1938

IEEE

TRANSACTIONS

ON

MICROWAVE

THEORY

AND

TECHNIQUES,

VOL.

38,

N0.

12,

DECEMBER

1990

Establishing the Minimum Reverse Bias for a p-i-n Diode in a High-Power Switch
ROBERT H. CAVERLY, MEMBER, IEEE, AND GERALD HILLER, MEMBER, IEEE

Abstract —Au p-i-n diode bias reverse voltage, paper the sion

important Until

circuit is the

design

parameter circuit

in

a high-power applied dc has This that diode’s expresselecting voltage parameter

performance ability must

parameters

such as loss, distortion,

and reli-

application voltage. either

selection using

of an appropriate the magnitude a possible for a p-i-n

not be compromised.

It is this conditionally than 1 kW) The applied

now, this

important

been chosen

conservatively, trials

of the peak RF lower value. diode operating

or by empirical explores

to determine

safe region where most high-power (greater p-i-n diode switches are designed to operate.

the reverse reverse dc voltage bias

bias requirement voltage similar voltage value for

in a high-power minimum for this self-geuerated verified setting.

RF and microwave under

environment. is equivalent is developed the applied

It demonstrates to the and reverse p-i-n A concise

dc reverse bias voltage must be large enough to prevent excessive conduction during the positive portion of the RF signal. If excessive conduction does occur, the p-i-n diode loss will failure. lines, equal increase and the diode will be subject design in to

RF conditions.

self-generated minimum

experimentally bias

-and will

assist the design

engineer

in more accurately

an appropriate

In the absence of any theory the design to the engineer RF peak voltage,

or analytic resulting

guide-

may choose

a dc bias voltage extremely frequently, howmatching a p-i-n presents a guide

I.

INTRODUCTION property large radio much lower are design current requirements of a p-i-n diode is its

conservative and costly designs; more ever, the design is based on empirically diode to an available voltage. This paper

ability to control microwave signals with and voltage. minimum ohmic no existing While there level

A

FUNDAMENTAL

frequency (RF) and values of dc current rules for selecting based on allowable are of [1], [2], there a

for the p-i-n diode circuit designer, similar to the forward bias case, for selecting a minimum applied reverse bias voltage based on diode and circuit operating parameters. The investigation requirement of p-i-n circuit diode conditions, of the relationship under zero of the reverse bias observations bias open applied was prompted distortion by experimental

of forward

loss and distortion

design rules on which

to base the selection

the minimum level of applied dc reverse bias voltage. The instantaneous voltage across the p-i-n diode (both RF and dc) must never exceed its avalanche breakdown voltage ( VBR in Fig. l(a)), where high densities may cause p-i-n diode failure. reverse current Safe operation

using a test circuit the effective voltmeter

of the type shown in in the voltmeter internal resistance

Fig. 2. The

109 Q resistor

was inserted

line to increase

will result if the instantaneous voltage never forces the p-i-n diode into forward conduction or into avalanche breakdown (Fig. l(b)). However, this requires that the applied dc voltage be at least equal to the peak RF voltage and that the breakdown voltage be at least twice the peak RF voltage (VR~). In many applications, high applied reverse bias voltages are often not available or are too expensive to implement. Frequently, p-i-n diodes are operated in the so-called conditionally safe region, where an instantaneous excursion of voltage into the forward conducting region may be tolerated (Fig. l(c)). If a dc reverse bias voltage in this region is chosen, circuit
Manuscript received March 30, 1990; revised August 13, 1990. This work was supported by the Semiconductor Products Division of M/ACOM, Inc.. Burlington, MA. R. H. Caverly is with the Department of Electrical and Computer Engineering, Southeastern Massachusetts Univers@, North Dartmouth, MA 02747. G. Hiller N with the Semiconductor Products Division, M/A-COM, Inc., South Avenue, Burlington, MA 01803. IEEE Log Number 9039361.

from its nominal 107 0 value to better approximate an open circuit across the diode. The voltage read was then approximately 1Yo of the diode voltage. A self-generated its high-impedance reverse bias dc voltage was developed in priand of an The magniacross the p-i-n diode that allowed state with the diode to operate

good stability.

tude of the self-generated dc voltage was influenced marily by the peak RF voltage level, the frequency, the thickness of the i region. Upon application

equivalent externally applied dc bias, the distortion generated was identical to the self-generated dc voltage. However, when the applied dc voltage was lower than the self-generated voltage, unstable performance would occur, manifested by large increases in distortion signals and by heating of the p-i-n diode, which often led to device failure. Published experimental results by other investigators [3] show similar device and circuit parameters that affect the degree of forward conduction in the p-i-n diode. An analysis of the p-i-n diode leading to a concise expression for the safe minimum operating dc reverse bias voltage is presented. The expression indicates how the

0018-9480/90

/1200-

1938$ 01.00

01990

IEEE

CAVERLY

AND

HILLER:

ESTABLISHING

THE

MINIMuM

REVERSE

BIAS

FOR

p-i-n DIODE

1939

I

I

VDC I L

‘PEAK ‘BR

~
I

I

J 1‘BR

‘DC

1
q

‘PEAK

—v

1
v

UNSAFE BIAS VDC + IVRFI > VBR

I
SAFE BIAS VDC + VRF e VBR VDC > VRF

(a)

(b)

‘DC

‘PEAK ~
‘BR

1;

1’
v > -.

I
I CONDITIONALLY SAFE BIAS VDC + VRF < VBR VDC < VRF (c) Fig. 1. (a) Unsafe operating region, (b) safe operating region, and (c) conditionally biased p-i-n diode. safe operating region for the reverse.

1940

IEEE

TRANSACTIONS

ON MICROWAVE

THEORY

AND

TECHNIQUES,

VOL.

38, NO. 12, DECEMBER

1990

R = 1000 M

ohm

the i-region positive the p-i-n

stored

charge is never established never achieved.

during

the

portion diode

of the RF signal. The conducting is therefore

state of

.,

Fr’””1

‘R=-’’=!G dR=’OMOhm
Between region for
)

these two limiting cases lies an operating the p-i-n diode where prevention of forward

conduction of the diode may be achieved by reverse biasing the diode at a voltage less than the peak RF voltage. Experiments [7] have shown that the dc voltage developed by the p-i-n diode in its open-circuit, zero bias state is directly to prevent analysis in p-i-n diode. of the developed diodes open-circuit, flowing zero bias of both through (1) is based on the existence currents related to the dc reverse bias voltage in the needed p-i-n The voltage significant forward conduction

SPECTRUM ANALYZER

I

conduction and displacement the p-i-n diode [8]:
dc voltage

Fig.

2.

Test set used for measurement of the self-generated of p-i-n diodes.

J(t) where carrier density elemental charge velocities),

= 2rrq~ + edE/dt

J(t)is the total
density (where of electrons electronic
q

p-i-n

diode

i region factor, the of

and circuit and minimum is verified peak

parameters RF using voltage bias

such as freaffect the The meavoltage.

current density, n is the i-region the density of holes equals the the hole analysis), q is the of the drift u is the drift equal permittivity, velocity and

quency, magnitude derived

duty

to simplify charge,

reverse

expression

experimental

carriers

(assuming

and electron

surements of the developed open-circuit zero bias voltage across a variety of p-i-n diode geometries and circuit operating conditions. An application example is also presented. 11. In the presence an ideal rectifier the state. diode RF voltage In most ANALYSIS RF or microwave conduct its forward diode into signal, when bias p-i-n can biased

is the dielectric

E is the

electric field. Assuming that the flux density through the diode’s cross section (A) yields

D is uniform

@dA=bdv=DA ‘QA’E
charge as J(t) density. Substituting current the results J(t), of (2) into shows that the total density,

(2)

of an applied places the

where Q is the sum of both the RF and dc components of the i-region stored charge and p is the total i-region (1) may be written

diode will instantaneously diode where applications avoided

a reverse-biased in the p-i-n is reverse

is used, forward

conduction

=2nqu

+(1/A)

dQ/dt. W then,
using

be unconditionally

if the diode

(3)
(3),

such that the peak RF voltage swing remains between the device turn-on voltage (typically 0.7 V for silicon diodes) and the reverse bias breakdown voltage. This requirement is the basis for a conservative “worst case” p-i-n diode dc reverse bias point, where IV~cl =

A p-i-n diode with i-region thickness has a total diode current of

Z= I~~+Z~c=Q/T+dQ/dt
where the i-region transit time,

(4)
as W/2 u [8].

IVRFI.
p-i-n diode time as the starts time is

T, is defined

In real circuit applications, however, the does not show the same instantaneous turn-on ideal rectifier diode. Rather, tive for a finite amount of time before

If a time variation of the form ejat for all quantities is assumed, the RF component of the total diode current may be written as

the RF signal must be posithe diode This turn-on

1~~ = Q~~(l+jcoT)\T
and the dc component written as of the total diode current

(5)
may be

conducting

in the forward

direction.

the time required for the i region to fill with charge carriers (both holes and electrons from the heavily doped end regions) during the forward cycle of the applied RF signal. Depending on the frequency, the peak RF voltage level (1V~~l), and the i-region thickness (W), the excursion into the forward direction of the RF signal may be too short to allow pletely traverse will then dependent sufficient time for the carriers the i region, hence preventing its conducting state [4]-[6]. to comthe p-i-n The diode

Idc = Qd. /T.

(6)

The resistance (or conductance) of the p-i-n diode can be computed from the amount of stored charge in the i region [3]. The RF and dc voltages are determined from these resistances and the corresponding currents:

diode from entering

act as a Iossy capacitor on the i-region thickness

with its capacitance and p-i-n diode cross

and

section. In this case no externally applied dc reverse bias voltage is necessary to prevent forward conduction since

‘RF”KIRF=(W+ ‘7a) ‘dc=ba’dc=(a
(7b)

cAvERLy

AND

HILLER:

ESTABLISHING

THE

MINIMUM

REVERSE

BIAS

FOR

p-i-n DIODE

1941

where electron

w is the mobility mobility

(again,

assuming

equal

hole and of (7) using

A

further

simplification

may be used

on (12)

if

the

values for simplicity). for the transit time

The magnitude as

applied

RF voltage

is low enough

so that velocity

satura-

the ratio

of these two voltages

may be written

tion does not occur:

and the definition

v~c

‘“ = [1+(7;,.
The velocity, transit time,

(8)
)2]’” ‘ drift In

‘v””=
At low frequency the developed

Iv’~]

[l+(o.0285fMHzw:i,
and/or

\vRF@)’]’/2” ‘“3)
with thin i regions, the peak RF of a pn junction i regions, holw-

T, is a function
a function

of the carrier

u, and, in turn,

of the electric

field.

for diodes

semiconductors such as silicon, the carrier velocity increases approximately linearly with applied electric field for only relatively low values of electric field (Fig. 1). At higher electric field values, the interaction of the charge carriers with the semiconductor lattice atoms serves to limit the velocity of the carriers to a value termed the saturation velocity, type, semiconductor approximately velocity which [9]. An approximation includes as [10] may be written U,.t. This velocity varies with material, and temperature, for electrons the effects in silicon field of the electric carrier but is on

dc voltage,

IV~cl, approaches

voltage, approximating the ideal behavior diode. For higher frequencies or thicker

ever, the dc voltage developed decreases as l/~ for a given RF voltage. The developed dc voltage will then decrease by increasing the i-region thickness. Increasing the i-region thickness effectively increases the transit titne proportionally since at high RF voltages the carrier vek]city is limited of the applied to its saturation of the ratio RF electric value, field (E= us~l. Fig. 3 illustrates 0.4751 VR~l~/ the dependence IVdc1/ IVR~I on the rms value

107 cm/s

at 290 K saturation

dependence

of velocity

W)

@=[l+ /*]
where The operating the term p-i-n and I-L is the low-field is often is able to used handle carrier in mobility. diode high-pulsed-power kilowatts applications

‘9)
of power field

with the factor m 2/ D1/2 as a parameter. A significant reduction in the developed dc voltage is indicated for low peak RF electric fields (low applied IVR~l), high-frequency

operation, large i-region widths (W), and short duty cycles. At large RF voltages, the dc voltage developed approaches the limiting factor

Ivdcl =

in this mode. The rms value of the electric

under these conditions is needed to estimate the carrier drift velocity (9) and may be computed assuming a halfwave-rectified pulsed RF waveform where the carrier frequency (~) is much larger than the inverse of either (T~). Assumthe pulse period (TP) or the pulse duration ing that the electric value of the electric field field, across the i region of the p-i-n as diode may be approximated by E = VR~/ W, then the rms III. Experimental voltage (reverse arm.
/r I
11 I

m
7TJ’W

lv’~1

( 114)
2“

1+

— v sat

EXPERIMENTAL RESULTS measurements of the dc self-generated to tlhat arm up to had in each the isolating diode levels

were

performed diode) were

using

a test set similar

shown in Fig. 2. This test set simulates bias p-i-n Measurements made diode

E, may be calculated

of a SP2T p-i-n at power specimens 1 to 60 MHz

(lo)
where D = T~ / TP is the RF pulse duty cycle. The dc voltage developed across the p-i-n diode may be written using (8), (9), and (10) as

100 W at. frequencies from

from

and duty factors selected

0.07 to 1.0. The p-i-n

i-region widths ranging from 50 to 200 ~m and encc~mpassed various cross sections and carrier lifetime values. Fig. 4 graphically illustrates a comparison between experimental measurements of the ratio of the peak RF voltage

Iv”.1 = [[
Equation (11) may be simplified Yalue of u~~t: 1+

IV’F1
Tjw’/o.95@’@ [ 1 + ~1+
mobility

[0.95~v~’@/w&]2]]
of p = 0.15 cm2/V-s

2]0”5

o
mentioned

if one assumes a carrier

and the previously

Ivdcl = [[
1+ 0.0142~~”zW~,l/V’’@[l+

IvRFl
~ 1+ [ 0.056VR,@/

( 12)
W~,,]2]]2]o”5 “

1942

IEEE

TRANSACTIONS

ON MICROWAVE

THEORY

AND

TECHNIQUES,

VOL.

38, NO. 12, DECEMBER

1990

1000 7

DF = 0.01

1
PEAK RF F
w2 /D112 MIL

1

VOLTAGE
---

I
2

1

I
4

I

I
6

I

I
8

I

1
10 i-region

0

MHz

—.----Fig. 3. Graph of themagnitude of thepeak the magnitude as a parameter. RF voltage with

1:0
1000 10000 dcvoltage versus the term ~~~ZW~,l/I@ Fig. 5. Calculated thickness

I-REGION

THICKNESS

(MILS)

generated dc reverse bias voltage versus and duty cycle for a 1 kW signal at 1 GHz.

of theself-generated

100

previously reported by the authors [11]. It was also observed that the time it took the self-generated voltage to reach it final value was virtually instantaneous upon the initial application of RF power. However, when the RF power was changed, as several seconds there until appeared the a time lag of as much dc voltage value. self-generated

10 $’ o >< 1 + ++ THEORY 0, ~ 0.1 ++ + + +

stabilized

and reached

its new final

IV.
+ EXPERIMENT

APPLICATIONS EXAMPLE

1

10 100 F ~HzIV&lL / VRF D1/2 and theoretical values

1000

As a result of this study, the designer may now analytically determine the minimum dc reverse bias requirement for a p-i-n diode in a high-power RF environment. It also affirms that the value of the dc voltage may be significantly lower than the peak RF voltage. a p-i-n with The following signal
ratio

Fig.

4.

Comparison

of measured

of the

lJ’’acl/lvdcl.

example

will illustrate

this. Consider at 1 GHz

diode operata 1 kW

ing as a switch

element

(lV~~l = 316.2 V across 50 Q). Fig. 5 shows the calculated to self-generated dc voltage (lv~~l/lv~Cl) and the expresvalue (using (12)) of the generated reverse bias voltage as a function of i-region thickness and RF duty cycle. For p-i-n diodes with i-region thicknesses less mately 1 mil (25 ~m), an external dc reverse the peak RF voltage is required, regardless This reverse bias requirement is relaxed, thicker p-i-n diodes 10 mil (250 ~m) to keep within under p-i-n
verse bias the applied at disdc thinner p-i-n will

sion for the ratio indicated by (12). A significant experimental observation indicates that the p-i-n diode may be operated at high RF power without applying any external dc bias. The p-i-n diode here is operating in a zero bias open-circuit mode where the self-generated ate in the voltage open-circuit becomes mode, the reverse bias. To operany external resistance higher than in this mode

that approxibias equal to of duty cycle. however, for so that a

and low RF duty cycles (D) requires safe reverse shows that, require
appear however, forward

diode

at most 35 V reverse bias bias region in general, lower
preferable that resistance thicker and

the conditionally cycle. The figure thicker
would should also have speed.

across the diode

must be very high, generally distortion
the carrier. dc dc When often

any duty diodes
voltages diodes. diodes may at

108 Q. Typical levels of harmonic were approximately 20 dB below
Upon identical tortion reverse higher proved the ated would application values measured bias than was the “of the was increased self-generated by as much was of an self-generated identical. slightly,

with
and It

i regions
therefore be noted, higher

dc reto

external

reverse voltage, the

switch

a slower

as little the distortion

as 10 V imlf

voltage, as 60 dB lower would to p-i-n

significantly, reverse the often of

below the

carrier. self-generand failure. the

V.

CONCLUSIONS

applied

voltage distortion leading bias with

that degrade diode on

dc voltage, increase,

loss The re-

In many situations, the design engineer is unsure of the exact RF voltage stress on the p-i-n diode or its i-region width. The experimental method described in this paper may also be used to measure the self-generated dc voltage. This value may then be considered as the minimum

dependence verse bias

reverse

distortion observations

increasing experiments

is consistent

and

CAVERLY

AND

HILLER:

ESTABLISHING

THE

MINIMUM

REVERSE

BIAS

FOR

p-i-n DIODE

1943

applied voltage

reverse bias voltage safe” established

that will bias the diode in the

[10] [11]

“conditionally

region. In many situations the dc must be able to support RF signals may increase to a

where the SWR of the load termination

Semiconductor Theory and Device NJ: Prentice Hall, 1989. “Distortion in Microwave and RF R. Caverly and G. Hiller, switches by reverse biased PIN diodes,” in 1989 IEEE MTT-S Int. Microwaue Symp. Dis., June 1989, pp. 1073-1076. S. Wang, Fundamentals Physics. Englewood Cliffs,

of

large value. Under this situation, the RF voltage will be double that of a perfectly matched circuit. In applying (12) or (13) the inserted value of RF voltage the peak value at maximum voltage stress. Another tionally devices. vices below benefit safe may the be region Depending avalanche of biasing may on the a p-i-n desired diode in the level be used must reflect
E13

in the condiscreening the significantly the of deRobert H. Caverly (S’80-M’82) was born in Cincinnati, OH. He received the M. S.E.E. and B. S.E.E. degrees from North Carolina State University, Raleigh, in 1978 and 1976, respectively. He received the Ph.D. degree in electrical engineering from The Johns Hopkins University, Baltimore, MD, in 1983. Dr. Caverly has been employed at Southeastern Massachusetts University since 1983 where he is an Associate Professor. He has also bean a consultant for M/A-COM during that time. In 1990, he was a Visiting Research Fellow with the Microwave Solid State Group at the University of Leeds, in the United Kingdom. In 1987, Dr. Caverly was awarded the Dow Chemical Outstanding Young Faculty Award by the American Society of Engineering Education for his educational activities. In 1985, he was appointed Director of the University’s Computer Aided Engineering Laboratory, and he is currently involved with the Massachusetts Microelectronics Center. In addition to his interest in computer-aided engineering in microelectronics, he is interested in the application of computer software to microwave engineering problems. Dr. Caverly’s current efforts are involved with cha~racterizing p-i-n diodes and MESFET’S in the RF and microwave control

application,

screened

at a voltage

breakdown

voltage,

increasing

reliability and lowering the cost of p-i-n diodes. It would be a valuable contribution if a circuit technique could be devised that would make it possible, by applying only a small incremental dc reverse voltage to the self-generated voltage, to obtain distortion performance similar to that obtained when applying the full dc reverse voltage. This would significantly simpli~ the design and lower drivers now phase shifters. ACKNOWLEDGMENT The ronic authors wish to thank the and Electrical Engineering Group of the assistance during Department of Electand the Microwave of Leeds, Leeds, of this the preparation the cost of the higher used in higher voltage power p-i-n switches diode and being

Solid-State

University

U. K., for their manuscript.

REFERENCES [1] [2] G. Hiller, “Design with PIN diodes,” RF Design, vol. 2, no. 2, FH

[3]

[4]

[5]

[61

[7]

[8]

Mar./Apr. 1979. R. Caverly and G. Hiller, “Distortion in p-i-n diode control circuits,” IEEE Trans. Microwave Theory Tech., vol. MTT-35, pp. 492-501, May 1987. M. Caulton, A. Rosen, P. Stabile, and A. Gombar, “p-i-n diodes for low-frequency, high-power switching applications,” IEEE Trans. Microwave Theory Tech., vol. MTT-30, pp. 876–881, June 1982. A. Ward, “Calculations of second breakdown in silicon diodes,” Harry Diamond Laboratories Tech. Rpt. HDL-TR-1978, Aug. 1982. A. Ward, “Calculations of high-current characteristics of silicon diodes at microwave frequencies,” Harry Diamond Lab. Tech. Rpt. HDL-TR-2057, Oct. 1984. A. Ward, J. Deppe, and R. Garver, “Spike leakage, limiting and rectification in silicon PIN diodes,” presented at High Power Microwave Tech. for Defense Appl. Conf., Dec. 1986. G. Hiller and R. Caverly, “The reverse bias requirement for PIN diodes in high power switches and phase shifters,” in 1990 IEEE MTT-S Znt. Microwaue Symp. Dig., May 1990. G. Lucovsky, R. Schwarz, and R. Emmons, “Transit time considerations in p-i-n diodes,” J. Appl. Phys., vol. 35, no. 3, pt. 1, p. 622, Mar. 1964. S. M. Sze, Physics Interscience, 1981,

[9]

of

Semiconductor

Devices.

New York:

Wiley-

Gerald Hiller (S’57-M’58) was born in New York City, NY. He received the B.E.E. degree from the City College of New York in 1958 and the M. S.E.E. degree from the University of Pemlsylvania in 1963. From 1958 to 1962, he was employed by Philco Corporation, Philadelphia, PA, where he developed microwave circuits using semiconductor devices focusing on mixer diodes and millimeter receivers. From 1962 to 1972, he was with the Micro State Electronics Operation of Raytheon Company, Murray Hill, NJ, where he was involved with control circuits, amplifiers, and multipliers and served as Semiconductor Applications Manaf!er. From 1972 to 1984. he served as Director of Microwave Engin~ering at Unitrode Corporation, Watertown, MA, where he was active in the development of Unitrode’s p-i-n diode product line. In this capacity, he wrote many technical papers pertaining to the use of p-i-n diodes at RF and UHF. Since 1984, Mr. Hiller has been Manager of Semiconductor Applications at the Semiconductor Products Divisicm of M/A-COM, Inc., Burlington, MA.