Wafer Probing System Parametric Evaluation Files

Transcription

1 Application Note Innovating Test Technologies Introduction Evaluation Test Summary Wafer Probing System Parametric Evaluation Files Accuracy of on-wafer semiconductor electrical measurements is often limited by the poor signal path integrity of the typical probing system. Stray capacitance, current leakage, high electrical noise, dielectric absorption can all reduce the effective measurement floor and compromise accuracy. Cascade Microtech has found several electrical tests quite useful for assessing the parametric measurement capability of a wafer probing system. These system evaluation tests can be executed using the industry standard Agilent 4156B/C series DC Parameter Analyzer instrument setup files(.dat files) are available from Cascade Microtech on a 4156 readable floppy disk. NOTE - It s highly recommended that you first read the Cascade Microtech Appnote entitled Recent Advances in DC Parametric Measurements before proceeding further. This appnote can be viewed or downloaded at Test - Measurement Significance Parameters Filename Effected Measurement Wafer chuck or probe noise External and internal noise Idoff, gm, Channel current with an applied from chuck heaters, drive hfe, Vth, Ig, Noise voltage bias motors, etc. can obscure Vbe Noise.dat measurement values IV-Error Probe current with a Capacitive charging current Ig, gummel IV-Error.dat ramping voltage applied can add an error current when plots, Vf a voltage ramp is used Bias Settling Time for wafer chuck or Settling time effects low Idoff, Ig, Time probe to settle to zero current measurements - Isub, BJT Setlb.dat current after a voltage wait time can be added but gummel is applied. this slows the measurement plots and reduces throughput Channel Calculates residual Measurement channels should Vth, Capacitance capacitance of wafer chuck be low capacitance gummel Cap1.dat or probe. settling time can add noise plot, and hysteresis in swept swept Ig measurements. Leakage - Measures open circuit wafer Swept voltages may exhibit Vth, dielectric chuck current with a voltage current hysteresis due to gummel absorption sweep from -100V to +100V dielectric absorption plots, Leakage.dat swept Ig 1

2 Following are detailed instructions for performing each of these five tests as well as a discussion of their overall relevance to on-wafer device measurements. NOTE After turning on the 4156 a 20 minute warm-up period is recommended. For 4156 button and control locations, you may want to refer to the 4156 Users Manual. 1. Measurement Channel Noise: noise.dat Measures wafer chuck or probe needle current leakage and noise. Figure 1. Force Terminal fig. 1 NOTE Cascade parametric stations have two triaxial inputs on the rear of the prober either can be used. (See Figure 2) Make sure that the wafer chuck is open and not in contact with a probe (If the wafer prober does not have triaxial input(s) an individual probe needle may be used to make electrical contact to the wafer chuck top surface). fig Insert the diskette with the test files in the 4156 floppy drive c. Use the 4156 MARKER/CURSOR to highlight Noise.dat file g. Press Graph/List - View the stored display of a thermal parametric chuck 2

3 In this test, 10 volts is applied to the wafer chuck (or probe); noise and leakage current is then measured over a period of time. In an ideal case, no leakage or noise would exist. The stored measurement display was taken on a Cascade parametric thermal chuck noise peaks and leakage are less than 50 fa even at elevated temperature. h. Press green shift (green key on the lower right of the 4156 front panel) i. Press Stop (zero) Green measurement light should light for about 1 second. This zeros inherent 4156 SMU offset current. j. Press Append A live test begins running new measurement data is added to the existing example display. You can see how your system compares to the stored display of chuck noise on the Cascade Microtech system. If necessary, re-scale the vertical current axis to view the measurement result on your system. NOTE If you are testing a probe (rather than a chuck) the observed current should be no more than a few femtoamps. For high parametric accuracy, the measurement area should be as noise free as possible and exhibit very little leakage current. This is critical for good measurement of device off-state currents, and junction and/or oxide leakage. For vertical wafer measurements of gate or substrate current, a good low-noise parametric chuck is highly desirable. 2. IV-ERROR iv-error.dat Measures open circuit probe current with a ramping voltage applied. probe channel to be tested. See Figure 1. c. Use the 4156 MARKER/CURSOR to highlight iv-error.dat file g. Press Graph/List View the stored display - note level of DC offset current In this test, a voltage sweep from 0 V to +1 V and back is applied. This linear sweep may cause an error current to flow that is proportional to probe capacitance(i = C* V/ T). This shows up as a positive or negative DC offset current depending on sweep direction. h. Press green shift ( green key on the lower right of the front panel) i. Press Stop (zero) Green measurement light should light for about 1 second. This zeros inherent SMU offset current. j. Press Append -The live test begins running - new measurement data is added to the existing example display for comparison. To view at a slower ramp rate, you can change the 4156 Integ Time to Medium or Long. Measurements like gate current, junction forward voltage, or BJT gummel plots depend on measuring device current with a ramping voltage applied. But probing channels that are highly capacitive reduce measurement accuracy due to the C* V/ T effect. 3

4 3. Bias Settling Time: Setlb.dat 4. Channel Capacitance: cap1.dat Measures wafer chuck or probe needle measurement settling time. Figure 1 and Figure 2. c. Use the MARKER/CURSOR to highlight Setlb.dat file g. Press Graph/List. View the stored display of current vs time. In this test, a 20 volt step is applied to the wafer chuck. Open circuit current versus time is measured with the application of the voltage step. Note that in the stored display the chuck settles very quickly to a very low open circuit current value. h. Press Append A live test begins running new measurement data is added. Note how your system compares to the stored display. Ideally no current should flow thru on open chuck. But many chucks are high in stray or residual capacitance. When a voltage bias is applied, this capacitance charges adding an error current to device measurement values for a finite interval. A good parametric chuck should be low capacitance and settle to a low value (< 500 fa) quickly. If chuck capacitance is high, then step recovery may take many seconds (or even minutes.) Slow recovery or settling time results in a time related measurement uncertainty. Measurement wait time can be added, but this may slow measurement time significantly or make some measurements impractical or impossible. Settling time error may effect the measurement of off-state drain current, gate and/or substrate current. Measures wafer chuck (or probe) capacitance. Figure 1 and Figure 2. c. Use the MARKER/CURSOR to highlight file named cap1.dat g. Press Graph/List View the stored display of thermal chuck capacitance at the top area of the 4156 display. Note the example capacitance value is < 2 pf. 4

5 In this test, a ramping voltage is applied and capacitance is calculated using the I =C* V/ T formula. Residual or stray capacitance effects measurements in several ways: 1. Adds error to capacitance measurements 2. May require frequent re-zeroing of the LCR meter -slows wafer throughput. 3. Adds an error current to DC measurements. 4. May cause V/ T current errors in swept measurements. 5. Results in slower measurements h. Press Append Live test begins running. Note your systems performance. For high resolution parametric measurements, the measurement area should exhibit low stray or residual capacitance. This is critical for accurate and fast measurement of device off-state currents, junction and/or oxide leakage. For vertical wafer measurements such as gate or substrate current, a good low-capacitance parametric chuck is highly desirable. 5. Leakage - dielectric absorption: Leakage.dat Measures wafer prober chuck or probe needle dielectric absorption and leakage. Figure 1 and Figure 2. NOTE This test uses +/- 100 volts - for 100 V output, the 4156 safety interlock must be enabled. The 4156 interlock cable, Cascade part number (See Figure 3), can be used with the interlock connection on Cascade probers. If interlock cannot be enabled then reduce the 4156 test range to +/- 40 V or less. fig. 3 c. Use MARKER/CURSOR to highlight file named Leakage.dat g. Press Graph/List View the display taken on a parametric chuck 5

6 In this test, the voltage is swept from 100V to + 100V and then back in 0.5V increments. Ideally, current should be zero (with no separation or hysteresis in display trace) since no current path should exist. A very good parametric displays very low current (< 100fA/100V) and hysteresis (< 100fA). h. Press Append - Live test begins running. Note your systems chuck performance High capacitance, dielectric absorption, or leakage show up respectively as display offset, hysteresis or slope over the course of the sweep. Summary Many factors must be considered in assuring high quality electrical contact to on-wafer semiconductor devices. The entire measurement path and surrounding environment must be designed to compliment the broad capability of today s precision test instrumentation. For more info on Cascade Microtech products, please visit us at Cascade Microtech, Inc., 2430 N.W. 206th Ave., Beaverton, OR Toll Free: Phone: Fax: Europe: Japan: sales@cmicro.com 6 Copyright 2004 Cascade Microtech, Inc. All rights reserved. No part of this manual may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopy, recording, or any information storage and retrieval system, without permission in writing from Cascade Microtech, Inc. All trademarks are the property of their respective owners. 4156EVAL-APP-0404 Data subject to change without notice