SEMICONDUCTOR MEASUREMENT TECHNOLOGY: Safe Operation of Capacitance Meters Using High Applied-Bias Voltage by Alvin M. Goodman Abstract: The use of capacitance meters (C-meters) to This report describes a technique for using a commercial The use of the BIU imposes certain limitations on the range Construction details of the BIU are appended. Key Words: Bias-Isolation Unit; capacitance measurements at high applied-bias voltage; capacitance-meter; extended-range capacitance measurement; high-voltage C(V) measurements; modified MIS C(V) measurements. 1. INTRODUCTION The use of capacitance meters (hereinafter abbreviated as C-meters) to determine small-signal (differential) capacitance as a function of applied-bias voltage is commonplace today in many research, development, and manufacturing applications. Many, if not most, of these applications are connected with the semiconductor industry. A variety of commercial instruments* is available to meet most existing measurement requirements. Typically, an instrument of this type uses a crystal-controlled 1-MHz test signal whose amplitude is ~15 mV. The test signal is applied to the unknown capacitance; the resulting current is amplified, and its quadrature component (with respect to the applied voltage) is determined using some form of phase-locked synchronous detector. The quadrature component of the current is directly proportional to the measured capacitance, and the C-meter is usually calibrated to read directly in picofarads. In addition, an analog output voltage is generally made available to enable the plotting of capacitance on a recorder. Means are usually provided for applying a quasi-dc bias voltage to the unknown capacitance through the C-meter. This allows the capacitance to be recorded as a function of the applied-bias voltage using the arrangement shown in figure 1. The magnitude of the bias voltage that may be applied in this way is limited. Although the limit differs for different instruments, it is in no case greater than 600 v. Figure 1. Conventional high-frequency C(V) mea surement. The inset is a schematic illustration It is sometimes necessary to apply a bias voltage larger than 600 V to a capacitor sample [1,2]. This can be accomplished by using an arrangement which applies the bias directly to the sample but not to the C-meter. An example (based on a circuit described in reference 3) is shown in figure 2. The isolation box allows the bias voltage to be *Some examples of commercial C-meters (or instruments containing C-meters) are as follows: BEC (Boonton Electronics Corporation) Models 71A, 71AR, 72A, 72AD, 72B, 72BD; and PAR (Princeton Applied Research Corporation) Model 410. Figure 2. 0.5-ImH (COARSE TUNING) (b) ISOLATION BOX CIRCUIT Isolation box arrangement for measuring C(V) with an applied bias voltage larger than the C-meter limit: (a) block diagram, and (b) isolation box circuit. applied to the sample while keeping it out of the C-meter; at the same time, it isolates the bias-voltage supply from the high-frequency test signal. It is assumed that there is a dc path through the C-meter to allow the high-voltage-blocking capacitor to charge and discharge as the bias voltage is (slowly) varied. The parallel resonant circuits are tuned to the test frequency (1 MHz in this case) to provide the necessary ac isolation. This isolation can also be obtained by using sufficiently high-value resistors instead of the tuned circuits. The blocking capacitor between the sample and the C-meter must be sufficiently large that it does not introduce unacceptable error into the measurement. For the value shown (0.02 μF), the maximum error for samples with C 100 pF would be 0.5%. The voltage rating of the blocking capacitor must, of course, be at least as large as the greatest anticipated bias voltage. The arrangement shown in figure 2 does not, however, provide complete protection for the C-meter in the case of sample failure. If the sample develops a short-circuit while a large bias voltage is applied to it, the blocking capacitor (which is also charged up to the bias voltage) must discharge through the C-meter. If the voltage across the blocking capacitor just before the sample "shorts" is sufficiently large, the C-meter will be damaged. In order to prevent this type of damage to the C-meter, a "bias-protection circuit" has been developed. The basic principles of this circuit and some of the design considerations are discussed in section 2. actual circuit and operation of a C-meter "Bias-Isolation Unit" (BIU), which allows safe capacitance measurements at bias voltage up to + 10 kV, are described in section 3. In section 4, some of the experimental results obtained using the BIU with commercial C-meters are presented. Finally, in section 5, present and possible future applications of capacitance measurements at high applied-bias voltage are discussed. 2. SIMPLIFIED CIRCUIT AND PRINCIPLES OF OPERATION The operation of the C-meter bias-protection circuit can best be described by considering a simplified version: first, in the normal operating mode, as shown in figure 3 (a), and second, the equivalent circuit after a sample "breaks down" with a large bias voltage applied to it, as shown in figure 3(b). The sample capacitance being measured is represented by Cs. The series combination of L and CB is tuned to resonance (at the measurement frequency of the C-meter) so that there is no reactance in series with Cs. The diodes D exhibit very small capacitance and conductance at the measurement signal level (~ millivolts) and thus do not interfere with the measurement of Cs. The shunt capacitance and conductance of the diodes can be reduced still further by applying a small reverse bias to each of the diodes. The value of R is much less than XCs, and in a first approximation its effect on the measurement of Cs may be ignored. A detailed consideration of its effect will be presented later. The bias voltage is supplied to Cs through high-value resistances (r >> Xcs); they serve two functions: (i) to effectively isolate the bias supply from the measurement circuit, and (ii) to limit the current in case of a sample breakdown (short circuit). The C-meter is represented by a parallel RLC circuit at the measurement frequency. There is a low-resistance dc path between the terminals of the C-meter. This allows the dc voltage at the C-meter terminal to remain effectively zero during slow variations of the applied-bias voltage. Figure 3. Simplified version of C-meter bias-protection circuits: (a) normal operating mode, and (b) after a capacitor sample breaks down with a large bias voltage applied to it. During large rapid variations of the sample bias-voltage, the voltage at the C-meter terminals would start to become significantly different from zero; one or the other of the diodes would then start to conduct heavily when biased in the forward direction. One rather important example of rapid variation of sample bias-voltage is the case in which the sample develops a short-circuit during application of a large bias voltage; the voltage across the sample drops (almost) instantaneously zero and CB must discharge through R, L, and the parallel combination of the diodes and the C-meter. This is illustrated schematically in figure 3(b). The forward resistance of the conducting diode rf must be sufficiently low that the voltage at the input terminals of the C-meter never exceeds the maximum allowable value; i.e., it remains "clamped" within some allowable range. |