1

MATERIALS SCIENCE AND TECHNOLOGY NEWSLETTER

Vol. 3, No. 4 FALL-WINTER 2006

Dr. Robert H. LacombeChairmanMaterials Science and TechnologyCONFERENCES, LLC3 Hammer DriveHopewell Junction, NY 12533-6124Tel. 845-897-1654, 845-227-7026FAX 212-656-1016E-mail: rhlacombe@compuserve.com

FOCUSING ON HIGH TEMPERATURE POLYMERS, THIN FILMS AND

ADHESION

IN THIS ISSUE

EDITORIAL COMMENTS . . . . . . . . . . . . . . . . 2

TALES OF THE DEVELOPMENT LABORATORY: THE CASE OF THE WRONG OVEN . . . . 2

Fracture Mechanics 101 . . . . . . . . . . 3

The Mystery Unwrapped . . . . . . . . . . 5

Solving the Riddle of ExcessiveStress in the Polyimide . . . . . . 5

Statistical Physics of Polymer Liquids 6

Epilog:

More Statistical Physics and the Valueof Stress Modeling . . . . . . . . . 7

Statistical physics of rigid rods

Value of stress modeling andfracture mechanics (8)

COURSES AND CONSULTING ON ADHESIONAND ADHESION MEASUREMENT . . . . 9

Adhesion Courses . . . . . . . . . . . . . . . 9

Consulting . . . . . . . . . . . . . . . . . . . . . 9

SYMPOSIA ON HIGH TEMPERATUREPOLYMERS AND ADHESION ASPECTSOF THIN FILMS . . . . . . . . . . . . . . . . 10

CALL FOR PAPERS:

FIFTH INTERNATIONAL SYMPOSIUMON POLYIMIDES AND OTHERHIGH TEMPERATURE POLYMERSAND THE

THIRD INTERNATIONAL SYMPOSIUMON ADHESION ASPECTS OF THINFILMS INCLUDING ADHESIONMEASUREMENT AND METALLIZEDPLASTICS . . . . . . . . . . . . . . . 10

2

EDITORIAL COMMENTS

This issue of the Newsletter will introduce theupcoming FIFTH INTERNATIONAL SYMPOSIUMON POLYIMIDES AND OTHER HIGHTEMPERATURE POLYMERS and THE THIRDINTERNATIONAL SYMPOSIUM ON ADHESIONASPECTS OF THIN FILMS (INCLUDINGADHESION MEASUREMENT AND METALLIZEDPLASTICS). These two topics are of particularinterest to the conference Director Dr. Mittal andmyself as we spent many years together in thedevelopment laboratory of the IBM Corporationwrestling with the problem of using the polyimidematerials as dielectric layers in microelectronicstructures. This application of the polyimidematerials was quite challenging in that it involveda whole gamut of materials related questionsincluding:

< Chemical stability of polyimides< Thermal stability of polyimides< Electrical properties of polyimides< Mechanical stability of polyimides< Adhesion of polyimides to metals,

semiconductors and ceramics

What better list of challenges could a materialsscientist want? Furthermore, in addition to thehighlights just mentioned there were furtherpressing questions concerning the thermodynamicbehavior of the polyimides with regard to thesolubility of moisture and their barrier propertieswith regard to a whole rogues list of nastycorrosives such as sulphur dioxide, nitrous oxideand a host of other contaminants commonly foundin the air we breathe. At low concentrations thesegases are relatively innocuous to humans but caneasily destroy a metal line on the order of onemicron in width. All of this gave rise to excitingtimes in the development of polyimides as thin filmmaterials for electrical insulation and out of thiswork, both within the IBM company and elsewhere,arose the very first international symposium onpolyimide materials which was held in Ellenville,New York in November 1982. The meeting washeld under the auspices of the Mid-Hudson Sectionof the Society of Plastics Engineers and featured60 papers from 130 authors worldwide. A quickperusal of the Proceedings Volumes shows that1

already by 1982 the polyimide materials had founda wide range of applications outside themicroelectronics industry including highperformance adhesives for aerospace applications,

protective coatings for service up to 700 F andinsulating layers for implantable electrodes. Theupcoming Fall 2007 symposium to be held inOrlando Florida will cover these and many othertopics dealing with this most interesting and usefulclass of materials.

Of all the problem areas listed above, the one thatmost bedeviled the microelectronic applications wasthe question of adhesion of the polyimides tometals, semiconductors and ceramics. Thisquestion was touched upon in a previous issue ofthe Newsletterwww.mstconf.com/Vol2No2-2005.pdf and is oneof the central themes of the companion symposiumon Adhesion Aspects of Thin Films being held backto back with the polyimide symposium. The twosymposia thus mutually reinforce each other andmany participants will have a keen interest in both.

By way of introducing the topic of adhesionproblems with polyimide coatings I present thefollowing personal reminiscence typical of the typeof problem commonly faced by the developmentlaboratory especially when introducing newmaterials to the manufacturing line. The eventsdepicted are true but names have been withheld orchanged to protect the guilty.

TALES OF THE DEVELOPMENT LABORATORY: THE CASE OF THE WRONG OVEN

I believe it was an afternoon in late Summer orearly Fall during the lunch hour and I was at mydesk gazing at the turkey vultures wheeling inlanguid circles high above the Southern hills andcontemplating the best way to use up someunspent vacation time when a knock came at thedoor. Upon signaling the caller to enter one of thetechnicians from the manufacturing lineapproached holding a plastic box commonly usedfor transporting ceramic substrates. Even from adistance of several feet I could see through theclear plastic cover that these parts had a coatingwhich was badly delaminated. Since I was one ofthe Lab’s resident experts on delamination andcracking problems it was immediately clear thathere was big trouble and thoughts of vacation timerapidly evaporated. Closer examination indeedconfirmed what a quick glance at a distanceforeboded. Here was a collection of ceramicsubstrates coated with a polyimide material thathad delaminated on an impressive scale. Figure 1gives a schematic view of the nature of these parts.

The substrates in question were destined tosupport an advanced line of microchips to be used

“Polyimides: Synthesis, Characterization1

and Applications, Volumes I and II”, Ed. K. L.Mittal (Plenum Press, New York, 1984).

3

Figure 1 Schematic of ceramic substrate coated

with polyimide and containing several layers ofcopper wiring.

in the fabrication of a state of the artsupercomputer. The underlying ceramic providedmechanical support and an insulating matrix forthe basic power requirements of the chip. On topof the ceramic was a much thinner layer ofpolyimide which supported a dense mesh of thinfilm copper wiring for transmitting fast digitalsignals. The high electrical conductivity of thecopper and the low dielectric constant of thepolyimide made this combination ideal for meetingthe stringent requirements of digital signals. Oneunusual feature of the design shown in figure 1 isthe large thickness of the polyimide layer. At 175microns the thickness of the polyimide was indeedexceptionally large according to prevailingstandards since in the past the laboratory hadmany problems with delamination and cracking offilms only 10 to 15 microns in thickness. Thebasic problem arose from the fact that thepolyimide had a thermal expansion coefficientnearly 10 times larger than the ceramic. Thiscoupled with the fact that the polyimide materialhad to be cured at temperatures ranging up to 400C gave rise to very high thermal strains when thecoated substrate laminate was cooled back to roomtemperature. However, the substrates in questionused a newer variety polyimide with a very lowthermal expansion coefficient closely matchingthat of the underlying ceramic so only relativelyminor thermal strains should develop on curing. In fact a number of calculations were performedbased on fracture mechanics studies of Suo andHutchinson which indicated that the polyimide2

coating in question should be stable in layers from4 to 8 millimeters in thickness?!

At this stage it was time to form a task force andstart rounding up the usual suspects. In additionto the polyimide layer there were also severallevels of copper thin film wiring imbedded in thepolyimide. The case against the copperstrengthened when the analytical laboratoryreported X-ray measurements which showed thatthe stress in the copper was something like100MPa. This is some 3 times higher than whatone would ever expect in even the most highlystressed polyimide. If the copper were indeed the

culprit then significant adjustments would have tobe made to the copper deposition process whichwould not have been a very appealing prospect. I,in fact had serious doubts that the copper wiringwas the problem since even though the stress washigh there was not that much of it in the thin filmwiring layers compared to the volume of thepolyimide. The precise reasoning for thisreservation will be made clear later. Rather, myfirst impulse was to doubt the initial Suo-Hutchinson calculations which indicated that theexpected stress in the polyimide was not enough tocause delamination or fracture. These calculationswere basically back of the envelope type estimatesbased on the relatively simple formulae developedby Suo and Hutchinson and perforce harbored anumber of simplifications which might not be validfor a full blown wired substrate. Now the Suo-Hutchinson formulae may be relatively simple andallow for straight forward hand calculations butthey are by no means trivial and to get a betterestimate means solving the full set of contiuummechanics field equations for the structure inquestion. This means rolling out the big guns andperforming detailed numerical evaluations of thestress/strain equations using finite elementmethods.

Fracture Mechanics 101

Luckily I happened to be prepared for just such aneventuality since I had spent a number of yearsupgrading the computational capabilities of thelaboratory using high powered workstations. Inaddition, I had also developed a number of fracture

“Steady State Cracking in a Brittle2

Substrate Beneath Adherent Films”, Z. Suo and J.Hutchinson, International Journal of SolidsStructures, 25, 1337 (1989). An adaptation ofthis paper more friendly for purposes ofengineering calculations may be found inADHESION MEASUREMENT METHODS: THEORYAND PRACTICE, Robert Lacombe (CRC, Taylor andFrancis, Boca Raton, Florida, 2006). See inparticular chapter 4 section 4.2.2.

4

Figure 2 Typical section of a finite element model

depicting delamination of a polyimide coating off aceramic substrate. Such models are quite general andcan simulate not only a clean delamination at thepolyimide/ceramic interface but also the unexpectedlycommon case where the crack actually propagates inthe ceramic slightly below the true interface. This typeof failure mode is called a decohesion though it caneasily be confused with a true delamination since thecrack can propagate quite close to the true interface.

Figure 3 Driving force for decohesive failure of a

thin film wiring layer of a low stress polyimide ona ceramic substrate. The horizontal axis ismeasured in meters.

mechanics routines based on the finite elementmethod as implemented using the commerciallyavailable ANSYS program. An example of a®

typical model is displayed in Fig. (2) which shows asmall section of a larger model near the edge of acoated substrate where a small crack has initiated. The module shown in Fig.(1) was in fact cut out ofa larger brick using a dicing saw. The saw canleave small nicks in the edge of the module whichserve as ideal weak spots for a delamination or afracture to initiate. In the case underconsideration the crack always propagate justbelow the true interface and was therefore adecohesive failure as opposed to a true interfacialdelamination. The finite element routines Ideveloped allowed one to simulate the propagationof a crack anywhere desired and thereby alsodetermine the driving force propelling the crack. The ultimate source of the driving force is ofcourse the residual stress in the thin film wiringlayer which would like to relax itself by anymechanism available. Propagating a crack is onesuch mechanism that is available if the material issufficiently weak. The driving force is measured asan energy per unit area or effectively the energyrequired to create a unit increment of new surfacearea by propagating the crack length by somesmall distance. The cohesive forces holding thematerial together of course try to resist thepropagation of the crack since they demand that aminimum amount of energy be supplied in orderbreak the cohesive bonds. If the driving force istoo weak, insufficient energy is available and thecrack cannot form. The driving force available topropagate a crack or delamination is called theStrain Energy Release Rate and given the symbolG. It is measured in units of energy per unit areaor Joules/meter (J/m ). Thus if G=5 (J/m ) the2 2 2

driving force can supply 5 joules of energy tocreate one square meter of new surface area. Onthe other hand, every material can be assignedwhat is known as a Surface Fracture Energysignified by the symbol (. This is the energy

required to create one square meter of newsurface by breaking the cohesive bonds which holdthe material together. Thus if ( = 5 (J/m ) then2

the driving force for crack propagation must be atleast this large to drive a crack. With all thesetools in hand it was fairly straight forward to pulltogether a model similar to the one shown inFig.(2) and test the original Suo-Hutchinsoncalculations mentioned above. The result shownin Fig.(3) essentially confirmed that the originalhand calculations were correct. There should havebeen no way for the thin film layers to delaminatefrom the ceramic substrate. Figure (3) plots thecrack driving force vs crack length for a crackpropagating 50 micro-meters below the true thinfilm/ceramic interface. The calculations

determined that the maximum driving forceoccurred for a crack propagating at that depth.Thus the failure mode was decohesive in theceramic. Figure (3) shows that the maximumdriving force for this crack is roughly 2.4 (J/m ). It2

was known from experiment that the surfacefracture energy of the ceramic was about 15(J/m ). Thus the data in Fig.(3) clearly indicates2

that the crack should not propagate. About theonly sinister aspect of the data in Fig.(3) is that thedriving force is increasing with crack length making

5

Figure 4 Repeat of analysis shown in Fig.(3)

but now using actual stress level measured inpolyimide coated on real parts. The drivingforce for a 70 micron crack is now very close tothe surface fracture energy of the ceramicmaterial which was measured to be 15 (J/m ).2

this an example of an unstable but definitelyarrested crack. At this stage any thoughts ofusing up unspent vacation time had completelyevaporated. The best calculations we couldperform were saying that the thin film wiringlayers definitely could not decohere. The onlyremaining theory was that the high stress in thecopper was somehow at fault, an idea which wasnot readily compatible with any fracture mechanicsargument that we could plausibly think of. In thefamous words of Winston Churchill we were facedwith a “riddle wrapped in a mystery inside anenigma” . Subsequent events would unwrap the3

mystery and reveal that the riddle was reallyrather more mundane that we could have imaginedat the time.

The Mystery Unwrapped

In this sort of situation one is reminded of thesagacious words of Sherlock Holmes that it “is afundamental error to try and theorize in theabsence of data” . Thus we recommended that4

more date be gathered, and in particular the stresslevel in the thin film wiring layer be measured. Since this layer was predominantly polyimide weneeded to gather more data on the stress in thepolyimide material. At length this task wasaccomplished and the result was not what wasexpected. The measured stress in the polyimidematerial was some 6 times higher than it shouldhave been. This data, though unexpected, clearlyexplained why the substrates were decohering asshown in Fig.(4). The driving force for a 70 microncrack was very close to the measured surfacefracture energy of the ceramic material and sincethe driving force was unstable in that it increasedwith increasing crack length the prediction wasclear. At this point we had unwrapped the enigmaof why the substrates decohered but the riddle ofwhy the polyimide stress was so much higher thanexpected remained unsolved. Since it was nowclear that excessive stress in the polyimide wasthe real culprit destroying the parts and not thestress in the copper thin film wiring the task forcecould now focus attention on why the stress levelin the polyimide was so high.

Solving the Riddle of Excessive Stress in thePolyimide

A bit of detective work was now required to trackdown to cause of the high stress in the polyimideinsulator layers. At length one of the task forcemembers was quietly monitoring the curing ovensand observed a line worker approach with a cartloaded with parts to be cured. These parts had justbeen coated and were about to go through the firstcuring step which was what is called a “soft bake”.This is essentially a low temperature heating atroughly 80 C which was required to drive off excesssolvent and allow the film to settle. Onapproaching the ovens our worker noted much tohis annoyance that the oven he was supposed touse was tied up curing other parts. This was notgood since management needed to get a largevolume of parts through the line in order to meetearly user requirements such as electrical testprocedures, machine test bed needs, qualityassurance demands and a host of other pressingnecessities. Because of this all line workersreceived bonus rewards based on the number ofparts processed and this busy oven was clearlystanding in the way of such rewards. Not to bethwarted at this critical juncture our enterprisingworker quickly noted that a neighboring oven wasnot in use and from his point of view an oven wasan oven and the need to press on critical so intothe idle oven go the parts. Now if were in thebusiness of baking bread or cookies theconsequences of this change in ovens would havequite likely been of little consequence. However,

Winston S. Churchill, Radio broadcast3

(October 1, 1939). Commenting most likely onStalin’s peace pact with Hitler on the eve of WWII.

Sherlock Holmes in “Hound of the4

Baskervilles”

6

Figure 5 Film stress vs cure temperature for

polyimide coating with a high temperature/shorttime soft bake. Note that on cooling back toroom temperature this film is in a high stressstate close to 30MPa.

Figure 6 Stress vs temperature cure cycle for

polyimide coating which has undergone therequired 80 C/30 minutes soft bake cure step. Note that final stress after cooling to roomtemperature is somewhat less than 5 MPa whichis dramatically lower than the 30MPa achievedby the sample shown in Fig.(5).

as anyone familiar with curing polymide films willtell you, the curing process highly critical andextremely finicky. The idle oven in questionessentially carried out the “soft bake” at too high atemperature for too short a time giving rise to afilm stress cure cycle similar to the one shown inFig.(5). From this figure we see that at the end ofthe cure cycle when the sample is returned toroom temperature the stress in the polyimidecoating is a very substantial 30MPa.

Figure (6) shows a sample which has received theproper soft bake cycle of 80 C for 30 minutes. What jumps out you immediately is that the stressin the finally cured coating is somewhat less than5MPa which immediately explains the unexpectedfactor of 6 increase in the polyimide stress in thefailed parts. So the riddle of the decohering partscame down to a case of too few ovens to meet thevolume demands of the development cycle or toobig a rush to get out parts which placed anexcessive strain on the capabilities of thedevelopment line. A rather more commonplaceoccurrence in the fast paced world ofmicroelectronics than one would like to imagine.

Statistical Physics of Polymer Liquids

An interesting aspect of this episode in the trialsand tribulations of the development laboratory isthe nature of the polymide material that was beingused for the thin film wiring structure. Most ofwhat one might call “standard polyimides” exhibitstress/temperature behavior similar to what isshown in Fig.(5). These materials are applied torigid substrates as a highly viscous polyamic acidprecursor liquid using a spin coater. After dryingthe polyamic acid is converted to polyimide via anumber of intermediate cure steps. The final curestep is typically a high temperature bakesomewhere near 400 C in order to develop the fullphysical properties of the final coating. The stressin the coating comes from a variety of sourcesincluding shrinkage due to chemical reactions andsolvent loss. However, the predominant factorcontributing to stress buildup tends to be thethermal expansion mismatch between thepolyimide and the substrate which can be as muchas a factor of 10 depending on the materialsinvolved. The polyimide we were using at the timewas of a special variety known as a low thermalcoefficient of expansion (low TCE) material. Whereas your standard garden variety polyimidehas a thermal expansion coefficient somewhere inthe range of 30-40 ppm/C the low TCE materialsare on the order of 6 ppm/C or roughly 5 timessmaller. This low TCE matches nicely withsubstrate materials such as silicon (3 ppm/C) and

ceramic (3-6 ppm/C) and accounts for the very lowresidual post cure stress shown in Fig.(6). Whatmakes the low TCE materials different from thestandard polyimides is the long and stiff chain unitswhich make up the polymer chain. The effect ofthis chain architecture difference is exhibited inFig.(7). The top half of the figure depicts astandard flexible chain in solution showingessentially random chain orientation. On drying

7

Figure 7 Schematic comparison of the curing

behavior of flexible and rigid chain polymers. Flexible chain systems assume randomorientatons in solution and more or lessmaintain this configuration on drying. Rigidchain systems are also random in solution at lowchain densities, however, upon drying the rigidchain segments force the chains to align parallelto the substrate surface due to simple stearicinterference effects.

the material shrinks down to a solid film whichmore or less maintains the same random chainorientations as existed in the solution state. Oneconsequence of this is that in-plane inter-chaininteractions are governed mainly by relativelyweak van der Waals interactions and this gives riseto high thermal expansion behavior inconsequence of the weak chain to chain coupling. The bottom half of Fig.(7) shows the correspondingcase of polymer chains containing long stiff chainsegments. In dilute solution the chain orientationsare again random since there is plenty of room forthe chains to spread out and there is minimalinterference between neighboring chains. Assolvent is lost due to drying, however, the chaindensity increases and the chains begin to bepacked together more closely. In particular thesegments of chains close to the substrate surfacebegin to feel the presence of this barrier and dueto their length and stiffness are forced to adopt anorientation closely parallel to the surface. Anyonecan easily simulate this effect by holding a pencilin the air and noticing how it can take on anyorientation. However, as you bring the pencilclose to a flat surface such as a table you quicklynote that due to the rigidness of the pencil it ofnecessity must assume configurations which areparallel to the obstructing surface. This all impliesthat in the finally dried polymer coating the chainsnow have a preponderant tendency to be alignedparallel to the substrate surface. In consequencethe thermal expansion behavior is now governedmore by the intra-chain carbon-carbon bonds asopposed to the much weaker inter-chain van derWalls interactions. The net result is a much lowerin plane thermal expansion behavior.

So what happened when the wrong oven was usedto cure the coatings? Here we must rememberthat these are viscous coatings and as solvent isremoved through drying it takes time for thechains to assume their preferred packing order. Ineffect the high temperature in the wrong ovendrove the solvent off faster than the chains couldreorient and worse the imidization reaction also setin at the much higher drying temperature. Afterthe imidzation reaction occurs the polyimidematerial freezes up making any further chainrelaxation processes impossible. Thus if dried inthe wrong oven the polyimide essentially froze intothe random chain configuration typical of thestandard materials giving it the same kind of postcure thermal expansion behavior. This of courselead to the high curing stresses which so nicelydestroyed the parts the development line wastrying to put out.

Epilog: More Statistical Physics and the Valueof Stress Modeling

Statistical physics of rigid rodsAll of these problems with stiff chain polymersystems brought back memories of my previous lifeas a post doctoral student at the University ofMassachusetts. I’m not going to reveal how longago that was but in those days there were twoburning issues in polymer physics, one relating tothe problem of short range order in amorphouspolymers and the other the problem of polymer-polymer compatibility. The short range orderproblem was definitely the more glamorous onesince there was a raging controversy between twoschools of thought one which contended that shortrange order was essentially nonexistent inamorphous polymers and the other whichcontended that even in amorphous systemsinterchain interactions would bring about a certainamount of order on the scale of a nanometer or so. On the no short range order side was no less a lightthan Paul Flory who I believe was one of only 2Nobel laureates in the polymer field, the otherbeing Carruthers for the discovery of NYLON. Onthe other side were no less impressiveinvestigators such as Philip Geil then at CaseWestern Reserve University who had doneconsiderable work on the morphology ofpolyethylene and a variety of biopolymers polymers

8

Figure 8 Plot exhibiting separate contributions

of each material to the overall driving force fordecohesion of copper/thin film wiring off aceramic carrier.

such as collagen and elastin. My thinking at thetime was that if I could bring off a rigorousstatistical mechanical calculation that wouldresolve the issue one way or the other that wouldbe a definite coup after which I would be able towrite my own ticket to further fame and fortune. However, I was not so naive as to think that sucha calculation was going to be easy since I had beentoiling in the vineyards of polymer statisticalmechanics for some 5 years up to that point andwas well aware of the difficulties. Nonetheless itwould be worth a try, so step one was toinvestigate the existing literature to see what hadalready been done which turned out to be notmuch. Prof. Flory himself had published an articledealing with the problem but there was anessential difficulty with his paper in that he usedmean field theory to compute the free energy ofthe system. Now mean field theory is a veryhandy computational method for working out thethermodynamic properties of liquids but isnonetheless an approximation which does not takeinto account the subtleties of the correlations which are involved in more rigorous approaches. In fact mean field theory can be shown to becompletely wrong as one approaches critical pointswhere the long range correlations become not onlyimportant but actually dominate the problem. Thus if I were to resort to the use of mean fieldtheory any result derived would immediately beopen to the criticism that an approximatecomputational method was used and thus theresults should be considered suspect.

There was, however, one other paper by LarsOnsager, also a Nobel Laureate for his work onirreversible thermodynamics, which was entirelyrigorous in that it involved a term by termevaluation of the virial series for the liquid. Thisessentially amounts to estimating the pressure,volume, temperature (PVT) behavior of the liquidas an expansion in the density. At any stage onecan look at the higher order terms and estimatethe residual error. Thus this approach wasrigorous in the sense that you could always knowwhat error you were committing which was not thecase for mean field theory where the error wasunknown but assumed to be small at least awayfrom critical points. Using this approach Onsagerwas able to compute the PVT the behavior of acollection of rigid rods thus modeling the behaviorof liquid crystal materials which are known toundergo a variety of interesting phase transitionsdue to the restrictions on their packing behavior asthe density of the fluid increases. I believe hecarried out the series to third order and at thatstage was able to conclude that the liquidunderwent a packing alignment transformationwhich he concluded was a second order phase

transition. This was quite an accomplishment andrelevant to the problem I was considering since itdealt directly with the problem of ordering inliquids due to short range stearic effects. However,Prof. Robert Zwanzig, then at the University ofMaryland, felt that this was not enough. With theaid of a succession of graduate students, Zwanzigwas able to push the expansion out to seventhorder, with each successive term in the virialexpansion being a PhD thesis for some hardygraduate student. At the end of the day Zwanzigbasically concluded that Onsager was in fact right,the rigid rod solution does undergo a second orderphase transition as you increase the density to acertain critical value. Such a heroic effort wascertainly impressive, but as a post doctoral fellowwith 2 maybe 3 years to show someaccomplishment I did not feel that I would haveenough time to go the virial expansion route. Thusconstraints of time and energy forced me to fallback on the polymer-polymer compatibility problemwhich turned out quite nicely if not so gloriously asthe short range order problem would have could Ihave solved it. Nonetheless, the case of stiff chainpolyimide materials brought back memories of thatproblem which might be approachable today viacomputer simulation given the tremendousadvances in available compute power that haveoccurred in the intervening decades.

Value of stress modeling and fracturemechanics

As pointed out above the ability to perform adetailed stress analysis on the module along withthe associated fracture mechanics simulationgreatly expedited the progress of the task force in

9

solving the problem of the decohering thin filmwiring structures. Figure (8) exhibits a plot of the driving force for decohesion vs crack length for afully developed steady state decohesion which hasprogressed from the edge of the substrate well intothe copper/polyimide thin film wiring layers. Onenice feature of this type of plot is that it allows oneto look separately at the contribution each materialmakes to the overall driving force for decohesion. Recall from the discussion above that early on inthe investigation it was thought that the copperwiring should be a prime suspect due to therelatively high stress level in this material asmeasured by X-ray techniques. Figure (8) showsthat, to the contrary, the contribution of thecopper to the overall driving force is negligible andthat the true culprit is the polyimide material. Having this data in hand therefore put the taskforce on the correct track as opposed to wasting alot of time going down a blind alley. Theexplanation as to why the much lower stresspolyimide material dominates the decohesionprocess as opposed to the much higher stresscopper is quite straightforward once one is awareof the basic mechanism driving the decohesionprocess. Recall that it is the residual stress in thethin film layer that is the source of strain energydriving decohesion. The units of stress are forceper unit area, however simply multiplyingnumerator and denominator by length we see thatthe stress can also be thought of as an energy perunit volume i.e. forcexlength/volume. Thus thepent up stress in the structure can also be thoughof as a reservoir of pent up elastic energy whichputs the structure in a thermodynamically unstablestate. It would somehow like to release all thisenergy and go into a lower energy state and oneway of doing this is by propagating a decohesioncrack. Thus the system drives the decohesioncrack by drawing on energy pent up in the thinfilm structure and since the volume of thisstructure is overwhelming dominated by thepolyimide material it is the stress in the polyimidethat dominates the decohesion process. Thussome fairly elementary fracture mechanics analysisgreatly expedited the work of the task force andallowed me to return to the pressing problem ofhow best to use up my remaining vacation time.

COURSES AND CONSULTING ON ADHESIONAND ADHESION MEASUREMENT

Adhesion Courses

Since the theme of this issue of the Newsletter isdevoted to Adhesion Aspects of Thin films andHigh Temperature Polymers this is a goodopportunity to bring to everyone’s attention that in

addition to organizing International Symposia, theConference Director Dr. Mittal and I also give ajoint course on Adhesion and AdhesionMeasurement Methods. The AdhesionMeasurement course lasts for 1 day and is given inconjunction with all of the MST Symposia. Thenext scheduled class will be on June 16, 2007 inCincinnati, Ohio in conjunction with the SIXTHINTERNATIONAL SYMPOSIUM ON POLYMERSURFACE MODIFICATION: RELEVANCE TOADHESION www.mstconf.com/surfmod6.htm andthe SIXTH INTERNATIONAL SYMPOSIUM ONSILANES AND OTHER COUPLING AGENTSwww.mstconf.com/silanes6.htm . Actually, both ofthese symposia are closely related to the two Fall2007 symposia on POLYIMIDES and ADHESIONASPECTS OF THIN FILMS and a number of readersmay want to participate in both of theseconferences. However, for those who have neitherthe time nor the wherewithal to go to conferencesthere is another option in that MST can come toyou. Dr. Mittal and I can come to your location andgive a full 3 day course covering all aspects ofadhesion science and adhesion measurement. Thefirst 2 days are given by Dr. Mittal and present anoverview of the science of adhesion coveringeverything from surface analytical methods toadhesion improvement strategies. Day 3 isdevoted to the critical topic of adhesionmeasurement since this is a universal tool inimproving the adhesion of existing processes anddeveloping adhesion strategies for new ones. Thebasic philosophy comes down to the fact if youcannot appropriately measure what you are tryingto develop you will have no idea where you are atany point much less whether you are makingprogress or not. Also for large classes of 10 ormore students the on site 3 day course is very costeffective on a per student basis since there are nostudent travel expenses involved and there is asignificant discount for every student above 10 thatjoins the course. The course schedule is quiteflexible and we strive to adapt it to the needs ofour clients. The course length can be abbreviatedor lengthened as required and since the adhesionmeasurement course stands alone it can be givenback to back with the 2 day overview or if moreconvenient given at later date as a follow upcourse. Full details of the course syllabus areavailable on the conference website at(www.mstconf.com/MSTCourses.htm ).

Consulting

If it should happen that you find yourself facedwith a real nasty adhesion related problem that iseither thwarting your development program or,heaven forbid, threatening to shut down yourproduction line then with our 40 some odd years

10

collective experience with adhesion relatedproblems, we can be of direct assistance in gettingyou on track again. The MST staff have addressedall manner of adhesion and surface relatedproblems for companies both large and small. From large Fortune 500 corporations to small startup enterprises with only a dozen or moreemployees. Further details are available on theConference Website atwww.mstconf.com/MSTconsult.htm .

SYMPOSIA ON HIGH TEMPERATUREPOLYMERS AND ADHESION ASPECTSOF THIN FILMS

CALL FOR PAPERS: FIFTH INTERNATIONALSYMPOSIUM ON POLYIMIDES AND OTHERHIGH TEMPERATURE POLYMERS AND THETHIRD INTERNATIONAL SYMPOSIUM ONADHESION ASPECTS OF THIN FILMSINCLUDING ADHESION MEASUREMENT AND

METALLIZED PLASTICS

It has been my pleasure in these pages to give abrief reminiscence of my own experiences indealing with the polyimide materials and theiradhesion to ceramic substrates. As their tends tobe a certain universal nature to these types ofproblems I’m sure many readers have come acrosssimilar situations whereby the specific details maydiffer widely but nonetheless the underlyingphysical processes that tend to make or break anygiven development or manufacturing enterprisetend to follow nearly parallel tracks. Thus wewould like to most cordially invite all readers tojoin us in Orlando Florida this coming Fall for whatpromise to be stimulating and rewarding symposiaon these topics. This offers an opportunity to notonly catch up on the latest developments but toalso personally interact with the investigatorsinvolved and learn first hand where things areheaded in the future both in terms of fundamentalinvestigations and novel applications.

11

CALL FOR PAPERS

FIFTH INTERNATIONAL SYMPOSIUM ON

POLYIMIDES AND OTHER HIGH TEMPERATURE POLYMERS

SYNTHESIS, CHARACTERIZATION AND APPLICATIONSTo be held November 5-7, 2007 in Orlando Florida, USA

This symposium is the fifth in a series the first ofwhich was held in Newark, NJ in 1999. As with itspredecessors, this symposium will be concernedwith all aspects of polyimides and other hightemperature polymers. These materials havefound applications in such diverse areas as theaerospace industry and microelectroniccomponents. A unique combination of physicaland chemical properties make these materialshighly attractive for demanding applications wherechemical inertness, high temperature stability, lowdielectric constant, mechanical toughness andprocessability are primary concerns. Thissymposium is organized to bring togetherscientists, technologists and engineers

interested in all aspects of high temperature polymers, to reviewand assess the current state of knowledge, to provide a forum forexchange and cross-fertilization of ideas, and to define problemareas which need intensified efforts. The invited speakers havebeen selected so as to represent widely differing disciplines andinterests, and they hail from academic, governmental andindustrial research laboratories. This meeting is planned to be atruly international event both in geographic coverage as well as inspirit. The technical program will contain both invited overviewsand contributed original research papers. It is planned tochronicle the transactions in a hard-bound volume of archivalquality (to match or exceed the standards of the journalliterature) which will serve as a reference work for futuregenerations of investigators.

TOPICS OF INTEREST INCLUDE:< Chemistry, synthesis and

characterization of polyimides andother high temperature polymers.

< Surface chemistry and surfacemodification

PHYSICO-CHEMICAL PROPERTIES< Thermal-mechanical properties< Electrical properties< Adhesion properties and adhesion

improvement< Encapsulation and barrier properties< Effects of aging and environment on

long term stability, reliability anddurability

APPLICATIONS< Polyimides as adhesives and

insulators.< Polyimides as dielectrics, photoresists

and encapsulants in microelectronicand biomedical structures

< Metallization of polyimide andinvestigation of interfaces.

NOVEL AND ADVANCED FORMULATIONS< Ultralow dielectric materials, low

thermal expansion liquid crystals,polyimide blends, nanocomposites,copolymers, foams,... etc.

This symposium is being organized by MST Conferences,LLC under the direction of Dr. K. L. Mittal, Editor, Journal ofAdhesion Science and Technology. A proceedings volumeis planned for this symposium and further details will beprovided in due course. Please notify the conferencechairman of your intentions to present a paper as early aspossible. An abstract of about 200 words should be sentby June 15, 2007 to the conference chairman by any ofthe following methods:

E-mail: rhl@mstconf.com

FAX: 212-656-1016Regular mail:

Dr. Robert H. LacombeConference Chairman3 Hammer Drive

Hopewell Junction, NY 12533

Contact by phone: 845-897-1654Full conference details and registration via the Internet willbe maintained on our web site:

http://mstconf.com/polyimd5.htm

Or mail response form below to the conference chairman atthe address above.

12

CALL FOR PAPERS

THIRD INTERNATIONAL SYMPOSIUM ON

ADHESION ASPECTS OF THIN FILMS (INCLUDING

ADHESION MEASUREMENT AND METALLIZED PLASTICS)To be held November 7-9, 2007 in Orlando, Florida, USA

This symposium is the third in a series dealing withadhesion aspects of thin films, adhesion measurement andmetallized plastics. The first symposium with this title washeld in Orlando, FL in 2003 with the intent of integratingkey aspects of three separate symposia which treated thesetopics singly in the past. The main idea was to provide abroader venue for the discussion and exploration of thesethree closely related fields of endeavor. The main part ofthe symposium focuses on those aspects of thin filmtechnology that have a direct bearing on film adhesion tothe substrate. This is a topic of both fundamental interestto all aspects of thin film technology and of great practicalconcern in applications where films of high stress areinvolved. The coating of diamond films onto machine toolsis one of many applications where thin film adhesion is acritical factor in coating durability. The second part of thesymposium will deal with the ability to accurately measurethe adhesion of coatings to surfaces. This is always acrucial part of development and manufacturing processesdealing with coatings and films. Finally, metallized plasticsare a burgeoning technology

heavily dependent on thin film adhesion withapplications ranging from decorative design tooptical coatings to advanced thin film wiringschemes in the microelectronics industry.Metallized plastic films allow the technologist tocapitalize on the favorable properties of twodisparate classes of materials to create new andunique products which transcend theperformance and usefulness that can beobtained by either class alone.

The invited speakers have been selected so asto represent widely differing disciplines andinterests, and they hail from academic,governmental and industrial researchlaboratories. This meeting is planned to be atruly international event both in geographiccoverage as well as in spirit. The technicalprogram will contain both invited overviews andcontributed original research papers.

TOPICS OF INTEREST INCLUDE:

Adhesion Aspects of Thin Films< Factors influencing adhesion - Residual stress,

mechanical properties, contamination ... etc.< Long term bond durability, corrosion prevention< Adhesion promoters Fundamental Issues< Role of surface chemistry, wettability and

morphology< Fundamental adhesion mechanisms including role

of surface roughness/morphology and film/substrate interactions

Applications of Adhesion Measurement< Adhesion measurements in quality control and

manufacturing< Adhesion measurements in support of coating

process research and development< Adhesion measurement instrumentation for

laboratory and manufacturing environmentsFundamental Aspects of Adhesion Measurement < Mechanics of adhesion testing, the role of film

stresses< Fracture mechanics of adhesion testing< Physico-chemical aspects of adhesion testing, the

role of film morphology and chemistryMetallized Plastics< Metallization techniques and properties of metal

deposits< Metal diffusion during deposition< Morphology and properties of metal depositsInvestigation of Interfacial Interactions< Influence of polymer surface functional groups< Metal-polymer interactions< Fundamental adhesion mechanisms including

coating-substrate interactions at nanoscale

This symposium is being organized by MSTConferences, LLC under the direction of Dr. K.L. Mittal, Editor, Journal of Adhesion Scienceand Technology. An archival volume is plannedfor this symposium and further details will beprovided in due course. Please notify theconference chairman of your intention topresent a paper as early as possible. Anabstract of about 200 words should be sent byJune 15, 2007 to the conference chairman byany of the following methods:

E-mail: rhl@mstconf.comFAX: 212-656-1016Regular mail:

Dr. Robert H. LacombeConference Chairman3 Hammer Drive

Hopewell Junction, NY 12533

Contact by phone: 845-897-1654; 845-227-7026Full conference details and registration via theInternet will be maintained on our web site:

www.mstconf.com/adhfilm2007.htm

Or mail response form below to the conferencechairman at the address above.