Application and Improvement of the spreading resistance method for p-type 6H-SiC

Since the end of the 1960's spreading resistance (SR) measurements have become a routinely used technique for determining charge carrier profiles in silicon. For wide band gap semiconductors however the application of this method is difficult because of the high barrier at the interface between probe tips and the semiconductor surface. In order to lower the barrier two different approaches can be taken. First, the material of the tips could be changed with respect to its work function. But there is a limited choice because of the required mechanical properties of the tips concerning high hardness and low brittleness. The other way is to lower the barrier by influencing the surface states of the semiconductor material. This is actually more promising.
A strong dependence of the SR values on the polishing material for beveling was reported in investigations for the wide band gap semiconductor GaAs [1]. For laser annealed 6H-SiC [2] and 3C-SiC [3] SR measurements have been reported, but very high measuring voltages (1 to 5 V) were used compared to the usual measurements on Si (10 mV). Ahmed et al. found an activation of more than 100% which points to problems in the interpretation of the measurement [2].

In this work the influence of mechanical processing, further annealing of the bevel and finally sputter cleaning on the SR measurements was investigated. The SiC - surface was beveled with a diamond emulsion on a rotating glass plate. Grain sizes of 1.0 µm and 0.1 µm were used. It was found that polishing with the smaller grain size leads to lower resistances. This is comparable to the results which are known for GaAs with Al2O3 emulsion from Ref. [1]. After beveling several of the samples were annealed under vacuum for 5 min at temperatures of 1300 to 1400°C. A modified thin layer (1 to 2 nm) at the surface was formed which resulted in a lowering of the barrier resistance. The measured resistance was lowered up to a factor of three but statistical variations increased.
In order to lower the barrier further, the influence of ion sputtering (Ar+, 2 keV) on the bevel before the measuring procedure was tested. The sputtering was carried out at a current density of 50 µAcm-2 for 300 s. During the process the temperature increased to 57 to 69°C. Subsequent SR measurements showed up to two orders of magnitude lower resistance. The shape of the depth profile remained the same and the statistical error of the SR dropped significantly. Samples with resistivities larger than 0.2 cm were not measurable after beveling but after subsequent sputtering. The resistivity range in which SR measurements can be carried out was extended to 3 cm. First investigations of SR depth profiles using this sputtering method show a very good reproducibility at remarkably small variations of the measured values.
The samples were implanted with Al+ at multiple energies of 450, 240, 115 and 50 keV to form a 500 nm thick homogeneously doped layer with plateau concentrations in the range of 5 1019 to 5 1021 cm-3. Various post annealing processes were carried out to cover a broad range of resistivities with the samples. Resistivities and mobilities were obtained from Hall measurements [4]. For SR profiling a SENTECH SR-210 device was used. The probe tips were made of sintered tungsten carbide with a tip diameter of 5 µm. The measuring voltage was kept constant at 10 mV allowing to measure resistance values up to 1 G . The highest spatial resolution of the apparatus was 5 nm and the probe tip load during the measurements 9 g.

References:

[1] G. Queirolo, J. Electrochem. Soc., 125 (10), 1672 (1978).
[2] S. Ahmed, C.J. Barbero and T.W. Sigmon, Appl. Phys. Lett. 66 (6), 712 (1995).
[3] J. A. Edmond, S.P. Withrow, W. Wadlin, R.F. Davis, Mat. Res. Soc. Symp. Proc.,
Vol. 77, 193 (1987).
[4] D. Panknin , H. Wirth, M. Mücklich, W. Skorupa, Mat. Sci. Engin. B56 (1999), in print