STRUCTURE A

It wou]d be an advantage to be ab]e to make Schottky contacts to p-GaSb as this wouid avoid the

need to dope the materia], Schottky diodes have been made to n-GaSb [Nagao e t al, 1981; Chin

e t al, 1980; Sukegauuo e t a!., 1978] but the main limitation to both these devices has been the high dark current. To date only ohmic contacts have been made to p-GaSb due to the low barrier height o f ~0.2eV. The Fermi level is pinned 0.2eV above the valence band edge due to the large density o f surface states. We investigated various methods o f enhancing the barrier height to carrier transport Many other groups have used various principles, incorporating insulators

[Schmidt e t ai. 1988], highly doped surface layers [Pearton e t a!.. 1989], and other

semiconductors [Cheeks eta!., 1991 ; UUaldrop eto L , 1988]. Since u-GaSb grows p-type when not

intentionally doped, these method have concentrated on increasing tiie hole barrier. The large arsenide/antimonide valence band offset suggested the use o f an interfacial layer o f GaAs between the metal and the GaSb. The band diagram is shown in fig. 7.6 with band offsets calculated from the model solid theory o f Van der Walle, 1989, assuming complete relaxation o f the strain. From this it can be seen that the barrier to holes is significant.

CtiAJ'TKR 7 E le ctric a l S tu dy o f G a S h -h a sed P h o to d e te cto r S tru ctu res 0 1 3 , V 1 .4 2 4 e V 0 . 7 5 e V O .S l c V G a A s c a p 1---- <250A 4 p m I 4 2 4 c \ ' G a S b a c t i v e r e g i o n , G a A s : S i S u b s t r a t e

Figure 7.6 Band diagram showing the hand offsets o f GaSb. GaAs heterojttnction with the addilioit o f the capping layer at then front jttnction

Sample A had the structure shown in fig. 7 7 where the GaAs cap thickness was <250Â. This tliickness was chosen as the GaAs grown at 560®C was foimd to be p-type witli a carrier concentration o f 2xlO"*cni'^. At tins concentration and tliickness the GaAs should be fully depleted which may not be the case for a thicker layer Tlie ~8% mismatch between the

GaSb/GaAs lattice constants results in the epilayer bemg highly dislocated [Chidley e t oL, 1989]

The cap appears to act as an insulator and enhances the barrier to hole conduction

Ti/Au Schottky contact

fiaA s cap |250AJ p-(>aSb absorber [~4pm |

p-GaSb buffer

GaAs:Si

substrate

Au O hm ic Contact

Figure 7.7 Stntctiire o f sample A

The structure o f fig 7.7 was considered. The metal/GaAs cap/GaSb absorber layer acts like a MIS [metal-insulator-semiconductor] type junction Tlie cap was added in a second growtli run. Between the substrate and the p-GaSb active region is a buffer layer grown at 400°C. When growing on substrates with a large mismatch it is usually necessary to grow a low temperature buffer layer to improve the initial growth conditions [Nishimura e t a/., 1991] It presents a good surface for growth and stops the dislocations from propagating too far into the active region.

C lI. lP TliR 7 E le ctric a l S tu d y o fG a S h -h a s e d P h o to d e tec to r S tru ctu res

From the I-V characteristics, fig 7 8, there is rectification on both sides The forward bias represents the depletion o f the GaSb active region under the GaAs capping layer and the reverse bias represents depletion o f the p-n junction at the back

0.010 0.005

I

Figure 7.8 I-V clianiclefistics of sample A. The inset shows the log I-]’ plot and the ettergy hand diagram of this de\>ice

Breakdown occurs at around 4V reverse bias and the dark current is less than lOpA up to -2V. From the energy band diagram it has been calculated [L. Ponnampolom, 1992] that tlie valence band offset is 0.8 leV compared to tlie conduction band offset which is 0.3 leV and it is believed that It is the high valence band offset between the GaAs and GaSb which causes good rectification Some o f the devices showed ohmic characteristic and one reason for this could be due to the fact that the top metal contact may have diffiised down to the p-GaSb through the cap Suice the cap is <250Â this is entirely possible.

Capacitance-voltage measurements were carried out on sample A and showed a carrier concentration o f lO’^cm'^ for the GaSb active region At this order o f carrier concentration and for a reverse bias o f 4V, the depletion region can be calculated to be 800Â [0 08pm ]. The theoretical external quantum efficiency near the bandedge can be calculated to be -2.6 % using the fact that -33 % o f the incident light would be reflected at the front surface.

The zero bias barrier height and ideality factor can be calculated from the forward bias I-V data. In the case o f structure A when the front junction is forward biased, the back junction is reverse biased. This makes the calculation o f the ideality factor and barrier height more complicated.

Butcher el al., 1993, show a similar problem in their GaAs diodes. TTiese showed that forward-

biased Schottky diodes on n-type epilayers were accompanied by depletion from the substrate metallisation, while forward biasing barriers on n-type epilayers reverse-biased the junction

CHAPTER E le ctric a l Stu dy o fG a S b - b a s e d P h o to d e te c to r S tru ctu res

between the epitaxy and the n-type substrate. A similar effect was observed in sample A where depletion o f active layer under the top contact is accompanied by depletion o f p-n junction between the active layer and the substrate.

It is usual to measure ideality factors from the forward biased I-V data, but this is not possible in

this case as explained. Butcher et al., 1993, calculated the ideality factors in both forward and

reverse bias using a graph similar to fig. 7.9.

B la v V

-2S0 ■ .

l n |I /|l- « p ( - q |V l/1 tT ) ll

Figure 7.9 Plot o f lu[I] as a function o f the applied voltage to calculate the ideality factor fo r both the back and fro n t junctions o f structure A

For the front Schottky contact the ideality factor was calculated to be 1.124 and for the back p-n junction the ideality factor was calculated to be 1.109. This indicated that the transport mechanism which causes the current to flow for the front junction dominates over the back junction. In forward bias the diffusion current dominated over the drift current.

The Schottky resulting from device A was not found to be repeatable and only one device gave this result. This was not promising from the point o f view o f growing structures and using them as Schottky detectors. The main reason for the other devices exhibiting non-Schottky contacts could be due to the top metal contact diffusing down through the cap layer and hence making an ohmic contact to the epilayer. One very important point which came about due to looking at the 1- V characteristics o f structure A is the comparatively good characteristics o f the back n^-p junction. This result was repeatable in all the devices measured and it was decided to investigate this further.

CH APTER 7 E le ctric a l S tu dy o f G a S h -h a sed P h o to d e tec to r S tru ctu res

7.4.2 HETEROJUNCTION [p-GaSb/n-GaAsiSi SUBSTRATE] STRUCTURES A, B, C, D, E, AND F

The 6 samples presented in the final chapters were selected on the basis that they showed the best