SLIDE 14 Alternative: InSb APD
- Making an avalanche device out of a
new material is hard. Practice on an easier material
- InSb has a gap of ~235 meV and was
first made into an APD in 1967 (!!)
- Large gap isn’t great for lightest DM,
but still sensitive to DM down to 200 keV
- To do: make a device, test at mK
temperatures, calculate DM sensitivity
- Novel elements: cryogenic
temperatures, SQUID current sensor readout in linear gain mode
4 2
K Γ L W X Γ
Energy [ E-Ef (eV) ]
E0 E’0 ∆SO ∆’SO Σ4 Σ3 Σ4 Σ3 Σ4 Σ3 Σ4 Σ3
(c)
JOURNAL OF APPLIED PHYSICS VOLUME 38, NUMBER 11 OCTOBER 1967
Noise and Multiplication Measurements in InSb Avalanche Photodiodes
General Electric Research and Development Center, Schenectady, New York (Received 15 May 1967) Multiplication and noise measurements on InSb avalanche photodiodes have been made from 77° to 125°K. For the diodes studied, the electron ionization rate is much larger than the hole ionization rate and depends only weakly on the field for fields between 5X loa and 1()4 V
- fcm. The electron ionization rate in-
creases and the breakdown voltage decreases with increasing temperature. The dependence of the ionization rate on field and temperature is explained by the weak dependence of the distribution function on field for high fields. The noise measurements are consistent with McIntyre's theory providing that the electron ionization rate is much greater than the hole ionization rate. A current-controlled negative resistance is
- bserved in InSb avalanche diodes at breakdown.
- I. INTRODUCTION
Improvements in the signal-to-noise ratio in an avalanching silicon diode used as a broadband detector for high-frequency modulated light were reported by Johnson.! It was thought at that time that the state analog of the photomultiplier had been discovered. The calculations of the noise generated in an avalanch- ing photodiode by Tager2 and McIntyre3 showed, how- ever, that the noise associated with avalanche multi- plication increases more rapidly than the signal. Thus the gain in signal obtained in an avalanche photodiode is only useful if the system under consideration is limited by amplifier noise or the thermal noise of the diode load resistance. This limitation is obtained only in broadband systems. The noise generated in an avalanche photodiode depends critically on the ratio of a to (3, the electron and hole ionization coefficients, respectively.3-6 The least noise is obtained if the ratio of the ionization rates is very large and the primary photocurrent is carried by the carrier with the higher ionization rate. .The maximum frequency response of an avalanche photo- diode is also obtained if the ratio of the ionization rates is large.7 The ratio of the electron to hole ionization rates in silicon varies between 10 and 50 depending on the elec- tric field,S while in germanium the hole ionization rate is a factor of two larger than the electron ionization rate.9
1 K. M. Johnson, IEEE Trans. Electron. Dev. ED 12, 5S
(1965) .
2 A. S. Tager, Soviet Phys.-Solid State 8, 1919 (1965). 3 R. J. McIntyre, IEEE Trans Electron. Dev. ED 13, 164
(1966) .
4 H. Melchoir and L. K. Anderson, 1965 Intern. Electron
Devices Meeting, Washington, D.C.
5 R. D. Baertsch, IEEE Trans. Electron. Dev. ED 13, 383
(1966) .
6 R. D. Baertsch, IEEE Trans. Electron. Dev. ED 13, 987
(1966) .
7 R. B. Emmons and G. Lucovsky, IEEE Trans. Electron.
8 C. A. Lee, R. A. Logan, R. L. Batdorf, J. J. Kleimack, and
- W. Wiegmann, Phys. Rev. 134, A761 (1964).
9 S. L. Miller, Phys. Rev. 99, 1234 (1955).
The measurements in GaAsI0,l1 and GaP12 show the hole ionization rate to be equal to the electron ioni- zation rate. This paper reports on measurements of ionization rates and noise in InSb avalanche photodiodes at tem- peratures from 77° to 125°K. From the temperature and field dependence of the ionization rate some sug- gestions as to the distribution function of hot electrons in InSb for very high fields are made. The calculation of the ionization rates from the multiplication data is given in the Appendix.
Single-crystal n-type indium antimonide with a con- centration of approximately 2 X 1014 donors/ cm3 was sawed into slices 1 mm thick and oriented in the (100)
- plane. The slices were chemically polished in a meth-
anol-bromine solution in contact with a rotating wheel covered with a fine polishing paper.I3 The polished slices were sealed in an evacuated quartz tube with an indium-candmium alloy and diffused at 400°C for 60 to 200 h. The slices were then back lapped and etched to a thickness of 150 }J., and cleaved along the [110J directions into dice approximately 1 mm on a side. The dice were alloyed to a gold-plated Kovar header with pure tin. A platinum wire was soldered to the player with pure indium. Care was taken that the indium did not alloy through the p layer which was 3 to 6}J. thick. A portion of the p layer was masked with Apiezon W black wax and a mesa was etched electrolytically in
- NaOH. A final clean-up etch was used to restore surface
stoichiometry.14
In some cases the diode was then anodized in a solu-
tion of O.lN KOH to form a passivating oxide film over the diode. In other cases silicon monoxide was evapo-
10 R. A. Logan, A. G. Chynoewth, and B. G. Cohen, Phys.
11 C. R. Crowell and S. M. Sze, App!. Phys. Letters 9, 242
(1966) .
12 R. A. Logan and H. G. White, J. App!. Phys. 36, 3945 (1965). 13 M. V. Sullivan and G. A. Kolb, J. Electrochem. Soc. 110
585 (1963). '
14 H. L. Henneke, J. App!. Phys. 36, 2967 (1965).
4267
*
Freeze-In Xenon10 Stellar
10-46 10-44 10-42 10-40 10-38 10-36 10-34 10-32 10-3 10-2 10-1 1 10 102 103 104
[Griffin et al., arXiv:1910.10716]
