Au-Cr Deposited on n-GaAs(100)

Ohmic contacts in semiconductor devices had to be homogeneous and uniform in their structure, with good thermal stability in order to avoid degradation due to electrodes and bonding problems. From this point of view, our interest is related to the diffusion problems in Au/Cr/GaAs contacts. It is known [46] that Au based contacts show a poor thermal stability because during annealing the volatile V-components (As) out diffuse easily and Au penetrates into the bulk. Regarding the interest in rapid thermal annealing (RTA) of compound semiconductors, is to be remarked that in GaAs have been generated due to the inherently short annealing times relative to conventional furnace annealing a less stringent restrictions on the method for protecting the semiconductor surface from thermal degradation [47].

The (100) GaAs wafers were prepared prior Cr, Au, and Au/Cr vacuum deposited contacts by cleaning in organic solvents e.g.: C2HCl3, CH3COCH3 and CH3OH and chemical etched in HCl:H2O(1:1). Cr and/or Au films were deposited on GaAs substrate by vacuum deposition at medium pressure p ~6•10^-5 torr followed by an annealing procedure. The thicknesses of Cr films were recorded by the use of a quartz oscillator during metal deposition; the Cr film thickness was monitored to be 400 Å. The Au films were projected to have a thickness in (1000-4000) Å range. Rapid thermal annealing (RTA) procedure used a standard furnace, at relatively low temperatures, i.e. upper limit 500 ^oC, in low vacuum (10^-4 torr). There were implied two sets of samples, annealed for 60 sec. at temperatures of 400 ^oC and 500 ^oC.

The characterization technique implied RBS measurements that were conduced using a

7Li⁺⁺ beam, provided by the Van de Graaff tandem accelerator of NIPNE. An ordinary backscattering set-up was used for the experiment. The backscattered particles were detected with a passivated ion implanted silicon detector. The energy resolution for ⁷Li at 4 MeV was about 30 KeV. The backscattering angle was 145 ^o. The sample surface was perpendicular to the beam direction. The beam spot was collimated down to 2 mm in diameter. The scattering chamber was at around 510^-6 torr. For the quantitative analysis of the RBS spectra, we used the code RUMP [23]. From the analysis of the spectra, we extracted information about Cr and Au layer thicknesses, and about the mass transport.

There are presented experimental data concerning the analysis of Cr/n-GaAs, Au/n-GaAs and Au/Cr/n-GaAs samples non-annealed and annealed at 400 ^oC in vacuum, for 60 sec. RBS spectra for Au/GaAs samples, as deposited, and annealed at 400 ^oC, respectively, are presented in Fig. 5.29 and Fig. 5.30 [48] The RBS spectrum in Fig. 5.30 clearly shows the transport of Au in the annealed sample. The peak near channel 600 corresponds to a concentration of Ga on the top surface of the Au layer in good agreement with previously reported results [46, 49]. The Auger electron spectroscopy (AES) spectra reported in [46] show that this surface layer is a thin Ga oxide layer. The Au layer remains essentially intact during the annealing process. In the attempt to use our RBS data for a quantitative analysis of the transport of Au in GaAs, Au transport is assumed to be the result of a low temperature diffusion process, and in this regard it is assumed a concentration profile of the form:

2

0exp .

4 C C x

Dt

 

  

  (5.11)

Fig. 5.29. RBS spectrum from an as-deposited Au/GaAs sample. Also shown the RUMP simulation, with continuous line.

Fig. 5.30. RBS spectrum from an Au/GaAs sample, annealed at 400 ^oC, for 60 sec. Also shown the RUMP simulation, with continuous line.

which is the solution of the diffusion equation for the conditions of constant total mass, for a thin film deposited on a semi-infinite substrate [46], where D (cm²/sec) is the temperature dependent diffusion coefficient and t is the annealing time, in sec. The RUMP code can compute the spectrum for diffusion profiles, by generating relatively fine layers with a slowly varying composition for each layer, and as a consequence RUMP code was used to determine the diffusion coefficient e.g. it was obtained a value of

1 10 12

D  ^ cm²/sec for the diffusion coefficient at 400 ^oC. It is remarked that this result

compare favorably with those previously reported for the Au/GaAs system, as will be discussed i.e. as is well known, the temperature dependence of the diffusion coefficient is given by:

 

D T D

kT

 

  

  (5.12)

In the case of following values D0 = 10^-11 cm²/sec, Q = 0.27 eV from reference [46], is obtained the value of D = 0.9510^-12 cm²/sec. for diffusion coefficient at 400 ^oC. The values for D0 and Q have been obtained in reference [46] as a result of a fit, using equation (5.12) and the data reported in reference [46] for five diffusion coefficients and the four of the five coefficients were measured for the AuGe/GaAs system, while the fifth was determined for the Au/GaAs system. If there are used the values D0 = 8.910^-9 cm²/sec, Q = 0.65 eV, obtained in reference [46] (as a result of a least squares fit of the data of diffusion coefficients measured by RBS using equation (5.12)), it is obtained the value of D = 1.210^-13 cm²/sec for the diffusion coefficient at 400 ^oC. It is worth to mention that in reference [46], the RBS spectra were analyzed using the “dilute alloy” approximation,, an approximation that neglects the influence of the Au distribution in GaAs substrate on the stopping cross sections, a fact that could give errors for assigned depth scale and for the Au concentration in the substrate. Regarding the data of RBS spectra for an as-deposited Cr/GaAs sample, and the same sample after a 400 ^oC, 60 sec. is remarked no detectable difference between the two spectra, and as a conclusion, the thermal stability of the Cr/GaAs interface contrasts with the low thermal stability of the Au/GaAs contact [48].

In Fig. 5.31 and Fig. 5.32 are presented RBS spectra of an as deposited Au/Cr/GaAs sample, and of the same sample after 400 ^oC, 60 sec. anneal, the general observation is that it was not observed Au or Ga or As mass transport, with the specification that Cr layer deposited on GaAs surface acts as a diffusion barrier for both Ga and Au.

Fig. 5.31. RBS spectrum from an as-deposited Au/Cr/GaAs sample. Also shown the RUMP simulation, with continuous line.

Fig. 5.32. RBS spectrum from an Au/Cr/GaAs sample, annealed at 400 ^oC, for 60 sec. Also shown the RUMP simulation, with continuous line.

5.5. Schottky Contacts Thin Films Deposited on III-V Semiconductor