&liedsignal AEROSPACE. of DuPont 2000 Series Resistors. Characterization of Printing and Laser Trimming. on DuPont 951 Green TapeTM

Transcription

1 Ch&-9bi66-5 Characterization of Printing and Laser Trimming of DuPont 2 Series Resistors on DuPont 951 Green TapeTM Federal Manufacturing & Technologies Howard Morgenstern, Sam Bandler, and Gregg Barner KCP Published October 1996 Approved for public release;distribution is unlimited. Prepared Under Contract Number DE-AC4-76-DP613 for the United States Department of Energy DIgRIBmIO! d OF?HIS DOCUMNT 1s &liedsignal AEROSPACE R

2 DISCLAIMER This report was prepared as an account of work sponsored by an agency of the United States Government Neither the United States Government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade names, endorsement, trademark, manufacturer, or otherwise, does not necessarily constitute or imply *%s recommendation, or favoring by the United States Government or any agency t h d. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof. Printed in the United States of America. This report has been reproduced from the best available copy. Available to DOE and DOE contractors from the Office of Scientific and Technical Information, P.. Box 62, Oak Ridge, Tennessee 37831; prices available from (615) , FTS Available to the public from the National Technical Information Service, U. S. Department of Commerce, 5285 Port Royal Rd., Springfield,Virginia Copyright 1996 by AlliedSignal Inc. The Government is granted for *&elf arc6 others acting on its behalf a paid-up, nonexclusive, irrevocable worldwide license in this data to reproduce, prepare derivative works, and perform publicly and display publicly. Aprime contractor with the United States Department of Energy under Contract Number DE-ACOd76-DP613. AlliedSignal Inc. Federal Manufacturing &T&mologies P.8-Box KansasCity, Missouri M44-6%9

3 KCP Distribution Category UC-76 Approved for public release; distribution is unlimited. CHARACTERIZATION OF PRINTING AND LASER TRIMMING OF DUPONT 2 SERIES RESISTORS ON DUPONT 951 "GREEN TAPETM" Howard Morgenstern, Sam Bandler, and Gregg Barner Published October 1996 Paper submitted to International Society for Hybrid Microelectronics 29"' International Symposium on Microelectronics October 6-1, 1996 Minneapolis, MN &lliedsig nal AEROSPACE

4 DISCLAIMER Portions of this document may be illegible in electronic image products. Images are produced from the best available original document.

5 Characterization of Printing and Laser Trimming of DuPont 2 Series Resistors on DuPont 951 Green TapeTM Howard Morgenstern, Sam Bandler, and Gregg Barner AlliedSignal Federal Manufacturing & Technologies* P.O.Box Kansas City,MO Phone: ; Fax: ; hmorgenstem@cp.com Abstract DuPont 2 series resistors were reviewed andfound to come closest to our requirement of I % resistor tolerance over the expected 3-year li$e of our products. The evaluationpeflormed involved the characterization of both the printing and trimmingprocesses. The printing process was characterizedfor firing temperature,print thickness, print direction, resistor geometry and encapsulant eflect. Laser trimming was characterized by Jirst finding an operating envelope and then selecting an operating point. The envelope was located by varying the trimming parameters and determining their acceptability to electrical and visual criteria. Samplesfiom both the envelope and operatingpoint were environmentallyconditioned m e conditioning included thermal shock temperature cycle, 1-hour temperatureaging, 1-hour humidity aging, and a simulated goldtin solder reflow. Key Words: Multichip module, Multichip module ceramic (MCM-C), LTCC substrate, manufacturability, DuPont 2 series resistors, resistor trimming All data prepared, analyzed and presented has been developed in a specific context of work and was prepared for internal evaluation and use pursuant to that work authorized under the referenced contract. Reference herein to any specific commercial product, process or service by trade name, trademark, manufacturer, or otherwise, does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government, any agency thereof or AlliedSignal Inc. *Operated for the United States Department of Energy under Contract No. DE-AC4-76-DP613. Introduction As part of an effort to develop a capability for low-temperature cofred ceramic (LTCC) networks, an evaluation was initiated to characterize DuPont 2 series resistors on DuPont s 951 LTCC material, This evaluation established design and manufacturing parameters for this system. Included in this evaluation were all six resistivity compositions (1, 1, lk, 1% 1% and 1M ohm per square) and both printing and laser trimming processes. Resistor Printing and Firing Evaluation The objective of this printing and firing evaluation was to determine the following characteristics: 1. Verification of manufacturer s print thicknesses recommendations and compatibility with resistor firing and trimming processes. 2. Compatibility of resistor physical design requirements with manufacturing processes. 3. Sensitivity of resistance to variations in manufacturing processes including firing, encapsulation (overglaze), and print thickness. 4. Effects of geometry variations on untrimmed resistance.

6 Test Sample Configuration and Fabrication The evaluation consisted of printing test samples using a two square resistor test pattern with each resistor having a length-to-width ratio of two. The test pattern shown in Figure 1 provided resistors of.635 mm (25 mils), 1.27 mm (5 mils), and 2.54 mm (1 mils) widths oriented with the resistor lengths parallel and perpendicular to the direction of squeegee travel to simulate actual production. The two square test patterns provide resistors which have nominal untrimmed resistances of twice the nominal as-fred sheet resistance. Each test pattern contained two resistors of each width in each orientation. Test patterns were printed on substrates containing nine test patterns each. The test sample substrates were comprised of six layers of DuPont 951AX LTCC tape laminated and cofred to a nominal thickness of 1.27 mm (5 mils). Figure 1. Two Square Resistor Test Pattern Test samples for each resjstor composition were prepared with and without encapsulation and fred at 825, 85, and 875OC. The complete sample size for each resistor composition, encapsulation, and fring combination was eight substrates (72 test samples) generating a total of 144 resistors for each geometry. The test samples were printed to the dried target thickness range recommended by DuPont of 18 to 2 microns. This lower thickness range was recommended to enhance post laser trim stability by minimizing the laser power required to trim the resistors. Test samples were prepared both with and without glass encapsulation for comparison of laser trim parameters. Sintering (firing) was performed using a 3-minute, 85 C, time and temperature profile and a 6-minute, 85OoC, time and temperature profile that are currently used at AlliedSignal Federal Manufacturing & Technologies (FM&T). -Both profiles provide a peak temperature of 85OoC with a time at peak temperature of 1 minutes. The 3-minute and 6-minute profiles maintain approximate ascent and descent rates of 1 and 5OoC per minute, respectively. Sensitivity of the resistor compositions to firing process variations was further evaluated using 825OC and 875"C, 3-minute profiles. Print Thickness The manufacturer's recommended target dried thickness for DuPont 2 series resistors is 18 to 2 microns. The evaluation established target wet printed thicknesses needed to obtain a dried thickness of 18 to 2 microns. A comparison was made among wet (as-printed), dried, and fired thickness and resistivity for each paste composition. Normalized Sheet Resistance and Design Considerations In order to eliminate the effects of print thickness and print width variations on fred sheet resistivity determinations, the untrimmed resistance measurements are normalized to account for these variations. This calculation is used to determine typical as-fired sheet resistivities for each resistor composition and subsequently the resistor nominal design values incorporated into product definitions. The form of the equation is to equate the normalized resistance to the resistance measured divided by the number of squares in the pattern, multiplied by a term to normalize for printed thickness and multiplied by a term to normalize for printed resistor width. This is expressed mathematically in the following equation: EM = The Average As-Measured Resistance?;, = Average Dried Thickness TR = Reference Dried Thickness (2 microns) W, AverageFiredWidth WR = Reference Fired Width where the fired width depends on the pattern being examined and is either.635 mm (25 mils), 1.27 mm (5 mils), or 2.54 mm (1 mils) =

7 The normalized resistance data was collected for each resistor composition as both a percent fiom nominal and actual resistance. Effects of Manufacturing Variations As part of the overall evaluation of the sensitivity of DuPont 2 series resistor compositions to variations in manufacturing processes such as firing, encapsulation, thickness, and geometry, electrical resistance measurements were made on as-fred (untrimmed) resistors. The untrimmed resistance data was collected for each resistor composition as both a percent fiom nominal and actual resistance. Firing Sensitivity Data was collected for the untrimmed sheet resistance as a percent of the nominal sheet resistance for each resistivity and firing profile combination. This data indicates DuPont 2 series resistors printed to a dried target thickness of 18-2 microns range fiom nominal to 37 percent below nominal when fred using a 3-minute, 85 C profile. These values shift by typically 1-2 percent when fired using 825 or 875'C profiles. Data on the normalized sheet resistivity as a percent of the nominal sheet resistance for. each resistivity and firing profile was collected. This data indicates DuPont 2 series resistors, normalized to a dried target thickness of 2 microns, range fiom 15 to 55 percent below nominal when fred using a 3-minute, 85'C profile. These values shift by up to 2 percent when fred using 825 or 875 C profiles. Resistivities normalized to 19 microns would range from 1 to 5 percent below nominal. Encapsulation Effects Virtually no difference in resistivities was observed between encapsulated and unencapsulated resistors for all six resistor decade compositions. However, trim parameters yielding acceptable visual and electrical results were significantly reduced with encapsulated samples. Thickness Sensitivity Data was collected on the variations in untrimmed resistance as a percent fiom nominal for various dried thicknesses for each resistor composition. It showed that the thickness is inversely proportional to resistance for all resistor compositions. The dried thickness target range of 18 to 2 microns was maintained by printing the wet thickness within the range of 32 to 38 microns for each resistor composition. Effects of Resistor Geometry and Orientation Data was collected on untrimmed resistance as a percent fiom nominal for each resistor geometry and orientation combination. The geometries tested were.635 mm (25 mils), 1.27 mm (5 mils), and 2.54 mm (1 mils) minimum resistor dimensions (resistor width). The orientation refers to the direction of the resistor length compared to the direction of the squeegee travel. The data shows that the effect of resistor orientation was negligible with all geometries exhibiting a spread of typically 5 percent between resistors parallel and perpendicular to the squeegee travel. The flatness of LTCC substrate material would contribute to this consistency Resistor geometry has an effect on resistance with most paste compositions displaying a 1 to 2 percent spread between the various resistor geometries. Overall Printinwiring Capability Data was collected for print variability within the lot for as-fired resistance of.635 mm (25 mils), 1.27 mm (5 mils), and 2.54 mm (1 mils) resistors. The data was analyzed for both parallel and perpendicular orientations. The 3 sigma limits were typically f 2% except for the 1-ohm composition which was f 5%. Accomplishments of Printing Evaluation The evaluation established the following material characteristics, design criteria, and manufacturing parameters for the six DuPont 2 series resistor compositions: 1. Target print thicknesses for each of the six pastes ranged fiom 32 to 38 microns to yield optimum as-fired resistivities and maximum post-trim resistor stability. 2. Resistors should be designed to 7% of trimmed nominal value for FM&T designs. 3. Average resistivities for all six DuPont 2 series pastes ranged fiom nominal to 38% below nominal. 4. Resistivities for all six pastes were unaffected by encapsulation (overglaze). 5. Resistivities were virtually unaffected by resistor orientation but were more dependent on resistor geometry. 6. Overall print lot consistency (f 3 sigma) was k 2% fiom average fired values for all pastes except 1-ohm,which was f 5%.

8 times on trimmed resistors. The 24-hour drift test was performed on these samples prior to the brazing test. Laser Trimming Characterization The laser trimming process was characterized by using samples printed with DuPont 211 (logi), 221 (loon),23l(lksz), 241(1KQ), 251 (look!2), and 261(1Ma) pastes on DuPont 951 LTCC substrates. DuPont 221,241, and 251 resistor pastes were used to print resistors on 144 test samples on DuPont 951. DuPont 211, 231, and 261 pastes were used to print resistors on 72 test samples on DuPont 951. The resistors were fired at 85 C for a 3-minute profile. Each test sample, shown in Figure 1, contains twelve resistors with a length-to-width aspect ratio of 2:l. There are three different resistor widths,.635 mm (25 mils), 1.27 mm (5 mils), and 2.54 mm (1 mils), on the sample. The samples of 221,241, and 251 were divided into two groups of 72. The samples fiom one of the groups were printed with DuPont 5415D laser trimmable glaze cover coat. The resistors on twenty samples each, glazed and unglazed, were laser trimmed using various combinations of pulse rate, beam speed, and power. Two resistors were cut using each combination. The.635-mm (25-mil) resistor was cut completely in half, and the 1.27-mm (5-mil) resistor was cut to raise its value 1.4 times its original value. The 2.54-mm (1-mil) resistor was left untrimmed to be used for reference. The trimming was accomplished on a Teradyne Model W419 laser trim system. Twenty samples printed with 211, 231, and 261 were treated the same way except there was no subset with glaze. The acceptable operating range of the laser was determined from these samples. The laser trims were evaluated using visual criteria, isolation resistance, and stability as determined by tests described later. Having established the operating range, an operating point was selected, and twenty additional samples were trimmed at this point. Glazed and unglazed samples printed with 221,241, and 251 were trimmed. Only unglazed samples printed with 21 1, 23 1, and 261 were trimmed They were tested and evaluated like the fmt group to determine the tolerance and yield that might be expected. Additional samples were laser trimmed to evaluate the effects of brazing temperatures and. Eighteen more samples were laser trimmed at the operating point to evaluate the effects on trimmed resistor stability of elevated temperature and humidity. The 24-hour drift test was performed on these samples prior to the elevated temperaturehumidity test being performed. Test Descriptions The samples trimmed for the operating envelope and the samples trimmed at the operating point were subjected to the following tests: The resistor values on all samples, trimmed and untrimmed, glazed and unglazed, were measured on a Teradyne W419 laser trim system withim a few seconds of having been trimmed. This measurement will hereafter be referred to as T,,. The samples were remeasured again 24 hours later, and the change in resistance was calculated. This was done to determine the ability to repeatably measure them and to obtain some measure of their off-the-shelf stability. In past studies of other resistor pastes, the largest change in resistance typically occurred during the first 24 hours. Each sample was then subjected to a solder pot shock test. This test involved placing each test sample individually * on molten solder heated to approximately 268 "C for 15 seconds. The sample was removed and placed on a stainless steel table top at approximately 25 "C and allowed to cool. All of the resistors were remeasured, and the change in resistance, compared to To, was calculated. The test samples were then subjected to a temperature cycle test. The test apparatus consisted of two interconnected test chambers. One chamber was set to -5 OC while the other was set to 125"C.A cycle consisted of moving the test samples fiom one chamber to the other and then back. The approximate transport time was five seconds. The samples were held at each temperature a minimum of 15 minutes. Upon completion of this test, the resistors were again measured, and the change fiom Towas calculated. One unglazed sample fiom each paste decade was removed fiom the samples trimmed at the operating point. Trimmed and untrimmed resistors were tested fiom each sample. These were used to determine the temperature coefficient of resistance

9 (TCR) for the resistors from each paste decade. The TCR test involved measuring the resistors at 25 C. The samples were then heated to 125OC and allowed to stabilize at that temperature for approximately 3 minutes. The resistors were measured at this temperature. The samples were then cooled to -55 C and allowed to stabilize at this temperature for approximately 3 minutes. The resistors were measured while at this temperature. The following formulas [I] were used Hot TCR(ppdC) = Cold TCR(ppd'C)= R12SLR25' R2S. x 1, (R- 5 5 ) - R25' x (-12,5) R25' This cycle was repeated three times. The calculated TCR was then averaged for the three readings. The samples were subjected to a 1-hour aging test. The samples were placed in an oven heated to 15 C. At intervals of 2, 6, and 1 hours, the samples were removed fiom the oven and allowed to cool to room temperature. The resistors were measured, and the change compared to Towas calculated. For the 85/85 test, samples were subjected to 1 hours at 85 C and 85% relative humidity (RH). They were removed fiom the oven at 2, 6, and 1-hour intervals to measure resistors. The resistors were measured, and the change in resistance, compared to To,was calculated. This test was undertaken to measure the effect of humidity on both the resistor material and the ceramic substrate material. The brazing test was used to determine the effect of brazing temperature on resistor stability. The resistors were measured after the test, and the change in resistance, compared to To, was calculated. A four-zone belt furnace was used. The four furnace zones were set at 36, 355, 355, and 34OoC, respectively. The belt speed was set at 4 inches per minute for 221, 241, 251 and 261 samples. It was set to 8 inches per minute for the 21 1 and 23 1 samples. In all cases the nitrogen flow was set to 2 liters per minute. Laser Trim Requirements Two main requirements are used when considering the ability of a combination of laser parameters to make an acceptable trim. These are visual requirements and isolation resistance. The visual requirements, when viewed at 3X magnification, include: e Laser kerf fiee of obvious debris. e Laser kerf fiee of bridges (continuous e laser kerf). No continuous grooving of the ceramic substrate in the kerf. The width of reflowed resistor material on either side of the kerf shall be less than the kerf width. Isolation resistance is the value of the resistor that has been cut completely in half. Depending upon the combination of laser parameters selected, the value of the isolation resistance will range fiom the design value of the resistor to some greater value (typically greater than 32 M ohms). As the combination of laser parameters is adjusted to cause increasing energy to be applied to the resistor, the value of the isolation resistance increases from a minimum to some lsirger value. Increasing laser energy cuts deeper into the resistor. At some value of laser energy, the resistor material is removed completely fiom the laser kerf, and there is a several orders of magnitude increase in resistance. This is the minimum laser energy considered sufficient to trim resistors. Operating Envelope The lower end of the operating envelope is determined by the combination of laser parameters that result in the minimum energy necessary to produce an isolated trim and meet the visual requirements for kerf cleanliness. The upper end of the operating envelope is determined by the combination of laser parameters that result in the maximum energy that can produce laser trims that meet all the requirements. Analysis of Resistor Trimming The following observations are made based on the data collected e The operating envelopes for resistors printed with 231 and 261 pastes are relatively large. This allows variations in the trim process in response to changes in the printing process, i.e. print thickness. e The operating envelope for resistors printed with 21 1 is much smaller. This allows for virtually no variation in the printing/firing process.

10 The TCR for the unglazed samples of 25 1 at cold were more negative than the other pastes (approximately -6 ppm/ OC versus approximately -35 ppm/oc). Resistors made from 251 were not as stable as those made fiom 221 and 241. This is not a totally unexpected result. The 1OOK-ohm paste in other series acts much the same. greater recommended tolerance is listed. It is expected that the recommended tolerance will approach the absolute tolerance as more parts and data points are measured with the following exceptions: TCR was determined to be within the manufacturer s specification of f1 ppmpc. Laser trimming had no significant effect on TCR Data analysis produced the following recommendations for FM&T DuPont 5415D glaze will not be used. It improves resistor stability but reduces the trimming operating range and the yield. Incomplete laser kerfs and poor isolation resistance were common on the samples trimmed at the operating point. The minimum resistor dimension shall be 1.27 mm (5 mils), because the.635-mm (25-mils) wide resistors are less stable than the 1.27-mm (5-mils) wide resistors. Preliminary estimates of resistor tolerances have been summarized in Table 1. The f absolute tolerance is the tolerance within which the values of all the resistors in the sample fell at the conclusion of the aging tests. Because of the small sample size in this study, a TABLE (1ohm) E S I 221 (1 ohm) S T,231 (IK ohm) R P A S T E S - If resistors made fiom the 261 paste are subjected to brazing temperatures and times, the tolerance will not be tighter than e%. Conclusions Analysis of the printing and laser trimming of DuPont 2 series resistors on DuPont 951 Green Tapem9 was completed. Design and manufacturing criteria were established for both processes. This material combination was found to come closest to our requirement of resistor tolerance over the expected 3-year life of our product.. Note: It is the author s understanding that DuPont will be changing its designation of resistors series for use on Green Tapem fiom a 2 to a 7 series designation. Reference: [11 W o n t Electronics, Resistor Test Method G RESISTOR TEST MATRIX FOR DUPONT 2 SERIES THRU 1HOURS R If resistors made fiom the 211 and 25 1 pastes are subjected to brazing temperatures and times, tolerances will not be tighter than 35%. BRAZETEST 5% 5% It TEST,~~~ 251 (1K ohm) 261 (1Mohm)

11 Sam W. Bandler received a B.S. degree in Industrial Technology fiom Pittsburgh State University. He is currently a Staff Process Engineering Specialist with AlliedSignal Federal Manufacturing & Technologies. He has worked with all phases of microelectronic network fabrication and has spent most of his career investigating and characterizing laser trimming of resistors. Mr. Bandler is a licensed Professional Engineer and is a member of the International Society of Hybrid Microelectronics. Gregg E. Barner received a B.S. degree in Mechanical Engineering fiom the University of Kansas. He is currently a Staff Engineer with AlliedSignal Federal Manufacturing & Technologies. He has worked in all phases of microelectronic network fabrication and most recently has worked on developing printing techniques for multichip module technology. Mr. Barner is a member of the American Society of Mechanical Engineers. Howard Morgenstern received a B.S. degree in Ceramic Engineering fiom Alfied University. He is presently a Staff Engineer with AlliedSignal Federal Manufacturing & Technologies. He has worked in all phases of microelectronics fiom production to advanced development. He has most recently worked on a multichip module technology development team. Mr. Morgenstern is a member of the American Ceramic Society, National Institute of Ceramic Engineers and International Society of Hybrid Microelectronics.