948 IEEE JOURNAL OF PHOTOVOLTAICS, VOL. 4, NO.

3, MAY 2014

Light Trapping in Thin-Film Cu(InGa)Se2 Solar Cells

James G. Mutitu, Uwadiae Obahiagbon, Shouyuan Shi, William Shafarman, and Dennis W. Prather

Abstract—A fundamental optical analysis of thin-film

Cu(InGa)Se2 solar cell structures is presented, wherein spectro-
scopic ellipsometry measurements were performed to acquire ma-
terial optical constants, which were then used as input parameters
to perform electromagnetic simulations. The accuracy of the elec-
tromagnetic simulation tools, and thus, the validity of the material
optical constants, were verified by comparing the values deter-
mined from the simulations with experimental measurements ob-
tained using a spectrophotometer. The verified optical modeling
tools were then used to analyze thin, <0.7-μm Cu(InGa)Se2 so-
lar cell structures, which do not absorb all incident light within a
single optical path length, and hence, the need to incorporate light Fig. 1. Structure of the Cu(InGa)Se2 superstrate design complete with diffrac-
trapping. To this end, a superstrate device configuration was em- tion gratings.
ployed in which the metallic back contact is deposited last, giving
rise to an opportunity to incorporate photonic engineering device
concepts to the back surface layer of the solar cell. Simulations
of superstrate Cu(InGa)Se2 solar cell designs, complete with light of groups have explored light trapping in Cu(InGa)Se2 solar
trapping structures were then performed and analyzed. cells, due in part to its high absorption coefficient, which en-
sures that 95% of the incident illumination is absorbed within a
Index Terms—Diffraction, photovoltaic cells, solar energy.
1-μm thick layer, i.e., within a wavelength band corresponding
to the material bandgap. However, when the thickness of the
I. INTRODUCTION Cu(InGa)Se2 absorber layer is reduced to below 1 μm, a single
optical pass of light through the material is insufficient to fully
GOOD understanding of the interaction between incident
A light and a solar cell device structure is critical to opti-
mizing device performance. To this end, improving thin-film
absorb as much light as compared with thicker cells, and thus,
the need for incorporating light-trapping schemes arises. The
advantages of having thin, <1-μm thick, Cu(InGa)Se2 absorber
solar cell performance by increasing the light-trapping capac- layers include, reduced material costs and shorter device pro-
ity has received significant attention in the recent past. Many cessing times, which ultimately lead to an increase in manufac-
of these studies have focused on c-Si and a-Si because of the turing throughput and lower capital requirements for deposition
ubiquitous use of these materials throughout the semiconduc- processes. In practice, however, the efficiency of superstrate de-
tor industry, in addition to the emergence and growth of the vices still lags behind those of conventional substrate devices,
silicon photonics industry [1]–[3]. In contrast, only a handful which further serves to prove the need for light-trapping studies
in thin Cu(InGa)Se2 solar cells.
In the past, studies by Orgassa et al. examined the effects
Manuscript received July 11, 2013; revised September 20, 2013; accepted of a variety of metallic back contact reflectors on Cu(InGa)Se2
October 20, 2013. Date of publication March 14, 2014; date of current version light trapping [4]. Malmström et al. investigated the effects of
April 18, 2014. This work was supported by the Department of Energy under a TiN back contact and reflector, in addition to front and back
Award DE-EE0005317. This report was prepared as an account of work spon-
sored by an agency of the United States Government. Neither the United States surface scattering [5]. However, many of these studies have been
Government nor any agency thereof, nor any of their employees, makes any focused on substrate device configurations, where all the device
warranty, express or implied, or assumes any legal liability or responsibility for components are deposited on top of a material substrate, e.g.,
the accuracy, completeness, or usefulness of any information, apparatus, prod-
uct, or process disclosed, or represents that its use would not infringe privately soda lime glass.
owned rights. Reference herein to any specific commercial product, process, or Traditional substrate configurations of Cu(InGa)Se2 solar
service by trade name, trademark,manufacturer, or otherwise does not neces- cells present a major problem when it comes to light-trapping
sarily constitute or imply its endorsement, recommendation, or favoring by the
United States Government or any agency thereof. The views and opinions of design. The main problem is that a substrate fabrication pro-
authors expressed herein do not necessarily state or reflect those of the United cess requires that the choice of back contact material be inert,
States Government or any agency thereof. in order to withstand the highly corrosive environment that is
J. G. Mutitu and U. Obahiagbon are with the Department of Electrical and
Computer Engineering and the Institute of Energy Conversion, University of generated during the Cu(InGa)Se2 deposition step. This lim-
Delaware, Newark, DE 19716 USA (e-mail: [email protected]; [email protected]). its greatly the choices of available back surface materials. To
S. Shi and D. W. Prather are with the Department of Electrical and Com- circumvent this hurdle, a superstrate design configuration is uti-
puter Engineering, University of Delaware, Newark, DE 19716 USA (e-mail:
[email protected]; [email protected]). lized, whereby the constituent solar cell materials are deposited
W. Shafarman is with the Institute of Energy Conversion, University of in the order of ITO/ZnO/Cu(InGa)Se2 /metal back contact, onto
Delaware, Newark, DE 19716 USA (e-mail: [email protected]). a glass superstrate, as shown in Fig. 1.
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org. The superstrate configuration enables a thinner absorber
Digital Object Identifier 10.1109/JPHOTOV.2014.2307487 layer and overcomes the inert metal constraint, which affects

2156-3381 © 2014 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications standards/publications/rights/index.html for more information.
MUTITU et al.: LIGHT TRAPPING IN THIN-FILM CU(INGA)SE2 SOLAR CELLS 949

equations to obtain the optical constants, which is not a trivial

process. In addition, some material systems consist of complex
compositional gradients, and hence, exhibit inhomogeneous re-
fractive index profiles. To this end, the optical characterization
using VASE may in some cases be used to obtain average, or
bulk, material refractive index values. Similarly, it may also be
impossible to completely represent the physical structure of a
solar cell device using simulation software, especially in sit-
uations where a considerable amount of surface roughness is
present, and again such situations may call for the use of some
approximations. However, to circumvent the accumulation of
many erroneous results, further characterization is required. In
this case, a spectrophotometer is used; the results obtained from
spectrophotometer measurements can be directly compared with
Fig. 2. Integrated Loop approach used to verify optical material characteristics simulation results, and hence, provide a reliable means by which
and accuracy of simulation tools. (Inset) Example of the comparison between to ascertain the veracity of both the VASE characterization
the simulation and experimental data from ITO. and simulation processes. The verified computational tools can
thereafter be used to design complex light-trapping structures,
traditional substrate devices, since the back contact is deposited such as the one shown in Fig. 1.
after the Cu(InGa)Se2 layer has been deposited. Additional ad-
vantages of superstrate solar cells include: the elimination of the III. EXPERIMENTAL METHODS
need for a glass back sheet, which leads to lower cost and lighter
weight modules; greater flexibility and simplification in the laser The Cu(InGa)Se2 films were deposited using an elemental
scribing process, which is ultimately necessary for monolithic coevaporation of Cu, In, Ga, and Se from independent sources
integration; and finally, the ability to process the TCO layers at in a Belljar evaporator system at a substrate temperature of
elevated temperatures, which leads to improved reliability and 550 ◦ C. A single-stage deposition recipe was used, whereby all
performance. the fluxes, of all the constituent materials, were held constant
throughout the process, producing films with uniform compo-
sitions and no Ga gradient [6]–[8]. The resulting films were
II. OVERVIEW
∼2-μm thick films with a [Cu]/[In+Ga] composition of about
This paper focuses on building a fundamental framework 0.9 and a [Ga]/[In+Ga] composition of 0.3, giving rise to a
for the study of light-trapping techniques in superstrate config- bandgap of 1.2 eV. This coevaporation process produces a film
ured Cu(InGa)Se2 solar cells. The goal is to develop a reliable with a high amount of surface roughness, which makes it dif-
electromagnetic simulation tool that can be used to optimize ficult to acquire accurate optical constants. To circumvent this
various design parameters in Cu(InGa)Se2 solar cells, such as hurdle, Paulson et al., peeled off the Cu(InGa)Se2 film by ex-
layer thicknesses, and thus, design better performing devices. ploiting the weakness that occurs at the MoSe2 layer between
To accomplish this, a customized integrated loop approach is the Cu(InGa)Se2 and Mo layers [8], [9]. Shafarman et al. also
employed, whereby samples are prepared and analyzed through determined the optical constants of Cu(InGa)Se2 by first reduc-
two distinct methods, the first being ellipsometry, which gives ing the roughness and then removing residual Se on the surface
material optical constants that can then be used to develop the using a Br-etch and KCN etch, respectively [7]. In this study,
electromagnetic modeling tools. The second method utilizes a the latter technique was employed using the optimized Br and
spectrophotometer to take alternate measurements that can then KCN etch processes. It is worthy to note that, typical values
be directly compared with the output of the electromagnetic of root-mean-square (rms) roughness measured by atomic force
models, thereby verifying the validity of the simulations, as is microscopy on Cu(InGa)Se2 films are between 50 and 100 nm
shown in Fig. 2. making it difficult to make valid ellipsometry measurements.
The first step in the integrated loop process is to accurately The rms surface roughness is reduced by the Br-Etch process
obtain the optical constants of materials in order to ensure that to less than 10–15 nm, which is significantly smaller than the
the subsequent analyses—including the development of simu- incident ellipsometer beam wavelengths, and hence, does not
lation tools—are of the highest accuracy. To achieve the desired adversely affect the measurements [10], [11].
precision in analysis, there is a strong need to be able to de- Thin films of ZnO and ITO with thicknesses of 162 nm and
posit repeatable material compositions and then analyze them 92 nm, respectively, were deposited using a sputtering process;
using variable angle spectroscopic ellipsometry (VASE) mea- the resultant layers had sheet resistances of 5 ×103 Ω/ (for
surements. The second step requires that the deposited samples ZnO) and 12.1 Ω/ (for the ITO). A 200-nm gold layer was de-
be prepared, which could entail the need for material etches to posited using electron beam evaporation, and a similar process
smoothen surfaces, in order to obtain reliable instrument read- was used to produce a 500-nm silver layer. Optical constants
ings during the VASE analysis. The VASE analysis, in turn, of these individual films were obtained using a J. A. Wool-
calls for the modeling of material properties using dispersion lam rotating analyzer variable spectroscopic ellipsometer that is
950 IEEE JOURNAL OF PHOTOVOLTAICS, VOL. 4, NO. 3, MAY 2014

equipped with an autoretarder. The VASE instrument works by

measuring the change in polarization that light waves experience
as they are reflected from surfaces and interfaces through pla-
nar multilayered materials [12], [13]. The VASE measurements
yield a phase difference (Δ) and an amplitude ratio (ρ), which
is the complex reflectance ratio of the parallel (rp ) and perpen-
dicular (rs ) wave polarizations, based on the Fresnel equations
for reflection and transmission. The complex reflectance ratio
(ρ) used in VASE is measured as function of both angle and
wavelength. The relationship between all these terms is shown
in
rp
tan(Ψ)eiΔ = ρ = (1)
rs
where tan(Ψ) represents the amplitude ratio of the reflected
waves (magnitude of reflectance ratio, ρ).
Since VASE measures the complex ratio between two values
it can be highly accurate in addition to yielding phase infor-
mation; the self-referencing nature of the measurement ensures
that the effects of fluctuations in lamp intensity, or degradation,
do not affect the final results, and hence, the measurements are
reproducible.
The reflectance and transmission measurements were ac-
quired using a Perkin–Elmer Lambda 750 spectrophotometer,
fitted with an integrating sphere. The spectrophotometer normal-
izes the results of reflectance and transmission measurements to
those achieved using calibrated reflectance standards.

IV. OPTICAL CONSTANT VERIFICATION

Throughout the optical characterization process, a number of
issues arose. Initially, each material was deposited on a soda-
Fig. 3. Comparison between experimental and simulated reflectance for
lime glass (SLG) substrate and then analyzed using VASE. It (a) ITO, (b)ZnO, (c) Cu(InGa)Se2 , (d) Gold, and (e) Silver.
would appear that obtaining the refractive index of a glass sub-
strate, such as SLG should be a straightforward process. How-
ever, it turns out that during the Pilkington glass manufacturing
process some of the molten tin, on which the glass floats, diffuses Kronig relations [16], even though a free carrier Drude model
into the glass, giving rise to the tin or float side of the SLG [14]. combined with a Lorenz oscillator could also have been used.
This tin side creates a nonlinear graded refractive index profile, The gold layer was modeled using a B-Spline layer, which gave
which extends about 900 nm into the 1.6-mm-thick SLG. In ad- an MSE of about 2 [16], whereas with silver, the values were
dition, the bulk of SLG exhibits birefringence, i.e., the refractive retrieved from the DeltaPsi2 software—from Horiba—based on
index depends on the polarization and propagation direction of modified data values obtained from the text Handbook of opti-
light. The air side also has a different refractive index [14]. This cal constants of solids by E. Palik [17], [18]. The comparison
complexity associated with the SLG analysis led to a substitution between experimental data, obtained from spectrophotometer
of c-Si substrates for the ITO and ZnO layers. Crystalline silicon measurements, and simulation data is shown in Fig. 3.
is a well understood material that exhibits a large refractive in- Throughout this study with Cu(InGa)Se2 , it was found that the
dex contrast between it and ITO or ZnO. Using the c-Si substrate material may have a graded index profile, depending on how the
led to the acquisition of lower mean square error values during deposition process actually occurred. Nevertheless, with multi-
the VASE analysis, which were 5.4 and 3.8 for the ITO and ZnO ple measurements, such as spectroscopic transmission intensity
films, respectively. The ZnO layer was modeled using a General measurements—performed on a Cu(InGa)Se2 layer on an SLG
Oscillator model, which allows for the combination of multiple substrate—which could be modeled simultaneously with data
oscillator models, in this case a P-Semi and Gaussian oscillator from more traditional VASE reflectance measurements, it was
were used; the General Oscillator model is one of the disper- possible to obtain better complex refractive index values. The
sion formulae that is released by the J. A. Woollam Company in Cu(InGa)Se2 was modeled using a B-Spline layer. Additional
their CompleteEASE software package, which is used to create ellipsometry, simulations and spectrophotometer characteriza-
the ellipsometer models [15]. The ITO was modeled using a B- tions were performed on a Cu(InGa)Se2 layer deposited on a
Spline dispersion layer, which is a dispersion equation based on crystalline silicon substrate. The resultant simulation data also
the B-Spline recursion relation and is consistent with Kramers– matched up well with experimental data, as is shown in Fig. 3(c).
MUTITU et al.: LIGHT TRAPPING IN THIN-FILM CU(INGA)SE2 SOLAR CELLS 951

Fig. 5. (a) Reflectance characteristics of different back surface materials into

a semiinfinite Cu(InGa)Se2 structure and (b) illustration of the structure under
analysis for each material shown in (a).

interval, using the optical constants that had been previously ob-
tained using the integrated loop analysis process, and the results
are plotted in Fig. 4(a) and (b).
Fig. 4. (a) Absorption and (b) transmission characteristics of Cu(InGa)Se2 This exercise enabled the observation of the absorption char-
layers of various thicknesses. (c) Cu(InGa)Se2 layers that are under analysis, acteristics of light in a single optical path length. The wavelength
in (a) and (b), are sandwiched between two semiinfinite structures of the same
material, and therefore, no front or back surface reflections occur. A light wave
range of this analysis was limited to 300–1100 nm, which cor-
propagating through the layer of interest is also shown. responded to the bandgap of the coevaporated Cu(InGa)Se2 ,
which is about 1.2 eV [6]–[9]. As can be seen in Fig. 4(b), for a
structure that is 700-nm thick, only light of wavelengths above
All the other materials properties used in this light-trapping
600 nm would hit the back reflector surface, and hence, that is
study were obtained from previous studies [3], [8]–[10].
the range where the back reflector design is focused.
The electromagnetic simulation tool that was used throughout
The reflectance characteristics were simulated for different
this study is centered on the scattering matrix algorithm, which
back surface materials, including soda-lime glass, aluminum,
is a modified version of a transfer matrix-based algorithm that
silver, gold, and molybdenum, with each of these materials
is used to solve Maxwell’s equations rigorously for each indi-
placed below a semiinfinite layer of Cu(InGa)Se2 . As can be
vidual wavelength in the spectrum being considered [20]–[22].
seen in Fig. 5, gold and silver exhibited the best reflectance char-
In this algorithm, the structure to be simulated is broken up into
acteristics especially at longer wavelength regions. Molybde-
many individual layers in the direction of light illumination. The
num, which is the commonly used back contact for Cu(InGa)Se2
fields in each layer are expanded into Fourier modes, whereby
solar cells, did not exhibit good reflectance characteristics.
redundant Fourier modes are also incorporated in order to ensure
An optimization of the thicknesses of the ITO and ZnO layers
the accuracy of the scattering matrix method; this also includes
was performed using the particle swarm optimization algorithm
cases where large index contrast materials are considered. The
[3] to reduce the device reflectance. It is pertinent to note that
expansion coefficients for each Fourier mode are then used to
that the soda-lime glass in this case was treated as a semiinfinite
form the scattering matrix. The scattering matrices from each of
superstrate, so as to focus on the ITO and ZnO thicknesses
the layers are then cascaded in order to attain an overall system
specifically. The optimal design parameters were found to be
matrix that can be used to provide both reflection and trans-
150 nm of ITO and 112 nm of ZnO, which gave an average
mission characteristics. The application of periodic boundary
reflectance of 6.5% in the 300–1100 nm range.
conditions makes it possible to significantly reduce the com-
Finally, the optimized ITO and ZnO coatings, along with the
putation time, thus making the S-Matrix method suitable for
silver or gold back reflectors, were applied to a superstrate so-
thin-film solar cell applications.
lar cell with 0.7-μm thick Cu(InGa)Se2 —in a structure similar
to Fig. 1 but without the diffraction gratings—and the light-
V. FUNDAMENTAL LIGHT-TRAPPING STUDY trapping characteristics were simulated. An additional diffrac-
In order to perform a light-trapping study of value, it is nec- tion grating, with a period of 860 nm and 50% fill factor (sim-
essary to identify the optical characteristics of a material when ilar to Fig. 1)—corresponding to feature sizes that have previ-
no optical enhancements or reflections occur. This is calculated ously been realized using a deep ultraviolet lithography process
by placing the Cu(InGa)Se2 between two semiinfinite struc- [19]—was included to the final design structure and simulated.
tures of the same material, i.e., a superstrate and substrate of The resultant absorption characteristics of the different struc-
Cu(InGa)Se2 in this case, and letting light flow from within the tures are plotted in Fig. 6. It was observed that the absorption
superstrate, through the structure and out into the substrate, as characteristics were almost the same, which is what was ex-
shown in Fig. 4(c). This way, it is possible to determine how pected, considering that the reflectance of gold and silver, into
much light is absorbed within a certain thickness of the mate- Cu(InGa)Se2 , is almost identical for wavelengths above 750 nm,
rial. This type of analysis was performed on Cu(InGa)Se2 layers as was shown in Fig. 5. Additionally, all wavelengths below
with thicknesses ranging from 100 to 2100 nm with a 200-nm 750 nm would be absorbed in the 700-nm Cu(InGa)Se2 layer
952 IEEE JOURNAL OF PHOTOVOLTAICS, VOL. 4, NO. 3, MAY 2014

TABLE II
ABSORPTION CHARACTERISTICS FOR ALL THE SIMULATED CU(INGA)SE2
SOLAR CELL STRUCTURES, ARRANGED IN ORDER OF PERFORMANCE

Fig. 6. (a) Percentage and (b) total power absorption characteristics of 700-
nm-thick Cu(InGa)Se2 solar cell structures with various optical enhancements;
silver and gold back reflectors and silver and gold diffraction gratings.

VI. CONCLUSION
In this paper, a fundamental optical analysis framework for
Fig. 7. (a) Percentage and (b) total power absorption characteristics of 300- designing light-trapping schemes for thin-film Cu(InGa)Se2 so-
and 50-nm Cu(InGa)Se2 solar cell structures with various optical enhancements;
silver back reflectors and silver diffraction gratings. lar cells was presented. The first part of the study examined
the process of obtaining accurate optical material constants us-
TABLE I ing a number of techniques including VASE, electromagnetic
OPTIMAL DESIGN PARAMETERS OF MATERIALS USED IN CU(INGA)SE2 SOLAR simulations, and reflectance measurements obtained using a
CELL STRUCTURE SIMULATIONS
spectrophotometer. The optical constants that were verified us-
ing all the aforementioned methods were then used to accu-
rately develop an electromagnetic simulation tool that was sub-
sequently used to analyze various aspects of light propagation
through Cu(InGa)Se2 solar cells. The first aspects to be con-
sidered were the absorption and transmission characteristics of
Cu(InGa)Se2 layers of varying thickness, thereby giving in-
sights into the wavelengths of interest in light-trapping designs.
Subsequently, the simulation tool was used to assess the reflec-
before ever reaching the back reflector, as can be inferred from tion characteristics of different candidate materials for back sur-
Fig. 4. face reflectors in Cu(InGa)Se2 solar cells. Finally, the tool was
It was finally decided that silver be used for the rest of used to analyze more complex light-trapping schemes, which
the study because of its wider bandwidth of high reflectance, included diffraction gratings, for thin, 300-, 500-, and 700-nm,
when compared with gold, which would work better for thin- Cu(InGa)Se2 solar cells based on superstrate cell configurations.
ner Cu(InGa)Se2 structures; this conclusion was derived from It was found that the structures that incorporated diffraction grat-
observing the reflectance graphs in Fig. 5(a). ings were the most effective at trapping light.
The same optimized design parameters, as used in the struc-
tures of Fig. 6, were applied to thinner, i.e., 0.5- and 0.3-μm-
REFERENCES
thick Cu(InGa)Se2 , solar cell structures, but this time only uti-
lizing silver as the back surface and diffraction grating material, [1] J. Zhu, C. Hsu, Z. Yu, S. Fan, and Y. Cui, “Nanodome solar cells with effi-
cient light management and self-cleaning,” Nano Lett., vol. 10, pp. 1979–
and the results are plotted in Fig. 7. 1984, 2010.
As can be seen in Fig. 7, the structures with the silver diffrac- [2] J. G Mutitu, S. Shi, A. Barnett, and D. W. Prather, “Hybrid dielectric-
tion gratings outperformed the other structures, of correspond- metallic back reflector for amorphous silicon solar cells,” Energies, vol. 3,
pp. 1914–1933, 2010.
ing thickness, which only had a planar reflector. [3] J. G. Mutitu, S. Shi, C. Chen, T. Creazzo, A. Barnett, C. Honsberg, and
The optimal design parameters used for the simulation results D. W. Prather, “Thin film silicon solar cell design based on photonic crystal
of Figs. 6 and 7, are summarized in Table I. and diffractive grating structures,” Opt. Exp., vol. 16, pp. 15238–15248,
2008.
The absorption characteristics, in terms of averaged percent- [4] K. Orgassa, H. W. Schock, and J. H. Werner, “Alternative back contact ma-
age and total power absorbed, for all the simulation results terials for thin film Cu(In,Ga)Se2 solar cells,” Thin Solid Films, vol. 431–
presented in Figs. 6 and 7 are summarized in Table II. 432, pp. 387–391, May 2003.
MUTITU et al.: LIGHT TRAPPING IN THIN-FILM CU(INGA)SE2 SOLAR CELLS 953

[5] J. Malmstrom, O. Lundberg, and L. Stolt, “Potential for light trapping in [14] R. A. Synowicki, B. D. Johs, and A. C. Martin, “Optical properties of
Cu(In,Ga)Se/sub 2/ solar cells,” in Proc. 3rd World Conf. Photovoltaic soda-lime float glass from spectroscopic ellipsometry,” Thin Solid Films,
Energy Convers., May 2003, vol. 1, pp. 344–347. vol. 519, no. 9, pp. 2907–2913, Feb. 28, 2011.
[6] W. Shafarman and J. Zhu, “Effect of substrate temperature and deposition [15] [Online]. Available: http://www.jawoollam.com
profile on evaporated Cu(InGa)Se2 films and devices,” Thin Solid Films, [16] B. Johs and J. S. Hale, “Dielectric function representation by B-Splines,”
vol. 361–362, pp. 473–477, 2000. Phys. Status Solidi, a, vol. 205, no. 4, pp. 715–719, 2008.
[7] W. N. Shafarman, R. S. Huang, and S. H. Stephens, “Characterization of [17] E. D. Palik, Handbook of Optical Constants of Solids. New York, NY,
Cu(InGa)Se2 solar cells using etched absorber layers,” in Proc. Photo- USA: Academic, 1985.
voltaic Energy Convers., Conf., May 2006, vol. 1, pp. 420–423. [18] Powerful DeltaPsi2 Software Platform drives HORIBA Scientific
[8] P. D. Paulson, R. W. Birkmire, and W. N. Shafarman, “Optical character- Ellipsometers–HORIBA. [Online]. Available: www.horiba.com. <http://
ization of CuIn1 − xGax Se2 thin films by spectroscopic ellipsometry,” www.horiba.com/us/en/scientific/products/ellipsometers/software/>
J. Appl. Phys., vol. 94, pp. 879–888, 2003. [19] J. G. Mutitu, S. Shi, A. Barnett, and D. W. Prather, “Light trapping in thin
[9] P. D. Paulson, S. H. Stephens, and W. N. Shafarman, “Analysis of silicon solar cells,” presented at the 37th IEEE Photovoltaic Spec. Conf.,
Cu(InGa)Se2 alloy film optical properties and the effect of Cu off- Seattle, WA, USA, Jun. 2011.
Stoichiometry,” MRS Proc., vol. 865, pp. F1.1.1–F1.4.6, 2005. [20] A. C. Marsh and J. C. Inkson, “Scattering matrix theory of transport in
[10] S. H Stevens, “Modeling Optical Properties of Thin Film Cu(In,Ga)Se2 heterostructures,” Semicond. Sci. Technol., vol. 1, pp. 285–290, 1986.
Solar Cells Using Spectroscopic Ellipsometry,” Master’s Thesis, Phys. [21] M. Auslender and S. Hava, “S-matrix propagation algorithm in full-
Dept., Univ. Delaware, Newark, DE, USA, 2006. vectorial optics of multilayer grating structures,” Opt. Lett., vol. 21,
[11] R. W. Birkmire and B. E. McCandless, “Specular CuInSe2 films for solar pp. 1765–1767, 1996.
cells,” Appl. Phys. Lett., vol. 53, pp. 140–141, 1988. [22] D. M. Whittaker and I. S. Culshaw, “Scattering-matrix treatment of pat-
[12] J. Hilfiker, R. Synowicki, and H. G. Tompkins, “Spectroscopic Ellipsome- terned multilayer photonic structures,” Phys. Rev. B, vol. 38, pp. 2610–
try Methods for Thin Absorbing Coatings,” in Proc. SVC 51st Annu. Tech. 2618, 1999.
Conf., 2008, pp. 511–516.
[13] J. A. Woollam, B. Johs, C. Herzinger, J. Hilfiker, R. Synowicki, and
C. Bungay, “Overview of variable angle spectroscopic ellipsometry
(VASE), Part I: Basic theory and typical applications,” in Proc SPIE,
vol. CR72, pp. 3–28, 1999. Authors’ photograph and biographies not available at the time of publication.