Comparison of Measurements at Kansas State and Stony Brook Introduction Kansas State University (KSU) and the State University of New York at Stony Brook (SB) have obtained similar systems for testing silicon microstrip sensors. Each system contains an Alessi (Cascade Microtech) 6100 automatic probe station, a Keithley 237 Source Measure Unit (SMU) picoammeter and voltage source (does both at once), a Keithley 487 picoammeter and voltage source (separate), and an LCR meter. For an LCR meter KSU uses the HP 4284, but SB uses the Agilent (HP) 4263B. Both also use the HP34904A 4 8 Matrix Switch to perform multiple electrical tests with the needle probes only touching the sensor pads once. KSU runs 3 coaxial probes with an additional probe riding on the chuck. SB runs 4 coaxial probes plus a fifth riding on the chuck. We will discuss the comparison of results from KSU and SB for two sensors which both labs have scanned. L2_014 is a perfect outer sensor from HPK. Both labs agree this sensor has no problems. L1_010 is an Elma sensor which is more interesting because it has problems, so we will discuss measurements on this sensor in detail. L1_010 IV Curves 450 400 350 Current (nA) 300 250 SB IVCV 200 SB KSU 150 100 50 0 0 100 200 300 400 500 600 700 800 900 -50 Bias Voltage (Volts) Figure 1. Total leakage current versus bias voltage for L1_010. IV Scans Figure 1 shows IV scans of L1_010. The important point is that the sensor does not break down with voltages as high as 800 V. Two SB measurements are shown with one (IVCV) only going up to 200 V. Temperature differences could account for the different current levels observed at SB on different days. (Fine temperature control at SB is being installed now.) Differences in shape of the IV curves taken at KSU and SB could be due to different settling times before the SMU measurement. (SB uses 1 second.) CV curve for L1_010 3 2.5 1/(C(nF)*C(nF)) 2 1.5 SB KSU 1 0.5 0 0 50 100 150 200 250 -0.5 Bias Voltage Figure 2. 1/Capacitance(nF)2 versus bias voltage for L1_010. CV Scans Figure 2 shows 1/C(nF)2 versus bias voltage for L1_010. SB obtained a full depletion voltage of 23 V for this sensor whereas KSU obtained 25 V. Idiel Figure 3 shows the current through the dielectric of the coupling capacitor for each strip. The purpose of this measurement is to detect pinholes, current greater than 10 nA. L1_010 does have some pinholes. A summary of bad strips is presented below for both sensors and both institutions. It should be noted that SB follows our QA document (“Silicon Sensor Quality Assurance for the D0 Run IIB Silicon Detector: Procedures and Equipment”, A. Bean et al., D0 note 4120, draft 4.0) more closely than KSU in performing many measurements. To measure Idiel, SB places a probe on a strip’s AC pad and measures the current out of the SMU at 80 V with the bias line grounded. KSU places the voltage of 20 V on a DC pad and measures the current to the corresponding AC pad at ground. Hence the KSU Idiel for L1_010 0.2 0.19 0.18 0.17 0.16 0.15 0.14 0.13 Idiel (nA) 0.12 0.11 SB 0.1 KSU 0.09 0.08 0.07 0.06 0.05 0.04 0.03 0.02 0.01 0 0 50 100 150 200 250 300 350 400 450 strip num ber Figure 3. Current through the coupling capacitor dielectric for each strip of L1_010. measurement is made by the K487 picoammeter, whereas the SB measurement is made by the K237 SMU picoammeter. SB observes a background current (current with the probe lifted) of typically 40 pA with the K237 SMU but less than 1 pA with the K487. So approximately 40 pA should be subtracted from the SB measurements on both plots for the sake of comparison. We have not bothered to do this because 40 pA is much less than the currents we care about (10 nA). The fact that SB uses a higher testing voltage than KSU could be significant. The differences in measurement technique do lead to one physical effect. On the L1_010 SB measures both of the end strips to be “pinholes”. Pinholes usually have abnormal values of the coupling capacitance, but not these two end strips. So it is suggested that these end strip “pinholes” actually result from surface conduction to the bias line. KSU probably should not see this because their AC pad is at ground, close to the same potential as their bias line. KSU does see an increase in Idiel by a factor of 7 on one end strip of L1_010, but this current does not exceed 10 nA. Coupling Capacitance for L1_010 100 95 90 85 80 75 70 Capacitance (pF) 65 60 55 SB 50 KSU 45 40 35 30 25 20 15 10 5 0 0 50 100 150 200 250 300 350 400 450 strip num ber Figure 4. Coupling capacitance versus strip number for L1_010. Cac Figure 4 shows the coupling capacitance for each strip. At this writing the absolute SB values are in question because differences from the hand measurements at SB have been observed. Nevertheless measurements at SB and KSU usually agree in the identification of bad channels (Cac greater than 1.2 times the typical value (short) or less than 0.8 times the typical value (open)). A summary of bad channels is provided below. The pinholes usually also qualify as opens or shorts. KSU and SB usually agree on whether the coupling capacitance is good or bad but for the bad channels, they often disagree on whether the bad value is an open or a short. This could be due to the differences in method. SB measures capacitance from the AC pad to the bias line, whereas KSU measures the capacitance between the DC pad and AC pad. The regular structure seen in the SB plots is real. It is due to the fact that every 10th AC pad is larger than the others. Strip Leakage Current for L1_010 1.2 1 Istrip (nA) 0.8 SB 0.6 KSU 0.4 0.2 0 0 50 100 150 200 250 300 350 400 450 strip num ber Figure 5. Leakage current to each strip for L1_010. Istrip Figure 5 shows the leakage current from each strip. Both KSU and SB use the K487 to measure this current and the agreement between the two measurements is quite good. There is a quantization in the SB data (about 5 pA for L1_010) which KSU does not see. For L1_010 the SB data are more spikey than the KSU data. The KSU data follow some spikes, but not all. All of the spikes are much less than the bad channel limit of 10 nA. Polysilicon Resistance of Each Strip on L1_010 1400 Rpoly (kOhm) 1350 SB 1300 KSU 1250 1200 0 100 200 300 400 500 strip num ber Figure 6. Polysilicon resistance on each strip on L1_010. Rpoly Figure 6 shows measurements of the polysilicon resistor on each strip. On L1_010 both labs find that ALL strips have resistors which are out of spec. Since all the channels are bad, we do not mention this in the summary. We arbitrarily call a channel bad if its resistance is greater than the average value (1272 kOhm) plus 300 kOhm or less than the average value minus 300 kOhm. SB sees one bad channel which is not confirmed by KSU. Interstrip Capacitance at 100 kHz for L1_010 4.5 4 3.5 Cint(pF) 3 2.5 SB 2 1.5 1 0.5 0 0 50 100 150 200 250 300 350 400 450 strip num ber Figure 7. Interstrip capacitance between adjacent strips starting with an even number. Cint Figure 7 shows the SB measurements of the interstrip capacitance of the even pairs of strips. KSU does these measurements on test structures. SB performs this measurement at 100 kHZ. Other labs see that the frequency dependence of Cint is rather flat between 100 kHz and 1 Mhz, at which Cint should be less than 1/2 1.2 pF/cm or 4.6 pF for a L1 sensor. There are no bad channels (among half the pairs). Interstrip Conductance of L1_010 70 60 50 Interstrip Conductance (pA/V) 40 30 20 SB 10 0 0 50 100 150 200 250 300 350 400 450 -10 -20 -30 strip num ber Figure 8. Interstrip conductance between adjacent strips starting with an even number. Rint Figure 8 shows SB measurements of the interstrip conductance for even pairs of strips on L1_010. KSU does these measurements on test structures. Between any pair of strips the interstrip resistance must be greater than 2 GOhm. This limit corresponds to an interstrip conductance of 500 pA/V. All even pairs of strips pass. The quantization of the SB data is not currently understood and reflects a systematic error. Some negative values of the conductance are observed, but all are much less than 500 pA/V in absolute value. The average interstrip conductance of L1_010 is 14.8 pA/V. This average corresponds to interstrip resistance 68 GOhm. Summary of Bad Channels Sensor strip L2_014 L1_010 0 80 239 240 241 242 243 247 316 376 380 383 Kansas State Stony Brook none none OK open pinhole, short pinhole, short pinhole, short pinhole, short pinhole, short pinhole, short OK open OK OK “pinhole” OK pinhole, open pinhole, open pinhole, open pinhole, open pinhole, open pinhole, short Rpoly high open short “pinhole” SB and KSU agree that L2_014 is perfect. For L1_010: SB saw 6 pinholes which KSU confirmed. SB saw one open (without pinhole) which KSU confirmed. SB saw 2 “pinholes” which KSU did not confirm. SB saw one short which KSU did not confirm. SB saw one Rpoly high which KSU did not confirm. KSU saw one open which SB did not confirm. In the future both labs plan to remeasure all bad channels, to confirm the measurements.