100

Cs

150

200

which the largest undulation takes place, indicated that the major culprit was probably a third-harmonic component in the test signal. To test this hypothesis, a crude filter was assembled for use between the BIU and the C-meter (fig. 13). The purpose of the filter was to attenuate the third-harmonic component of the current flowing from the test capacitor Cs into the C-meter HI terminal. The measurements of S' (Cs) were repeated for the BEC Model 72AD and PAR Model 410 Cmeters using the filter. The results are shown as the squares in figures 11 and 12. It was clear that the filter provided a substantial improvement in the 0 to 100 pF range of Cs; it was equally clear that further improvement was desirable. For this reason, a spectral analysis of the test signal of each of the three C-meters was carried

The results (shown in table 2) indicate that the BEC Model 71A has the "cleanest" test signal, while the test signals of the other two are relatively rich in harmonics. These results confirmed that the unexpected behavior of S' (Cs) was indeed due to the presence of undesirable harmonics in the test signal.

It was felt that the best place to eliminate the harmonics was in the C-meter, either in the test signal generator or in the amplifier chain preceding the phase-sensitive detector. Accordingly, arrangements were made with the PAR Corp. to obtain a modified version of the PAR Model 410 C-meter with extra filtering to eliminate the unwanted harmonics. The results of a measurement of S' (Cs) using the BIU with this C-meter are shown as the filled circles in figure 12. The deviation of S' (Cs) from 1.0 is less than 1% over the range 0 to 200 pF. The experimental points do not fall below 1.0 as rapidly with increasing Cs as would be expected from eq (6); this is consistent with the behavior expected' from a small excess inductance provided by the connecting leads. It is reasonable to expect that similar results could be obtained using the BIU with a BEC Model 72AD C-meter suitably modified to filter out the test signal harmonics.

Table 2.

Test-Signal Spectral Analysis for Three C-Meters

These results may be summarized as follows: It has been demonstrated that the BIU can be used with any of three commercially available C-meters at applied voltages up to 10 kV. For less than + 1% error in relative sensitivity S(Cs), the measurable range of Cs is from 0 to greater than 400 pF; for less than 1% error in relative incremental sensitivity S' (Cs), the measurable range of CS is from 0 to greater than 130 pF.

It is useful to consider the results which have been achieved thus far with the BIU and to place them in the broader perspective of the requirements of present and possible future applications.

The two applications of high-voltage capacitance measurements which have already been investigated in some detail are:

(i)

measurements of C(V) of metal-glass-silicon capacitors for the purpose of characterizing the interface between silicon and a passivating glass layer (tglass ≈ 10 to 100 μm), and

(ii)

measurements of C(V) of metal-sapphire-silicon capacitors

for the purpose of characterizing SOS and the silicon-
sapphire interface (tsapphire ~ 100 to 150 μm) .

In these applications the range of values of Cs encountered was 5 to
55 pF. Clearly, the equipment described herein is entirely adequate
for these measurements. It is anticipated, however, that future measure-
ment applications (using thicker sapphire wafers) will require bias-
voltage capability up to about + 25 kV, and current efforts are directed
toward the development of equipment that will be capable of operating
at these higher voltage levels.

Two other applications have been investigated only in sufficient detail to show that the measurement technique is a useful one. These applications are: (i) characterization of the interface between heavily doped silicon and a thick (> 1 μm) overlying insulator, (ii) measurement of the base-collector capacitance of a very high voltage transistor (max Vc 1500 V) in order to nondestructively characterize the doping profile of its base-collector region. Other possible future applications include the characterization of new passivating and encapsulating layers for semiconductor devices (e.g., plastics, resins, and epoxies), studies of the electric field dependence of dielectric polarization under high fields, and studies of dielectric-electrolyte interfaces. In each of these applications, the equipment described appears to be adequate for all current and future applications.

In conclusion:

(1)

(2)

(3)

A technique has been described which allows the safe operation of commercially available C-meters for capacitance measurements with applied-bias voltage up to 10 kV.

The technique requires a bias-isolation unit for which the circuit, theory, and practical results have been presented.

Details of construction of the bias-isolation unit are provided in the appendix.

ACKNOWLEDGMENT

I am indebted to Chester J. Halgas for the mechanical design and the construction of the bias-isolation unit described in this report, to Paul Kuczer for construction of an earlier developmental version, and to James M. Breece for assistance with some of the measurements.

APPENDIX

This appendix contains information which is intended for use by anyone who wants to duplicate the BIU described in the body of the report. It consists of:

(i)

(ii) (iii)

A list of the mechanical parts which are not commercially available,

detailed drawings of those parts, and

a drawing and photographs showing the placement of all parts (the circled numbers refer to the mechanical parts and electrical components identified in tables 1 and 3).