- Email: [email protected]
Reflection type microwave modulators
NOTES
induced by the scanning beam goes through a minimum as the boundary is traversed. We have confirmed that these spikes and the dark structure seen in the p-regions follow the mosaic structure by subsequent etching experiments which revealed this structure by surface markings. In addition these experiments :
665
Current experiments in this laboratory the latter approach is feasible.
Acknowledgement-This note is published by permission of the Ministry of Defence.
D. A. SHAW K. A. HUGHES
(u) enabled the spikes to be studied in depth; (b) revealed that twin boundaries enhanced diffusion; (c) established that this structure unconnected with surface scratches.
also lead to is completely
We suggest that these observations indicate that current laser crystals contain small volumes (linear size N 10 to 200~) which are centred on the p-n junction plane and in which the zinc concentration varies in a cellular like manner. The very existence of such a structure raises the question as to whether it can effect laser performance. In our very small sample the poorer laser material has a more extensive structure than the better material. In support we have the observation that in the ten faces of good lasers we found only three diffusion spikes, although there were probably more in the bulk. If we accept this correlation we have a natural explanation for “filament” formation. Let us assume that the spike regions are “dead due possibly to excessive absorption, spots”, scattering or poor injection properties. Then, the greater the density of such regions, the less chance of wide active filaments, because each operating mode has to pass between dead spots or be highly attenuated in traversing them. If we have small spike regions then minor changes in filament pattern can occur as the current is varied. This idea has the merit of simplicity. It is applicable to the other kinds of function irregularities previously observed,@*@ such as those due to variations in doping level in the starting crystal or due to clusters of defects. The relationship between any of these effects and laser performances can be studied by photographing the filament positions and then examining the whole of the function plane by these techniques with the electron beam directed along the normal to the junction plane. This is possible either on lasers with shallow diffusions (4 to 5~) or on lasers with deep diffusions which have been subsequently etched down.
show that
N.
F. B. NJZVE
D.
v.
SULWAY
P. R. THORNTON
Dept. of Materials Science Uniwrsity College of North Wales Bangor, U.K.
C. GOOCH
S.E.R.L., Bala’ock Herts., U.K. References
1. R. F. BLOOM, C. H. GOOCH, C. HILSUM and D. J. OLIVER, Nature, Land. 198, 368 (1963). 2. T. E. EVERHARTand 0. C. OATLEY, J. Electron. 2,568 (1957). 3. T. E. EVERHART,0. C. WELLS and R. K. MATTA, J. Electrochem. Sot. f 11, 929 (1964). 4. I. G. DAVIES, K. A. HUGHES, D. V. SULWAY and P. R. THORNTON, Solid-State Electron. 9,275 (1966). 5. J. C. MARINANCE, J. Electrochem. Sot. 110, 1153 (1963). 6. M. H. PILKUHN and H. RUPPREZCHT,Trans. Met. Sot. A.I.M.E. 230, 296 (1964).
Solid-State Electronics Pergamon Press 1966. pp. 665667. Printed in Great Britain
Reflection type microwave (Received
Vol.
9,
modulators
10 December 1965 ; in revised form 27 January 1966)
A NEW type of microwave modulator has been recently described by JACOBS et al.(l) The device described in the above reference is a practical example of the general class of microwave modulators that consist of a one port resonant cavity that is filled with a variable conductivity semiconductor. The purpose of this note is to indicate the conclusions about reflection type modulators that can be made by considering this general problem. The expression for the impedance presented to a waveguide by a one port cavity is given by SLATER.@) From
this expression
it is possible
to
666
NOTES
obtain the reflection coefficient of the cavity. The reflection coefficient, r, is approximately I ~
U3--l)+WQo (13+1)+2j~Qo
(1)
where
00 = the unmodulated
conductivity,
01 = the change of the to the modulation. Equation
conductivity
@a-l)-01/~0+2jSQo (Bo+1)+~d~oo~j~Q~
WO
wo = the resonant frequency of the cavity, Qs = the Q of the lossless cavity filled with the lossy medium, Qext = the Q of the lossless cavity loaded by the external waveguide terminated in its characteristic impedance, QolQext.
Equation (1) is valid for values of 8 much smaller than unity. ‘The experimental methods for determining ,f3are well established.(s) The above result may be cast in the form desired for the present discussion by noting the Qs of the filled cavity is inversely proportional to the medium conductivity, and defining
due
(1) may then be put in the form y=
w--w0
s=------
B=
where CC= the total conductivity,
(3)
Where /3s is the coupling coefficient, Qs/Qext, in the absence of the conductivity modulation. It has been assumed that the loss of the unmodulated cavity is determined by the semiconductor rather than by the wall losses of the cavity. Equation (3) is the general form for the reflection coeficient of a cavity filled with a variable conductivity semiconductor in the neighbourhood of the resonant frequency. It is clear that the use of a resonant configuration has made the modulation frequency sensitive, however, it has enhanced the dependence of the modulation upon the conductivity of the medium. A plot of the reflection coefficient at resonance as a function of the conductivity change is given in Fig. 1 for several values of
FIG. 1. A plot of the voltage reflection coefficient as a function of conductivity modulation for various coupling coefficients. The intercept at q/oo = 0 gives the reflection coefficient with no modulation present, and the intercept at r = 0 gives the peak conductivity change for optimum operation.
NOTES
the coupling coefficient. The conclusions about a resonant modulator that may be made on the basis of equation (3) and Fig. 1 are indicated below. (a) The undercoupled case, in which /3s < 1, is not very sensitive to conductivity modulation unless the conductivity is reduced by the modulation process. (b) There is an optimum relation coupling coefficient and the maximum increase in the case of the overcoupled relation is such that the reflection reduced to zero at the peak of the modulation. (c) The maximum change in coefficient for the overcoupled coupled cases is
between the conductivity cavity. This coefficient is conductivity
the reflection and critically
(d~O)peak
(4
max = 2+ (+O)peak
where (Q/os)p@ is the peak value of the conductivity increase due to modulation. (d) The variation of Y with conductivity is a non-linear function, although it is approximately linear over a wide range of Y for the overcoupled case. On the basis of the above results it is apparent that the heavily overcoupled case described by JACOBS et al.(l) is a nearly optimum situation. However, it may be desirable to place an iris in front of the semiconductor to vary the coupling coefficient so that the maximum reflection coefficient modulation may be obtained with a given conductivity modulation. The optimum coupling coefficient is determined by means of equation (3). R. E. HAYES Electrical Engineering Department University of Colorado, Boulder Colorado, U.S.A. References 1. H. JACOBS, R. W. BENJAMIN and D. A. HOLMES, Solid-State Electron. 8. 699 (196.5). 2. J. C. SLATER, Microwabe E&o&s, p. 81, Van Nostrand, New York (1950). 3. E. GINZTON, Microwave Measurements, Chap. IX, McGraw-Hill, New York (1957).
667
Solid-State
Electronics Pergamon Press 1966. pp. 667-668. Printed in Great Britain
Gallium
Vol. 9,
arsenide four-layer device
(Receiwed
24 January
1966)
GALLIUM ARSENIDE npn- and pnp-transistors have already been produced by means of the planar technology.(lpsJ) In this note the fabrication of a planar four-layer device by triple-diffusion will be reported.
Basic considerations
Due to the high electron-hole mobility ratio in GaAs it is possible to inject electrons from an n-region of lower doping level into a p-region of higher doping level. Consequently, doublediffused planar GaAs npn-transistors can be operated in the reverse mode (bulk material serving as emitter) with an appreciable current gain. It has been shown by STATZ,@) on the other hand, that hole injection is possible from a p-region obtained by a high-concentration zinc diffusion process. A planar structure according to Fig. 1 p,(lowcmcentmtii
p2Ii-i@ comntr0tion Zn-diffusion) nz En-diffusion)
v-diffrsan’ \
,L_L_llJ
/
“I n - GoAs
FIG. 1. Schematic
cross-section of triple-diffused layer device.
four-
can therefore exhibit an overall current amplification greater than unity if the doping levels are properly chosen. The device then features the typical thyristor characteristic with a negative resistance separating high- and low-resistance regions. The fabrication process
The basic material was Czochralski grown GaAs, doped with selenium (n = 3 x 101s cm-a). Masking for all three diffusion steps was done with pyrolytically deposited silicon dioxide layers containing some phosphorus in order to avoid cracks and to improve the masking efficiency.@) The diffusions were carried out in sealed quartz tubes, with some arsenic added. A 0.1 per cent
induced by the scanning beam goes through a minimum as the boundary is traversed. We have confirmed that these spikes and the dark structure seen in the p-regions follow the mosaic structure by subsequent etching experiments which revealed this structure by surface markings. In addition these experiments :
665
Current experiments in this laboratory the latter approach is feasible.
Acknowledgement-This note is published by permission of the Ministry of Defence.
D. A. SHAW K. A. HUGHES
(u) enabled the spikes to be studied in depth; (b) revealed that twin boundaries enhanced diffusion; (c) established that this structure unconnected with surface scratches.
also lead to is completely
We suggest that these observations indicate that current laser crystals contain small volumes (linear size N 10 to 200~) which are centred on the p-n junction plane and in which the zinc concentration varies in a cellular like manner. The very existence of such a structure raises the question as to whether it can effect laser performance. In our very small sample the poorer laser material has a more extensive structure than the better material. In support we have the observation that in the ten faces of good lasers we found only three diffusion spikes, although there were probably more in the bulk. If we accept this correlation we have a natural explanation for “filament” formation. Let us assume that the spike regions are “dead due possibly to excessive absorption, spots”, scattering or poor injection properties. Then, the greater the density of such regions, the less chance of wide active filaments, because each operating mode has to pass between dead spots or be highly attenuated in traversing them. If we have small spike regions then minor changes in filament pattern can occur as the current is varied. This idea has the merit of simplicity. It is applicable to the other kinds of function irregularities previously observed,@*@ such as those due to variations in doping level in the starting crystal or due to clusters of defects. The relationship between any of these effects and laser performances can be studied by photographing the filament positions and then examining the whole of the function plane by these techniques with the electron beam directed along the normal to the junction plane. This is possible either on lasers with shallow diffusions (4 to 5~) or on lasers with deep diffusions which have been subsequently etched down.
show that
N.
F. B. NJZVE
D.
v.
SULWAY
P. R. THORNTON
Dept. of Materials Science Uniwrsity College of North Wales Bangor, U.K.
C. GOOCH
S.E.R.L., Bala’ock Herts., U.K. References
1. R. F. BLOOM, C. H. GOOCH, C. HILSUM and D. J. OLIVER, Nature, Land. 198, 368 (1963). 2. T. E. EVERHARTand 0. C. OATLEY, J. Electron. 2,568 (1957). 3. T. E. EVERHART,0. C. WELLS and R. K. MATTA, J. Electrochem. Sot. f 11, 929 (1964). 4. I. G. DAVIES, K. A. HUGHES, D. V. SULWAY and P. R. THORNTON, Solid-State Electron. 9,275 (1966). 5. J. C. MARINANCE, J. Electrochem. Sot. 110, 1153 (1963). 6. M. H. PILKUHN and H. RUPPREZCHT,Trans. Met. Sot. A.I.M.E. 230, 296 (1964).
Solid-State Electronics Pergamon Press 1966. pp. 665667. Printed in Great Britain
Reflection type microwave (Received
Vol.
9,
modulators
10 December 1965 ; in revised form 27 January 1966)
A NEW type of microwave modulator has been recently described by JACOBS et al.(l) The device described in the above reference is a practical example of the general class of microwave modulators that consist of a one port resonant cavity that is filled with a variable conductivity semiconductor. The purpose of this note is to indicate the conclusions about reflection type modulators that can be made by considering this general problem. The expression for the impedance presented to a waveguide by a one port cavity is given by SLATER.@) From
this expression
it is possible
to
666
NOTES
obtain the reflection coefficient of the cavity. The reflection coefficient, r, is approximately I ~
U3--l)+WQo (13+1)+2j~Qo
(1)
where
00 = the unmodulated
conductivity,
01 = the change of the to the modulation. Equation
conductivity
@a-l)-01/~0+2jSQo (Bo+1)+~d~oo~j~Q~
WO
wo = the resonant frequency of the cavity, Qs = the Q of the lossless cavity filled with the lossy medium, Qext = the Q of the lossless cavity loaded by the external waveguide terminated in its characteristic impedance, QolQext.
Equation (1) is valid for values of 8 much smaller than unity. ‘The experimental methods for determining ,f3are well established.(s) The above result may be cast in the form desired for the present discussion by noting the Qs of the filled cavity is inversely proportional to the medium conductivity, and defining
due
(1) may then be put in the form y=
w--w0
s=------
B=
where CC= the total conductivity,
(3)
Where /3s is the coupling coefficient, Qs/Qext, in the absence of the conductivity modulation. It has been assumed that the loss of the unmodulated cavity is determined by the semiconductor rather than by the wall losses of the cavity. Equation (3) is the general form for the reflection coeficient of a cavity filled with a variable conductivity semiconductor in the neighbourhood of the resonant frequency. It is clear that the use of a resonant configuration has made the modulation frequency sensitive, however, it has enhanced the dependence of the modulation upon the conductivity of the medium. A plot of the reflection coefficient at resonance as a function of the conductivity change is given in Fig. 1 for several values of
FIG. 1. A plot of the voltage reflection coefficient as a function of conductivity modulation for various coupling coefficients. The intercept at q/oo = 0 gives the reflection coefficient with no modulation present, and the intercept at r = 0 gives the peak conductivity change for optimum operation.
NOTES
the coupling coefficient. The conclusions about a resonant modulator that may be made on the basis of equation (3) and Fig. 1 are indicated below. (a) The undercoupled case, in which /3s < 1, is not very sensitive to conductivity modulation unless the conductivity is reduced by the modulation process. (b) There is an optimum relation coupling coefficient and the maximum increase in the case of the overcoupled relation is such that the reflection reduced to zero at the peak of the modulation. (c) The maximum change in coefficient for the overcoupled coupled cases is
between the conductivity cavity. This coefficient is conductivity
the reflection and critically
(d~O)peak
(4
max = 2+ (+O)peak
where (Q/os)p@ is the peak value of the conductivity increase due to modulation. (d) The variation of Y with conductivity is a non-linear function, although it is approximately linear over a wide range of Y for the overcoupled case. On the basis of the above results it is apparent that the heavily overcoupled case described by JACOBS et al.(l) is a nearly optimum situation. However, it may be desirable to place an iris in front of the semiconductor to vary the coupling coefficient so that the maximum reflection coefficient modulation may be obtained with a given conductivity modulation. The optimum coupling coefficient is determined by means of equation (3). R. E. HAYES Electrical Engineering Department University of Colorado, Boulder Colorado, U.S.A. References 1. H. JACOBS, R. W. BENJAMIN and D. A. HOLMES, Solid-State Electron. 8. 699 (196.5). 2. J. C. SLATER, Microwabe E&o&s, p. 81, Van Nostrand, New York (1950). 3. E. GINZTON, Microwave Measurements, Chap. IX, McGraw-Hill, New York (1957).
667
Solid-State
Electronics Pergamon Press 1966. pp. 667-668. Printed in Great Britain
Gallium
Vol. 9,
arsenide four-layer device
(Receiwed
24 January
1966)
GALLIUM ARSENIDE npn- and pnp-transistors have already been produced by means of the planar technology.(lpsJ) In this note the fabrication of a planar four-layer device by triple-diffusion will be reported.
Basic considerations
Due to the high electron-hole mobility ratio in GaAs it is possible to inject electrons from an n-region of lower doping level into a p-region of higher doping level. Consequently, doublediffused planar GaAs npn-transistors can be operated in the reverse mode (bulk material serving as emitter) with an appreciable current gain. It has been shown by STATZ,@) on the other hand, that hole injection is possible from a p-region obtained by a high-concentration zinc diffusion process. A planar structure according to Fig. 1 p,(lowcmcentmtii
p2Ii-i@ comntr0tion Zn-diffusion) nz En-diffusion)
v-diffrsan’ \
,L_L_llJ
/
“I n - GoAs
FIG. 1. Schematic
cross-section of triple-diffused layer device.
four-
can therefore exhibit an overall current amplification greater than unity if the doping levels are properly chosen. The device then features the typical thyristor characteristic with a negative resistance separating high- and low-resistance regions. The fabrication process
The basic material was Czochralski grown GaAs, doped with selenium (n = 3 x 101s cm-a). Masking for all three diffusion steps was done with pyrolytically deposited silicon dioxide layers containing some phosphorus in order to avoid cracks and to improve the masking efficiency.@) The diffusions were carried out in sealed quartz tubes, with some arsenic added. A 0.1 per cent