IPC/NEMI Symposium 0n Lead-Free Electronics. I. Boguslavsky September 19, 2002 Montreal, Canada

Transcription

1 Whiskers: Truth and Mystery IPC/NEMI Symposium 0n Lead-Free Electronics I. Boguslavsky September 19, 2002 Montreal, Canada

2 Pure Tin Grows Whiskers Bulk tin Vapor deposited tin films on inert substrates: paper, mica, plastics Metallurgically prepared polished 50%Sn/50%Al discs Electroplated foils delaminated from the substrate Electroplated tin and tin alloys No whiskers on high purity tin such as zone refined tin or on single crystal of tin 2

3 Whisker Types Column Needle Hillock Flower or OSE Needle growing out of hillock Needle growing out of OSE SEM photos courtesy of P. Bush, SUNY at Buffalo 3

4 Whiskers on 90 Sn/ / 10 Pb Secondary Electron Image (SEI) Backscattered Electron Image (BEI) Substrate: C194 Test conditions: 1500 thermocycles 55C to +85C SEM photos courtesy of P. Bush, SUNY at Buffalo 4

5 Literature Review Most of the studies on whiskers were carried out and published before the end of 1970 s The following information is available from those papers Theories of the whisker growth mechanism 1,2,3 Effect of plating process parameters on whisker growth Current density 4,5 Temperature 6 Mechanical and chemical preparation of the substrate 7 Substrate material 6,8,9 Concentration of the bath components and particularly additives 1,4,5,6 5

6 Literature Review (continue) Additional information available Deposit thickness 6,10 Underlayers 8,11 Post-plating annealing 2,4,11 Crystallographic structure of whiskers 2,9,12 Deposit structure 4, 13,14 Effect of alloying tin 10,15,16,17 Lead reduces propensity to whisker as low as 1% Even deposits with 10% Pb grow whiskers, although, less than pure tin Only 60%Sn/40%Pb is immune to whiskering 6

7 Literature Review (continue) Most recent publications presented in the NASA website list maybe divided into two categories: About 60%: Case studies for whisker growth and warning against using pure tin deposits from high reliability users (J. Brusse 18 and J. Kadesch 19 dominate the list) About 20%: studies related to whisker testing of various components plated with tin and tin alloys About 20%: Publications from suppliers of plating chemicals and industry consultants promoting their processes and expertise The last type of studies created mystery and confusion 7

8 Current Status of Plating Processes Plating chemicals suppliers significantly improved plating processes by: Testing various additives Utilized recommendation from various publications (improved substrate pre-treatment, post-plating annealing, Ni underlayer) However, no reliable deposit characterization has been done to substantiate those improvements. As an example, there is no method available for deposit internal stress (IS) measurement (probably, the most important deposit property) Standard contractometers are not sensitive enough for low value of IS in tin (<10 Mpa for Sn vs ~100 MPa for Ni and ~300 MPa for hard Au) XRD method 23 has % experimental error due to low IS values, large grain structure (crystal distribution is not random) and anisotropy (different material properties in different directions) for tin deposits 8

9 Production Trials Another source of conflicting results are production trials. Apparent reasons for that contradiction are No standard whisker test procedure No standard inspection procedure Different acceptance criteria. No whiskers as an acceptance condition leaves uncertainty for results interpretation Not so obvious reasons are: Variation in process parameters Pre-treatment: stamping versus etching; de-flashing, de-scaling Cell design: current density variation; agitation Contamination Possibly more than one mechanisms for whisker growth may explain different effect of accelerating tests on different deposits 9

10 Problem: to resolve whiskering and create whiskerfree lead-free finish, the industry rely on suppliers and component manufacturers who in most of the cases do not have enough expertise and capabilities (equipment, resources, etc.) to address this issue at an appropriate level. Need multi-disciplinary team of experts in the relevant fields and advanced equipment. 10

11 NEMI Fundamental Group Approach Short-term practical solution: establish whisker performance for best in class pure tin plating processes (supplied by component manufacturers) utilizing three test methods and combination of those Thermal cycling Ambient temperature and humidity Elevated temperature and humidity Use that information as a basis for specification for pure tin production processes Long-term study: Fundamental understanding of whisker growth mechanism and quantitative characterization of the factors affecting whiskering. Predict incubation period, growth rate and maximum length of whiskers based on measurable deposit properties and modeling 11

12 Work Done by NEMI Modeling Group 12

13 Verification of XRD Method Use of synchrotron radiation to determine the state of strain in tin coatings on integrated circuit packages. Dr. Peter W. Stephens Department of Physics & Astronomy Stony Brook University 13

14 XRD Study Advances in x-ray technology, especially use of synchrotron radiation, permit microscopic studies of materials that were previously impossible/difficult. The premise of powder diffraction is to look at a statistical sample of grains. Complements measurements of individual grains/whiskers such as recently discussed by K.N. Tu. Preliminary work was done on sample N2, QFP package (14mm x 14mm, 100 leads), ~10 _m matte tin over alloy A42 (fcc, a=3.588). Temp. and humidity cycled 1,500 times. Microscopy (Peter Bush) shows grains 5 50 microns. Films are evidently one grain thick. 14

15 Is the sample a good powder (observe many small grains)? Rock the sample at fixed diffraction angle Sn (220) reflection _ = 0.83 Angstrom _ X-ray counts per second NO! Large fluctuations imply that sample has relatively large grains. Also known from microscopy Sample angle theta (degrees) typ _m 15

16 Diffraction Pattern (low penetration) FWHM = 0.016º (220) I=3 10 mm Sn film on Cu IC lead l = 1.15 Å. X-ray Intensity (arb units) (200) I=100 (101) I=88 Substrate (111) not observed (211) I=59 (301) I=14 (112) I=18 (400) I=8 (321) I=16 Substrate (220) theta (degrees) 16

17 Diffraction Pattern (medium penetration) X-ray Intensity (arb units) (200) I=100 (101) I=88 10 mm Sn film on Cu IC lead l = 0.65 Å. Puzzle: why did (220) get so weak? The rest of the pattern is about the same as 1.15A. (211) I=59 (220) I=32 (301) I=14 (112) I=18 (400) I=8 (321) I= theta (degrees) 17

18 Diffraction Pattern (high penetration) X-ray intensity (arb units)40000 (200) I=100 (101) I=88 (220) I=32 Substrate (111) (211) I=59 10 mm Sn film on Cu IC lead l = 0.500Å. (301) I=14 (112) I=18 (400) I=8 (321) I=16 Substrate (220) theta (degrees) 18

19 Stress/Strain Measurements Diffraction patterns have been collected at three different wavelengths. Used this data to measure strain at different depths of the sample. Re-plotted three diffraction patterns as a function of sin(_)/_ = 1/2d, which should be the same for each diffraction line. Larger angles => shorter d-spacing 19

20 XRD Study 3000 Tin (200) Motorola 14mm square pack Nominal position, d = X-ray intensity (arb units) Å 0.65 Å 0.5 Å Deeper penetration -> Smaller d, negative strain. dd / d = sin(theta) / lambda = 1/2d 20

21 XRD Study Motorola 14mm square pack Tin (220), d = Å 0.65 Å 0.5 Å X-ray intensity (arb units) dd / d = Ni 40 Fe 60 (111) sin(theta) / lambda = 1/2d 21

22 Preliminary Conclusion of XRD Study 1. Sn films are severely textured. (220) or (321) appears to be the main growth direction. 2. Films are made of rather large grains, which makes accurate powder diffraction measurements difficult for characterization of both crystal orientation and deposit strain/stress. 3. It appears that the deeper material is compressed, perpendicular to the surface. Cannot tell if there is such compression along the surface direction. Estimated deep strain is equivalent to a stress of x 40 GPa = -40 MPa (compression) 4. More work requires more manpower. 22

23 Deposit Orientation: effect of current density 5 ASD 10 ASD 20 ASD Slide is courtesy of Shipley Co 23

24 XRD Method for Crystal Orientation In addition to internal stress, deposit crystal orientation is a critical parameter that may influence whisker growth There are two reasons that complicate the use of XRD orientation data for deposit characterization and prediction of whisker behavior: Large experimental error when powder diffraction method is used (similar to the interpretation of XRD data for strain stress measurements) Influence of plating process parameters on crystal orientation 24

25 XRD Method for Crystal Orientation To obtain reliable XRD data for deposit orientation by powder diffraction method, statistical approach should be utilized (multiple measurements of the same sample) XRD micro-probe can be used to determine orientation of individual grains The effect of plating process parameters on crystal orientation needs to be studied to define the operating window of a plating process which would provide the best whisker behavior 25

26 Auger Analysis of Tin Oxide Film Thickness: Effect of Aging Conditions Peter J. Bush SUNY at Buffalo 26

27 Auger Study of Oxide Film (sample after ambient conditions) Sn C O 27

28 Auger Study of Oxide Film (sample after 50C/85% RH) Sn C O 28

29 Whisker Growth Theories Surface oxidation develops stress that causes screw dislocations to move up. Counter-argument: whiskers grow in vacuum Stress caused by hydrogen incorporated in the deposit during electroplating drives whiskers. Counter-argument: whiskers grow on bulk tin and deposits produced from vapor phase. IMC formation, particularly, SnCu intermetallics, induce stress and increase propensity to whisker. Counterargument: whiskers grow if tin has no interface with copper Most of the theories emphasize only one factor influencing whiskers 29

30 Diffusion theory Diffusion theory provides the most comprehensive approach Electrodeposits have higher amount of crystal defects (vacancies and dislocations) than metallurgical material. Those defects in combination with high mobility of tin atoms explain fast diffusion in tin deposits Diffusion mechanism: for each tin atom leaving the deposit, the vacancy is formed below the root of the whisker which is then absorbed by long-range diffusion of tin atoms Macro- and micro-stress, surface energy, stored strain energy and crystal orientation may influence vacancy diffusion 30

31 Atomic Diffusion Mechanism Simulation of the vacancy mechanism Simulation of the direct interstitial mechanism 31

32 Diffusion theory Two-stage model: 1 st stage dislocation loop formation and expansion within the deposit (gliding). Vacancies are formed and absorbed by lattice and grain boundary diffusion and through dislocation pipes SEE 34 2 nd stage dislocation loop climbing (upwards) by motion of grain interior atoms, grain boundary atoms and grain boundary sliding. The latter mechanism is dominant for deposits with columnar grains and flat grain boundaries like tin SEE 35 If the 1 st stage is the rate-limiting process, the whisker growth rate maybe described by the equation: h 2Ds s z v = R kt w a Where D s is self-diffusion coefficient s z is stress normal to the surface V a is atomic volume R w is whisker radius 32

33 Dislocation Loop Gliding Slide courtesy of Ohio State University web-site 33

34 Dislocation Loop Climbing b d Slide courtesy of Ohio State University web-site 34

35 Diffusion theory For the 2 nd stage (climbing), the main obstacle is intersection with forest dislocations that creates opposing pressure - s i. The whisker growth rate maybe described by the equation h = k( s -s ) i n Where k and n depend on the temperature but not stress s is the stress driving whisker growth s i depends on deposit structure For low stress, the 2 nd stage is rate-determining and whisker growth strongly depends on internal and external stress and somewhat on temperature For high stress, the 1 st stage is dominant and the whisker growth rate has reverse dependence on temperature (higher rate at RT) 35

36 Features Grown on the Scratched Area Scratching is plastic deformation process that creates numerous dislocations.. Step-wise growth visible on the side correlates with diffusion theory SEM photos courtesy of P. Bush, SUNY at Buffalo 36

37 Step Growth and Voids Around Whisker Base Whisker test conditions: 55C dry heat, 4 weeks Substrate: A 42 Note steps at the whisker base 37

38 Calculation of Stress along Grain Boundaries (Modeling) Balasubramaniam Radhakrishan Oak Ridge National Lab 38

39 Calculation of Stress along Grain Boundaries (Modeling) Estimation of normal force was done for the existing deposits with various degree of whiskering based on their XRD spectra (crystal orientation) The normal strain, e 33, was calculated for each grain orientation The difference in normal strain between two grain orientations is proportional to the stress acting to promote whiskers Larger the difference in normal strain, greater is the probability for initiating whisker growth (Lee and Lee, 1998) 39

40 Calculated Results Orientation e'_33 Intensity Normal force Good practice whiskers per lead No whiskers

41 Discussion of the Results For the good practice texture, set 1, the magnitude of normal strain difference is significantly smaller than for the set 2 (40 whiskers per lead) and set 3 (no whiskers). Sets 2 and 3 represent the same plating process The presence of <200> component (which has the highest normal strain) in set 2 and set 3 results in a larger magnitude of strain difference that in set 1 and, thus, higher probability of whiskering Volume fraction of <200> component in set 3 is lower than in set 2; therefore the probability of finding grains with high stress is lower Based on the components that have the highest volume fractions (probability of finding such pairs of grains in the samples is high), the stress value for set 1 is significantly lower than in sets 2 and 3. Good correlation of the modeling results with practice confirms the feasibility of the approach 41

42 Future Work Determine deposit characteristics that affect whisker growth and techniques for their measurements. Develop a model for whisker growth mechanism. Based on the modeling, predict long-term whisker performance: Incubation period, growth rate and maximum length (whisker seizure mechanism) Use the knowledge of whisker growth mechanism to identify accelerated test conditions and determine accelerating factor To coordinate various types of activities and thoroughly conduct the experiments, long-term planning and appropriate funding should be done 42

43 Capabilities of NEMI Modeling Group NEMI Fundamental Group has a unique set of expertise and capabilities and unbiased attitude necessary to resolve such a complex problem as whisker growth. It includes experts in the following areas: Electroplating chemistries and processes SEM characterization of whiskers (quantitative approach) XRD for macro- and micro-stress measurements and crystal orientation Crystal mechanics, modeling and computation Metallurgy Material science Engineering/manufacturing 43

44 References 1. R. Kawanaka, K. Fujiwara, S. Nango and T. Hasegawa, Influence of Impurities on the Growth of Tin Whiskers, Japanese Journal of Applied Physics, vol. 22, pp , March 19, B.-Z. Lee, D. N. Lee, Spontaneous Growth Mechanism of Tin Whiskers, Acta Met., vol. 46, pp , U. Lindborg, A Model for the Spontaneous Growth of Zn, Cd, and Sn Whiskers, Acta Met., vol. 24, p. 181, A. Selcuker, and M. Johnson, Microstructural Characterization of Electrodeposited Tin Layer in Relation to Whisker Growth. Capacitor and Resistor Technology Symposium: CARTS, pp 19-22, October, L. Zakraysek, et. Whisker Growth from a Bright Acid Tin Electrodeposits, Plating and Surface Finishing, vol. 64, pp , V.K. Glazunova and N.T. Kudryavtsev, An Investigation of the Conditions of Spontaneous Growth of Filiform Crystals on Electrolytic Coatings Translated form Zhurnal Prikladnoi Khimii, vol 36, no. 3, pp , March

45 References 7. D. Endicott, K.T. Kisner, A Proposed Mechanism for Metallic Whisker Growth, Proceedings of AESF SURFIN Conference, July, M. Endo, S. Higuchi, Y. Tokuda and Y. Sakabe. Elimination of Whisker Growth on Tin Plated Electrodes, Proceedings of the 23 rd International Symposium for Testing and Failure Analysis, pp , October 27-31, P. Harris, The Growth of Tin Whiskers, ITRI, pp. 1-19, N. A. J. Sabbagh, H.J. McQueen, Tin Whiskers: Causes and Remedies, Metal Finishing, March S.C. Britton, Spontaneous Growth of Whiskers on Tin Coatings: 20 Years of Observation, Transactions of the Institute of Metal Finishing, vol. 52, pp , April 3, N. Furuta and K. Hamamura, Growth Mechanism of Proper Tin-Whisker, Journal of Applied Physics, vol. 8, no. 12, pp , December 1,

46 References 13. K. Fujiwara, M. Ohtani and T. Isu, Interfacial Reaction in Bimetallic Sn/Cu Thin Films, Thin Solid Films, vol. 70, pp , T.Kakeshita, R. Kawanaka and T. Hasegawa, Grain Size Effect of Electro-Plated Tin Coatings on Whisker Growth, Journal of Materials Science, vol. 17, pp , P.L. Key, Surface Morphology of Whisker Crystals of Tin, Zinc and Cadmium, IEEE Electronic Components Conference, pp , May 1970 or S. M. Arnold, Repressing the Growth of Tin Whiskers, Plating, vol. 53, pp , K.M. Cunningham and M.P. Donahue, Tin Whiskers: Mechanism of Growth and Prevention. 4 th International SAMPE Electronics Conference, p.569, June, J. Brusse, et, Tin Whiskers: Attributes and Mitigation, CARTS, March 2002, pp

47 References 19. J. S. Kadesch, J. Brusse, The Continuing Dangers of Tin Whiskers and Attempts to Control them with Conformal Coat, NASA s EEE Link Newsletter, July B. Hom, S. Winkler, Back to the Future: A Look at the Past Reveals a Lead-Free Drop-In Replacement, MEPTEC Report, March-April Y. Zhang, et., An Alternative Surface Finish for Tin/Lead Solders Pure Tin, SMI 96 Proceedings, Sept. 1996, p R. Schetty, Tin Whisker Study Experimentation and Mechanistic Understanding, Proceedings of AESF SURFIN Conference, June 2002, pp Y. Zhang, et., Understanding Whisker Phenomenon: Whisker Index and Tin/Copper, Tin /Nickel Interface, Proceedings of IPC SMEMA Counsel APEX, pp. S061-1 through 11, January