AN ABSTRACT OF THE THESIS OF

Albert Sears Crane for the Master of Science in

(Name) (Degree)

Electrical Engineering presented on / -)0

(Major) (Date)

Title: The High Frequency Performance of a Schottky

Diode Detector

Abstract Approved:
Redacted for Privacy
James C. Looney
This paper analyzes the theory and performance of a

Schottky barrier diode detector used over a frequency range

extending from a few kilohertz into the gigahertz region.

The process of large signal rectification is analyzed with

the simplified detector circuit. Phenomena which can
affect the rectification process at high frequencies, such

as minority carrier lifetime, relaxation time, conductivity

modulation and the parasitic component elements are exam-

ined. Measurements are made on each of the detector com-
ponents and values derived for a network model. The model
is used to predict input impedance and frequency response

which is then compared to measured 'results.

THE HIGH FREQUENCY PERFORMANCE OF A SCHOTTKY
DIODE DETECTOR

by

Albert Sears Crane

A THESIS

submitted to

Oregon State University

in partial fulfillment of
the requirements for the
degree of

Master of Science

June 1971
APPROVED:

Redacted for Privacy

Associr Professor of Electrical)and Electronics
Engineering
in charge of major

Redacted for Privacy

He of Department of Electrical and Electronics
12 Engineering

Redacted for Privacy

Dean of Graduate School

Date thesis is presented 73

Typed by Erma McClanathan for Albert Sears Crane

 TABLE OF CONTENTS

I. Introduction 1

II. Theory of Rectification 3

Diode Rectification Model 3

Rectification Efficiency 9

Minority Carrier Lifetime 10

Conductivity Modulation 12
Relaxation Time 14
Parasitic Elements 15

III. Experimental Determination of Detector Elements 20

The Diode Detector 20

Diode Package and Junction Capacitance 22
Diode Series Resistance 23
Diode Series Inductance 26
Bypass Capacitor 29

IV. Composite Detector Performance 31

Rectification Efficiency versus Frequency 31

Network Analyzer Measurements 34
ECAP Results 37

V. Summary and Conclusions 42

Bibliography 44

Appendix 46
 LIST OF FIGURES

Figure Page

1. Basic Diode Detector Circuit 3

2. Linearized Diode Model 4

3. Diode Equivalent Circuit during Forward and

Storage Conduction 5

4. Diode Equivalent Circuit during Reverse Bias 5

5. Equilibrium Waveforms of the Rectifier Circuit 7

6. Equilibrium Values for the Rectifier Circuit 8

7. Effect of Series Resistance on Forward Drop 13

8. Diode Equivalent Circuit 16

9. Parasitic Elements of Diode Detector 18

10. Ladder Equivalent of Detector 19

11. Schottky Diode Detector in 50 Ohm Transmission

Line Environment 21

12. Junction Capacitance versus Reverse Voltage 24

13. Diode Log I versus Forward Voltage 25

14. Detector with Stub Line 27

15. Diode Impedance with Vr = 0 28

16. Impedance Plot of Erie Feed-through Capacitor 30

17. Frequency Response of Diode Detector 33

18. RLC Equivalent to Give e.3 /e 32

19. Series Diode and Capacitor Impedance Using

Stub Line 35

20. Equivalent Input Impedance Circuit 36

21. Equivalent Ladder Circuit 36

22. Impedance Plot of Diode Detector 38

 LIST OF TABLES

Table Page

1. Calculated Value for Capacitor Impedance 29

2. Calculated and Measured Values for Detector 40
 THE HIGH FREQUENCY PERFORMANCE OF A
SCHOTTKY DIODE DETECTOR

INTRODUCTION

High frequency, high level voltage measurements are

necessary for signal generator leveling and calibration.

The high frequency characteristics of a Schottky diode

detector are analyzed here to better understand the prob-

lems which exist in designing a high frequency voltage
detector. A flat frequency response and a constant input
impedance are the primary design objectives.

Semiconductor diodes have been in use for several dec-

ades in high frequency mixing and detecting circuits. Par-
ticular attention has been given to choosing a condition

which will optimize the square law detecting region for

accurate microwave measurements. In recent years, however,
swept frequency measurements and the desirability of cali-

brated and leveled generators has increased the need for

precision high level detectors with low VSWR and extended
frequency flatness over broad bands (7). The advent of the
Schottky barrier diode with its near-ideal I-V characteris-

tics has opened possibilities for high level voltage detec-

tion on a precision scale suitable for RF standards work.

A metal-semiconductor diode was used by Ondrejka (12) at

the National Bureau of Standards for peak pulse measurement.

Driver and Arthur (4) have described a differential voltage

 2

comparitor using hot carrier diodes which agree with NBS

voltage standards within 2% to 1 GHz.

The diode detector as a voltage standard has several

advantages over other types of RF voltage standards. For

high level signals its output is two orders of magnitude

greater than micropotentiometers or thermal convertors

which use vacuum thermocouples. The diode is also much

less susceptible to burn out due to over load. The diode

itself is smaller than a thermocouple element and inherent-

ly has lower parasitic elements to affect its high fre-

quency performance.

As with any physical device which is used for preci-

sion measurements, the parasitic elements inevitably influ-

ence electrical performance and require evaluation. This

paper will evaluate the high frequency limiting mechanisms

in a diode rectifier and determine the magnitude of their

effects.
 3

THEORY OF RECTIFICATION

Diode Rectification Model

The basic detector circuit is shown in Figure 1 with

the diode in series between the generator and the bypass

capacitor with its load resistor. The output voltage of

E sin wt

Figure 1. Basic diode detector circuit.

this low frequency model is dependent on the input voltage

E sin tit and the characteristics of the diode. At low
p
levels the output voltage is proportional to the input

power level, and the detector is described as operating in

the square law region. In the square law region a 1%

change in input voltage will cause a 2% change in detector

output voltage.

With the input voltage greater than a few tenths of a

volt, the ratio of forward to reverse diode current becomes

large enough that the voltage approaches a linear detection

 4

region. A graphical solution for finding the dc output

voltage and diode conduction angle based on the linearized

diode model shown in Figure 2 has been developed by Kra-

kauer (10).

I
/ Forward conduction

Storage conduction
Reverse
conduction

Figure 2. Linearized diode model.

The first three of the following diode parameters can

be chosen to give the best fit to the diode I-V character-

istics in the actual operating range.

R = forward resistance
f
R
r
= reverse resistance

7//
o
= nominal electrostatic junction potential

T = minority carrier life time. (The rate at which

minority carriers spontaneously recombine to

to return to thermodynamic equilibrium.)

From this model two equivalent circuits for the

 5

detector can be drawn: one for forward conduction, Figure

3, and the other for reverse conduction, Figure 4.

Rf

Figure 3. Diode equivalent circuit during forward

and storage. conduction.

Rg Rr

E sin w

Figure 4. Diode equivalent circuit during reverse

bias.

During forward conduction the effective charging re-

sistance R c is the combined source and diode resistance.
 6

Rc = Rg + Rf (1)

There are two currents acting to discharge C. The

current through RL is continuous and constant. The cur-

rent through Rr flows when the diode is in reverse conduc-

tion and the generator emf drops below EDC. Summing these

two currents gives the effective leakage current, i.e.

E 2v+0 E - E sin wt
DC DC r
dint (2)
RL /- 2v J R + Rr
8+0

where:

8 = diode current initiation angle

0 = diode conduction angle

E and E are related, Figure 5, by the sin 8 :

DC

E + * E
DC o 7 DC
E - (3)
sin 8 sin 8

Substituting and integrating yields:

0 cos 8 - cos (8+0)

ri 1 - 2v 2v
(4)
EDC 1RL Rg + Rr

Since C is assumed large enough so that ripple is negli-

gible, E
DC is constant and an effective discharge resist-
ance, R d may be calculated.

1 -
0 cos 8 - cos (8+0)
+
1 1 2v 2v
= + (5)
Rd R R + R
L g r
 7

E sin EDe+yo

Voltage o
e__1

Diode current
Recombination current
Current o
k-Leakage current

Recombined charge component

Total stored charge
Charge o Net charge transferred from C

*-Leakage charge

Figure 5. Equilibrium waveforms of the rectifier

circuit.

The ratio of charging resistance to discharging re-

sistance will determine the output voltage from the recti-

fier circuit. Combining Equations (1) and (4):

R Rq + R Rq + + R
c f $ cos 8 - cos (8+4))
1
Rd RL R + R 2v 271- sin 8
g r
(6)

Figure 6 shows the relationship between Rc/Rd, sin 8 ,

and wr.1 Knowing these four parameters, a value for

sin 8 can be found from the graph, and the value of EDC

determined from:

1 See Appendix for equilibrium equations.

 8

EDC = E sin 8 - *0 (7)

P

10
I
\O=ltrt
\v`

10

473159
=1-\-
1
M2 2"

1 0- ETIO

10-

I0

16 0. 1417 45°-
0 02 OA 0,6 0.8 1,0

SIN 8= E-9-
EP

*0 Diode electrostatic potential

Diode minority carrier lifetime
R Effective charging resistance for C
c
R
d
Effective discharging resistance for C
8 Diode conduction initiation angle
Diode conduction angle

Figure 6. Equilibrium values for the rectifier

circuit.

Since R /R sin 8 and (0 are not initially known, an

c d'
approximation for Rc/Rd can be used to find approximate
 9

values for sin 8 and 4) . Letting 0 = 0 in equation (6):

Rc R + R Rq + R
q f f
R R Rg + R ( 8 )
d L r

Using the approximate value of R /R values for sin

c d'
8 and I) can be read from Figure 6. These values for 0 and
4) can then be used in Equation (6) to re-evaluate Rc/Rd.

With this new value for R /R Figure 3 can again be used to

c d
find more precise values for 8 and sin 4). Since To is

given and sin 8 now determined, Equation (7) can be solved

for the detector output voltage.

Rectification Efficiency

Rectification efficiency is defined as the ratio of dc

output voltage to peak input voltage.

E
R.E. = DC 71jo

E
= sin 8 - (9)
p p

From this model analysis it is seen that the rectifi-

cation efficiency is dependent on several parameters. The
barrier potential *0 must be small compared to E for high
rectification efficiency. Reverse leakage also will have a

strong influence on the output. The ratio of forward to

reverse resistance of the diode should be low. This is

difficult to achieve when the peak voltage E is comparable

to the barrier potential, since for small signals forward

and back resistance approach each other.

 10

High speed switching limitations may be present in the

diode which would reduce the output at high frequencies.

If minority carriers are present, there will be a stored

charge in the junction. As this analysis has shown, the

presence of stored charge acts as a discharge path, reduc-

ing rectification efficiency. Stored charge will become

more detrimental as the time required to sweep out the

stored charge becomes a larger fraction of the period. The

Schottky barrier diode will now be examined to determine

the extent to which stored charge and other mechanisms can

limit frequency response.

Minority Carrier Lifetime

In evaluating the high frequency performance of the

Schottky diode detector, a number of physical aspects of

the diode exist which may place limitations on its high

frequency rectification efficiency. The Schottky diode

consists of a metal-semiconductor junction. In theory the
conduction is by majority carriers which should be able to
switch in a picosecond (2). However, actual measurements
indicate additional limitations.

In a metal-n silicon diode the main conduction is by

majority carriers, electrons. In forward conduction elec-

trons are transported from the semiconductor material over

the potential barrier and into the metal by thermionic

energy. However, to a much lesser extent, hole flow from

 11

the metal to the semiconductor is also present as it would

be in a p-n junction. Yu (16) gives quantitative expres-

sions for a Schottky barrier diode as follows:

-5
(3 x 10 A) [exp (qVFkT)-1] amperes (10)
ITE

I = (3 x 10-11A) [ exp (qVF/kT)-11 amperes (11)

where :

I = thermionic emission current

TE
A = diode area
q = electronic charge

V = forward bias
F
k = Boltzmann's constant

T = absolute temperature

I = hole current
p

These expressions show that the hole current is six

orders of magnitude smaller than electron current, and from

this standpoint can be neglected.

The presence of minority carriers will manifest itself

in two particular ways. In high speed switching applica-
tions any minority carriers present in the junction must be
swept out before the diode will stop conducting. Minority
carriers present in the junction are referred to as stored
charge. The effect of this stored charge on rectification

was evaluated earlier in the rectification model. From a

pulse switching point of view minority carrier life time
 12

and switching speed are related by the diode current

(8, p. 18-9):
I
F
t = T in (1 + I
) (12)
R

where:

is = switching time

T = minority carrier life time

I = forward current
F
IR = reverse current

Conductivity Modulation

The presence of minority carriers in a Schottky bar-

rier diode can also be observed on the forward I-V charac-

teristics of the diode. When the diode is conducting ac-

cording to the Schottky thermionic model, current can be

expressed as

I = Is Exp (qV/nkT) - (13)

where I is the reverse saturation current and n is the

s

diode ideality factor which equals 1.0 for an ideal diode.

If log I is plotted versus voltage in the forward

direction a straight line will be obtained. Any departure

from this line will show the diode's departure from theo-

retical performance. At higher currents (on the order of 10

mA), the diode's ohmic resistance starts contributing with

its IR drop. A departure to the right is observed as shown

 13

in Figure 7.

Ideal. Minority carrier

H
4-1
V7-
/
diode . injection
/

Series resistance

Figure 7. Effect of series resistance on

forward drop.

From this graph values for Rs, the diode series re-

sistance, can be calculated. Rs can also be observed on a

linear I-V curve tracer as that portion of the curve past

the knee where the trace assumes a straight line IR drop.

In an ordinary Schottky diode, as the current is in-

creased to several tens of milliamperes it is observed that

there is a decrease in the effective R. This is attrib-

uted to minority carrier injection which contributes to the

conduction, hence the name, conductivity modulation. Lep-

selter and Sze (11) and Zettler and Cowley (18) have re-

ported successful use of p-diffused guard rings at the

metal edge. These reduce early reverse voltage breakdown

and reverse leakage. In addition, the guard rings give

 14

ideal emission over eight decades of forward current to

1 mA. Unfortunately, the guard rings add undesirable

capacitance. Saltich and Clark (13) describe an n-p double

guard ring Schottky diode which reduces some of the un-

desirable effects of the p-guard ring and gives substantial

improvement in the forward direction. Whereas the single

p-guarding showed minority carrier injection at 35 mA, the

double guard ring diode showed none at the highest test

current of 50 mA.

From the I-V characteristics, then, it is possible to

evaluate the magnitude of minority carrier current, and the

level of current at which the minority carriers and their

associated stored charge will start to affect the perform-

ance of the diode as a detector. For a detector operating

in the microampere range there should be no problem with

stored charge.

Relaxation Time

One ultimate limiting mechanism to the Schottky bar-

rier diode switching speed is the relaxation time of the

semiconductor and the metal. The relaxation time is de-

fined as the time constant required for a crystal lattice

to return to normal equilibrium after being excited by

some disturbance (1). In the case of the metal in a

Schottky barrier, T, the relaxation time, would be the time

required for the "hot electron" to lose its thermal energy,

 15

so that it can not return over the potential barrier.

The relaxation time for any crystal is given by:

T = = Ep (14)
a
where:

= dielectric permittivity

a = dielectric conductivity

p = dielectric resistivity

For a metal T is about 10 -17 seconds. For silicon with

-12
E = 12(8.8 x 10 ) F/m and = 1 ohm cm (0.01 m), t is
-12
about 10 seconds. Hence, the semiconductor would have
the slower speed of the two materials. However, 1 ps is
not a severe restriction as compared to practical limita-

tions of realizable detector circuits with their parasitic

inductance and capacitance elements.

Parasitic Elements

So far, only the diode junction itself has been ana-

lyzed. Before an RF voltage wave can reach the junction,

it must travel through the diode structure. The diode

package is placed between the center conductor and the
outer wall of a 50 ohm coaxial line. The glass package of

the diode represents shunt capacitance on the diode, C in

Figure 8. Small leads within the package have inductance,

L . Inside the diode chip there are the ohmic contacts and
s

the semiconductor bulk resistance, lumped together as Rs.

 16

The barrier junction itself is a voltage variable capacitor,

C.. Figure 8, then, represents an equivalent circuit for

the diode.

Ls Rs

CP

Figure 8. Diode equivalent circuit.

R- is the voltage dependent junction resistance. For

smallvoltageswingsthedynamicresistanceofR.can be
found by differentiating the diode equation (Equation 12).

dV 26 -V
R. = dI
exp ( )

where:

nkT
= 26 for ideal diode at 300 K (15)

V is in millivolts

I is in amperes

Forward voltages much greater than 26 mV, this expression

simplifies to:
26
(16)
 17

where:
I is in milliamperes

When the diode detector is used for rectifying signals

several times larger than the barrier potential, the diode

is back biased most of the cycle. On peaks the forward

resistance can be determined from Equation (15). During
reverse biasing on a Schottky diode, Rj will be quite high.
For practical purposes it can be neglected.

The junction capacitance, Cj, is determined by the

equation:
C.
C = (17)
(1- )1/2
Vb

where:

C. = zero voltage junction capacitance

V = junction voltage

V = barrier potential
b

For the diode detector with its peak detection circuit, the
average voltage across the diode is approximately equal to
the do output voltage. Although Cj is not a constant, meas-
urements the dc bias voltage
could be used for predicting performance.

Perhaps the most significant parasitic element is that

found in the bypass capacitor. In the detector unit shown

in Figure 11, a high quality 1000 pF feed-through capacitor

was used. The lead between the diode body and the capacitor
 18

was the minimum length, about 3 mm, required by the small

socket. Even this small lead inductance combined with the

residual inductance of the capacitor was sufficient to give

a series resonant circuit at about 100 MHz. Consequently,

although the capacitor still acted as a low pass for dc

current, the equivalent circuit for the capacitor over the

operating range was a lossy inductor.

Figure 9 shows the high frequency equivalent circuit

for the detector.

Figure 9. Parasitic elements of diode detector.

neactualRFvoltage,e.] ,at the junction is recti-

fied and fed through the bypass capacitor. The reactance

of C is so low above resonance that for RF analysis it can

be neglected. Since Rc and Lc are common to the input

circuit only, for purposes of analysis they can be

 19

redrawn as shown in Figure 10.

Lc
L Rc Ls Rs
rign,vvvr1515P"snA,--
e1 cp Cj ej

o..----.

Figure 10. Ladder equivalent of detector.

 20

EXPERIMENTAL DETERMINATION OF DETECTOR ELEMENTS

The Diode Detector

The detector configuration used in this analysis is

shown in Figure 11. A 50 ohm transmission line environment

was chosen to optimize the accuracy of the high frequency

measurements and to reduce unwanted distributed impedance

effects in the diode circuit. The transmission line is the

standard 14 mm dimension with General Radio 874 connectors

on each end. This size line lends itself well to the

physical dimensions of a DO-7 diode package, and allows

space for a diode lead socket to be embedded in the center

conductor and on the bypass capacitor. The block forming

the outer conductor is cut somewhat above the center line

to give access to the center conductor.

This detector in the feed-through configuration per-

mits a precision termination to be attached to the output

port for impedance measurements. In addition, accurate

frequency response measurements can be made by placing a

power meter at the output port.

 21

Figure 11. Schottky diode detector in 50 ohm

transmission line environment.
 22

Diode Package and Junction Capacitance

In order to determine values for Cp and C., the zero

capacitance of the detector with the termination and the

diode removed was measured on a Boonton 75D capacitance

bridge. The diode was then inserted, and total capacitance

versus reverse voltage measured from 0 to -15 volts. Sub-

tracting the zero capacitance value gave total diode capac-

itance versus reverse voltage.

Equation (17) for junction capacitance can be rewrit-

ten in the form:

1 1 1
)17 + (18)
2 2
C
3(0)
Vb c
j(0)
.j

where:

V is positive for reverse voltage

If C. is the total diode capacitance, and C the pack-

age capacitance, Equation (18) becomes:

1 1 1
)v + (19)
2 2
(C - C ) C. Vb C
0)
t p 3 (0) j

2
Equation (18) is a linear equation when 1/C, is

plotted against V. Equation (19) will be a straight line

only when the correct number is chosen for Cp which will

make C. = Ct- Cp.

-2
Trial values were assumed for C and (C - C p ) was
t
tabulated versus voltage from 0 to 15 volts for each trial.

A linear regression was made for each line and the

 23

correlation coefficient calculated. The best fit was ob-

tained with C = 0.19 pF (correlation coefficient =

0.99995).

With the value of C determined, the graph, Figure 12,

of junction capacitance versus reverse voltage was plotted.

Diode Series Resistance

The diode series resistance was measured from the for-

ward I-V characteristics as shown in Figure 13. The de-

parture from theoretical performance not only evaluates Rs,

but also can indicate the presence of minority carrier

injection. The average value for Rs from this graph is 7.3

ohms. For forward currents up to 70 mA, Rs appeared to be

constant, indicating no minority carrier injection.

 pF

.5

.4
.

.3

.
0
4J

.2

.1 ---.

Package Capacitance = 0.19 pF

0
1 2 3 4 5 : 11 12 13 14 ' IS -Voif

Figure 12. Junction capacitance versus reverse voltage.

 25

IR drop due to Rs

10

10

.1 .2 .3 .4 .5 .6 .7 .8
Diode forward voltage, Volts
 26

Diode Series Inductance

The diode series inductance was measured using a

Hewlett-Packard network analyzer. The diode was measured

in the same electrical environment as it is used. A stub

center conductor was made which stopped at the diode loca-

tion as shown in Figure 14. This not only eliminated re-

flections from the end of the line, but also gave an accu-

rate open circuit reference for the network analyzer with

the diode removed.

Figure 15 is the impedance plot for the diode at zero

bias from 2.0 to 4.0 GHz. The trace follows an equivalent

series RLC line over the entire frequency range very well.

At 2.61 GHz the series circuit is purely resistive with

Rs = 6 ohms. Taking the zero capacitance value from Fig-

ure 12 as 0.435 pF, Ls calculates to be 8.6 nH.

 27

Figure 14. Detector with stub line.

 28

Figure 15. Diode impedance with V = 0.

Cathode grounded, stubrline,
2.0 to 4.0 GHz.
 29

Bypass Capacitor

The electrical characteristics of the feed-through

capacitor were also measured on the network analyzer. A

GR-874 to female N adaptor provided a convenient plug-in

fixture with which a coaxial connection to the capacitor

could be made. Figure 16 a and b show the RF impedance

plot of the capacitor.

Note the capacitor appears as a fairly high quality

inductor up to 1.0 GHz. At 2 GHz the plot loops through a

lossy resonance. Table 1 lists the values calculated for

the series equivalent circuit of the impedance plots of

Figure 16.

Table 1. Calculated Values for Capacitor Impedance

X L R
Frequency L c c

0.25 GHz 3.2 ohms 2.04 nH 0

0.50 8.0 2.54 0

1.0 17.5 2.78 2

2.0 31.0 2.47 20

3.0 34.0 1.81 12
4.0 43.0 1.71 21
Mean 2.6

The values for R c are not completely constant over the

frequency range of interest. However, since 3.0 GHz turned

out to be the maximum output for the detector, the most

accurate value for this frequency was chosen, Rc = 12 ohms.

 30

Figure 16a. Impedance plot of Erie feed-through

capacitor. 0.1 to 2.0 GHz.

Figure 16b. Impedance plot of Erie feed-through

capacitor. 2.0 to 4.0 GHz.
 31

COMPOSITE DETECTOR PERFORMANCE

Rectification Efficiency versus Frequency

One of the primary objectives for examining the mech-

anisms involved in the diode detector is to be able to

develop a model that will predict the rectification effi-

ciency versus frequency. Often it is more important to

know relative frequency response than the actual value for

rectification efficiency. After an accurate low frequency
ac voltage measurement has been made, this voltage level

can be extended to high frequencies with the detector of

known frequency response.

In order to accurately evaluate the model, it was

necessary to make a high frequency response measurement on

the detector. The output of the detector was fed through a

6 db attenuator to an HP 431C power meter. Originally the

input level to the power meter was held constant at a power

equivalent to 1.0 volt peak on the transmission line. The

disadvantage of having a constant incident voltage response

is that the changing dc output voltage affects the junction

capacitance, and the equivalent circuit is not constant.

By holding EDC = 1.0 V constant for the entire fre-

quency range, the variation in C. is avoided. The incident

voltage on the detector is allowed to vary and the equiva-

lent response from the power meter reading is calculated

from the relationship:
 32

V1
1 1
(20)
V2 P
2

Figure 17 shows the results of these frequency re-

sponse measurements. The predominant landmark on the curve

is the nearly six times increase in output voltage at 2.8

GHz. One wonders if this is simply a series resonant RLC

with Q 5.7.
EDC/ei
If a value for Ci(0) = 0.255 pF is used to find R and

L from Q = X /R and L = 1/w2C, the values in Figure 18 are

c
obtained.

L R
TIRPwv-
12.7 nH 34.8
ei ej
CJ
0.255pF

Figure 18. RLC equivalent to give e.3 /e..

The voltage transfer function for this circuit is:

- 2
m 9
= 1 + ((It- ) - 2) + 41 (21)
e.
0 Q
2
0j
6 I T 1 T T-T T T -1-.1
Calculated peak at 8.3

I.- -3 dB
4

DC out held constant at 1.0 volt

100 Megohm dc load
o Measured response
)( Calculated from RLC equivalent
2
ar Calculated from ladder equivalent

1 0

L 1 1 L_Ifill 1 1 I 1 I 1 I 1

10 100 1000 NMZ

Figure 17. Frequency response of diode detector.
 34

Several points were calculated from this equation and

plotted on the measured response as an "X" in Figure 17.

These points fit the curve very well over the whole range.

Note, however, that the actual frequency response measure-

ments show some aberrations between 1 and 2 GHz. These

were probably caused by imperfections in the bypass capa-

citor which were observed also in Figure 16.

Having examined the voltage transfer model, the im-

pedance measurements will now be analyzed to see if they

agree with this model.

Network Analyzer Measurements

If frequency response is the only result desired, Fig-

ure 18 is a good model. It does not tell the whole story,

however. First, the 3db points in Figure 17 give an equiv-

alent Q of 7.4. This indicates a higher order function.

Also, the input impedance for the detector calculated from

this model does not agree well with measurements.

Figure 19 shows the impedance plot of the diode in

series with the bypass capacitor. Figure 20 is an equiv-

alent circuit for this combined impedance which fits very

well over the 2 to 3 GHz range.

 35

Figure 19a. Series diode and capacitor impedance

using stub line. V r = 0, 0.1 to
2.0 GHz.

Figure 19b. Same as above, 2.0 to 4.0 GHz.

 36

22 6.8nH

O To.396pF

Figure 20. Equivalent input impedance circuit.

Note that the impedance equivalent circuit is not the

same as the e.3 /ei circuit of Figure 18.

A complete model which can predict both impedance and

rectification efficiency flatness is the ladder network of

Figure 10. Figure 21 is this circuit with the various ele-

ments shown as measured.

r - - -
1 R c

12 2.6nH1 1 7.4 8.6nH

L 1

e. C 1-- C. I e
P
0.19pF;
T 0.225pF
a O

L_ _ _

Figure 21. Equivalent ladder circuit.

 37
2
Figure 22 shows the network analyzer plot of the de-

tector terminated in 50 ohms. The impedance values shown

in Figure 19, when paralleled with 50 ohms, yield the val-

ues measured in Figure 22.

ECAP Results

To check the validity of the composite model, a Tym-

share computer program for AC analysis was run on the cir-

cuit of Figure 21. The results of this analysis are shown

in Table 2 in a comparison with measured data. The calcu-

lated values for e.3 /e. are shown as " " points on the
I

curve in Figure 17.

2
Impedance measurements of the detector were made with
the network analyzer which has an incident signal in the
neighborhood of 50 mv. External bias was used to set
E DC = 1.0V. To verify that this was equivalent to the
input impedance with self bias equal to one volt, a slotted
line measurement was made with ei = 1 volt. The two
measurements, taken at 1 GHz, agreed within 5% of each
other, well within system error limits.
 38

Figure 22a. Impedance plot of diode detector.

1.0 volt bias, 0.2 to 1.0 GHz.

Figure 22b. Same as above, 1.0 to 2.0 GHz.

 39

Figure 22c. Impedance plot of diode detector.

1.0 volt bias, 2.0 to 4.0 GHz.
 40

Table 2. Calculated and Measured Values for

Detector

Freq. e.3 /e.

1
Normalized Input Impedance

GHz ECAP Measured ECAP Measured

0.1 1.001 1.00 1.000 -j0.014 1.0 -j0.01

1.0 1.148 1.16 0.970 -j0.149 1.0 -j0.15
2.0 1.982 1.64 0.784 -j0.340 0.75-j0.38
2.6 4.806 3.80 0.409 -j0.294 0.35-j0.30
2.8 8.107 5.70 0.293 -j0.126 0.28-j0.15
3.0 8.304 4.31 0.292 -j0.095 0.30-j0.08
4.0 1.440 0.74 0.858 -j0.259 0.85-j0.38

The calculated and measured values for e./e. in Table

2 show good agreement for predicting the resonant frequency

for the network. The computed output voltage is generally

higher than the measured value. The ECAP program predicted

a peak of 8.3 whereas 5.7 was the maximum measured. This

could be due to some error in the measurement of R which

s

made the diode look less lossy than it actually was. One

wonders if perhaps there is some stored charge reducing the

rectification efficiency slightly. This seems unlikely
since it was not observed at all on forward I-V measure-
ments.

Comparing calculated input impedance with measured

values shows agreement to about the accuracy limit of the

network analyzer. At 100 MHz both methods give Zi = 50

ohms with about 1(,'/9 capacitive reactance. At 1.0 GHz

 41

the reactance is up to about 15%, but resistance is still

within 1%. Both real and imaginary parts have close

agreement up to 4 GHz. Note that ECAP yielded a purely

resistive input between 2.8 and 3.0 GHz. The analyzer
measured zero phase shift at 2.92 GHz.
 42
SUMMARY AND CONCLUSIONS

This paper has examined the rectification process in-

volved in a detector using a Schottky barrier diode. Since

stored charge is essentially negligible in a Schottky

diode, the height of the barrier potential is the most sig-

nificant parameter in determining rectification efficiency.

It was shown that the presence of minority carriers,

which would degrade the high frequency performance, can be

observed on the logarithmic I-V plot of the DC forward

characteristics of the diode. The TI Schottky diode used

in the experiment did not show evidence of conductivity

modulation even at a forward current of 70 mA. Relaxation

time for the silicon crystal in the Schottky diode is on

-12
the order of 10 seconds. This represents an insig-

nificant limitation when operating in the gigahertz fre-

quency range.

Of all the high frequency limitations discussed here,

it is obvious that the parasitic elements play the largest
role in limiting frequency response. The equivalent cir-
cuit for the detector was shown to be a two-section LRC

ladder which had a peak resonance output at 2.8 GHz. The

lead inductance of the diode and the internal inductance

of the bypass capacitor are primarily responsible for this

peaking.

If improvements are to be made, it should be in

 43

reducing these parasitic effects. The package capacitance

could be designed into the transmission line, and diode

packages with lower inductance are available. Finally,

the bypass capacitor should have as little inductance as

possible. However, capacitors with sufficient size for

the low frequency range are likely to have inductance when

used at microwave frequencies.

By using the equivalent circuit developed with the

methods presented here, it is possible to predict the fre-

quency response of a precision detector. Design improve-

ments then can be made to the detector or corrections

calculated for its response.

 44
BIBLIOGRAPHY

1. Alder, R. B., A. C. Smith and R. L. Longini. Intro-

duction to semiconductor physics. New York, John
Wiley, 1964. 247p.

2. Cowley, A. M. and H. 0. Sorensen. Quantitative com-

parison of solid state microwave detectors. IEEE
Transaction on Microwave Theory and Techniques
MTT-14 (12):588-602. 1966
3. Cowley, A. M. Titanium-silicon Schottky barrier
diodes. Solid State Electronics 12:403-414. 1970
4. Driver, L. D. and M. Gerald Arthur. A wideband RF
voltmeter comparator. I.R.E. Transactions on Instru-
mentation and Measurement IM-17 (Z):146-150. 1968
5. Gray, Paul E. et al. Physical electronics and circuit
models of transistors. New York, John Wiley, 1964.
262p.

6. Hewlett-Packard. The hot carrier diode theory, design

and application. Application note 907. Palo Alto,
California. 15p.

7. Hosking, M. W. Coaxial wide band detector type 6060.

Marconi Instrumentation 12 (3):45-48. 1969

8. Hunter, Lloyd P. Handbook of semiconductor electron-

ics. 3d ed. New York, McGraw-Hill, 1970. 21
Sections.

9. Krakauer, Stewart M. and S. W. Soshea. Hot carrier

diodes switch in picoseconds. Electronics 36:53-55.
July 19, 1963.

10. Krakauer, Stewart M. Harmonic generation, rectifica-

tion, and lifetime evaluation with step recovery
diode. Proceedings of the I.R.E. 50 (7):1665-1676.
1962

11. Lepselter, M. P. and S. M. Sze. Silicon Schottky

barrier diode with near-ideal I-V characteristics.
The Bell System Technical Journal 47 (2):195-208.
1968

12. Oudrejka, A. R. and P. A. Hudson. Measurement stand-

ards for low and medium peak pulse voltages. Journal
of Research of National Bureau of Standards
70C (1):13-18. 1966
 45

13. Saltich, J. L. and L. E. Clark. Use of a double dif-

fused guard ring to obtain near ideal I-V character-
istics in Schottky barrier diodes. Solid State
Electronics 13:857-863. 1970

14. Shive, John N. The properties and physics of semi-

conductor devices. Princeton, Van Nostrand, 1959.
487p.

15. Sorenson, Hans 0. Using the hot carrier diode as a

detector. Hewlett-Packard Journal 17 (4):2-5. 1965

16. Yu, A. Y. C. The metal-semiconductor contact: an

old device with a new future. IEEE Spectrum
7 (3):83-89. March 1970

17. Yu, A. Y. C. and E. H. Snow. Minority carrier injec-

tion of metal-silicon contacts. Solid State
Electronics 12:155-160. 1969

18. Zettler, R. A. and A. M. Cowley. P-n junction

Schottky barrier diode. IEEE Transactions on
Electron Devices Ed-16 (1):58-63. 1969
APPENDIX
 46

APPENDIX

Steady-state Equilibrium Equations for Rectifier

Continuity of stored charge for steady-state equilib-

rium (1) is:

dQ Q
dt T d

where:

i = diode current
d
E
= 2 sin (cult © sin 8)
Rc
r-
-1 E + 7n i
= sin
P j

Solution to charge equation:

Ep sin 8 w1 (cot 8
Q + 1 sin cult +
R w 2 pr.

c 1
1 + T L\

( 1
- cot 0 cos Cult +
1T J

1
(cot 8 + co T)e- wit/tali - (77-- + w T)
1 1
1T

Equilibrium conditions:

1. Q = 0 at w1T = 8 +

2. Capacitor added charge per cycle = diode

recombination charge = charge lost through RL

 47

Parameter relationships:

2 2 (1-E-4/(1)1T)
W T sin 4) - cos 4) + T + 1
1 1
cot 0 =
(I)/W T
1
sin 4) - T cos (0 tali E
1

Rc co
1 1T Cot 8+1)
Rd = 2 11 (031 2 2+1 (1-cos 8) + 1 - cot 8 sin 4) +
T w1T 1T

-(0/WiT
1
w1T (cot 0 + culT) (1 -E ) W,J- T)
T
1