the open circuit voltage.
SCAPS doesn’t a priori know the open circuit voltage. Hence it will perform an I-V simulation from 0 V up
to a predefined maximum voltage with the ‘stop at Voc’ option on. This predefined maximum voltage can be
set on the numerical panel and has a default value of 2.00V. The accuracy of the determination of the cell
characteristics is determined by the increment voltage used in the I-V simulation. SCAPS uses the value
which is present on the action panel unless it is larger than a predefined minimum value. This value can also
be set on the numerical panel and has a default value of 0.05V.

Figure 8.2 Recorder settings on the numerical panel

8.3 Analysing the recorder results

The results of a recording can be accessed through the ‘Recorder Results’ button on the action panel (or on
the EB- or AC-panel). Recorded properties of the type IV characteristics, General properties or Interfaces
are immediately plot as a function of one of the batch parameters. Record properties of the other types are
plot as a function of the position in the cell (the mesh). The results can also be saved to a file, see Figure 8.3.
For each type of property type a different file is saved. The user can chose which files are to be made using
the checkboxes seen in Figure 8.3. The show option has the same meaning as everywhere in SCAPS. But as
not all property types can be showed in one window, only the last one (= the lowest checkbox checked on
Figure 8.3) will be showed.

Figure 8.3 Saving and showing recorder results

8.3 Analysing the recorder results 81
 Chapter 9: Curve fitting

The purpose of curve fitting is to vary one or more parameters in the cell definition to obtain a fit between
one or more measured curves and the simulation. The numerical algorithms are taken from [7]. If there is
only one parameter to fit: the golden section search ([7], p. 401). When there are 2 to 9 parameters to fit: the
downhill simplex method, or Nelder-Mead method ([7], p. 408).

9.1 General principles

The parameters to fit are the parameters set in the batch set-up, together with their range and linear or
logarithmic character. Not all batch parameters are usable as curve fitting parameters: parameters that are file
names (e.g. definition files, spectrum files, absorption files), and parameters of an on/off nature (e.g.
‘illumination on/off’, ‘tunnelling on/off’) are not usable. The curve fitter can only be set up when at least one
parameter allowable for curve fitting has been defined in the batch set-up.
The measurements to be fit should be selected in the curve fitting set-up. Currently, a maximum of 9
measurements is possible. A fixed parameter can be attached to each measurement. This can be useful when
you want e.g. to fit 4 dark I-V measurements at 4 different temperatures, or 5 illuminated I-V curves at 5
illumination intensities. A limited selection of parameters is offered by the user interface as a fixed parameter
of a measurements: only those which can be controlled in an experiment (as we think: we do e.g. not believe
that one can make different samples that differ by their electron capture constant cn of some defect…).

9.2 Setting up the curve fitter

The curve fitter can be set by clicking the ‘Curve fit set-up’-button on the action panel. This button is
however only available when a batch is set. Once the curve fitter is set, it can be launched by clicking the
‘calculate: curve fitter’-button.
As already mentioned, the curve fitting set-up takes the allowable parameters from the batch set-up,
complete with their range and linear/logarithmic nature. In the curve fitting set-up, the starting value of this
parameter should be set, see Figure 9.1: by default it is the central value in the range (taking into account the
linear or logarithmic character). You can also start from the actual value of a parameter in the present cell
definition.

Figure 9.1 Curve fitting parameters taken over from the batch set-up; setting the start value to the central value in the
range, or to the actual value in the cell definition.

Addition SCAPS 3.3.03, february 2016. The ‘simultaneous’ option in the batch settings is now also used by
the curvefitter. The first parameter of a group of ‘simultaneous’ parameters is treated as a master curvefit
84 Chapter 9: Curve fitting

parameter, and is varied independently by the curvefitter. The other simultaneous parameters are treated as
slaves of that master; each time the master is given a next value in the curvefitting process, the slave values
follow automatically. Thus, a group of one master + several slaves is treated as only one curvefit variable.
When a parameter was defined as a table list in the batch set-up, you did not have the option to set its
linear or logarithmic character (the batch calculation does not need it, it just takes the values from the list).
But the curvefitter needs to know the linear or logarithmic character. Therefore, you have the possibility to
set the lin or log property of a parameter in the curvefit set-up,when this parameter was defined as a table list
in the batch set-up; even when this curvefit parameter was a slave parameter, you can set its lin/log property
then.
The slave parameter values, that must always follow the master values, are found from interpolation. As
usuall in SCAPS, the lin/log property of a variable affects the interpolation.
To assist you with the curvefit set-up, groups of curvefit variables (thus one master and one or more
slaves) are set in an own color, see Figure 9.2. The first curvefit variable is a group of two thicknesses d1 (the
master) and d2 (the slave); the second variable is a group of 3 variables: NA3 (the master) and n3 and p3 (the
slaves). With these settings, d1 + d2 = d = 1 m for all values of master d1. This is thus a way to vary the
thickness of two layers, while the total thickness remains constant. In the second group, doping-depend
mobilities n(NA) and p(NA) are set. The values of n(NA) and p(NA) are for each value of master NA
interpolated between the table values set in the batch set-up; note that we have set here the lin/log property of
NA to ‘logarithmic’.

Figure 9.2 Example of a batch and curvefit parameter set-up. There are two independent curvefit variables: one is a
group with two thicknesses; and the other is a group with one doping density NA and two mobilities n and p.

Then the measurements should be set, see Figure 9.3. If ‘show’ is clicked, the measurement is shown in
the simulation panel (I-V, C-V, C-f or QE).

Figure 9.3 Measurement set-up in the curve fitter: the measurement type and illumination, the linear/logarithmic
character, and a possible fixed parameter; the (relative) weight of a measurement is available from SCAPS3.3.02 on.

The correspondence between a simulation and a measurement is expressed by the 2 sum (‘Chi square’),
that we define here as:

  ymeas  ycalc    log ymeas  log ycalc 

2 2

2  i (linear) or 2  i (logarithmic) (31)

  ymeas    log ymeas 
2 2

i i
9.2 Setting up the curve fitter 85

where ycalc is the simulated property and ymeas the measured property (thus: I, C or QE), and only the points i
are considered from the overlap in the simulation and measurement range of x (thus: V, f or ). In these
expressions, the simulations are linearly or logarithmically interpolated at the xmeas values. Our normalisation
of 2 ensures that the 2 of all measurements are dimensionless and are scaled to 1: a fit is good if 2  1. It
also allows that the 2 values of the individual measurements/simulations can be compared and summed.
Please note that the definition Eq. (31) of 2 deviates somewhat from the more standard definition as e.g.
given in the Recipes:

 ymeas  ycalc i2

 Recipes  
2
(the Recipes, p. 660) (32)
i i2
where i is the standard deviation of measurement point i. This definition Eq. (32) is also dimensionless, but
it does not show the ‘scaled to unity’ property as discussed above, and it can cause numerical inaccuracies or
problems if the measurement is thought to be nearly perfect or even perfect (  0 or  = 0). Also,
attributing an individual standard deviation to each measurement point, even if it were known, is laborious
for the user, and (more important ?  ) very laborious for the programr. For this reason we prefer our own
definition (31) of 2 above the mathematically more standard definition (32).

We use tot
2
as the figure of merit of the whole curve fitting action:

 wi i2 1
tot
2
 i   i2 if all wi  1 (or no weights set, SCAPS  3.3.01) (33)
 wi n i
i

where i2 are the 2 values of the individual measurements, n is the number of measurements, and wi the
relative weight that the user attributed to each measurement in the Curvefit Set-up Panel (Figure 9.3). In
SCAPS versions  3.3.01, no weight could be set, and Eq. (33) then reduces to its most right-hand member. A
user can interpret the weight wi of measurement i as a parameter for the standard deviation, if all
measurement point in a measurement had the same standard deviation (then w   ymeas
2
 2 ).

Some numerical parameters should also be set in the curve fitting set-up, see Figure 9.4.

Figure 9.4 Numerical parameters in the curve fitting set-up

The step size of a parameter variation is relative to the range of this parameter (taking into account the
linear or logarithmic nature of this parameter). Do not expect a perfect curve fitting after 25 steps only: 100
steps or so is more realistic, … but takes more time. The curve fitter walks around though the parameter
space, and stops when any of the stop criteria is met (maximum number of iterations exceeded, or step size
small enough, or total tot 2
low enough). The parameter set then obtained is only written in the cell
definition when the button at the bottom right is clicked. When you visualise the results of the curve fitting
(§9.4), you will be able to judge the quality of the curve fitting, and you will get another chance to write the
‘fitted’ parameters in the cell definition. The curve fitting settings can be saved and loaded, and are also
included in the ‘save/load all SCAPS settings’ actions.
86 Chapter 9: Curve fitting

9.3 Defining groups of simultaneous curve fitting parameters

Figure 9.5 Example of a batch set-up. The settings of d1 and d2 in parameters 2and 3 is such that always d1+ d2 = d =
2 m. Parameters 5, 6 and 7 contain lists that define a relationship n(NA) and p(NA). In this batch set-up panel, there is
no possibility to set the linear or logarithmic character of these variables NA, n, p, as this is not relevant for the batch
calculation. When passing to the curve fitting set-up, you will have to set the lin/log character of these variables.
Parameter 8 is of the on/off type, and thus will not be used as a curve fitting parameter.

The ‘simultaneous’ option in the batch settings is now also used by the curve fitter. A group of
simultaneous batch parameters can also form a group of simultaneous curve fitting parameters (remember
that not all possible batch parameters are allowed as curve fitting parameters: parameters of the on/off type
(e.g. illumination on/off), and parameters that are a list of files (e.g. spectrum files) are discarded by the
curve fitter. Figure 9.5 is an example of a batch set-up, with detailed comments in the figure caption.
The first parameter of a group of ‘simultaneous’ parameters is treated as a master curve fit parameter, and
is varied independently by the curve fitter. The other simultaneous parameters are treated as slaves of that
master; each time the master is given a next value in the curve fitting process, the slave values follow
automatically. Thus, a group of one master + several slaves is treated as only one curve fit variable.
So finally there are two types of curve fitting variables: independent ones, consisting of one variable only;
and groups of dependent ones, consisting of one master and one or more slave parameters. To assist you with
the curve fit set-up, independent variables are set in black as before, but groups of dependent curve fit
variables are set each in an own colour, see Figure 9.6.

Figure 9.6 Curve fitting parameters derived from the batch set-up of Figure 9.5. There are two groups of dependent
variables, the blue group and the pink group. Two parameters (temperature T and ND filter transmission) are
independent ones. The illumination left/right parameter is not in the curve fitting parameter list. You can set the lin/log
character of the variables that were derived from a list in the batch set-up, thus NA, n and p of layer 1.

The slave parameter values, that must always follow the master values, are found from interpolation. As
usual in SCAPS, the lin/log property of a variable affects the interpolation. When a parameter was defined as a
table list in the batch set-up, you did not have the option to set its linear or logarithmic character (the batch
calculation does not need it, it just takes the values from the list). But the curve fitter needs to know the
linear or logarithmic character. Therefore, you have the possibility to set the lin. or log. property of a
9.4 Analysing the curve fitting results 87

parameter in the curve fit set-up, when this parameter was defined as a table list in the batch set-up; even
when this curve fit parameter was a slave parameter, you can set its lin/log property then, see Figure 9.5
When the curve fitter varies d1, it simultaneously varies d2; and when it varies NA, it simultaneously varies
n(NA) and p(NA); here we have given NA a logarithmic character, and n and p a linear character, meaning
that the (NA) data are first ‘plotted’ in a  vs. log(NA) plot, and then linearly interpolated. Our example thus
illustrates a way to warrant a constant total thickness over two layers, and a way to take the proper
dependencies of some materials parameters into account.

9.4 Analysing the curve fitting results

The curve fitting results can be accessed by clicking the ‘Curvefitting results’-button on the action panel,
which opens the curve fitting results panel, that allows you to select a variety of graphs to visualise and
evaluate the entire curve fitting, see Figure 9.7.

Figure 9.7 Choices of abscissa (x-axis, top) and ordinate (y-axis, bottom) to visualise the curve fitting results.

The examples shown in Figure 9.8 to Figure 9.11 illustrate the Numos exercise 1 example. The file Numos
CIGS baseline.def first has been simplified to simplified Numos CIGS baseline.def, by taking a single
energy defect level wherever a band of defects was presents: this is speeding up the calculations
substantially. All settings, including the batch settings and the curve fitting settings, are saved in ‘simplified
Numos curvefitting.scaps’. This calculation took 24 minutes on my computer… The iteration was stopped
after 73 steps because the required accuracy ( tot
2
< 10-7) was met. Please note:
 do not expect a decent curve fit after e.g. 10 iteration steps only! In the example here, 50 – 60 steps seem
to be necessary (Figure 9.8, Figure 9.9).
 do not expect such a very nice fit in a real situation! Here, the measurements were ‘constructed’, that is,
calculated with SCAPS. Thus, an exact fit is possible. With real measurements, it can very well be that 2
never gets below 10-2 or 10-3, meaning that your measurements can not be fitted very well with the
present cell definition.
88 Chapter 9: Curve fitting

 it is also clear that you should not exaggerate with the number of parameters and the number of
measurements. Though 9 is (now) the allowed maximum for both, I would recommend to use much
more modest numbers, e.g.  3, certainly to start with.
 The recipes state that it is a good idea to start a curve fitting over again starting with the parameters
obtained from a previous curve fitting calculation. You can do so by saving the fitted parameter values in
the cell definition, and running the curve fitting calculation again starting from the ‘actual values’,
possibly with a smaller initial step, and/or with a narrower range for the parameters.

Figure 9.8 Some examples of curve fitting results: Numos exercise 1, with a simplified problem file. The evolution
of all χ2 values with iteration index. When you are satisfied with the results, you can insert the fitted parameter values in
the cell definition by clicking the grey button.

Figure 9.9 Some examples of curve fitting results: Numos exercise 1, with a simplified problem file. The evolution
of the parameters dCdS and NA, CIGS with the iteration index
9.5 Curve fitting résumé 89

Figure 9.10 Some examples of curve fitting results: Numos exercise 1, with a simplified problem file. Walking in the
parameter space to find optimum values for NA, CIGS and Nt, CIGS.

Figure 9.11 Some examples of curve fitting results: Numos exercise 1, with a simplified problem file. The
measurements (dots), the initial calculation (simplified Numos CIGS baseline.def) and the calculation after 73 curve
fitting iterations: I-V (left), C-V (middle) and QE (right).

9.5 Curve fitting résumé

1. Set up a rough SCAPS model for your cell
2. Verify that scaps is reading your measurements correctly (units, normalisation per area,…): display your
measurements and check if all is OK.
3. Refine the model, by hand. As there are probably > 100 parameters to input in your model, you can’t
leave this to the computer, you should do it yourself. Consider the scaps curve fitting facility only as a
fine-tuning.
4. Define the parameters you want to fit in the batch set-up, together with their range and
linear/logarithmic nature. Do a quick batch calculation (only very few steps per parameter) to check if
your measurement indeed could be fitted by the parameters you have selected
5. Set-up the curve fitter: the parameters were already set in the batch set-up; now only specify the starting
value (mid-range or actual value). You have already checked that the measurements are available and
readable by scaps. Now select the measurements to curve-fit, set their linear/logarithmic nature, and set
one fixed parameter for a measurement if necessary.
6. First use a modest number of iteration steps, e.g. < 10, and run the curve fitting calculations, to see if
everything is going as expected.
7. If it looks promising, set a higher number of steps, run the curve fitter, be patient and wait, and hope
that the computer will have solved your problem. Do not be too disappointed when the result is less than
you hoped for…
 Chapter 10: Scripting

SCAPS is and has been designed to be a user-interactive program. The most important computer should be
based on neurons instead of on transistors. It is important that the user understands what is physically
happening rather than performing more simulations than (s)he ever could analyse. Nevertheless, SCAPS also
provides the possibility to write a script and run it.
There are several levels of sophistication for the SCAPS script. One can use it to create a personalized
version of SCAPS, to automate actions within the user interface, to use SCAPS in symbiosis with another
program, to run SCAPS without mouse-clicks…
An overview of the SCAPS script language is given in §10.4, its use is explained in the preceding sections.

10.1 Running SCAPS externally

Normally SCAPS is started by clicking an icon on the desktop. Internally in Windows, a command line
attached to this icon is executed. This command line just contains the full path of the SCAPS .exe file (thus
scaps2902.exe or scaps3002.exe or so). Now you can add extra arguments to this command line: a list of
filenames to be loaded/executed before SCAPS starts: these can be one or more of the definition, action,…,
script files listed above; also a spectrum file and a generation file can be given. The order of execution is:
first .def, then .act, .scaps, .spe, .gen, .sbf, .srf, and finally .script. By doing so, you can ensure that SCAPS
starts in the condition that you prefer, not in the condition fixed by the SCAPS developers: good news for
those who had to set e.g. the wavelength range appropriate for CIGS each time again and again.
Here are several ways to start SCAPS from a command line, and to edit this line:
 Make, e.g. with Notepad, a batch file with extension .bat (this stems from the very old MS-DOS times,
but is still supported in Windows). Write the command line in one line, e.g.
scaps2903.exe all CIGS.scaps AM0.spe
A .bat file is run by double clicking it; you can also make a shortcut to it on the desktop. To edit the .bat
file, right-click the name and select ‘edit’, or directly open from Notepad.
 Right-click the normal SCAPS-icon on your desktop, select ‘properties’, and edit the third line (‘target’).
This is well suited when you do this once and for all. When you would alter the command line more
frequently, the previous method might feel more comfortable.
 Some programs (e.g. Matlab) provide options to run external programs. Of course you can benefit from
these options to run SCAPS or a SCAPS script within such a program environment. Example ‘*.m*’-files
(for Matlab users) which allow communication with SCAPS are available for users on request.

10.2 Running a script

A script-file is a normal text file, that contains commands that are equivalent to a mouse-click. During
execution of the script, SCAPS will not respond to user interventions (mouse or keyboard). After execution of
the script, SCAPS returns to it’s normal interactive mode, or is switched off (to be set with a set
quitscript… script command). [SCAPS 3.3.04, october 2016] Also, the updating of the screen (e.g.
92 Chapter 10: Scripting

drawing of all curves in the Energy Bands panel) can be suppressed with a set
script_display_mode… script command.
When no script file was found in the command line, SCAPS remains the same interactive program as it has
always been. A few SCAPS scripts are distributed with SCAPS to serve as an example.
There are several levels of sophistication to use scripts.

10.2.1 Automate mouse clicking

The basic use is that you write down all actions you would do in the interactive mode, and then can leave the
lab. One advantage of doing so would be that you could split up your to-do list in several smaller batch jobs
instead of one gigantic batch job (by giving several load batchsettings commands), and that you
save the results in between (by giving save results or save graphs commands). This is safer when
there is a risk that your computer (or SCAPS ) would hang up underway; also, all results are waiting for you
when you come back to the lab.
This automatisation of interactive mouse clicking can be initiated from a batch file, an external program,
but also from within SCAPS. Hereto set up a script using the ‘script set-up’-button and run it using the
‘execute script’-button.

Figure 10.1 Running a script from within the SCAPS user interface.

10.2.2 Run external programs within SCAPS

A more sophisticated use is that you start and run an own (or another) program somewhere during the
execution of the script, by giving a runsystem command. For example, you could load a problem, do just
the equilibrium calculation, and save the results of the generation panel to a file; this also contains the mesh.
You could then open an own program, read this mesh-file and use it to do an own calculation of the optical
generation, and save this in a file with the format of a SCAPS generation file. When your function returns, the
script is continued. You could then load this generation file, and do all calculations you want. This procedure
is more convenient then the full-manual method that several SCAPS users have intensely used.
In this way, the communication between scaps and your own program is only via the file system: both
SCAPS and your program read and write files, but do not communicate directly with each other. Also, reading
and especially writing files in a SCAPS format might be cumbersome for a programr.

10.2.3 Running dynamically linked libraries

For advanced users, a more direct communication method between SCAPS and an own program has been
implemented. The user program should have been compiled as a dynamically linked library, and should be
placed in the SCAPS mother directory as a .dll and a .lib file. Also, these filenames, the name of the dll-
program(up to now there are only two), and the definition (the argument list) of this function is fixed. This
functionality seems to address well skilled programmers only. However, one such dll, with a flexible
functionality, is distributed with SCAPS, and using it does not require high level programmer skills: see the
description of the rundll command below, and the examples distributed with SCAPS. In contrast, we are
confident that the basic use of SCAPS scripts can facilitate the simulation work of a broad class of SCAPS
users.

10.3 The script editor

A script-file can be written in any text editor, e.g. notepad. A typing error is however quickly made. In order
to avoid this, it is strongly recommended to use the script editor, which is launched when clicking the ‘script
set-up’-button on the action panel, see Figure 10.1, which opens the panel shown in Figure 10.2.
10.4 The scaps script language 93

Figure 10.2 Screen shot of the SCAPS script editor.

 When placing the cursor in an existing line of a script (as shown in Figure 10.2), the components of this
command line (thus: the command, the arguments and the value) are shown in the six blocks at the
bottom of the panel. When typing a new line in a script, the parts of the command line available so far
are proposed in these blocks, and can be selected and placed in the editor box of the script, either by
double-clicking on the desired argument or by pressing F3.
 Existing script files can be saved and loaded.
A tip:
When developing a script, frequently insert a show scriptvariable statement. The execution of a
script stops at such statement, and you can inspect the script variables to see if all is going as you intended.
When everything is checked and OK, you can remove or out comment these show statements.

10.4The SCAPS script language

10.4.1 General
The SCAPS-directory, this is where the scaps.exe file resides, is noted as ‘SCAPS\’.
A comment line in a script is a line that cannot be interpreted as a command line. E.g. any line starting with a
punctuation character is treated as comment. You can also add comment at the end of a command line. The
Script Editor will recognize such in-line comment when it starts with a double punctuation, except ‘]]’ (thus
e.g. ‘//’ or ‘!!’ or ‘>>’ are OK …).
All command lines in a script consist of up to three parts:
command argument value
where command and argument are reserved words, and value is free with some restrictions, depending
on the command line. The three components of the command line are separated by whitespace (spaces,
tabs,..), but should be on one line. They are not case-sensitive (upper case or lower case letters do not
matter). The possible commands are given in Table 10.1.
94 Chapter 10: Scripting

Table 10.1 Available script commands

load action set math calculate

save clear get loop run

extract show rundll

plot runsystem

Whilst processing a script, SCAPS internally maintains a few variables, as specified in Table 10.2. The user
can use these variables in set and get commands, and some are used internally in a loop. Also, these
variables are passed to an external dll function, that can be made by the user.

In this list (and in this entire manual) the notation {m} should be replaced by x, y, z, u, v, w, in order to
get script variables like xvector, wvector, uvalue, ny , nv, zname, uindex…SCAPS  3.3.09, december
2020: {m} can be any letter of the Latin alphabet (without accents), thus any letter from (a, b, c, d, e, f, g, h,
i, j, k, l, m, n, o, p, q, r, s, t, u, v, w, x, y, z). Hence, also variables like na, nm, kindex, pvalue,
qvector, bname… are accepted.

Table 10.2 SCAPS script variables

name C-type default value max value

{m}value double 0
{m}vector array of double 0
n{m} int 0
{m}name character string empty max size 256 bytes
{m}index int 0
loopcounter int 0
maxiteration int 25
looperror double 1E30
maxerror double 1E-3
status int 0
mode int 0
filename character string empty max size 256 bytes

10.4.2 Load commands

Syntax:
load argument value
Where load is the reserved command word, argument can take 8 reserved values, and value is a
filename, without path. The filename can contain spaces. The files are supposed to reside in their default
directories. There is (exceptionally) some freedom allowed in the name of the argument: just writing
definition, action, batch, record, allscaps, spectrum or generation will also
do.
10.4 The scaps script language 95

Table 10.3 SCAPS load commands

command argument value default-directory

load definitionfile a filename scaps\def
load actionlistfile a filename scaps\def
load initialworkpointfile a filename scaps\def
load batchsettingsfile a filename scaps\bdf
load recordersettingsfile a filename scaps\bdf
load allscapssettingsfile a filename scaps\def
load spectrumfile a filename scaps\spectrum
load generationfile a filename scaps\generation
load definitionfile.filenamelist listfile[index]
Here, listfile is a file that contains a list of filenames, in this
example, a list of deffiles. The index can be a number, or one of
the SCAPS script variables loopcounter, maxiteration or mode. This
format works for the load statements definitionfile to
generationfile. Example:
load generationfile.filenamelist my genfilelist.gen[loopcounter].
load singleshotbatch scaps\bdf

The last argument (load singleshotbatch) is slightly deviating from the others as it does not take a
value. The purpose of this command is to work together with the command get recorder. When load
singleshotbatch is called the batch settings file singleshotbatch.sbf is loaded. This file sets a batch
calculation with one calculation at the working point temperature. So it enables you to perform a recording
of a singleshot calculation. This option is very useful as a lot of properties can only be accessed in the script
through performing a record calculation and taking the value via get recorder. In this way you can
access e.g. the electrical field distribution in the structure and do calculations with it. The temperature in this
batch is set to the working point value when the command load singleshotbatch is called. Hence
when you vary the temperature afterwards you should repeat the command again.
With SCAPS 3.3.10, March 2021, a few more load commands are introduced. They are used to load the
temporary definition and generation files that were generated by the new command math
split_tandem, without having to know the exact names of these files (see next Chapter)
command argument value default-directory
load definitionfile.temporary.tandem_cell a filename scaps\def
load definitionfile.temporary.top_cell a filename scaps\def
load definitionfile.temporary.bottom_cell a filename scaps\def
load generationfile.temporary.tandem_cell a filename scaps\generation
load generationfile.temporary.top_cell a filename scaps\generation
load generationfile.temporary.bottom_cell a filename scaps\generation

10.4.3 Save commands

Syntax:
96 Chapter 10: Scripting

save argument value

Where save is the reserved command word, argument has a compound syntax; the first part can take 3
reserved values (settings, results or graphs). The value is a filename, without path. The
filename can contain spaces. The files are supposed to reside in their default directories.
Commands in blue are new in SCAPS version 3.3.02, june 2015

Table 10.4 SCAPS save commands

command argument value default-directory

save scriptvariables a filename scaps\results
save scriptvariables.xyzuvw a filename scaps\results
SCAPS  3.3.09, december 2020: From now on, save scriptvariables will save all 26 indices, values
and vectors. The new command save scriptvariables.xyzuvw acts as the traditional(SCAPS 
3.3.08) command save scriptvariables: only xindex,… windex; xvalue,… wvalue and
xvector…wvector are saved, in the order x, y, z, u, v and w. This is to support users that have developed e.g.
MatLab tools that import and process the traditional file exported by save scriptvariable (SCAPS 
3.3.08)
save settings.definitionfile a filename scaps\def
save settings.actionlistfile a filename scaps\def
save settings.batchsettingsfile a filename scaps\bdf
save settings.recordersettingsfile a filename scaps\bdf
save settings.allscapssettingsfile a filename scaps\def
save results.eb a filename scaps\results
save results.genrec a filename scaps\results
save results.SCAPSgenerationfile a filename scaps\generatio
n
contains only an G(x)
table, in the format of
a SCAPS generation
file (SI units for G)
save settings.definitionfile.filenamelist listfile[index]
Here, listfile is a file that contains a list of filenames, in this example, a list of deffiles. The index can be a
number, or one of the SCAPS script variables loopcounter, maxiteration or mode. This format
works for the save statements settings.definitionfile to results.SCAPSgenerationfile.
save scriptvectors.asSCAPSgenerationfile SCAPS 3.3.07 of 4-5-2018 (and
scriptvectors.asSCAPSfilterfile SCAPS 3.3.08 of 18-5-2020 for the
save
first command “…
save scriptvectors.asSCAPSabsorptionfile
as…generation…”). The save
save scriptvectors.asSCAPSoptcapt_n_file scriptvectors command takes
save scriptvectors.asSCAPSoptcapt_p_file as value AB filename or AB
listfile[…]. Here A and B are
save scriptvectors.asSCAPSfilterfile.filenamelist
scriptvectors (only use the letters x,
save scriptvectors.asSCAPSabsorptionfile.filenameli
y, …, w; SCAPS 3.3.09: all letters
st
10.4 The scaps script language 97

save scriptvectors.asSCAPSoptcapt_n_file.filenameli a… z are allowed). If saved as a

st filter, absorption or optical capture
save scriptvectors.asSCAPSoptcapt_p_file.filenameli file, vector A is assumed to contain
st wavelengths in nm. If saved as a
save scriptvectors.asSCAPSgradingfile.grading(compo grading file or as a generation file, A
sition_y) contains the position within a layer
save scriptvectors.asSCAPSgradingfile.grading(abs_p (left = 0; x (in m) if “absolute”
osition_x)
(then a line “position: absolute” is
save scriptvectors.asSCAPSgradingfile.grading(rel_p added to the file) and x/d if “relative”
osition_x_d)
(then “position relative” is added,
save scriptvectors.asSCAPSgradingfile.filenamelist.
see Section 3.5.5.1 page 15.). The
grading(composition_y)
user should have checked that the A
save scriptvectors.asSCAPSgradingfile.filenamelist.
and B contain the appropriate
grading(abs_position_x)
information, SCAPS cannot know
save scriptvectors.asSCAPSgradingfile.filenamelist.
grading(rel_position_x_d)
that. If no “.filenamelist” appears in
the argument, the information in A
and B is saved in a file with name
filename (old files with this name
are overwritten!). If the argument
contains “.filenamelist” appears, then
the last part of the value is a list file
specification, as explained above.
save results.ac a filename scaps\results
save results.iv a filename scaps\results
save results.cv a filename scaps\results
save results.cf a filename scaps\results
save results.qe a filename scaps\results
save results.recorder a filename scaps\results
save graph.eb.wholepanel always .png ! scaps\results
save graph.eb.energybands always .png ! scaps\results
save graph.eb.carrierdensities always .png ! scaps\results
save graph.eb.currents always .png ! scaps\results
save graph.eb.ftraps always .png ! scaps\results
save graph.ac.wholepanel always .png ! scaps\results
save graph.ac.currents.amplitude always .png ! scaps\results
save graph.ac.currents.phase always .png ! scaps\results
save graph.ac.potentials.amplitude always .png ! scaps\results
save graph.ac.potentials.phase always .png ! scaps\results
save graph.genrec.wholepanel always .png ! scaps\results
save graph.genrec.genrec always .png ! scaps\results
save graph.genrec.ftraps always .png ! scaps\results
save graph.iv.wholepanel always .png ! scaps\results
98 Chapter 10: Scripting

save graph.iv.iv always .png ! scaps\results

save graph.iv.recombination always .png ! scaps\results
save graph.cv.wholepanel always .png ! scaps\results
save graph.cv.cv always .png ! scaps\results
save graph.cv.gv always .png ! scaps\results
save graph.cv.Mott-Schottky always .png ! scaps\results
save graph.cv.dopingprofile always .png ! scaps\results
save graph.cf.wholepanel always .png ! scaps\results
save graph.cf.cf always .png ! scaps\results
save graph.cf.gf always .png ! scaps\results
save graph.cf.Nyquist always .png ! scaps\results
save graph.cf.G(f)/f-f always .png ! scaps\results
save graph.qe.wholepanel always .png ! scaps\results
save graph.qe.qe always .png ! scaps\results
save graph.recording.wholepanel always .png ! scaps\results
save graph.recording.resultsgraph always .png ! scaps\results
save graph.grading.wholepanel always .png ! scaps\results
save graph.grading.gradinggraph always .png ! scaps\results
SCAPS 3.3.10, March 2021: new save commands to save temporary tandem cell components, with their
appropriate (fixed) filename)
save definitionfile.temporary.tandem_cell a filename scaps\def
save definitionfile.temporary.top_cell a filename scaps\def
save definitionfile.temporary.bottom_cell a filename scaps\def
save generationfile.temporary.tandem_cell a filename scaps\generation

save generationfile.temporary.top_cell a filename scaps\generation

save generationfile.temporary.bottom_cell a filename scaps\generation

10.4.4 Action commands

Syntax:
action argument value
Where action is the reserved command word, argument can take the values in Table 10.5, and value
is a numerical value or a script variable or a filename, without path. The filename can contain spaces. The
files are supposed to reside in their default directories. Some values can take two values only (0 or 1). There
is a (very) limited degree of freedom in the exact arguments. E.g. instead of iv.checkaction, you can
also write iv.doiv or iv.iv. Instead of batch.checkaction, you can also write batch.dobatch
(as in the user interface of SCAPS < 2.10); and alike with recording.dorecord. When the value of these
commands is omitted, the value 1 is assumed (giving a clear meaning to the form doiv, docv,…,
dobatch…).
10.4 The scaps script language 99

Table 10.5 SCAPS action commands

command argument value remark

action workingpoint.temperature Kelvin
action workingpoint.kT Volt or eV
action workingpoint.voltage Volt
action workingpoint.frequency Hz
action workingpoint.numberofpoints 2
action dark none overrides light
action light none overrides dark
action illumination.fromleft none overrides fromright
action illumination.fromright none overrides fromleft
action generationfrominternalmodel none overrides generationfromfile
action spectrumfile filename scaps\spectrum

action spectrumcutoff.on none overrides spectrumcutoff.off

action spectrumcutoff.off none overrides spectrumcutoff.on
action spectrumcutoff.shortlambda Nm
action spectrumcutoff.longlambda Nm
action intensity.ND
action intensity.T %
action generationfromfile none overrides
generationfrominternalmodel
action generationfile filename scaps\generation

action generationfromfile.attenuation these 4 script commands have the

same effect, as there is only one
action generationfrommodel.attenuation
generation attenuation defined; the
action generation.attenuation attenuation is expressed as a
transmission in %
action generationattenuation

action spectrum_defined.frommodel

action spectrum_defined.fromfile

action spectrum_defined.file filename

action generation_defined.frommodel

action generation _defined.fromfile

action generation_defined.file filename

action generationmodel.constantG

action generationmodel.exponentialfromL

action generationmodel.exponentialfromR

action generationmodel.rectangular

action generationmodel.gaussian
100 Chapter 10: Scripting

action generationmodel.rect_exponentialtails

action generationmodel.rect_gaussiantails

action generationmodel.constantG.generation

action generationmodel.exponentialfromL.generation

action generationmodel.exponentialfromL.decaylength

action generationmodel.exponentialfromR.generation

action generationmodel.exponentialfromR.decaylength

action generationmodel.rectangular.generation

action generationmodel.rectangular.central_position

action generationmodel.rectangular.full_width

action generationmodel.gaussian.generation

action generationmodel.gaussian.central_position
action generationmodel.gaussian.standard_deviation
action generationmodel.rect_exponentialtails.generation

action generationmodel.rect_exponentialtails.central_position

action generationmodel.rect_exponentialtails.rectangle_full_width

action generationmodel.rect_exponentialtails.decaylength

action generationmodel.rect_gaussiantails.generation

action generationmodel.rect_gaussiantails.central_position

action generationmodel.rect_gaussiantails.rectangle_full_width

action generationmodel.rect_gaussiantails.standard_deviation

action spectrummodel.white_constant_flux

action spectrummodel.white_constant_power

action spectrummodel.blackbody

action spectrummodel.monochromatic_constant_flux

action spectrummodel.monochromatic_constant_power

action spectrummodel.white_constant_flux.total_flux

action spectrummodel.white_constant_flux.short_wavelength

action spectrummodel.white_constant_flux.long_wavelength

action spectrummodel.white_constant_power.total_power

action spectrummodel.white_constant_power.short_wavelength

action spectrummodel.white_constant_power.long_wavelength

action spectrummodel.blackbody.total_power

action spectrummodel.blackbody.temperature

action spectrummodel.blackbody.short_wavelength

action spectrummodel.blackbody.long_wavelength

action spectrummodel.monochromatic_constant_flux.total_flux
10.4 The scaps script language 101

action spectrummodel.monochromatic_constant_flux.central_wavelength

action spectrummodel.monochromatic_constant_flux.full_bandwidth

action spectrummodel.monochromatic_constant_power.total_power

action spectrummodel.monochromatic_constant_power.central_wavelength
action spectrummodel.monochromatic_constant_power.full_bandwidth

action generationmodel.nx

action spectrummodel.nwavelengths

SCAPS 3.3.05, december 2016: the spectrum and generation model commands above also work for the ‘initial workpoint’
(metastable defects), e.g. action initialwp.generationmodel.gaussian.central_position
action initialwp.temperature Kelvin
action initialwp.kT Volt or eV
action initialwp.voltage Volt
action initialwp.numberofpoints 2
action initialwp.dark none overrides light
action initialwp.light none overrides dark
action initialwp.generationfrominternalmodel none overrides generationfromfile
action initialwp.spectrumfile filename scaps\spectrum

action initialwp.spectrumcutoff.on none overrides spectrumcutoff.off

action initialwp.spectrumcutoff.off none overrides spectrumcutoff.on
action initialwp.spectrumcutoff.shortlambda nm
action initialwp.spectrumcutoff.longlambda nm
action initialwp.intensity.ND
action initialwp.intensity.T %
action initialwp.generationfromfile none overrides
generationfrominternalmodel
action initialwp.generationfile filename scaps\generation

action initialwp.generationfromfile.attenuation %
action iv.startV Volt
action iv.stopV Volt
action iv.points 2
action iv.increment Volt
action iv.doiv none equivalent to action iv.checkaction 1
action iv.checkaction 0 or 1 1 is the default
action iv.stopafterVoc none
action iv.continueafterVoc none
action cv.startV Volt
action cv.stopV Volt
action cv.points 2
102 Chapter 10: Scripting

action cv.increment Volt

action cv.docv none equivalent to action cv.checkaction 1
action cv.checkaction 0 or 1 1 is the default
action cf.startf Hz
action cf.stopf Hz
action cf.total.points 2
action cf.points.per.decade 1
action cf.totalpoints 2
action cf.pointsperdecade 1
action cf.checkaction 0 or 1 1 is the default
action cf.docf none equivalent to action cf.checkaction 1
action qe.startlambda nm
action qe.stoplambda nm
action qe.points 2
action qe.increment nm
action qe.checkaction 0 or 1 1 is the default
action qe.doqe none equivalent to action qe.checkaction 1

10.4.5 Clear commands

Syntax:
clear argument
With clear scriptvariables, all script variables (or all but 2 or 3 elements) are set to their defaults.
clear simulations is equivalent to pressing the ‘clear all simulations’ button in the SCAPS action
panel.

Table 10.6 SCAPS clear commands

command argument value remarks

clear scriptvariables.all no value see text above
clear scriptvariables.allbutfirst3 no value leaves {m}vector[i] with i = 0, 1, 2.
n{m} are set to 3. The other script
variables are not affected.
clear scriptvariables.allbutfirst2 no value idem, but with i = 0, 1
clear scriptvariables.allbutlast3 no value idem, but shifts elements i = n{m}-1,
n{m}-2, n{m}-3 to i = 0, 1, 2 and leaves
them
clear scriptvariables.allbutlast2 no value idem, but shifts elements i = nx-1, nx-2
(or with ny) to i = 0, 1 and leaves them
clear simulations no value see text above
clear plot no value clears all script graphs; identical with
plot clear
10.4 The scaps script language 103

clear scriptgraphs no value identical with plot clear or

clear plot
clear actions no value unchecks all 4 actions (IV, CV, Cf and
QE) and restores the workpoint settings
to a default (300 K, 0 V, 1 MHz, 5 pts)
clear all no value clears all simulations, all scriptvariables
and all plots: equivalent to clear
simulations plus clear
scriptvariables.all
but not clear actions

10.4.6 Set commands

Syntax:
set argument value
where set is the reserved command word, argument can take the reserved values from Table 10.8. The
set command can also be used to set the script variables. The third part of the set command line is
value: this is a numerical value, a script variable or a filename, without path. The filename can contain
spaces. The files are supposed to reside in their default directories.
Some values can take two values only (0 or 1). When the value is a numerical value, you can specify a
number, e.g. 1.25E16, or one of the internal script variables mode, loopcounter, maxiteration, {m}index,
{m}value, {m}vector and n{m}. Here {m} can be one of the letters x, y, …, w., and n{w} is the number of
elements in the corresponding {w}vector. SCAPS  3.3.09, december 2020: {m} can be any of the 26 letters
in (a, …, z).
The values of the internal variables {m}value, {m}vector, … can be set directly with a set-command;
also, they are used and possibly changed in SCAPSUserFunction.dll (see §10.4.14). The value of n{m} can
be set directly with the set command; it is also updated in some commands: get, math and clear, see
later. The allowed indices in SCAPS script vectors are listed in the Table 10.7.
When you set a new value of n{m}, the length of the corresponding vector is updated. If the new value is
smaller than the previous one, data gets lost, if it is larger, the vector is extended with non-initialised
(random) numbers. Before setting a script variable, you might want to re-initialise them with one of the
clear commands, see later.
These conventions for the use of scriptvectors in the set and get (see further) commands are
summarised in the Table below.

Table 10.7 Allowed formats for SCAPS script vector elements

script vector format index meaning; remarks

{m}vector no index Only as an argument of set scriptvariable… or
as the value of get characteristics… The
value of n{m} is incremented, all existing elements of
{m}vector are shifted one up, and the value of the set
scriptvariable… command, or the parameter to
get, is placed at {m}vector[0]
{m}vector[-1] -1 Only as an argument of set scriptvariable… or
as the value of get characteristics… The
value of n{m} is incremented, and the value of the set
scriptvariable… command, or the parameter to
get, is placed as the new last element of {m}vector
104 Chapter 10: Scripting

{m}vector[i] a number i is an integer number and should be 0  i  n{m} -1

{m}vector[last] For your comfort: internally, last is replaced with the
appropriate n{m}-1
{m}vector[loopcounter] a scriptvariable
{m}vector[mode] a scriptvariable
{m}vector[maxiteration] a scriptvariable
{m}vector[{n}index] a scriptvariable m and n can differ: you can specify e.g. zvector[yindex]
{m}vector[{n}value] a scriptvariable m and n can differ: you can specify e.g.
uvector[wvalue]. The value of {n}value is first rounded
to the nearest integer.
{m}vector[{n}vector[…]] a scriptvariable Here the index … of the inner {m}vector takes one of the
forms allowed in this Table. You can nest many vectors,
but that should not be a reason to exaggerate

Table 10.8 SCAPS set command-arguments

command argument value remark

set the script variables
set scriptvariable.maxiteration integer
set scriptvariable.status integer
set scriptvariable.mode integer
set scriptvariable.looperror
set scriptvariable.maxerror
set scriptvariable.xvalue
set scriptvariable.xvalue
set scriptvariable.{m}vector[i] 0  i  nx - 1, or i = -
1, or no index
set scriptvariable.n{m} integer
set scriptvariable.{m}name character string length < 256
set scriptvariable.filename character string length < 256
set scriptvariable.filename.filenameli listfile[…] the format of the list
st file with an [index] is
explained under the
save commands above
set scriptvariable.filename.SCAPSpath character string length < 256
set scriptvariable.filename.SCAPSpath
The filename is completed to (or changed to) the full default SCAPS path. E.g. the command
scriptvariable.filename.SCAPSpath mycell.def will set filename to (e.g.) c:\MB\SCAPS
try-outs\def\mycell.def. If no value is given, the actual filename is completed to the SCAPS default path. This
command is useful to pass a filename to another program, that might need to know the full path (e.g. the SCALSdll
function). script_display_mode… and batch_display_mode… are new in SCAPS 3.3.04 (october 2016).
general set commands
set quitscript.interactiveSCAPS no value the default
set quitscript.quitSCAPS no value
10.4 The scaps script language 105

set script_display_mode.normal no value the default

set script_display_mode.not_suppressed no value same as .normal
set script_display_mode.mostly_suppressed no value screen updating is
mostly suppressed
during script execution
set script_display_mode.fully_suppressed no value screen updating is fully
suppressed during script
execution
set batch_display_mode.normal no value the default
set batch_display_mode.not_suppressed no value same as .normal
set batch_display_mode.suppressed no value screen updating is
suppressed during
batch execution
set errorhandling.toscreen no value
set errorhandling.appendtofile no value the default during script
execution
set errorhandling.overwritefile no value
set errorhandling.outputlist.truncate no value the default
set errorhandling.outputlist.fillzeros no value
set errorhandling.outputlist.fillwhite no value
set external.Rs cm2
set external.Rsh cm2
set external.Gsh Scm-2
set internal.reflection fraction, not %
set internal.transmission fraction, not %
illumination set commands
set generation.from_scriptvectors AB scriptvector A should
contain the position x in
m, and vector B the
generation G(x) in
#/cm3/s. SCAPS store
these values in an
internal generation file
(internalSCAPS.gen),
sets the illumination
conditions to
“generation specified”,
“generation from file”,
and sets the internal gen
file a generation
filename. (SCAPS 3.3.08
of 18-5-2020). This is
equivalent with saving
(A,B) in a generation
file to be specified (with
save
scriptvectors_as
106 Chapter 10: Scripting

scapsgenerationf
ile AB filename),
setting the generation
conditions as should
and loading the
generation file just
created: but now all this
with one command.
set illumination.fromleft no value
set illumination.fromright no value
set qe.photonflux a scalar value #.cm-2s-1
set qe.photonpower a scalar value Wcm-2
set qe.constant_photonflux no value QE calculation mode
set qe.constant_photonpower no value QE calculation mode
numerical set commands (see Numerical Panel, and CV-analysis panel) (new in SCAPS 3.3.02 version 2-10-2015)
set numerical.CV-analysis.layer first layer = 1
set numerical.CV-analysis.side -1=LEFT; +1=RIGHT
set numerical.CV-analysis.points # points in local
neighbourhood
set numerical.CV-analysis.order_middle polynomial order within local
neighbourhood
set numerical.CV-analysis.order_edge polynomial order at edges of
local neighbourhood
set numerical.include_G_mesh_in_SCAPS_mesh 0 = NO; 1 =YES;
set numerical.recalculate_mesh 0 = NO; 1 =YES;
set numerical.recalculate_mesh.points maximum # of mesh points
introduced/deleted per
iteration cycle
set numerical.recalculate_mesh.min_ratio see Numerical Panel
set numerical.recalculate_mesh.max_ratio see Numerical Panel
set numerical.recalculate_mesh.generation_limit see Numerical Panel
set numerical.recalculate_mesh. see Numerical Panel
recombination_limit

set contact.Sn cm.s-1

set contact.Sp cm.s-1
set contact.opticalfilter.on no value
set contact.opticalfilter.off no value
set contact.opticalfilter.transmission no value
set contact.opticalfilter.reflection no value
set contact.opticalfilter.value fraction, not %
set contact.opticalfilter.file a filename scaps/filter
10.4 The scaps script language 107

set contact.opticalfilter 0 or 1
set contact.workfunction V or eV
set contact.flatband.off no value
set contact.flatband.once no value
set contact.flatband.always no value
layer set commands: replace layer with layer1, layer2, … layer7
set layer.remove S
set layer.duplicate
set layer.split.leftthickness m
set layer.split.rightthickness m
set layer.split.leftfraction -
set layer.split.rightfraction -
set layer.add_default a layer with default
properties is added to
the right of the cell
structure; the
layernumber in the set
layer1… command is
ignored
set layer.name character string
set layer.thickness m
set layer.Eg eV
set layer.chi V or eV
set layer.epsilon -
set layer.NC cm-3
set layer.NV cm-3
set layer.vthn cm.s-1
set layer.vthp cm.s-1
set layer.mun cm2V-1s-1
set layer.mup cm2V-1s-1
set layer.NA cm-3
set layer.ND cm-3
set layer.radiative cm3s-1
set layer.Augern cm6s-1
set layer.Augerp cm6s-1
set layer.absorption.file a filename scaps\absorption
set layer.absorption.A eV-1/2cm-1
set layer.absorption.B eV+1/2cm-1
108 Chapter 10: Scripting

From SCAPS  3.3.07, january 2018: absorption model script commands. Replace layer with layer1,
layer2,… and absorptionA with absorptionB as necessary
set layer.absorptionAfile a filename
set layer.absorptionBfile a filename
set layer.absorptionmodelA.background equivalent with
…background.on
(and alike for the next
5 commands)
set layer.absorptionmodelA.Egstep
set layer.absorptionmodelA.Egsqrt
set layer.absorptionmodelA.power1
set layer.absorptionmodelA.power2
set layer.absorptionmodelA.subbandgap
set layer.absorptionmodelA.background.on -
set layer.absorptionmodelA.background.off -
set layer.absorptionmodelA.background.alpha_background cm-1
set layer.absorptionmodelA.Egstep.on -
set layer.absorptionmodelA.Egstep.off -
set layer.absorptionmodelA.Egstep.alpha_constant cm-1
set layer.absorptionmodelA.Egsqrt.on -
set layer.absorptionmodelA.Egsqrt.off -
set layer.absorptionmodelA.Egsqrt.alpha0 cm-1
set layer.absorptionmodelA.Egsqrt.beta0 cm-1
set layer.absorptionmodelA.power1.on -
set layer.absorptionmodelA.power1.off -
set layer.absorptionmodelA.power1.alpha1 cm-1
set layer.absorptionmodelA.power1.beta1 cm-1
set layer.absorptionmodelA.power1.Eg1 eV or -multiple
set layer.absorptionmodelA.power1.exponent1 -
set layer.absorptionmodelA.power2.on -
set layer.absorptionmodelA.power2.off -
set layer.absorptionmodelA.power2.alpha2 cm-1
set layer.absorptionmodelA.power2.beta2 cm-1
set layer.absorptionmodelA.power2.Eg2 eV or -multiple
set layer.absorptionmodelA.power2.exponent2 -
set layer.absorptionmodelA.subbandgap.on -
set layer.absorptionmodelA.subbandgap.off -
set layer.absorptionmodelA.subbandgap.Ee0 eV
10.4 The scaps script language 109

In the layer defect commands below, give layer and defect their appropriate number, e.g.:
set layer3.defect2.Et 0.6
Note: as of now(7-11-2018) a defect level is not implemented in the SCAPS script: the property you set
will be applied to all levels… not as it should be for multivalent defects …
set layer.defect.remove
set layer.defect.duplicate inserts a defect after
the defect with the
given number, and with
identical properties as
this defect
set layer.defect.add_default adds a defect with
default properties after
the last defect in this
layer; the defect number
in the command is
ignored
set layer.defect.singlelevel no value
set layer.defect.uniform no value
set layer.defect.gauss no value
set layer.defect.CBtail no value
set layer.defect.VBtail no value
set layer.defect.neutral no value
set layer.defect.singledonor no value
set layer.defect.doubledonor no value
set layer.defect.singleacceptor no value
set layer.defect.doubleacceptor no value
set layer.defect.amphoteric no value
set layer.defect.aboveEV no value
set layer.defect.belowEC no value
set layer.defect.aboveEi no value
set layer.defect.Et eV
set layer.defect.Echar eV
set layer.defect.capture_cross_section.electrons cm2
set layer.defect.capture_cross_section.holes cm2
set layer.defect.Ntotal cm-3
set layer.defect.Npeak cm-3eV-1
Metastable defect set commands: replace layer with layer1, …,
set layer.metadefect.E_HE eV
set layer.metadefect.E_EE eV
set layer.metadefect.E_EC eV
set layer.metadefect.E_HC eV
110 Chapter 10: Scripting

set layer.metadefect.E_TR eV
set layer.metadefect.aboveEV no value
set layer.metadefect.belowEC no value
set layer.metadefect.aboveEi no value
set layer.metadefect.allowmeta 0 or 1
set layer.metadefect.Ntotal cm-3
interface set commands: replace interface with interface1, interface2, …
interface6
set interface.name character string
set interface.IBtunneling.off no value
set interface.IBtunneling.on no value
set interface.IBtunneling.me --
set interface.IBtunneling.mh --

interface defect set commands: replace interface with interface1,… and IFdefect with
IFdefect1, IFdefect2, IFdefect3
set interface.IFdefect.remove
set interface.IFdefect.duplicate as for bulk defects
set interface.IFdefect.add_default as for bulk defects
set interface.IFdefect.singlelevel no value
set interface.IFdefect.uniform no value
set interface.IFdefect.gauss no value
set interface.IFdefect.CBtail no value
set interface.IFdefect.VBtail no value
set interface.IFdefect.neutral no value
set interface.IFdefect.singledonor no value
set interface.IFdefect.singleacceptor no value
set interface.IFdefect.abovehighestEV no value
set interface.IFdefect.aboveEVleft no value
set interface.IFdefect.belowlowestEC no value
set interface.IFdefect.aboveEileft no value
set interface.IFdefect.aboveEiright no value
set interface.IFdefect.aboveEiGap no value
set interface.IFdefect.Et eV
set interface.IFdefect.Echar eV
set interface.IFdefect.Ntotal cm-2
set interface.IFdefect.Npeak cm-2eV-1
10.4 The scaps script language 111

set interface.IFdefect.capture_cross_section.electrons cm2

set interface.IFdefect.capture_cross_section.holes cm2
set interface.IFdefect.tunneling.on no value
set interface.IFdefect.tunneling.off no value
set interface.IFdefect.tunneling.me --
set interface.IFdefect.tunneling.mh --

10.4.7 Get commands

Syntax:
get argument variable
Here, variable is one of the internal script variables.
When you ask for a scalar property, you can use {m}value or {m}vector[index]: the actual value of the
variable will then be overwritten with the result of the get action. Here index is one of the allowed formats
for indices in the SCAPS script. Other scalar script variables that can be used are looperror and maxerror.
When you ask for a vectorial properties, like a full I-V or QE curve, these are placed in two vectors: e.g. I
in {m}vector. and V in {n}vector. If no vectors are specified, xvector and yvector are assumed: thus get cv
is identical to get cv xy. Also note that only the result of the last simulation is acquired: the last single shot
simulation, or the last simulation in a batch run.
The get command updates {m}name as well.
The purpose of the get command is that the script file, or the program launching the script file (e.g. from
within SCAPS, from MatLab, another C-program, Windows script or MS-DOS command language…) would
have access to variables such as Voc , Jsc, , … or even arrays as J(V), … in a more convenient way then
having to retrieve them from a SCAPS output file.
Also, these internal variables can be passed to and updated by the SCAPSUserFunction, that is under the
control of the SCAPS user, see §10.4.14.

Table 10.9 SCAPS get commands

command argument value and remarks

get calculated solar cell characteristics
get characteristics.eta a scalar script value:
get characteristics.voc {m}value or {m}vector[i] where the
get characteristics.jsc index i should be in the range 0  i 
get characteristics.ff n{m}-1. Using i = -1 means that the
value is appended at the end of
get characteristics.vmpp {m}vector, and that n{m} are
get characteristics.jmpp incremented with one. Using
{m}vector[0] {m}vector (thus
without index) means that the size
n{m} is incremented with one, all
elements of the vectors are shifted one
position up, and the value returned by
characteristics… is placed at
{m}vector[0].
get general calculated characteristics
112 Chapter 10: Scripting

get iv Two letters should be passed for

the value, corresponding with two
get cv vectors. The abscissa is saved in
get gv the first, the ordinate in the second.
(e.g. get cf zw, saves the
get cf
frequency in the vector zvector
get gf and the capacitance in the vector
wvector). If the value is omitted,
get qe
xy is assumed. The sizes n{m} are
get gx set automatically, and also the
names {m}name are set.
SCAPS 3.3.10, March 2021. New get commands to get properties that are a function of both position x
and wavelength : the light flux in the cell (x, ) in photons/cm2.s, the eh generation rate G(x,) in
#/cm3.s, and the optical absorption constant of the materials (x, ), in 1/cm2.
1- Example:
get lightflux.all_internalwavelengths.at_position LF 0.25
You can get these properties at a single position, either to be specified by the position x (m) (in the
example: 0.25 m) or by the mesh index i. Hint: you can specify interesting positions by first calling
commands like get.layer2.leftindex, get layer3.rightx, get celllength… Then
you get these properties as a function of wavelength , either at all internal wavelengths (this will be in
most case the wavelengths where the spectrum is defined (file or model)), or at the wavelengths you
have specified in a scriptvector (in the example, in lvector). The result (, G or ; here ) is written in
scriptvector fvector (in the example)
2. Example:
get generation.at_wavelength.all_mesh_positions XG lvector[loopcounter]
You can also get these properties as a function of position x (no user choice, the x positions of the
internal mesh are filled in the scriptvextor xvector (in this example). These properties are at one
specified wavelength (in nm). You can specify any number (e.g. 532.0) or scriptvariable like the scalar
lvalue or any script vector format (in this example: lvector[loopcounter]). The property (in this example
G) will be filled in gvector.
3. Example:
get incidentlightflux.all_internalwavelengths.just_outside_cell lf
The incident light flux will be placed in fvector, the wavelengths in lvector. When you opt for
all_internalwavelengths, these will be placed in lvector. When you opt for all_userwavelengths, you
should first have filled lvector with the  values (in nm) of your choice. When the light was set to be
incident from left, the light fluxes are at position 0, thus (0-) for just_outside_cell and (0+) for
just_inside_cell. And when the light was incident from the right, the incident fluxes are (L+) for
just_outside_cell and (L-) for just_inside_cell, when L is the total cell length or thickness. The ratio
between inside and outside is the transmission of the front contact filter:
(inside,) = Tfront()(outside, ).
4. Example:
get SCAPSinputfile.filter LT my fancy top filter.ftr
You can also access the contents of SCAPS input files without having to load or set these input files in
your cell definition: you can directly go from the filename (here my fancy top filter.ftr) to the script
vectors. in this example, the wavelengths of the file are written in lvector (in nm), and the filter property
(transmission or reflection, in %) is written in tvector. You have then the occasion to modify these
properties, e.g. T() in the script, and only after manipulation load or set them in your cell problem,
10.4 The scaps script language 113

with, in this example), e.g. by

save scriptvectors.asSCAPSfilterfile LT some filename.ftr
and then setting it as a contact filter file by
set rightcontact.opticalfilter.file some filename.ftr.
… and now the list of the new commands…
argument value
get lightflux.all_internalwavelengths.at_position position x in
the cell (m)
get lightflux.all_internalwavelengths.at_meshindex mesh index i
in the cell
get lightflux.all_userwavelengths.at_position
get lightflux.all_userwavelengths.at_meshindex
get lightflux.at_wavelength.all_mesh_positions wavelength 
(nm)
get generation.all_internalwavelengths.at_position
get generation.all_internalwavelengths.at_meshindex
get generation.all_userwavelengths.at_position
get generation.all_userwavelengths.at_meshindex
get generation.at_wavelength.all_mesh_positions
get absorption.all_internalwavelengths.at_position
get absorption.all_internalwavelengths.at_meshindex
get absorption.all_userwavelengths.at_position
get absorption.all_userwavelengths.at_meshindex
get absorption.at_wavelength.all_mesh_positions
get incidentlightflux.all_internalwavelengths.just_outside_cell

get incidentlightflux.all_internalwavelengths.just_inside_cell

get incidentlightflux.all_userwavelengths.just_outside_cell
get incidentlightflux.all_userwavelengths.just_inside_cell

get SCAPSinputfile.generation
get SCAPSinputfile.filter
get SCAPSinputfile.absorption
get SCAPSinputfile.grading
get SCAPSinputfile.optcapt_n
get SCAPSinputfile.optcapt_p
get measurement.iv In the same way, you can now
get measurement.cv also get a measurement. It
places the last calculated I-V,
get measurement.gv C-V,… measurement in the
get measurement.cf vectors specified (if none
114 Chapter 10: Scripting

get measurement.gf specified, xy is assumed)

get measurement.qe
get recombination.tot … and likewise the
get recombination.SRH recombination information on
the IV panel (at least one I-V or
get recombination.rad C-V calculation should have
get recombination.aug been done). The first vector will
contain the voltage V.
get recv.tot These commands are
equivalent to the get
get recv.SRH
recombination commands.
get recv.rad
get recv.aug
get energybands.EC Also the output of the energy
bands panel (obtained by
get energybands.EV
clicking show or save in the
get energybands.EFn EB-panel), can be accessed in
get energybands.EFp the script. The output goes to
the two vectors specified (or to
get energybands.n xvector and yvector if none
get energybands.p were specified). The first vector
get energybands.chargeindef will contain the mesh x in m.
The second vector the specified
get energybands.netdoping property, e.g. the electric field
get energybands.chargedensity E (note that get energy.E
get energybands.E and get energy.field are
equivalent).
get energybands.field
Remember that at least one
get energybands.Jn calculation (I-V, C-V, C-f or
get energybands.Jp QE) should have been done
before this information is
get energybands.Jntunnel
accessible.
get energybands.Jptunnel
get energybands.Jtot
get energybands.generation
get energybands.recombination
get energybands.cumulativeGeneration
get energybands.cumulativeRecombination
get energybands.Jtotal_scalar scalar All energy other band
output comes in the form of
two vectors, e.g. EFp(x) or
Jtot(x). In contrast, this
commands puts the value of
Jtot in a script scalar, e.g. into
xvalue, or
yvector[loopcounter],…
10.4 The scaps script language 115

get recorder You can get also the result

of a recorder-calculation.
You get the result of one
recorded property, that is on
position mode in the
recording list as you have
specified it. Here mode is the
script variable. It is somewhat
laborious to set up the
recorder list from a script
(you should set up a recorder
file, and load this in the
script), set the mode variable
to the property you want, and
then get it. But at leastyou
can access any variable in
SCAPS in thus way.

get mesh characteristics; layer should be substituted by layer1, layer2, … or layer7

get layer.leftindex the index of the leftmost point in the
specified layer
get layer.leftx the position x (in m) of the
leftmost point in the specified layer
get layer.rightindex the index of the rightmost point in
the specified layer
get layer.rightx the position x (in m) of the
rightmost point in the specified
layer
get numberoflayers the number of layers in the actual
cell definition
get celllength the total cell length x (in m) of the
cellength actual cell definition; both
celllength and cellength will work


From SCAPS 3.0.02 on (may 2011), the scalar cell parameters that are available in set are made available
in get. When your cell has graded properties, the parameters that you can set or get relate to the ‘pureA’
material (when grading is a function of composition) or to the left side of a layer (when grading is a function
of position) (see the SCAPS2.8 add-on manual on grading). The units and remarks are as for the
corresponding set commands.

Table 10.10 SCAPS get commands (cell definition related)

contact get commands: replace contact with either leftcontact or rightcontact

get contact.Sn get contact.opticalfilter.file

get contact.Sp get contact.opticalfilter

get contact.opticalfilter.on get contact.workfunction

116 Chapter 10: Scripting

get contact.opticalfilter.off get contact.flatband.off

get contact.opticalfilter.transmission get contact.flatband.once

get contact.opticalfilter.reflection get contact.flatband.always

get contact.opticalfilter.value

SCAPS 3.3.10, March 2021: you can also get the optical contact filter property T() or R() and place it
directly into scriptvectors ( in nm in the first scriptvector, T or R in % in the second). This can be at all
internal  values (the  values where the spectrum was specified), or at a specified user list of  values (then
interpolation between the internal  values at each user  is used). Examples:

get contact.opticalfilter.all_internalwavelengths LT ( in lvector, T or R in tvector)

get contact.opticalfilter.all_userwavelengths KR ( in kvector, T or R in rvector)

layer get commands: replace layer with layer1, layer2, … layer7

get layer.thickness get layer.NA

get layer.Eg get layer.ND

get layer.chi get layer.radiative

get layer.epsilon get layer.Augern

get layer.NC get layer.Augerp

get layer.NV get layer.absorption.file

get layer.vthn get layer.absorption.A

get layer.vthp get layer.absorption.B

get layer.mun get layer.composition

get layer.mup

defect get commands: replace layer with layer1, …, and defect with defect1, defect2 or
defect3
returns an integer that encodes for single, uniform,
get layer.defect.energydistribution
Gauß, …
returns an integer that encodes for neutral, single
get layer.defect.chargetype
donor, ….
returns an integer that encodes for above EV, below
get layer.defect.referencelevel
EC, above Ei
get layer.defect.Et

get layer.defect.Echar

get layer.defect.capture_cross_section.electrons
10.4 The scaps script language 117

get layer.defect.capture_cross_section.holes

get layer.defect.Ntotal

get layer.defect.Npeak

Metastable defect get commands: replace layer with layer1, …,

returns an integer that encodes for above EV, below
get layer.metadefect.referencelevel
EC, above Ei
get layer.metadefect.E_HE

get layer.metadefect.E_EE

get layer.metadefect.E_EC

get layer.metadefect.E_HC

get layer.metadefect.E_TR

get layer.metadefect.Ntotal

get layer.metadefect.allowmeta

About the Script and grading: The SCAPS script does not fully support grading, as of now (1-9-2013,version 3.2.01). The
properties that you set or get are either the property of the ‘pure A material’ (composition y = 0) if the property is not
graded, or graded as a function of composition y, or the property at the left side of the layer, if the property is graded as a
function of position x. You cannot access or set any other grading property (thus e.g. the value of the pure B material,
bowing parameters, characteristic exponential decay lengths…). However, from SCAPS 3.2.01 (2-9-2013) on you can set
a grading file for all the properties above that can be graded. The grading profile for this property is then automatically
set to ‘graded from file’. To do so, use e.g.
set layer1.Eg.file mypersonal_Eg_gradingfile.grd
or: set layer3.composition.file myfavourite_composition_gradingfile.grd.
The extension .grd is necessary, and the program assumes that the file resides in the …/grading subdirectory, unless you
specify a full file path. (And make sure that the file exists and can be found and read correctly: do a single shot
calculation before you launch complicated scripts  ).

interface get commands: replace interface with interface1, interface2, … interface6

get interface.IBtunneling.off get interface.IBtunneling.me

get interface.IBtunneling.on get interface.IBtunneling.mh

interface defect get commands: replace interface with interface1,… and IFdefect with
IFdefect1, IFdefect2, IFdefect3
returns an integer that encodes for single,
get interface.IFdefect.energydistribution
uniform, …
returns an integer that encodes for neutral, single
get interface.IFdefect.chargetype
donor, …
returns an integer that encodes for above EV left,
get interface.IFdefect.referencelevel
above highest EV, below lowest EC, …
get interface.IFdefect.Et
118 Chapter 10: Scripting

get interface.IFdefect.Echar

get interface.IFdefect.Ntotal

get interface.IFdefect.Npeak

get interface.IFdefect.capture_cross_section.electrons

get interface.IFdefect.capture_cross_section_holes

get interface.IFdefect.tunneling.on

get interface.IFdefect.tunneling.off

get interface.IFdefect.tunneling.me

get interface.IFdefect.tunneling.mh

Some action commands and some other set commands now also have their corresponding get command

get action.workingpoint.frequency get action.generationfromfile.at

tenuation

get action.workingpoint.temperature get action.initialwp.temperature

get action.workingpoint.kT get action.initialwp.kT

get action.workingpoint.voltage get action.initialwp.voltage

get action.spectrumcutoff.shortlambda get action.initialwp.spectrumcutof

f.shortlambda

get action.spectrumcutoff.longlambda get action.initialwp.spectrumcutof

f.longlambda

get action.intensity.ND get action.initialwp.intensity.ND

get action.intensity.T get action.initialwp.intensity.T

get action.Pin.from_lamp

get action.Pin.after_cutoff

get action.Pin.after_cutoff_and_ND

get action.Pin.in_cell (same as above)

get action.Jideal.in_genfile

get action.Jideal.after_attenuation

get action.Jideal.in_cell (same as above)

get action.initialwp.generationfromfile.attenuation
10.4 The scaps script language 119

get external.Rs

get external.Rsh

get external.Gsh

get internal.reflection

get internal.transmission

get numerical.qe.photonflux

get numerical.qe.photonpower

The command get recorder gets the data from the record results, and hence allows to access almost
any property in script mode. The recorded property is selected by the value of the script variable mode (the
first property in the record setting list is accessed when mode = 0, the next when mode = 1…). If the
recorded property is a scalar value as a function of the batch calculation (e.g. the open circuit voltage) the
abscissa consists of the numbers of the batch calculations. If the recorded property is a vector (e.g. the
conduction band profile) the abscissa value is the mesh. In this case only the recorded vector of the last batch
calculation is copied to the script variable. In this view, performing a batch with only one calculation using
load singleshotbatch is very recommended, see §10.4.2.

10.4.8 The extract command

SCAPS3.3.07 of 4-5-2018. There is only one extract command in the script:
extract layerx ABCD
Here x is the layer number (1 to 7 presently; the user should have checked that this layer exists). A to D
represent script vectors (use only the first letter x, y, … z). C is supposed to contain the mesh position x in
m over the whole cell, and D a cell property (obtained e.g. with a get command). The the command extracts
the information (A,B) of layer x from the whole cell information in (C,D). Also, the vectors A and B get the
same name as C and D, with the word “layer x” added to it.
To do the same, several commands were used in earlier versions. As an example, find n(x) in layer 2:
[after a calculation]
get energybands.n xy // now xvector contains x and yvector n(x) over the whole cell
get layer2.leftindex xindex // now xindex points to the leftmost point of layer2
get layer2.rightindex yindex // now yindex points to the rightmost point of layer 2
math extract ZXxy // the contents of xvector from xindex to yindex is copied (with an offset) into
zvector: zvectori = xvectori+xindex – xvectorxindex, for i=0 to (and including) i=yindex-xindex.
math extract UYxy // the contents of yvector from xindex to yindex is copied into uvector: uvectori
= yvectori+xindex, for i=0 to (and including) i=yindex-xindex.
From SCAPS 3.3.07 of 4-5-2018, this can be replaced with:
[after a calculation]
get energybands.n xy // now xvector contains x and yvector n(x) over the whole cell
extract layer2 ZUXY // now zvector contains x in layer2 only, and uvector contains n(x) in layer2
only
… thereby saving 3 commands and, more importantly, 2 integer variables (xindex and yindex).
120 Chapter 10: Scripting

10.4.9 Math commands

Syntax:
math argument value
The math commands allows to perform mathematical operations on the script vectors. The argument is
followed by a list of one to four letters form the set {a, …, z}. Uppercase or lower case do not matter;
however, for clearness in the description below, we will use upper case letters when vectors are meant, and
lower case letters otherwise.
If a variable is a vector, e.g. Y, it is interpreted as Yvector. If a variable is a scalar, e.g. z, it is interpreted
as zvalue. If a variable is an index, e.g. w, it is interpreted as windex.
Some operations are on vectors. Then operations are performed element by element and can be performed
‘in place’ (e.g. A A+B) where the original content of A is lost.

Table 10.11 SCAPS math commands

command argument value remark

A,B and C represent a vector variable
a, b, c represent a scalar variable
i represents an index variable
math add ABC Ai = Bi + Ci . (element-by-element)
Vector operation
math multiply ABC Ai = Bi * Ci . Vector operation
math subtract ABC Ai = Bi – Ci . Vector operation
math divide ABC Ai = Bi / Ci . Vector operation
math exp AB Ai = exp(Bi) . Vector operation
math log AB Ai = ln(Bi) . Vector operation
math square AB Ai = Bi2. Vector operation
math sqrt AB Ai = Bi. Vector operation
math power ABc Ai = Bi ^ c . Vector operation, A and B
are vectors, but c is a scalar
math abs AB Ai = |Bi|. Vector operation
math change_sign AB Ai = -Bi. Vector operation
math reciproque AB Ai = 1/Bi. Vector operation
math vectoradd ABC Ai = Bi + Ci . Vector operation.
Identical to add
math vectormultiply ABC Ai = Bi * Ci . Vector operation
Identical to multiply
math vectorsubtract ABC Ai = Bi – Ci . Vector operation
Identical to subtract
math vectordivide ABC Ai = Bi / Ci . Vector operation Identical
to divide
math vectorexp AB Ai = exp(Bi) . Vector operation
Identical to exp
math vectorlog AB Ai = ln(Bi) . Vector operation Identical
to log
10.4 The scaps script language 121

math vectorsquare AB Ai = Bi2. Vector operation. Identical to

square
math vectorsqrt AB Ai = Bi. Vector operation. Identical to
sqrt
math vectorpower ABc Ai = Bi ^ c . Vector operation, A and B
are vectors, but c is a scalar. Identical
to power
math vectorabs AB Ai = |Bi|. Vector operation, identical to
abs
math vectorchange_sign AB Ai = -Bi. Vector operation. Identical to
change_sign
math vectorreciproque AB Ai= 1/Bi. Vectoroperation,identicalt] to
reciproque
The vector and scalar scale and offset commands are new in SCAPS 3.3.07 of 4-5-2018
math vectorscale AB x Ai = xBi. x is a scalar number in any
format (a number, or a script(vector)
variable)
math vectoroffset AB x Ai = x + Bi. x is a scalar number in any
format (a number, or a script(vector)
variable)
math scalarscale ab x a = x * b. x is a scalar number in any
format (a number, or a script(vector)
variable). [the difference with
scalaradd abc below is that in
scalaradd, c can only be one of the
scalar script values xvalue,…, wvalue]
math scalaroffset ab x a = x + b. x is a scalar number in any
format (a number, or a script(vector)
variable). [the difference with
scalaradd abc below is that in
scalaradd, c can only be one of the
scalar script values xvalue,…, wvalue]
Examples:
To multiply all elements of xvector with 106, the traditional script was:
math fillConstant Y 1E6 nx // creates yvector with the same number of elements as xvector, all equal to
1E6: yvectori = 106, and ny = nx
vectormultiply ZYX // the result is in zvector: zvectori = yvectori * xvectori (and the size is nz = nx)
Now it is easier with one single new command (and saving a vector variable)
vectorscale ZX 1E6 // the result is in zvector: zvectori = 106 * xvectori
… where the multiplicator can take formats such as: 1E6, xvalue, yvector[0], yvector[-1], yvector[loopcounter], …
math scalaradd abc a = b + c . Scalar operation
math scalarmultiply abc a = b * c . Scalar operation
math scalarsubtract abc a = b - c . Scalar operation
math scalardivide abc a = b / c . Scalar operation
math scalarexp ab a = exp(b) . Scalar operation
math scalarlog ab a = ln(b) . Scalar operation
122 Chapter 10: Scripting

math scalarsquare ab a = a2. Scalar operation

math scalarsqrt ab a = a. Scalar operation
math scalarpower abc a = b ^ c . Scalar operation.
math scalarchange_sign ab a = -b. Scalar operation
math scalarabs ab a = |b|. Scalar operation
math scalarreciproque ab a = 1/b. Scalar operation
B
math integrate ABC
A  B    C  B dB . Vector
0
operation

dC  B 
math differentiate ABC A B  . Vector operation
dB
math interpolate aAbB When A is considered as a function of
B, thus Ai = A(Bi), it returns by
interpolation a = A(b)
SCAPS 3.3.03, february 2016. Four more interpolation functions: interpolate. linlin,. linlog, .loglin and. loglog
math interpolate.linlin aAbB Identical with math interpolate
math interpolate.linlog aAbB First “plot” data in a A vs. log(B) plot.
Then apply math interpolate
math interpolate.loglin aAbB First “plot” data as log(A) vs. B
Then apply math interpolate
math interpolate.loglog aAbB First “plot” data as log(A) vs. log(B)
Then apply math interpolate
math closestindex iaA Returns the index i for which the
element Ai is closest to a
math extract ABcd c and d are indices. The elements c up
to and including d of vector B are
copied into vector A, that now gets
dimension d-c+1; the previous
contents and dimension of A is lost.
The operation can be ‘in place’: A=B is
allowed. This function is useful to pick
out the information of one layer from
the full x-dependence, when the
indices c and d are obtained with get
layer.leftindex and get layer.rightindex
.
math increment i The index i is incremented with one.
When i is one letter from {x, y, …,
w}, the index is interpreted as xindex,
yindex,…, or windex. But you can also
use loopcounter, maxiteration, status,
mode, or one of the words xindex…
written in full.
math decrement i The index i is incremented with one. i
is a SCAPS script index, see the above
statement for the valid format.
10.4 The scaps script language 123

math min aA a = min(Ai)

math max aA a = max(Ai)

math sum aA
a   Ai
i
n A 1
1
math average aA a
nA
 Ai
i 0

math median aA a  A i  n A 2

math sumofsquares aA a   Ai2

i
n A 1
1
math averageofsquares aA a
nA
 Ai2
i 0

math product aA a   Ai
i
1
a    Ai 
nA
math geometricaverage aA
 i 
math chi_square aBC where B contains Xmeasured and C
DE contains Ymeasured; and D contains
Xcalculated and D contains Ycalculated. Then
chi_square is calculated as:

  ymeas  ycalc 
2

2  i The sum
  ymeas 
2

i
is taken at the measurement points
xmeas,i that fall within the range of the
calculated xcalc points. ycalc is linearly
interpolated between two calculated
points xcalc,j and xcalc,j+1 around the
measured point xmaes,i. Yhe 2 sum is
normalised: dimensionless, and should
ever become small compared to 1.
math chi_square_log The same as chi_square but first the
logarithm of (the absolute value of) all
ymaes and ycalc is taken.
math push ABC A = [B , C] A is a concatenation of B
and C. B is placed on top of C
math constant ABc A = c; Watch out: c is a scalar, A
gets the same length as B. B is only
used to set the length of A. AAc is
allowed as well.
math linear AB A = [1, 2 ,3…]; A gets the same length
as B. B is only used to set the length of
A. AA is allowed as well.
124 Chapter 10: Scripting

math rangeLin A The first point A[0] and the last point
A[nA-1] of the vector A are
conserved. The points in between are
scaled in a linear way between those
end points.
math rangeLog A The first point A[0] and the last point
A[nA-1] of the vector A are
conserved. The points in between are
scaled in a logarithmic way between
those end points.
math characteristics.eta aBC Vector B contains the voltage data,
vector C the current density data of a
math characteristics.Voc
calculated IV curve. Scalar a will then
math characteristics.Jsc contain  or Voc or Jsc or FF or Vmpp or
Jmpp (V and J at the maximum power
math characteristics.FF
point). If you acquired the IV curve
math characteristics.Vmpp with get iv, then you can better use
get characteristics.eta…
math characteristics.Jmpp
(the scalar that should receive e.g.  in
get.characterisctics can have the
general SCAPS script format, like
xvalue, yvector[0],
zvector[loopcounter]…).
Here, with math.characteristics, the
result can only be stored in the scalars
xvalue, yvalue,…. However, this
function is useful if the IV curve was
processed in the script (from series or
parallel connection, all kinds of
corrections made to IV curves, …)
math characteristics.all ABC Works the same as the previous
commands, but the results are all
stored in vector A. The order is fixed:
A[0] = Voc, A[1] = Jsc, and then FF, ,
Vmpp and Jmpp.
The five math commands below are special: they require a composed value. The first part of the fill… commands is
a vector: a letter from {x, y, z, u, v, w} (noted here as A), that stands for the corresponding script variable xvector,
yvector,…. The next parts of the value must be separated by whitespace (space of tab) from the first part and from
each other. They can be a number, or a SCAPS script variable. The first part of the force_in_range command is a
scalar: : a letter from {x, y, z, u, v, w} (noted here as a), that stands for the corresponding script variable xvalue,
yvalue, … The format of the series and parallel commands are explained below.
math fillConstant A constant n n is the number of
points
math fillLinear A startvalue stopvalue n n is the number of
points
fillinear.FixedNumberOfPoints

math fillLinear.FixedIncrement A startvalue stopvalue x is the increment

x
math fillLogarithmic A startvalue stopvalue n n is the number of
points
fillLogarithmic.FixedNumberOfPoints
10.4 The scaps script language 125

math fillLogarithmicPerDecade A startvalue stopvalue n n is the number of

points per decade
fillLogarithmic.PerDecade

math fillLogarithmic.FixedMultiplyer A startvalue stopvalue q q is the multiplier

math force_in_range a minvalue maxvalue the value of the scalar
a is forced in the range
[minvalue, maxvalue]
math series ABCDEF n The script vectors CD
and EF contain the IV
curves of two cells
(voltage in C and E,
current density in D
and F). The IV data of
the series connection is
placed in the script
vectors AB (voltage in
A, current density in
B); these vectors
Avector and Bvector
will contain n IV-
points.
math parallel ABCDEF n x y The script vectors CD
and EF contain the IV
curves of two cells
(voltage in C and E,
current density in D
and F). The area of
cell 1 is multiplied
with the scalar weight
factor x, the area of the
second cell with the
weight factor y (only
x+y=1really makes
sense). The IV data of
the parallel connection
is placed in the script
vectors Avector and
Bvector (voltage in A,
current density in
B);there will be n IV-
points.

Examples:
math fillinear x 1 3 5 results in xvector = [1.0,1.5, 2.0, 2.5, 3.0]
math fillinear.fixednumbeofpoints y 2 8 4 reults in yvector = [2, 4, 6, 8]
math fillinear.fixedincrement z 1 5 2.0 results in zvector = [1, 3, 5]
math fillinear.fixedincrement z 1 6 2.0 results in zvector = [1, 3, 5, 6]. Note that the last
increment is not the ‘fixed’ one (=2), but that the stop value (=6) is respected instead.
math filllogarithmic u 10 1000 5 results in uvector  [10, 31.6, 100, 316, 1000]
math fillogarithmic.perdecade v 10 2000 2 results in vvector  [10, 31.6, 100, 316, 1000,
2000]
126 Chapter 10: Scripting

math filllogarithmic.fixedmultiplyer w 1 20 2.0 results in wvector = [1, 2, 4, 8, 16, 20].

Note that the last multiplier is not the ‘fixed’ one (=2), but that the stop value (=20) is respected instead.
SCAPS 3.3.10, March 2021: Four new math-commands are introduced to support the study of tandem cells.
The details of their working principles, and application examples are given in a next Chapter.
math split_tandem_cell.only_split n1 n2

math split_tandem_cell.adapt_contact_filters n1 n2

math split_tandem_cell.adapt_generation n1 n2

math split_tandem_cell.with_electronically_inactive_parts n1 n2
The summary of their working is:
o the actual cell problem is saved as temporary tandem cell definition file.def
o this “tandem cell” is split between layers n1 and n2 into a top cell and a bottom cell (n1 +1 =
n2 is required)
o some strategy for this is applied (e.g. ‘adapt generation’).
o these single cells are saved as temporary top cell definition file.def and temporary bottom
cell definition file.def.
o when applicable, other relevant files (filter files, generation files) are saved with an
appropriate ‘temporary’ name.

10.4.10 Loop commands

Syntax:
loop argument variable
On encountering a loop start command line, the internal script variables are set to: loopcounter = 0 and
looperror = 1.0E30 (or the value of looperror that was set before).
The next script commands are executed until loop stop is met. Then, if loopcounter < maxiteration-1
and looperror > maxerror, the internal script variable loopcounter is incremented, and the script is retaken
from the preceding loop start command. Thus, when the error condition is never met, loopcounter will
successively be set to 0 … maxiteration-1, thus maxiteration values. The internal variables maxiteration
and maxerror can be set with set loop.maxiteration and set loop.maxerror at any time, see
§10.4.6.
There is no set command to set the internal script variable loopcounter. The variable loopcounter is
internally set to zero on starting a loop, and then incremented with one each times the loop is run. The
variable looperror can be set directly or be returned by the dll program SCAPSUserFunction.dll (§10.4.14),
that should be set-up by the user (one example of such dll is distributed with the SCAPS installation). Two of
the loop commands are equivalent with a set command:
E.g. loop maxiteration 20 is equivalent to set scriptvariable.maxiteration 20
E.g. loop maxerror 1E-6 is equivalent to set scriptvariable.maxerror 1E-6

Table 10.12 SCAPS loop commands

command argument value default-directory

loop start no value
loop stop no value
loop maxiteration min=5; max=100;
10.4 The scaps script language 127

default = 25.
loop maxerror min=1E-8; max=1E25;
default=1E-5

10.4.11 Show command

There is only one show command. The command line to do this is:
show scriptvariables
These are shown on the screen (see §Figure 10.3), if errorhandling.toscreen is set, or to the
standard error file, if errorhandling.appendtofile or errorhandling.overwritefile are
set. This command is very useful in debugging your script files. Also, you can stop the execution of a script
when the script variables are shown on the screen. When scripts are nested, you exit all scripts upon clicking
‘stop script execution’. You can comment out the show commands once the script is OK. The show
scriptvariables panel is also available from the action panel (the SCAPS main panel) after execution of
a script. The show command does not show all values of (very) long vectors, if the output is directed to the
standard error file. In order to access these, you should use save scriptvariables.

Figure 10.3 The script variables panel. When the panel is displayed after a show scriptvariables
command, the script is interrupted and can now be aborted or continued. SCAPS  3.3.09, december 2020: a few changes
since up to 26 index, value and vector variables could need to be shown: only the variables that were addressed in the
script (either as input or output) are shown, the others are hidden; there are left and right arrows to let you scroll if there
were more than 6 variables of a type (index, value or vector) that were addressed.
128 Chapter 10: Scripting

10.4.12 Plot commands

Syntax:
plot argument value
The plot command works in a similar way as the math command §10.4.9.

Table 10.13 SCAPS plot commands

command argument value remark

A,B and C represent a vector and should be chosen out of the set { x,y,z,u,v,w}
plot draw AB Plot A on the abscissa and B on the ordinate

plot AB Identical to plot draw

plot drawversusindex A Plot the index i on the abscissa and Ai on the ordinate

plot clear Clear the plots drawn by the script. Identical to clear
plot

Whenever a plot draw command was met, a graph is added to the ‘script plot panel’, see Figure 10.4.
This option allows you to make personalized graph of any data in SCAPS . This panel is shown immediately
after the script calculation is finished and can also be accessed from the action panel.

Figure 10.4 The script plot panel

10.4.13 Calculate commands

Syntax:
calculate argument
This is equivalent with pressing one of the “Calculate”-buttons in the interactive user interface. If no
argument is present the command gets interpreted as calculate singleshot
command argument remark
10.4 The scaps script language 129

calculate singleshot this argument can be omitted

calculate batch
calculate recorder
calculate equilibrium_only (SCAPS 3.3.10, march 2021) dark and V=0 only;
this is just enough to define the mesh and all
cell properties. After completion the action list
is restored to its original state (workpoint and
illumination settings, IV, C-V… settings)
calculate short_circuit_only (SCAPS 3.3.10, march 2021) light and V=0 only;
this is just enough to calculate the illumination
parameters (lightflux, generation, …). After
completion the action list is restored to its
original state (workpoint and illumination
settings, IV, C-V… settings)

10.4.14 The run commands

10.4.14.1 Running SCAPSUserFunction.dll or SCAPSUserFunctiondll26

These function are run by

rundll scapsuserfunction or rundll26 scapsuserfunction or

run dll scapsuserfunction or run dll26 scapsuserfunction or
run dll run dll26

Traditionally (SCAPS  3.3.08), only one user dll is recognized in SCAPS, named SCAPSUserFunction.dll.
With SCAPS  3.3.09, a second user dll is recognised, named SCAPSUserFunction26.dll.
The format of these commands allows possible later addition of more dll’s).
This dll is the method that SCAPS is using to implement two-way communication with the user. When you
do not (want to) know how to write an own program and make a dll (dynamic link library) of it, you are
restricted to use only the SCAPSUserFunction.dll and/or SCAPSUserFunction26.dll as delivered with
SCAPS, or not to use loops in a SCAPS script. The following information is for SCAPS users with programming
skills. By writing their own SCAPSUserFunction.dll, they now can realize the following (in the formulation
of an external SCAPS user):
“I would need the possibility to do a simulation, evaluate the result with an external program and let it
adjust the problem definition for the next simulation, do a simulation, and so on...”
… well, this external program should be named SCAPSUserFunction, and be present as a dll file in the
scaps/lib directory. When implemented in C or C ++, this function must comply with the function definition:
int DLLIMPORT SCAPSUserFunction (int mode, double *xvalue, double
*yvalue, double *svector[6], int sn[6], double *looperror, char
*filename);

and:
int DLLIMPORT SCAPSUserFunction26 (int mode, int sindex[26], double svalue[26], double
*svector[6], int sn[26], double *looperror, char *filename);
130 Chapter 10: Scripting

The keyword DLLIMPORT might be dependent on the development environment; here it is for LW/CVI
of National Instruments.
The meaning of the other items is:
 SCAPSUserFunction: the name of the traditional dll. The user must provide a SCAPSUserFunction.dll
and SCAPSUserFunction.lib with this name, in the scaps/lib directory. This function only accepts and
possibly changes the SCAPS script variables xvalue, yvalue, xvector, yvector, zvector, uvector, vvector
and wvector; and also mode, looperror and filename… but no indices are passed (such as xindex,
yindex), and no other values of vectors. SCAPS  3.3.09: this function still works, and is implemented as
explained below, even if there are now more variables defined
 SCAPSUserFunction26: the name of the new dll. It passes all scriptvariables: index, value and vector
with letter a-z. At the moment (17-12-2020), this function is empty, it returns without having done
anything.
 int SCAPSUserFunction(): the function should return an integer value, indicating the success of the
function evaluation. SCAPS interprets 0 as ‘success’ and a negative value as a failure. This value is
stored in the script-variable status, and shown in the error output (to screen or in the SCAPS error
logfile).
 int mode: an integer that can be used to implement several strategies in one dll function. In the example
delivered with SCAPS, mode = 1 or 2 means ‘find a root’ (e.g. find some NA such that Voc = 0.50 V), and
mode 3 or 4 means ‘find an extremum’ (e.g. find some Nt such that  is maximal).
 double *xvalue, double *yvalue: (pointers to) two scalar values, passed to the function by reference, such
that a new value of them can be returned by the function. Note with SCAPS 3.0.02: though there are now
6 scalar values xvalue, …, wvalue, only xvalue and yvalue are passed to the SCAPS dll. Also, the new
integer variables xindex, …, windex, are not passed to the dll: thus, this dll remains compatible with
earlier SCAPS versions.
 double *svector[6]: array of (pointers to) one dimensional arrays, with dimensions specified in sn[].
These vectors correspond to the vectors xvector (=0), yvector (=1), zvector (=2), uvector (=3),
vvector (=4), wvector (=5). These arrays can get new values in the function that is returned to SCAPS.
 int sn[6]: the dimension of the above vectors. These are passed by value, not by reference: their value
cannot be updated and returned by the function.
 double *looperror: a pointer to a scalar variable, that can be updated and returned by the function. In the
SCAPS script processor, ot is treated as the internal looperror variable. Returning its value by
SCAPSUserFunction.dll is the only way to change looperror in a loop. Since the script processor only
checks if |looperror| < maxerror, so you can also return a negative value here.
 char *filename: a pointer to a string variable of max. 256 characters. The SCAPS script processor will
treat it as a filename, that can be used to set e.g. a spectrum file, a generation file, a filter file,… with the
set command.
To set up your own dll, you can use other variable names; however, the type, size and order of the
variables must be exactly as specified here. Those not using C or C ++ should use variable types of the same
size (in bits) as the C types int, double, char, pointer.
The users who want to develop their own dll, or to modify the existing dll (that is easier to start with
), should ask us for the source code: SCAPSdll.c and SCAPSdll.h.

10.4.14.2 Executing system commands in a script

The command line to do this is:
runsystem systemcommand or
run system systemcommand
10.4 The scaps script language 131

where systemcommand is something that is recognized by MS-Windows as a valid command. These can
be .exe files, .bat files or WINDOWS commands. Here you can run any of your own programs (extension
.exe; the arguments on the command line can be included), or any of your batch files (extension .bat).
Examples are:
runsystem myownopticalprogram.exe inputfile1 inputfile2 outputfile
runsystem myownwindowsbatchprogram.bat
runsystem print ivresults.iv
(in the last command, it is likely that Windows will need to know the full path and not only the filename…).

10.4.14.3 Executing a script from within a script

The command line to do this is:
run script scriptfile
where scriptfile is a file containing a script. You can nest script files (that is, run a script file from
within another script file) as you like, but that should not be a reason for exaggeration. All the script
variables are transferred from one script to the other, with the exception of some loop-variables
loopcounter, looperror, maxerror, maxiterations, which are local to each script file.
132 Chapter 11: Script support for simulating multi-junction (tandem) cells

Chapter 11: Script support for simulating

multi-junction (tandem) cells

11.1 Introduction
The SCAPS application discussed in this document makes use of:
 SCAPS version 3.3.10 of March 2021, or more recent.
 The following files:
Definition files that must be present in the folder [your SCAPS]\def:
 any tandem cell (thus, a pnpn sequence, or an npnp sequence. Examples of these, delivered
with SCAPS are:
 test tandem cell scripts illum from left.def
 test tandem cell scripts illum from right.def
o Script files that must be present in folder [your SCAPS]\script:
 test tandem split generation.script
 test 2 adapt filters varying Rb.script
 test tandem split fake top and bottom.script

11.2 Problems with the direct simulation of tandem cells in SCAPS

SCAPS was originally intended to study single cells of CIGS or CdTe type. Also, the original program was
conceived to work well for a pn structure, where the leftmost layer is p-type, and the rightmost layer is n-
type, and there is only one p-n junction (however, several p layers and several n-layers could be present).
Indeed SCAPS runs well for such structures, even with parameters that are very different from CdTe or
CIGS: it has been successfully applied to CZTS, perovskite, c-Si, a-Si, bulk heterojunction, … solar cell
structures. Unfortunately, SCAPS often fails to converge when either more then one pn junction is present
(e.g. a pnp or pnpn structure), or when there are too strong reverse junctions at a contact. The convergence
could be improved (a little) by setting tunnel mechanisms on, or by playing with convergence settings (e.g.
more iterations; adaptive mesh,...) ... but often these tricks do not remedy the basic convergence problem.
This SCAPS problem is well known, and persisting since the very birth of SCAPS. However, it never was on
our priority list to work on tandem cells, as we wanted to enhance SCAPS with more physics of single thin
film cells, and we never worked on tandems, let be (real) multiple junctions. Recently, some work was done
on the convergence problem, but so far without any tangible success.
In recent years, SCAPS users are increasingly interested in simulation of multi-junction cells, mainly
tandem cells based on perovskite materials. To serve these users, even when direct simulation of tandem
cells is most often not possible is SCAPS, we developed script support to split a tandem structure into partial
cells (thus, a top cell and a bottom cell), to study the partial cell under ‘tandem illumination conditions’, and
to assemble the tandem cell again, this is, to place the two partial cells electrically in series. This chapter can
be read as an application note on this topic.

11.3 Basic strategy: calculate top and bottom cell separately, place them in series afterwards
For the time being (and perhaps a little bit longer), those wanting to study tandem cells should remedy
themselves with tricks: you can calculate the top and bottom cell separately, and series connect them yourself
in an external program or with a SCAPS script. Of course, this is by far not as user friendly as a direct
calculation within SCAPS...
11.4 Splitting the tandem cell into top and bottom cells 133

In the study of the partial cells (thus, top and bottom) the user should take care that these partial cells are
studied under the illumination conditions they have in a tandem cell configuration. There is now script
support for:
 splitting a tandem cell configuration into a top and bottom cell
 three strategies to study these partial cells under “tandem-light conditions”
 making the electrical series connection of these partial cells to obtain the IV characteristics of a 2-
terminal tandem cell.

11.4 Splitting the tandem cell into top and bottom cells
Splitting a tandem cell configuration manually into a top cell and a bottom cell is easy but a little bit
laborious, and maybe boring should you have to do it regularly. Therefore, this action is now automated with
the script command:
math split_tandem_cell.only_split n1 n2
This script command first saves the actual cell problem in a file named temporary tandem cell definition
file.def. Then, this “tandem cell” is split between layers n1 and n2 into a top cell and a bottom cell (with the
requirement n1 +1 = n2). SCAPS takes account of the side where the light is incident. Thus, when light is
incident from left, the top cell will contain the original layers 1 to n1, and the bottom cell the original layers
n2(=n1+1) to Nlayers. And similarly for light incident from the right. It might be redundant to ask the user to
specify two values n1 and n2, but it forces her/him to think well about the tandem cell configuration. When
the parameters n1 and n2 are omitted (or when only one is specified), SCAPS will make an own guess of where
to split the tandem, and notify you about this with a warning. The tandem split strategy of SCAPS is simple: it
starts looking to the doping type of the leftmost layer, and goes on to the right. When a second type change
(np or pn) is met, the split position is placed there. This algorithm will be fooled when truly intrinsic
layers are present (thus, where the shallow ND = NA, exactly), or when the doping is modelled by charged,
not-too-deep defects instead of by ideally shallow donors/acceptor. Therefore, it is better that you express
your ideas on the split position (between layers n1 and n2), than to leave it to SCAPS.
The command … .only_split just does the splitting, and nothing more. The other tandem split
commands (see below) also implement a strategy about the illumination/generation in the partial cells. In all
cases, the partial cell definitions are saved in your [SCAPS]\def directory, with names temporary top cell
definition file.def and temporary bottom cell definition file.def. You can use these def files to express your
own ideas/strategies on the illumination/generation conditions. To load these files, you can use e.g.
load definitionfile temporary top cell definition file.def
but then you will have to remember (or copy) the exact filename of the partial cells. Instead, you can use:
load definitionfile.temporary.tandem_cell
load definitionfile.temporary.top_cell
Since the script editor is proposing the correct form of these commands, the risk of mistyping is minimised.

11.5 Strategy 1: ‘adapt generation’

When using
math split_tandem_cell.adapt_generation n1 n2
SCAPS is doing the same as with …only_split, but before splitting the tandem cell, a generation profile
for the tandem cell and for each of the sub-cells, the top cell and the bottom cell, is calculated. First, the eh
generation profile Gtandem(x) in the tandem cell is determined, and saved as file temporary tandem cell
G(x).gen in the [SCAPS]\gen directory. Then, this generation profile Gtandem(x) is split between a generation
profile Gtop(x) for the top cell and one for the bottom cell Gbottom(x). For example, if xt is the position of the
top/bottom split (thus the position of the right side of layer n1 = the left side of layer n2, when n1 and n2 are
134 Chapter 11: Script support for simulating multi-junction (tandem) cells

the value parameters of the math split_tandem script command), and when light is incident from the
left, then
Gtop  x   Gtandem  x  0  x  xt
(34)
Gbottom  x   Gtandem  x  xt  0  x  d tandem  xt

where dtandem is the total thickness of the tandem cell. These generation profiles are saved as files temporary
top cell G(x).gen and temporary top cell G(x).gen in the [SCAPS]\gen directory. These generation files are
however not stored in the definition files for the top and bottom cells, as SCAPS definition files do not store
information on illumination conditions.
To use the definition and generation files of a partial cell (top or bottom), one thus should load the definition
file and the generation file of that partial cell, and set the illumination condition to ‘Directly specify G(x)’ in
the SCAPS action panel. In a script, this is automated as shown in the example below, for the top cell:
load definitionfile test tandem cell scripts illum from left.def
// we want the generation calculations to be done based on an incident spectrum
action lightflux_specified
math split_tandem_cell.adapt_generation 2 3 // the tandem split is between layers 2 and 3
action light
// to use the generation profiles constructed by split_tandem, the generation should be directly
specified from file
action generation_specified
load definitionfile.temporary.top_cell
load generationfile.temporary.top_cell
(note that there is a script command to load a generation file for tandem, top or bottom cell, as there was such
a script command for loading def files).
The generation conditions for the partial cells will be exact within the SCAPS optical model. They are only
valid for the spectrum and intensity that was set in the Action Panel, and for the optical properties of the
contacts (R() or T() filters), and for the internal reflection at the front side, and for the optical absorption
(x, ) and thickness that was set in each layer. However, this advantage of exactness has its price: should
the user, while studying or optimising a partial cell, change any of the optical relevant parameters (a
thickness or band gap of a layer is enough), the generation profile is no longer exact, and the above
procedure should be repeated: start from the tandem, set the generations Gtandem(x), Gtop(x) and Gbottom(x), and
split the tandem.
Some users have developed much more elaborated optical models for the calculation of G(x) profiles then
the SCAPS optical model, that is rather coarse. Their optical calculation goes through own or dedicated
programs, and a generation profile file calculated in this way can be passed to a SCAPS script manually or
automated e.g. via MatLab. This is definitely a much more sophisticated and (optically) exact way to do
tandem simulations – but it is more laborious…

11.6 Strategy 2: ‘adapt the optical filters at the contacts’

When using
math split_tandem_cell.adapt_ contact_filters n1 n2
SCAPS is doing the same as with …only_split, but before splitting the tandem cell, new contact filter
files are defined, containing reflection R() or transmission T(), for each of the sub-cells, the top cell and
the bottom cell. These adapted filter files are then included in the definition files for the top cell and bottom
cell.
11.6 Strategy 2: ‘adapt the optical filters at the contacts’ 135

The internal optical model of SCAPS was discussed in Section 4.3.1 and resulted in Eq. (15), that we repeat
here in Eq. (35) for reading comfort.
1  Rback    exp  2  d  x      
N phot  , x   N phot0     Tfront     exp   x      (35)
1  Rback    Rint exp  2d     

This equation is for a light flux Nphot0() incident from the left, on a cell with a uniform, but wavelength
dependent optical absorption constant (x, ) = (). Tfront() is the wavelength dependent transmission at
the front contact (here: left), and Rback() is the wavelength dependent reflection at the back contact (here:
right). Rint is the internal reflection at the front contact (here: left); no wavelength dependence for Rint is
implemented in SCAPS. Of course, SCAPS takes account of different () functions in the different layers (set
by an absorption file or an absorption model), and correctly treats grading, thus the possibility that within a
layer,  is also position dependent: (x, ). This extension is rather trivial: simple exponential forms are
replaced with a form with an integral in the exponent:
 x 
exp       x   exp      x,   dx
 0 
(36)
 d 
exp        d  x   exp      x,   dx
 x 

In this discussion, we will go on with the simple formulation of Eq. (35), in order not to obscure the
argumentation by complicated equations. The treatment below will only be exact when there is no internal
reflection in the cell, thus Rint = 0. Then Eq. (35) reduces to Eq. (37) below
N phot  , x   N phot0     Tfront     exp   x      1  Rback    exp  2  d  x      (37)

This Eq. (37) holds for all x in the tandem cell, thus 0  x  d. Assume that the tandem split is at position xt,
thus that the top cell is ranged in 0  x  xt, and the bottom cell is ranged in xt  x  d, see Figure 11.1.
The light flux in the top cell can now be written as (with d = xt + d’, d’ is the thickness of the bottom cell):
N phot  , x   N phot0     Tfront     exp   x      1  Rback    exp  2d   exp  2  xt  x   
    exp  2  xt  x      
N phot  , x   N phot0     Tfront     exp   x      1  Rback (38)

     Rback    exp  2d     

Rback

Thus, we can treat the top cell on its own (thus without bottom cell attached to it), when we keep Tfront as is,
and replace Rback() with R’back() as given in the third equation (38).
In the same way, we can write the light flux in the bottom cell (with x’= x - xt ) as:
xt  x  d or 0  x  d 
N phot  , x   N phot0     Tfront     exp   x  exp   xt    1  Rback    exp  2  d   x  
(39)
     exp   x   1  Rback    exp  2  d   x  
N phot  , x   N phot0     Tfront
     Tfront    exp   xt     
Tfront

Thus, we can treat the bottom cell on its own (thus without top cell attached to it), when we keep Rback as is,
and replace Tfront with T’front as given in the fourth equation (39).
Note that this strategy fails when Rint  0. And of course, SCAPS will not use the simple exponential forms
of Eqs. (38) and (39),but the more elaborate forms of Eq. (36). The new back contact reflection R’back() of
136 Chapter 11: Script support for simulating multi-junction (tandem) cells

the top cell alone is saved in [SCAPS]\filter as file temporary top cell back reflection.ftr and included in
temporary top cell definition file.def; and similarly for the new front transmission T’front of the bottom cell.

Tfront()
Rback() tandem cell

0 xt d x

Tfront()
R’back() top cell

0 d - xt x’

T ’front()
Rback() bottom cell

Figure 11.1 Splitting a tandem cell into a top cell and a bottom cell. The top cell has the original Tfront(), but an
adapted R’back(). The bottom cell has an adapted T ’front(), but the original Rback().

These optical conditions for the partial cells (thus the filters T’front and R’back()) will be exact within the
SCAPS optical model. They will remain valid when the settings of the spectrum and intensity in the Action
Panel are changed (new spectrum model parameters, new spectrum file, new settings for neutral density filter
and spectrum cut-off filters), and this is an advantage over the previous strategy (‘adapt generation’). It could
be a good idea to set the spectrum to a simple, smooth model (e.g. monochromatic with constant photon flux
at all wavelengths) when running the script command split_tandem.adapt_contact_filters
(then you get rid of the peaks and dips that are present in realistic spectrum files – due to our good old sun,
and especially to our good old earth atmosphere). Afterwards, set the spectrum conditions to whatever you
want to study. However the calculated filters T’front and R’back() will have to recalculated when the user is
changing optical properties of the contacts (R() or T() filters), or the optical absorption (x, ) and
thickness that was set in each layer. Also, the method cannot be applied when Rint > 0; however, when the
layers reasonably absorb the light (optical  and thickness sufficiently high), the influence of Rint will be
negligible anyway.

11.7 Strategy 3: ‘replace parts of the cell with electronically inactive layers’
This is the strategy that was proposed in a SCAPS FAQ or application note “Assembling a tandem cell from
two single cells in SCAPS”, february 2016. It is now automatised and implemented in the script command
math split_tandem_cell. with_electronically_inactive_parts n1 n2
Now SCAPS does not really ‘split’ the tandem cell, but is setting up a top cell and a bottom cell that both
consist of all layers of the tandem cell.
11.7 Strategy 3: ‘replace parts of the cell with electronically inactive layers’ 137

In setting up the top cell, the layers that belong to the top cell are taken as they are defined. However, the
electronic properties of the bottom cell layers are changed so that they remain optically active, but are made
‘electronically inactive’. Assume that light is incident from the left, so that the top cell is the left part, and the
bottom cell the right part of the tandem. Also assume that the rightmost layer of the top cell (thus the layer
adjacent to the bottom cell) has an effective n-type doping. To render the bottom cell ‘electronically
inactive’, some properties are changed, of all layers of the bottom cell:
1. the layers are made (relatively) heavily doped n-type: the script sets ND = 1019 cm-3 and NA = 101 cm-3.
2. discontinuities in the conduction band edge EC are removed, by setting the electron affinity  of all
bottom cell layers equal to the  value of the rightmost edge of the rightmost layer of the top cell.
3. the contact velocities Sn and Sp of the right contact are set to Sn = Sp = 107 cm/s; and the flat-band feature
of this contact is clicked on.
4. the layers are given a very high recombination:
 all defects are removed
 a new defect 1 is set up, with very high efficiency for recombination
 this is a neutral defect, single level at mid gap
 with n = p = 10-12 cm2, and Nt = 1017 cm-3.
Action (1) ensures that the ‘electronically inactive’ bottom cell does not add an extra pn junction to the top
cell being set-up. Together with actions (2) and (3), it ensures that convergence will (almost) always be
reached: the bottom cell acts more or less as a thick ohmic contact to the top cell. At the same time, nothing
has changed to the optical behaviour of the bottom all: all layers have their (x, ) as set in the tandem
definition, and also Rback() at the back contact remains unchanged. Action (4) ensures that (almost) no
electrons or holes of the eh pair generation in the bottom cell will contribute to the current: they all will
recombine before reaching the top cell or the right contact. Of course this strategy is adapted if the right most
layer of the top cell is p-type, and/or if light is incident from the right side. The partial cells are saved
automatically as temporary top cell definition file.def and temporary bottom
cell definition file.def, as before.
[in our application note of 2016, we called this strategy ‘top cell with fake bottom cell’, ‘bottom cell with
fake top cell’; since the word ‘fake’ has got an utmost unscientific connotation since that time, we now use
the terminology: ‘top cell with optically unchanged but electronically inactive bottom layer’, …].
This strategy is illustrated in Figure 11.2 where the energy band diagrams of the complete tandem cell, the
top-cell-with-electronically-inactive-bottom-cell, and the bottom-cell-with-electronically-inactive-top-cell
are shown, all in equilibrium, this is in dark and at V = 0.
It is clear that this strategy is not an exact treatment of the tandem cell, but it can be a good and very
convenient approximation. The advantages are clear: the top and bottom cell can be studied in arbitrary light
conditions (‘from illumination’ and ‘from generation’; all settings of intensity and spectrum cut-off). Even
more, one can study a sub-cell (top or bottom) with variation of optically important parameters (thickness d,
band gap Eg, optical absorption () of a layer; contact filters Tfront and Rback()) without having to set-up the
top and bottom cells each time again (by invoking math split_tandem_cell…). The drawback is that
it is not so easy to assess the validity of the basic approximation (‘electronically inactive counterparts’). One
will have to check that the result (the calculated IV and efficiency parameters) of the tandem cell are not very
sensitive to the assumptions in this script command (the precise value of the high ND and low NA values in
the above example, the value of the contact Sn and Sp that were used, and especially the assumptions on the
high recombination effectiveness of the ‘inactive counter parts’, especially the value of Nt).
It might make sense to do the vast amount of the exploring work with this strategy of inactive
counterparts, and now and then (and also finally) check with the more rigorous, but less versatile strategies
of adapt generation or adapt filters.
138 Chapter 11: Script support for simulating multi-junction (tandem) cells

Figure 11.2 Top: Equilibrium energy band diagram of a tandem cell, illuminated from the left; the top cell (left part)
is a wide band gap np junction, and the bottom cell is a narrow band gap np junction. Bottom Left: band diagram of the
top-cell-with-electronically-inactive-bottom-cell; the bottom cell behaves electronically as one single p+ layer, but
remains optically active as it was in the tandem. Bottom Right: band diagram of the bottom-cell-with-electronically-
inactive-top-cell; the top cell behaves electronically as one single n+ layer, but remains optically active as it was in the
tandem.

11.8 Connecting the top and bottom cells in series to a 2-terminal tandem cell
We illustrate the script for the tandem cell configuration test tandem cell scripts illum
from left.def, and we work with the strategy ‘adapt filters’. The script could start with:
load definitionfile test tandem cell scripts illum from left.def
// set the illumination conditions, and the parameters of the IV simulation
// (Vstart and Vstop, # points, possibly ‘stop after Voc’…)
// split the tandem, set-up the top and bottom cell
math split_tandem_cell.adapt_contact_filters 2 3
action light // if not already done
// calculate IV of the top cell; store V in vvector and I in tvector (t of “top”)
load definitionfile.temporary.top_cell
calculate singleshot
get iv vt
plot draw vt // IV of the top cell
// calculate IV of the bottom cell; store V in uvector and I in bvector (b of “bottom”)
load definitionfile.temporary.top_cell
calculate singleshot
get iv ub
plot draw ub // IV of the bottom cell
// do the series (tandem) connection
// the first IV curve (top cell) is in (vvector, tvector); the second (bottom cell) in (uvector, bvector)
// the tandem IV result will be placed in wvector (the voltage) and ivector (the current density)
11.9 Accessing the internal optical properties of a cell 139

// the tandem IV curve will contain 100 points

math series WIVTUB 100
plot draw wi // the (constructed) IV curve of the tandem cell
To do the series connection, the common current density range of the top and bottom IV curves is
determined. (In this example) 100 equidistant current density points are taken in this common I-range. For
each such current density, the corresponding voltages Vtop and Vbottom of the IV curves of top and bottom are
determined by interpolation. The voltage Vtandem = Vtop + Vbottom is then attributed to this current density of the
tandem cell. A few remarks are in order:
 when you want that the short circuit point of the tandem IV is included, thus Vtandem = 0, then in general
one voltage of (Vtop, Vbottom) should be positive, the other negative. It is thus advisable to set the Vstart
value of the IV calculation of top and bottom cells to a negative value, -0.5 V or so. Maybe some testing
is required to find a suitable value for each specific problem.
 when calling e.g. get iv vi, the script gives suitable names to the script vectors, in this case “V (V)”
for vvector and “J (mA/cm2)” for ivector. But now you are simulating three sets of IV curves (top,
bottom and tandem). Thus it is advisable to set suitable names explicitly by scripts commands as set
scriptvariable.iname J(top cell), mA/cm2: this will be of great help to interpret the
results (graphs, script variable tables, output tables, …)… and is definitely worth the effort of inserting
many such commands.
 you can get the efficiency parameters (, Voc, Jsc, FF, Vmpp, Jmpp) of the last calculated IV curve with e.g.
get characteristics.eta avector[0]
get characteristics.voc cvector[7]
get characteristics.jsc dvector[loopcounter]
The result (here , Voc and Jsc) can be placed in any scalar scriptvariable (e.g. hvalue, xvalue,…) and in
any element of a script vector, as done in the examples above. Refer to Table 10.7 for allowed formats.
 however, you cannot use these get characteristics commands to obtain the efficiency parameters of the
tandem cell: the IV curve of a tandem cell is a constructed IV curve (here with math series), and not
a directly simulated IV curve. Instead, you should use e.g.:
math characteristics.eta eWI // the -value of the IV curve I(W) is calculated in the script
scalar evalue
math characteristics.voc oWI // the Voc-value of the IV curve I(W) is calculated in the script
scalar ovalue
To place the result in a script vector, instead of a script scalar, you can subsequently use a command like
set scriptvariable.evector[loopcounter] evalue // evalue is placed in element
loopcounter of evector
(once there should come a single command to replace the two above).

11.9 Accessing the internal optical properties of a cell

Some optical properties of the cell are a function of both position x and wavelength : the light flux Nphot(,
x), the eh pair generation rate G(, x), the optical absorption constant (, x). Others are a function of
wavelength  alone: the optical transmission T() or reflection R() of the contact filters, the incident light
flux just outside and just inside the cell, given by (when light is e.g. incident from the left side)
N phot ,just_outside     N phot 0   
(40)
N phot ,just_inside     N phot 0     Tfront     N phot  ,0 
140 Chapter 11: Script support for simulating multi-junction (tandem) cells

And the total generation Gtot(x), that is the generation G(,x) integrated over all wavelengths, is a function of
x alone.
All -dependencies are evaluated internally in SCAPS at the wavelengths where the spectrum is defined (from
a file, or from a model). This also applies to the optical absorption () of a layer, and to the contact filter
properties T() or R() when these are taken from an input file: internally, SCAPS does not calculate at the -
values of for example an () file, but first interpolates these  values at the -values of the spectrum. Thus,
the internal  values meant here are those of the spectrum, but not those of the () … input files. All x-
dependencies are evaluated internally at the mesh points.
Since SCAPS 3.3.10, march 2021, there is script support to get all these properties in script vectors, for
displaying, plotting, manipulation…
One can list these properties as a function of wavelength , either by using all internal wavelengths, or at
all wavelengths listed in a script vector (in nm) (then interpolation is used). When these properties are also a
function of x, one should specify a single x value, either by specifying the mesh index, or by specifying the x-
value directly, in m. Tip: the position or index of the edges of a layer can be accessed by get commands like
get layer3.leftindex aindex or get layer2.rightx bvalue…
One can also list these properties as a function of position x, only by using all internal mesh values of x.
When these properties are also a function of , one should specify a single  value in nm.
Examples of such get commands are thus:
get incidentlightflux.all_internalwavelengths.just_outside_cell lo // the
result is in lvector () and ovector (Nphot, just_outside)
get lightflux.all_internalwavelengths.at_position la 2.50 // the internal  are set
in lvector, the flux Nphot in avector; this flux is at x = 2.5 m
get contactright.opticalfilter.all_userwavelengths lt // the user should have filled
lvector with the desired wavelengths ; the filter value T() or R() is set in tvector (this depends on the filter
mode, that could have been set (definition file, user interface or script) to transmission or reflection)
get absorption.at_wavelength.all_mesh_positions xa lvalue // the mesh x-values
are set in xvector, the (x) values in avector; these  are at the wavelength set in lvalue (in nm)
With these new get commands, the user has more insight and control over the optical conditions in a cell.

11.10 Application examples

11.10.1 Introduction
We present here a few application examples to illustrate the new SCAPS facilities described in the above
sections. The problem we treat is defined in test tandem cell scripts illum from
left.def. It consists of a (relatively) wide band np cell with Eg = 1.8 eV and cell thickness 2.5 m, and a
(relatively) narrow band np cell with Eg =1.1eV and cell thickness 3.5 m. Some cell properties are
summarised in Table 11.1. All layers have the optical absorption model ‘power law’ with exponent n = 1.0,
with 1 = 105/cm, 1 = 0 and with the band gap value Eg of the layer, and no sub-bandgap absorption.
In most cases of multi-junction cells, and also with the .def file under study here, any non-equilibrium
calculation (this is, V  0, or illumination, or both) will lead to convergence failure. Nevertheless, the light
conditions in the cell, Nphot(x, ) and G(x, ) have been calculated correctly, and can be used, before the non-
convergence message pops up. In a script, you can direct these annoying error messages to an error log file
(script command set errorhandling.appendtofile or .overwritefile). You can also
avoid these convergence failures at all by first simplifying the electronic properties of the whole tandem cell
as discussed in section 11.7 above (the optical results remain unaffected, but of course any electrical result is
useless).
11.10 Application examples 141

Table 11.1 Some parameters of test tandem cell scripts illum from left.def.

top cell bottom cell

emitter base emitter base
Eg eV 1.8 1.8 1.1 1.1
g nm 689 689 1127 1127
d m 0.5 2.0 0.5 3.0
-3 17 17
ND cm 10 10
-3 15
NA cm 10 1015

Figure 11.3 Left: light flux Nphot(x; ) (in #/cm2s) and Right: generation G(x, ) (in #/cm3s) as a function of x (in m),
and for 18 -values from 300 nm (blue curve bottom left) to 1150 nm (green curve on top in the Nphot graph left, but
absent in the G graph right), step 50 nm. The tandem split (this is, between the top cell and the bottom cell) is at x = 2.5
m. The light of the curves 300 nm    600 nm is absorbed almost completely in the top cell (0 x 2.5 m) and
quickly dies out in the bottom cell (2.5 m x 6.0 m). The curves 650 nm    1100 nm shows that light passes
without any absorption through the top cell, and is absorbed in the bottom cell, however less and less complete at longer
wavelengths. The curve for 1150 nm shows that the light passes without absorption through the complete tandem cell,
and does not cause any eh generation at all.

11.10.2 The illumination, absorption and generation conditions

Figure 11.3 illustrates the result of the script commands:
get lightflux.at_wavelength.all_mesh_positions xf lvector[loopcounter]
get generation.at_wavelength.all_mesh_positions xg lvector[loopcounter]
when lvector had been filled with 18 values 300, 350, …, 1150 before starting the loop.
And Figure 11.4 illustrates the result of the script commands:
get lightflux.all_internalwavelengths.at_position lf xvector[loopcounter]
get generation.all_internalwavelengths.at_position lg xvector[loopcounter]
when xvector had been filled with 13 values 0.0, 0.5, …, 6.0 before starting the loop.
These figures Figure 11.3 and Figure 11.4 are useful to visualise the illumination, light flux and eh
generation rate behaviour in the tandem cell under study, as explained in the figure captions.
142 Chapter 11: Script support for simulating multi-junction (tandem) cells

Figure 11.4 Left: light flux Nphot(x; ) (in #/cm2s) and Right: generation G(x, ) (in #/cm3 s) as a function of  (in nm),
and for 13 x-values from 0 (red curve top left) to 6.0 m (grey curve bottom right), step 0.5 m. The tandem split at 2.5
m is the yellow curve. The wavy shape of the top curves originates from the sinusoidal shape of the front transmission
filter R() at the front contact (left); this has no realistic or physical relevance at all, we just used this R() to clearly see
the influence of this filter… and we do see it .

11.10.3 Results of the tandem simulation with ‘adapt filters’

In the next illustration, we will vary the back reflection Rb() of the tandem cell from 0 to 100 %, 11 values
with 10 % steps. The calculated filters (Rback’() that is placed at the back contact of the top cell, and Tfront’()
that is placed at the front contact of the bottom cell, are shown in Figure 11.5, right. The resulting total (=
integrated over all ) generation G(x) is shown in Figure 11.5, left. The oversized figure caption explains
some of the features to be observed.

Figure 11.5 Problem file test tandem cell scripts illum from left.def. The back reflection Rback
is varied from Rback = 0 to Rback = 1.0, step Rback = 0.1.Left: the eh generation rate G(x) in #/cm3s integrated over all
wavelengths of the incoming spectrum; the top cell has 0  x  2.5 m, and the bottom cell has 2.5 m  x  6 m.
Right: the adapted back reflection Rback’() of the top cell calculated by Eq. (38) and the adapted front transmission
Tfront’() of the bottom cell calculated by Eq. (39). The bundle Tfront(; Rback) is seen as one curve (red); the sinusoidal
shape of the front transmission filter used in this problem is followed for  > 690 nm, but is efficiently damped for  <
690 nm due to strong absorption in the top cell. The bundle Rback(; Rback) is seen as a bundle of 11 curves in the long
wavelength range: Rback’() = Rback for light not absorbed at all ( > 1130 nm), and the effect of Rback is quickly dying
out for shorter wavelengths due to absorption of reflected light in the bottom cell. As a consequence, the generation
G(x) in the top cell does not depend on Rback at all, and G(x) in the bottom cell has some weak dependence on Rback in
the immediate vicinity of the back contact at x = 6 m.
11.10 Application examples 143

The IV and QE curves of the top and bottom cell, calculated with
math split_tandem_cell.adapt_contact_filters 2 3
and the IV curves of the tandem connection, calculated with (for example)
math series WIVTUB 100
are shown in Figure 11.6. The conclusions are consistent with Figure 11.5: the top cell is not sensible to
Rback, the bottom cell only slightly.

Figure 11.6 Left: the IV curves of the top cell (curves for all 11 Rback values almost coinciding, with Voc ~ 1.3 V), the
top cell (a bundle of slightly differing curves with Voc ~0.6 V) and the series connection (curves coinciding with Voc
~1.8 V). Right: the QE curves of the top cell (all curves coinciding; only > 0 for  < 690 nm) and of the bottom cell
(only > 0 for 690 nm <  <1130 nm; the Rback bundle is only noticeable at long wavelengths, say  > 1030 nm or so)

The conclusion would be different if the optical absorption of the layers would be much lower (lower
absorption constant  or thickness d). We changed constant 1 of the optical absorption model of all layers
from 1 = 105 /cm to 1 = 103 /cm, and redid the calculations that lead to Figure 11.5 and Figure 11.6: the
result is in Figure 11.7 and Figure 11.8.

Figure 11.7 As in Figure 11.5, but for very low optical absorption (in all layers, 1 was set to 103 /cm). The total
generation Gtot(x) now strongly depends on Rback in the bottom cell, and slightly but clearly visible in the top cell. The
adapted Tfront’() does not depend on Rback (in agreement with Eq. (39) ), but the adapted Rback’() does (in agreement
with Eq. (38), and much stronger with this low value for 1: the influence is visible at all wavelengths, also for  < 690
nm, that is almost completely absorbed in the top cell when 1 = 105 /cm.
144 Chapter 11: Script support for simulating multi-junction (tandem) cells

Figure 11.8 As in Figure 11.6, but for very low optical absorption (in all layers, 1 was set to 103 /cm). Note that now
the top cell and the resulting tandem cell have some noticeable dependence on Rback, and the bottom cell has a strong
Rback-dependence.

11.10.4 Comparison of the 3 script strategies for tandem cells

With the strategy “adapt generation”, the spectral response of course cannot be calculated. The results for
G(x) and IV are almost coinciding with the results obtained with strategy “adapt filters” (no illustration in a
figure here).
The third strategy “electronically inactive parts…” does allow for the calculation of QE. The results for
G(x) and QE() are almost coinciding with the strategy “adapt filters” (no illustration). The IV results are
similar, but here some deviation is visible. The IV curves of the top, bottom and tandem cells, calculated with
the three strategies, are shown in Figure 11.9.

Figure 11.9 The IV curves of the top cell (blueish curves with Voc ~ 1.25 V), the bottom cell (pinkish curves with Voc
~ 0.6 V) and the tandem cell (greenish curves with Voc ~ 1.8 V), for one value Rback = 1, and for the three script
strategies. The IV curves for “adapt filters” and “adapt generation” almost coincide, but the IV curves of the top cell
(and hence also the tandem cell) with strategy “electronically inactive parts” deviate somewhat (and have a Voc about
0.04 V lower than calculated with the two other strategies).

To explore why the strategy “electronically inactive parts…” is not performing too well, we check the
assumptions and algorithm that we used to make the bottom cell electronically inactive when studying the
top cell, and the other way around. These assumptions were discussed in Section 11.7 page 136. We limit us
11.10 Application examples 145

here to the “top cell with electronically inactive bottom cell”, and to the doping and defect settings in the
electronically active bottom cell (that we once called fake bottom cell in an Application Note – not very
scientific, but shorter to type, and maybe clearer). The top cell in our example is np type (n-layer left, p-layer
right), thus the fake bottom cell was set to p-type everywhere, with NA = 1019 cm-3, and was given one mid-
gap defect with density Nt = 1017 cm-3. These settings are internal in SCAPS when running the script command
math split_tandem_cell. with_electronically_inactive_parts n1 n2
and cannot be varied by the user. One can however load the definition file for the top cell with fake bottom
cell set-up by the script (with load definitionfile.temporary.top_cell) and then, outside of
the script, vary NA or Nt of all fake bottom cell layers simultaneously in the batch settings. We did so once for
our tandem cell example test tandem cell scripts illum from left.def as is, and then
again for this same tandem cell with a modification: the top cell was given a Back Surface Field layer (BSF):
the base layer (p-layer) of the top cell has NA = 1015 cm-3 over 1.5 m, but NA = 1017 cm-3 over a 0.5 m
distance from the bottom cell to the right.
That leads to the results of Figure 11.10. The efficiency parameter most affected by the variation of NA and
Nt is Voc; the influence on Jsc and FF is very minor.

Figure 11.10 Dependence of the open circuit voltage Voc of the top cell of a tandem cell with an “electronically inactive
bottom cell”, as constructed by the script command math split_tandem_cell.
with_electronically_inactive_parts… Left: dependence on the acceptor density NA set in the two layers
of the electronically inactive bottom cell. Right: dependence on the defect density Nt set in the two layers of the
electronically inactive bottom cell. The values of NA = 1019 cm-3 and Nt = 1017 cm-3 that are set in the script are indicated
by the magenta vertical line. Red curves: starting from test tandem cell scripts illum from
left.def as is, thus without back surface field layer (no BSF). Blue curves: the top cell of this tandem structure was
given a BSF layer (see text).

It is clear that “a mild variation of Nt around the set value Nt = 1017 cm-3” does not influence the results at all
(Figure 11.10 right). “At all” is somewhat exaggerated: when examining e.g. the detailed Voc(Nt) dependence
for the top cell with BSF, one sees some influence, but very, very minor, indeed negligible for all practical
purposes (Figure 11.11). In contrast, a variation of NA around its internal setting NA= 1019 cm-3 does influence
the IV results, see Figure 11.10 left: this influence is small and maybe negligible for the cell with BSF (blue
curve), but is substantial for the cell without BSF (red curve).
These examples show that:
 The strategy “partial cells with electronically inactive counterparts” is the most versatile of the three
strategies: one can change the illumination conditions (spectrum, intensity) and the optically relevant
parameters of the partial cell under study (absorption files or absorption model parameters, band gap,
thickness; one can add layers…) without having to run each time again the math
split_tandem_cell command in a script.
 However, the user should check the validity of the approximations used to obtain the “electronically
inactive counterpart”. These assumptions might be more or less valid, or totally not, depending on the
146 Chapter 11: Script support for simulating multi-junction (tandem) cells

cell settings. SCAPS cannot do this analysis automatically for you, and vary the settings (in our
illustration, the parameters NA and Nt) to obtain a better implementation of the “electronically inactive”
counterpart, the user is responsible for this. This illustrates again that a physical simulation program
cannot substitute for physical insight  …
 And thus, we repeat our conclusion at the end of Section 11.7: “It might make sense to do the vast
amount of the exploring work with the strategy of inactive counterparts, and now and then (and also
finally) check with the more rigorous, but less versatile strategies of adapt generation or adapt filters”.

Figure 11.11 Detail of the blue curve of Figure 11.10 right: the Voc(Nt) dependence for the top cell with BSF. There is a
very slight Voc(Nt) dependence, negligible in practice.

11.11 Conclusion
We can copy here the conclusion of our former Application Note on tandem cells:
 SCAPS cannot handle multi-junction structures directly, even not tandem structures.
 The way out is to calculate the top and bottom cells separately. This text gives hints and examples of
how these ‘top’ and ‘bottom’ cells should be set-up.
 Three strategies, and script support for them, are developed in SCAPS 3.3.10.
 The SCAPS script also handles the series connection for you, within the SCAPS environment.
 … but it still looks very much as a Plan B… and actually it is one.
 References

[1] M. Burgelman, P. Nollet, S. Degrave, Modelling polycrystalline semiconductor solar cells, Thin Solid
Films, 361 (2000) 527-532.
[2] K. Decock, S. Khelifi, M. Burgelman, Modelling multivalent defects in thin film solar cells, Thin Solid
Films, 519 (2011) 7481-7484.
[3] M. Burgelman, J. Marlein, Analysis of graded band gap solar cells with SCAPS, Proceedings of the
23rd European Photovoltaic Solar Energy Conference, Valencia, 2008, pp. 2151-2155.
[4] J. Verschraegen, M. Burgelman, Numerical modeling of intra-band tunneling for heterojunction solar
cells in SCAPS, Thin Solid Films, 515 (2007) 6276-6279.
[5] S. Degrave, M. Burgelman, P. Nollet, Modelling of polycrystalline thin film solar cells : new features
in SCAPS version 2.3, Proceedings of the 3rd World Conference on Photovoltaic Energy Conversion,
Osaka, 2003, pp. 487-490.
[6] A. Niemegeers, M. Burgelman, Numerical modelling of ac-characteristics of CdTe and CIS solar
cells, Proceedings of the 25th IEEE Photovoltaic Specialists Conference, Washington DC, 1996, pp.
901-904.
[7] W.H. Press, S.A. Teukolsky, W.T. Vetterling, B.P. Flannery, Numerical recipes in C, 2nd ed.,
Cambridge University Press, New York, 1992.
[8] J. Verschraegen, S. Khelifi, M. Burgelman, A. Belgachi, 21st European Photovoltaic Solar Energy
Conference, Dresden, Germany, Sept. 2006, 2006.
[9] S. Khelifi, M. Burgelman, J. Verschraegen, A. Belghachi, Impurity photovoltaic effect in GaAs solar
cell with two deep impurity levels, Solar Energy Materials and Solar Cells, 92 (2008) 1559-1565.
[10] S. Khelifi, J. Verschraegen, M. Burgelman, A. Belgachi, Numerical simulation of the impurity
photovoltaic effect in silicon solar cells, Renewable Energy, 33 (2008) 293-298.
[11] K. Decock, P. Zabierowski, M. Burgelman, Modeling metastabilities in chalcopyrite-based thin film
solar cells, Journal of Applied Physics, 111 (2012) 043703.
[11'] M. Burgelman, K. Decock, S. Khelifi and A. Abass, Advanced electrical simulation of thin film solar
cells, Thin Solid Films, 535 (2013) 296-301.
[12] A. Niemegeers, S. Gillis, M. Burgelman, A user program for realistic simulation of polycrystalline
heterojunction solar cells: SCAPS-1D, Proceedings of the 2nd World Conference on Photovoltaic
Energy Conversion, Wien, 1998, pp. 672-675.
[13] H.J. Pauwels, G. Vanhoutte, Influence of interface states and energy barriers on efficiency of
heterojunction solar-cells, J. Phys. D-Appl. Phys., 11 (1978) 649-667.
[14] J. Verschraegen, Karakterisering en modellering met SCAPS van de CISCuT dunne-filmzonnecel,
dissertation Universiteit Gent. Faculteit Ingenieurswetenschappen, 2006.
[15] S. Selberherr, Analysis and Simulation of Semiconductor Devices, Springer Verlag, Wien-New York,
1984.
[16] J. Marlein, M. Burgelman, Proceedings of NUMOS (Int. Workshop on Numerical Modelling of Thin
Film Solar Cells, Gent (B), 28-30 March 2007). p. 227-233 2007, 2007.
[17] T. Walter, R. Herberholz, C. Müller, H.W. Schock, Determination of defect distributions from
admittance measurements and application to Cu(In,Ga)Se 2 based heterojunctions, Journal of Applied
Physics, 80 (1996) 4411-4420.
[18] K. Decock, S. Khelifi, S. Buecheler, F. Pianezzi, A.N. Tiwari, M. Burgelman, Defect distributions in
thin film solar cells deduced from admittance measurements under different bias voltages, Journal of
Applied Physics, 110 (2011) 063722.