jr
Z LU
It'.O The Corrosive Effect of Soldering Fluxes f UJ and Handling on Some Electronic Material! 1;Q :"0 i., ^h >: s.J A variety of rosin-based liquid soldering fluxes are characterized *. tn as to halide content, and investigation indicates galvanic corrosion dmS < UJ stress corrosion cracking are mechanisms oi material degradation thy can cause premature failure of inadequately cleaned electronic devicy a. ,q pbJ > to BY B. D. DUNN AND C. CHANDLER «i o
ABSTRACT. A preliminary survey has craft projects are generally assembled initial pre-tinning of "diifie-.i**- ,.-".* • been conducted to assess the variety by hand-soldering methods. Compa ponent leads—particularly m ris •• - of rosin-based liquid soldering fluxes nies which have been contracted to of nickel-based alloys—to adW&we h•---•• r utilized by European Space Agency manufacture such equipment will fol ter solderability. After pre»l?«v ..H, »he contractors. These fluxes were charac low the general soldering require- flux residues must be thorotggjMy I •'•'''' terized according to their halide con ments specified by ESA1 in order to cleaned from the component \e-rn ,*, tents, and a limited number of the obtain an adequate standard oi soider face to preclude time-depef***;-- J cor more common commercially available joint reliability. This is achieved mainly rosive attack. 1 •"** types were further evaluated in order by the employment of trained and Notwithstanding the tight Sofefe to assess their performance in terms of certified operators and inspectors, but process controls, several p solderability and corrosiveness. The also by the control of materials and associated with the corrosive - investigation included both standard soldering techniques. The mass assem liquid-soldering fluxes arid If? O and ad hoc corrosion tests and the bly of components to printed circuit dues have cost certain ESA < v determination of flux halide content boards by wave soldering has b'jen much wasted time and «•-,- and pH values. The corrosion tests agreed for the ESA Spacelab project in majority of these problem-! 4 • •' were performed in a warm, damp view of the large number of identical as non-conformances to :S atmosphere following the deliberate circuits utilized by this project and, inspection requirements, ••>"'; contamination by the fluxes of sam following qualification programs, a the formation of corrosic^'. ;r • '12 ples of electronic materials. limited number of wave-soldering on the surfaces of both s; . lines have been approved. a. The corrosive effects of residual flux copper wires and fus © on the surfaces of stressed Kovar com The successful outcome ot all sol coated printed circuit L- ponent leads and silver-plated copper dering operations will depend on sev cases, soldering had been > 1 • J wires are correlated against the phys eral material factors. The choice of a with the additional applicat •:- ii '. O'-. : iochemical properties of these fluxes. suitable soldering flux is very impor uid fluxes, supplied by rr-; - ; The results were compared with those tant, because it is the flux medii.m ufacturers to both U.S." an.' . '••••• obtained from similar control samples which will provide for the imlial trans specifications, purporting .••'••, v. either in clean condition or after delib fer of heat from the hot-soldering iron, dues of a non-hygroscopk', mm • erate contamination during handling. or liquid wave of solder, *o the sur sive and non-conducpv- tgtytf Met Galvanic corrosion and stress corro faces being joined together. The ESA allographic examine: gi . r sion cracking are considered to be soldering specification limits the strands and printed wi i • .•••im mechanisms of material degradation choice of material finishes which may ported both bright g • >-.y- which can cause the premature failure be interconnected to those having an brown corrosion p odi :i : of inadequately cleaned electronic excellent solderability, so that highly reveal reduced co devices. The acceptance of supposedly activated fluxes are not needed during tional areas; it is as • ' r "non-corrosive" liquid soldering fluxes the actual assembly process. luxes of short term such disc •.:<.•<< ,.- on the basis of routine standard tests is high activity, which are potentially from surface corrosi; > unlikely to obviate all the potential more dangerous from a corrosion cosmetic defect. v ;- corrosion problems associated with viewpoint, are permitted during the spacecraft electron:.' . • - . •,: electronic hardware. harnesses have long •:; : • to launch. Once it . to assess the long te r B. D. DUNN and C CHANDLER are with corrosion mechar sim Introduction the Materials Section, Product Assurance conductors with.- -: ; i Group, European Space Research anc Tech* which have beer (I •• ••:. •._•• • • Electronic packages intended for nology Centre, Noordwijk, Ti:e Nether for at least 10 year
European Space Agency (ESA) space lands •
WEEDING RESEARCH Si 'IH Problems of a more serious nature 71 previously examined in the flux system for the measurement of ionic included the failure of various gold- survey; they represent typical commer- conductivity of flux residues. This plated Kovar leads on flat-packaged ical products ranging from non-acti equipment, often referred to as the components. These leads had been vated to fully activated rosin-based lonograph,* has been evaluated by formed and then prepared according fluxes containing halogenated addi Naval Avionics and pronounced as a to the ESA requirement1 for gold tives. The copper mirror test proce method which provides for the semi- removal prior to soldering. De-golding dure2-4 is now re-examined against a quantitative measurement of flux resi 1112 and pre-tinning were performed by new flux test proposal.13 Also, certain dues on printed circuit boards. dipping the leads into solder baths special corrosion tests have been The test is limited to the detection fluxed with a mildly activated rosin. devised to reproduce those material of ionizable constituents in solder After a few months of storage, the compositions and environments fluxes which are monitored by the leads from several batches of flat pack thought most likely to have promoted DCM on a scale based on a known ages were observed to fracture com both the previously mentioned space quantity of sodium chloride dissolved pletely when exposed to light han craft corrosion problems and the few in either pure water or a 1:1 solvent dling operations. Metallographic in problems which have been reported in mixture of isopropanol and water. Cal vestigations strongly indicated that the the literature.5-10 ibration of this system was based on failure mechanism of these leads was The ionic content of each flux was the conductivity of 1 /ig NaCI per 1 ^l one of stress corrosion cracking of the assessed together with the effect of water. Kovar alloy due to the combined excessive operator handling contami For the purpose of this test program, effect of residual stresses in the lead nation on the ensuing corrosion of a constant amount of flux (5 /xl) was material following the forming opera component leads. The various flux added to the equipment's solvent tion and the presence of a thin surface types which have been subjected to which was continuously pumped in a film of supposedly "non-corrosive" this test program are listed in Ta closed loop and then passed through flux residue. Other failures have ble 1. two conductivity cells. Values of con involved the fracture of mounted tran ductivity were measured for each of sistor leads following equipment level the 13 fluxes in their as-received con vibration testing; fractography re Experimental and Test Procedures dition. Each flux sample was then vealed that stress corrosion had ini Chemical Analysis boiled for 1 minute (min) at -l-200oC tiated a crack in the lead material and (392°F) in an attempt to simulate flux this had later propagated by a fatigue Chemical analysis of solder fluxes is composition modifications which may mechanism. extremely difficult to perform and is occur during a soldering operation; often found to be inaccurate. Rosin the DCM test was then repeated by Discussion of these sporadic corro fluxes can include a vast range of introducing 5 /xl of the boiled flux sion-related problems during Project additives known only to the flux man concentrate into the solvent. Material Review Board meetings has ufacturers themselves. These may in led to the supposition that, although clude solvents and wetting, foaming ESA contractors purchased both cored and viscosity agents which have been Effectiveness of Flux Based on and liquid solder fluxes to recognized chosen to strengthen the fluxing prop Solderability Tests specifications, the complex chemical erties of the rosin. composition of activators contained in The soldering efficiencies of the Full chemical analyses have been proprietary rosin-based fluxes may various fluxes were assessed by means attempted. Fluorine, chlorine, bro change from one batch to another of a standard solderability test method mine, iodine, sulpher and phosphorus with resultant modification of proper compliant with B.S. 2011, Part 2T (Phil have been detected by emission spec- ties such as corrosiveness and solder ips Globule Method). The solderability troscopy, X-ray fluorescence spectros- ability. test was applied to both gold-plated copy and activation analysis. Gas chro- copper wire and plain copper wire by Initially, a survey of soldering fluxes matography and infrared absorption the horizontal immersion of short used by some of the major ESA con spectroscopy have been used to sepa lengths of each degreased wire in a tractors was made; the results are pre rate and determine the volatile organic globule of liquid eutectic solder held sented in the Appendix. This survey components. Non-volatile compo at 4-235°C (455°F) on a heated steel established that all the liquid soldering nents have been identified by both block. fluxes employed for component as ionic and non-ionic chromatography Immediately prior to wire immer sembly work and wire interconnec on ion exchange columns. Those tions would satisfy the corrosion sion, a standard volume of the flux analyzed compositions contained under investigation was dripped onto requirements of recognized flux speci ethanol, methanol, water, fatty acid fications.2"4 It indicated that even the the molten solder surface. Once monoethanolamide, alkylbenzene- immersed, the wire split the solder strongly activated fluxes, normally sulphonate and many unidentified used for the pre-tinning of "poor sol globule into two halves and the time compounds derived from abietic was taken for each half to wet, flow derability" component leads, might acid. not cause extensive corrosion of stan and coalesce around the wire sur A full chemical analysis of each flux dard copper mirror test pieces; it was face. under evaluation was soon discontin recommended that the true corrosive This test was repeated 50 times for ued, particularly as it was thought nature of any flux can be realistically each combination of wire and flux unlikely that such details could ever be assessed only if the corrosion test type. New 200 mg pellets of solder related to the effectiveness or corro assembly reproduces the essential were applied to the heated block at sive properties of individual fluxes. characteristics of the individual metals the commencement of each test. The The chemical analysis was, therefore, which make up a soldered connec mean soldering times were calculated limited to the two simple checks used tion. for each flux when applied to the in the initial flux survey described in gold-plated wire (xU]) and the copper Additional test methods to the the Appendix—halide content deter wire (xc„). An arbitrary unit, termed screening tests reported in the Appen mination, and pH value determina flux efficiency (FE), was generated to dix have now been selected in an tion. attempt to assess the relative effective Dynamic Conductivity Monitor (DCM) ness of different liquid fluxes. Thirteen flux types have been chosen from the The DCM utilizes a solvent pumping *Trademark ol Alphametal.
290-sl OCTOBER 1980 Table 1—Identification of Fluxes
Identification Act vity as described code by manufacturer"'
A1 RA A2 RA A3 R A4 RMA rerspex Container K1 RA K2 RA K3 RA S = location oi' i tnndard scratch M1 R Fig. 1—Test configuration for possible corrosion of Kovar component leads under M2 RA constant deformation M3 RMA Z1 RMA Z2 RA 1 g pellet of 60:40 tin-lead solder to best as possible. These components Z3 R each depression. By means of tongs, contained glass-to-Kovar seals with
(a the test pieces were in turn lowered gold-plated Kovar leads and were of a 'R—pure or low activity rosin; RMA—mildly activated rosin; RA—activated rosin. onto the surface of a heating bath type known to pass regularly electron containing liquid solder at 235°C ic component environmental qualifi (455°F). Contact between test piece cation tests.14 A jewellers' tooling jig enable a comparison between the and bath was maintained until 5 sec was used to form a standard scratch on ability of different fluxes to effect onds (s) after the initial melting of the each component lead. The scratches solder wetting of these particular me solder pellet. The test pieces were were positioned mid-length across the tallic finishes: removed in the horizontal position upper surface of the leads; the jig was and cooled for 15 min. then set such that a diamond cutting FE = x , + x . 0l Au They were then transferred to a edge passed through the gold-plated humidity chamber and held, in a verti layer and was just deep enough to Copper Sheet Corrosion Test cal position, for 21 days at a tempera penetrate and expose a microscopical A new test which evaluates the cor- ture of between 38 and 42°C (100.4 ly thin band of the Kovar base materi rosiveness of flux residues has recently and 107.6°F) and at a relative humidity al. been proposed13 and may become of between 91 and 95%. Assessment of The test configuration was chosen applicable in the assessment of rosin- any corrosion product was made after to represent both residual and applied based fluxes. The test method involves the test period with an eyeglass at x7 stresses which may be expected of a melting a piece of solder on a copper magnification. The test pieces were component lead during service. Exam sheet in the presence of the flux under considered to have failed the test ples of the causes of such stresses are evaluation, and submitting this test should any blue-green corrosion prod lead bending prior to mounting the piece to a humid environment. The uct have formed on the copper sheet, component on a printed circuit board, test results are subjective and based on or if discrete white or colored spots the soldering operation which may visual inspection for corrosion or had appeared in the flux residue or on constrain the leads, particularly if chemical reaction between the copper the surface of the solder alloy. plated-through holes are employed, sheet, solder alloy and constituents in and differences in coefficient of the flux residue. expansion of interconnected materials Square copper sheet test pieces Copper Mirror Corrosion Test during thermal cycling. (50 X 50 mm, i.e., 2x2 in.) were This test was performed with flux in In order to reproduce these stresses, made from 0.5 mm (0.02 in.) thick both the as-received condition and small containers were accurately ma material in the half-hard condition. An after it had been boiled for 1 min at chined from thick perspex sheets. The indentor and die were applied to each approximately 200°C (392°F). Labora inside width of the containers was test piece to form a central 4 mm (0.16 tory production of the mirrors fol slightly less than the measured length in.) diameter depression. lowed the method prescribed in MIL- of each component—as sketched in Before this corrosion test could be F-14256C4 and the test accept/reject Fig. 1^so that all leads required to be initiated, it was necessary to calculate criteria are as described in the Appen deflected at their ends as they were the non-volatile content of each flux dix. slipped between the perspex walls. so as to enable a constant weight of Once released, each component was solid flux to be transferred into the test thus suspended within the container "Ad Hoc" Corrosion Tests piece depression. The solid content of sides by the spring properties of its each flux was found from weight-loss The tests were devised to reproduce Kovar leads. The dimensional mis calculations based on the evaporation and investigate the corrosion mecha matches were calculated to produce a of volatiles from flux samples situated nisms which have occurred during the constant tensile strain on the lead top in dried aluminum containers after fabrication of several ESA spacecraft. surfaces approaching that of the Kovar three hours in an oven held at The samples consisted of: material's yield strength.* 110 ± 2°C (230 ± 3.6°F). Desiccators 1. Insulated silver-plated copper The silver-plated copper wire and containing silica gel were employed wire, partly stripped to expose approx flat-packaged component test pieces for storage of the containers and solid imately 8 mm (0.31 in.) of stranded fluxes. The copper test pieces were wire. The materials have been ap thoroughly cleaned and pretreated in proved for space use; however, the * Metallography had revealed that the com line with the recommended proce silver-plated strands were nicked and ponent leads were in the fully annealed 13 dures and then a sample of each flux scraped with a stripping tool so as to condition. Mechanical testing of individual leads produced the following results: 29.2 was transferred to the indentations. reveal the underlying copper. 2 2 kg/mm yield strength; 43.4 kg/mm ulti The solid flux samples were heat- 2. High reliability, space quality, 14- mate tensile strength; 14 X 104 kg/mm 2 treated at 60°C (140°F) for 10 min, and pin flat-packaged components were Young's modulus; 43% elongation at frac this was followed by the addition of a chosen to reproduce service failures as ture and 150 VPN micro-hardness
WELDING RESEARCH SUPPLEMENT I 291-s OJ -o T3 c c .ti OJ 5 E_£
o OJ [j — o -*• ill ; Q. ~D0 •c-5 §L 1 C QJ C re > OJ "C QJ c 'o •£-* c QJ 0 ""> " nj u : U OJ ™ QJ ra " U t/i to u Sis ^— 3
in ^J ^ < -a O ro O l/l ^3" --N.O 4_ 8 1"° o >CO Q. u * (J T3 \j vD o tr> I DC "D C < 2 o -a S ro T— c QJ QJ c u 0
bein g 0 r e ini - > ambe r 1- c GJ _fr —' OJ i cu u 0 1/) <1J — °-< cu •2 E y 3 >• OJ ."Lt Zl Lo o — n: „_ : OJ — o o 5 3 2 fe ra •I SI a> E UJ I | CO V} CO E/J i ^; c T3 =1 — —. -—- — —-^ W, -—- t/i £r» LO •—- cn -—• •— - LT> -—to — to -—to -Tr-cn -r> to iT- ; - ° I» 0 OJ JZ E o t/1 ! 0 u JZ .in C < ea
292-sl OCTOBER 1980 o ~2 = o S3 2 -— re re JZ *rai QJ ^ CL CL u e "a 0) 01 ra oi 0) t/i O) E Ol 0) JZ JZ c ra i/) JZ 0 4-1 OJ ,:z JZ c rd CL —0 ' t/.1 ^> c 0) c E JZ 0) .0 ~ ra TI u 1 "t7 u VI u — t/>1^ 3 3 u ra 3 CO u c t/1 'X. 01 in = y • c C tn ra — ro o u 5 JZ t/1 t/i C t/i 3 XCi — c 0) t/1 3 c re QJ 0) E t/i LJ o 01 t/1 t/1 QJ QJ t/i OJ X Z_ —* 01 0) Ol JZ OJ OJ "5 CA >ro 01 t/1 t/i aj n 0> OJ rv
OpJ* o C O a. C u CL O •" o — re U J^„ X c 0) _i X CL fc c J2 > C j-. aj u UJ rO re tf QJ QJ m JZ — 2 a. > X LU X c 0) o ._ re g 01 I ~S ™ S3 £ t/i O OJ o C re c Q-.2 x cc c E ~ O t/i ra CD < B O U UJ O oj t: — 0C_>. o 5 C/5 '— t/i c OJ o C UJ m "a 0) 01 \3 OOJl .± te oz a «j CL^ O tZ t/1 —t E o o S u-i O OJ ^ QJ ts E . ro ri cj U c CL 0) OJ OJ C OJ u O OJ 01 JZ « o. C O) 3 •- X Q.
WELDING RESEARCH SUPPLEMENT | 293-s LO LU X
<
U) LJ X n
Q_
LO Q -Z O c_> UJ CO
LU < 2
o
CO cc LU O o to o' D
UD
CN O Q IX O
S1S31 JO ON
294-s I OCTOBER 1980 n DO c z 01 CJ OJ Z JZ E oo LJ T3 <: - g QJ c ro OJ t/1 re >- ro >- 0) 00 3 JZ 3 JZ '0 C U cn ra OJ r- J5 01 O ra C c > X LU r3 X •JZ t 0J > OJ u X c u JZ c Cy 0 3 rO tn W 3 X QJ o ra X 0 o i£ X E ci 0J p tn t/1 CJ x 2 O k- DO c ^ QJ E CJJ 3 3 < t/i 01 z c c >- E '4- .X ro QJ 0 n3 fc CC 01 cu t/1 -° QJ c o 01 00 X u x "9 JZ E X X 0 " si 3 co OJ -^ 3 (z0 E L3 C cu. "Eh ra c j_J E 01 X Si u o 00 CL OJ cn U n ooi C JC: •^ -3* _rr t W _ CO OJ u E JZ JZ o 5 DO 01 c J2 x y a. — Qj re p y •? f- •£ c > p OJ-TC K > c QJ 0 ra — QJ QJ 0) ZL: rd 01 QJ 32 5 5c5£ O 5= 0)
re 0) 0) >- t/i z c z JZ _Q 0) T3 11 0) 3 01 01 u QJ t/i QJ 0 0 z X •S3 £ ,5 QJ £ 0 >- 01 1 JZ CL u ZJ OJ 01 x JZ c JZ z JZ JZ 53 0J 5 _Q> X 01 11 ZL tn C QJ 1 c s X > 0 t/1 Z CO' to z JZ. 3 OJ U z z_ ~F t/1 ro i OJ t/1 c JZ CJ t/1 01 X 01 OJ 0 01 LJ C"! ZL JZ u c°z5 X X o 0 re ro a ra JZ •J3 3 u c z cj t/1 CL QJ Cu JZ s c 53.2- 3 01 - t/1 O X X u OO 5 QJ 01 u 01 0u ) 0 0 OJ ra z t/1 JZ E >- JZ JZ t/1 ro E 01 t/ci 3 z DO 00 15 X x o DO t/1 5 u c QJ Ji ra JZ c •a 0) OJ y o QJ C ~o o> c X CL O JZ c E x 2 11 t/i "JZ ,>- E 0J ^5 0 _0 o o o o c > 0 OJ o X 01 CJ c3 o X X 0 01 u "9 0 re C X JZ o E ra CL t/1 J5 5 re ro E '£ 1 T ca
*J> 0) w tfl *J tfi LO E .E ro QJ C •- I x ~ UJ C JZ _1 .Ex fi CLSJ-T : QJ re : X Si *" LL x OJ i %£• =3 rz c O tn < o o 2 I °y : > ,_ - ZL
>- OJ rz J_ CD JD JZ o ii CO gj P > IS >. c re Q O O O O O o O o o o O o c x c cr u c S C re DO C o c O JZ tn re O O .ti 'JZ 3 CL 2 xz ai 5 3 £ LU C X o£ U o QJ QJ E re CL U *2 ^OJ cr 3 5 >— re ulj OJ •- O" OJ — _CU rjj II Q- ? CL '•S 5 U 0) f= OJ TS rr §12 3 JZ o oi 01 J- E ^ > x c O X J_ "Jo -r. CN O ' > CJ u • •S OJ C QJ 0 OJ -rj ° o° 3 2 D_ u re O O O o o o O O O O O o C/l — • — ozE OJ °°x 5 cu 1 -Q 30 JZ _ 1= ' C JZ c O DC ., re O O CJ o LJ re ii c re JXL •• S1S31 JO °N o ce i - OJ s in .y c — "° JZ ! O QJ oi oi "> re o2 fi .y OJ QJ QJ LJ OJ 5JE JZ -2 "cD o 3 2 a
WELDING RESEARCH SUPPLEMENT | 295-s 35 © = R cl Ml • = RMA SLOW • = RA KEY MANUFACTURERS' 3 < DISCRIPTION OF FLUX ACTIVITY IX + ©A3 D CJ IX
2 c! AA
0 , 10 100 1000 10000 100 000 BOILED FLUX, DC M. TEST RESULT (rjg Net Cl /jjl FLUX ) Fig. 4—Relationship between flux efficiency and the ionic content of boiled fluxes
split solder globule to flow and co RMA and R-type fluxes on gold wires (see Table 2, footnote c) ought to have alesce around the wire. It should be are observed to be 0.14, 0.25 and 0.35 s, been recorded as "failures" according noted that, before molten solders can respectively. Repetition of these tests to his interpretation of the specifica wet and spread, the metallic surfaces on plain degreased copper wires tion.' All inspectors agreed that the must be free of any tenacious non- proved to be more discriminatory Class 3 failures, evident as corrosion metallic films such as oxides or sul owing to the additional time taken by and flux penetration of the copper, phides. One purpose of the applied the flux to remove an adherent surface were immediately apparent and that flux is to remove these barrier films film of tarnish. Typical mean wetting quality checks of the "goods inward" either by reducing them to the metal times for RA, RMA and R-type fluxes type, for this classification of corrosive lic state or by chemically dissolving are 0.5, 1.3 and 2.0 s, respectively. flux, would probably be both rapid them. The arbitrary unit of flux efficiency and straightforward to control. Two common, easily reproducible (FE) which is ascribed to each flux It is also noticed from the copper metal surfaces were chosen to repre listed in Table 2 has been plotted mirror results in Table 2 that identical sent typical material finishes that are against the ionic content, as indicated results were obtained for both the frequently interconnected by solder by the DCM test result for boiled flux, as-received fluxes and the same sam ing techniques in the electronics in Fig. 4. The fast acting fluxes are ples after boiling. This is somewhat industry. Gold-plated (extremely thin noted to be of the RA-type and gener surprising since the DCM results indi to preclude the formation of brittle ally contain a high ionic concentra cated that, once boiled, the ionic con tin-gold intermetallics) copper wire tion. However, the remaining R and tent of these fluxes increased; this represented a barrier-free surface of RMA fluxes show no particular rela would be expected to create a more high solderability: degreased plain tionship between FE and ionic con corrosive environment. It is probable copper wire represented a surface tent. that the copper mirrors have not been which, due to long atmospheric expo greatly affected by this increase in sure, supported a thin film of oxidation ionic content because stable basic products. Results of Standard Corrosion Tests chlorides, such as those green patinas The results of the solderability tests The flux corrosiveness performance which form on copper, will reduce the are presented in Fig. 3. It is seen from results show good agreement between Ck activity and may form barrier layers the solderability distribution curves for the standard copper mirror test and that retard corrosion. In fact, only the gold-plated wires that most of the the proposed copper sheet test, as fluxes which contain very high (greater soldering times are very short, particu shown in Table 2. However, the than 4.45%) halide concentrations and larly when the fully activated fluxes inspectors performing the visual exam extremely high ionic contents were were applied. In these instances when ination of both types of humidity- seen actually to penetrate the thin the fluxes are not expected to undergo tested samples regarded the specified copper film. The copper mirror test is, any surface chemical reactions, it is classes of acceptance and rejection however, limited in that it does not only just possible to differentiate criteria to be very subjective. indicate the danger of electrolyte cor rosion accelerated by direct contact between the respective wetting abili One inspector considered that the between different metals. ties of the various fluxes. "marginal passes" attributed to the Typical mean wetting times for RA, copper mirror test identified as Class 2 Under actual soldering conditions, a
296-sl OCTOBER 1980 liquid flux will flow to surround the cooled, some fluxes are extremely bi-metallic corrosion resulting from molten front of solder and will be in hygroscopic and may promote gal the applied solder. contact with both hot metallic and vanic corrosion between the various Cooper is more noble than solder. non-metallic materials that may sub interconnected materials. The pro When these are in contact in the pres stantially modify the chemical compo posed copper sheet test method may ence of a chloride concentration, this sition and activity of the flux. Once go some ua\ to assess this problem of factor will cause galvanic corrosion of a. O AIR AiB A2R A2B A2R IT"* > Q X o Csz < LU CO LLl ip • QC X uj
ft. A4R A4B Z1R Z1B Z2R Z2B o _l LU >
X o cc < UJ to UJ •>fls
MIR M1B M2R M2B M3R M3B u OJ >
X wm o 5 cc '^yy > UJ Q oX cc CONTROL HANDLED OMTROL SW Z3P 2313 ft. O —i •-If >LU ¥ LU Q E"*-?" or. • •' •' -?'iy...". 3 * - < UJ : CO Rift- f : .fff::;*!< UJ ip^a DC Fig. 5—Overall view of samples after 56 days exposure WELDING RESEARCH SUPPLEMENT I 297-s Table 3—Detailed Results of the "Ad Hoc" Testing Stress u corrosion b attributed t by flux L Corrosivity of f ux on scratched Corrosivity of flux on gok -plated from O t Total silver-pl ated copper wire based on Kovar component leads based micro- a corrosivity bl Id) 0 visual ins aection on visual ins section; sections I index"" Duration of exposure (days) Flux type"" 4 10 21 30 40 56 4 10 21 30 40 56 56 A1a 3 5 5 5 5 5 1 5 5 5 5 5 10 64 130 A1b 5 5 5 5 5 5 1 5 5 5 5 5 10 66 A2a 3 5 5 - 5 5 5 1 5 5 5 5 5 10 64 122 A2b 5 5 5 5 5 5 1 5 5 5 5 5 4 58 A3a 1 1 1 3 3 3 1 3 3 3 3 3 10 38 54 A3b 1 1 1 1 1 1 1 1 1 1 1 1 4 16 A4a 1 1 1 1 1 I 1 1 1 1 2 2 4 18 32 A4b 1 1 1 I 1 1 1 1 1 1 2 2 0 14 K1a 3 3 3 3 3 3 1 3 3 3 3 3 10 44 88 K1b 3 3 3 3 3 3 1 3 3 3 3 3 10 44 K2a 3 3 3 3 3 3 1 1 3 3 3 3 10 52 104 K2b 3 3 3 3 3 3 1 1 3 3 3 3 10 52 K3a 3 5 5 5 5 5 1 5 5 5 5 5 10 64 130 K3b 5 5 5 5 5 5 1 5 5 5 5 5 10 66 M1a 1 1 1 1 1 1 1 1 2 2 3 3 4 22 44 M1b 1 1 1 1 1 1 1 1 2 2 3 3 4 22 M2a 1 1 1 3 3 3 1 1 3 3 3 4 10 36 79 M2b 1 3 3 3 3 3 1 3 3 3 3 4 10 43 M3a 1 3 3 3 3 3 1 3 3 3 3 3 6 38 80 M3b 1 3 3 3 3 3 1 3 3 3 3 3 10 42 21a 1 1 1 1 1 1 1 1 1 1 1 1 0 12 30 Z1b 1 1 1 1 1 1 1 1 1 1 1 1 6 18 Z2a 3 3 3 3 3 3 1 3 3 3 3 3 4 38 80 Z2b 1 3 3 3 3 3 1 3 3 3 3 3 10 42 Z3a 1 1 1 1 1 1 1 1 2 3 3 3 4 23 50 Z3b 1 1 1 1 1 1 1 1 2 3 3 3 8 27 Contamin ated control el 1 2 3 3 3 3 10 - — Clean control' 1 1 1 1 1 1 0 — — 's'Code of flux types (see Table I for details), a) represents sample of fresh, uncontaminated flux, as-received and b) represents same flux boiled for one minute at 200 ± 5°C to simulate a soldering operation which may modify flux properties. lb'At regular intervals the samples were visually inspected and classified 1 = no corrosion, no discoloration or tarnish 3 = less than 50% of surface corrosion 5 = extensive surface corrosion with severe discoloration. "'Classification of depth of stress-corrosion cracking based on data appearing in Table II. ""The 'Corrosivity Index' presupposes the presence of both unhealed and boiled flux on these samples and represents the summation of all classifications for a particular flux type. ,e'These samples consist of stressed gold-plated Kovar leads either deliberately contaminated with hand perspiration or as-cleaned in I.P.A. and de-ionized water; no flux had been applied. the electronegative solder metal. The come contaminated by a flux and sub ver-plated wires and the scratched potential difference which will arise is sequently corrode. Assimilation of the flat-pack leads have invariably at difficult to estimate accurately, but as conditions which are considered to tained their worst visual appearance. a guide1"' is taken to be 0.3 volts (V). have promoted ESA spacecraft elec Samples supporting a typical RA type (Some fluxes can contain the same tronics corrosion problems are likely flux and one RMA type flux have been chloride concentration as seawater, 2.2 to have been achieved by the "ad hoc" detailed in Figs. 6 and 7 after 21 days of g/l, the basis of this guide.) As far as corrosion tests of this program. humidity exposure. These photo electronic circuit materials are con Visual Inspection Results. A general graphs are captioned according to the cerned, copper and solder are relative overall view of the component and inspectors' observations. ly "compatible," and this copper sheet stranded copper wire samples is seen It is to be observed that the fluxes test would probably be more selective in Fig. 5 which depict the corroded containing the higher ionic content if thin platings of a noble metal such as appearances of many items after 56 and having higher activity (RA) pro silver or gold were applied to the days exposure to 95% RH at 40°C duce the most severe forms of surface copper, these metals being separated (104°F). It is noted that the control corrosion. It is also observed that, des from solder by a potential of 0.5 and specimens show no evidence of corro pite the claims made by certain flux 0.65 V, respectively. sive attack at the end of the test manufacturers, many products are hy period. The progressive inspection groscopic. Such fluxes are likely to results and a description of the numer have become electrically conductive Results of the Ad Hoc Corrosion Tests ical classification system employed gels during the test period and are able It is believed that the standard tests throughout this evaluation are listed in to initiate and sustain various forms of for establishing the corrosiveness of Table 3. corrosive attack that are dependent on liquid soldering fluxes bear little It is important to note that, after electrochemical reactions. resemblance to those conditions in only 10 days subjection to the test Effect of Fluxes on Silver-plated which electronic materials may be environment, both the damaged sil Wires. The deliberately damaged si I - 298-sl OCTOBER 1980 .4.1 . lu» f/g. 6—Flux type Al (RA) after 27 days exposure. Corrosion products appear on all metallic surfaces, the wires support a green product, and the leads have several micro pits in their plating. A—as-received flux; B—boiled flux •i ri j I MI in Fig. 7— Flux type Z1 (RMA) after 27 days exposure. A non-hygroscopic flux showing absolutely no sign of corrosion on metallic parts A—as-received flux; B~boiled flux ver-plated wires show varying degrees that, once the flux and its residue had during soldering or by wicking of con of surface corrosion products after been cleaned from the strands during taminated cleaning fluid along the exposure to flux and humidity; the the preparation of samples for SEM stranded wires. results are listed in Table 3. Unfortu and metallographic inspection, the On the basis of these results, it is not nately, it was not possible to quantify remaining adherent corrosion product considered possible to predict the the extent or depth of corrosion by appeared to have a green-brown col long-term reliability of spacecraft means of either surface scanning elec oration. It is believed that this debris is wires which turn green shortly after tron microscope or metallographic a mixture of so-called red and green the introduction of flux during solder examination of individual samples. plagues which have been previously ing operations. The greatest danger The depth of the deliberate score reported in the literature.1'1 Red plague will occur when the flux is hygroscop varied from one wire strand to anoth is the oxidation product, Cu.O, which ic, contains a high ionic concentration er. While the non-surface-corroded forms because of galvanic action and is present on wires which will be samples failed to reveal any metal between silver and copper in the pres periodically subjected to a humid wasting, it was extremely difficult to ence of the electrically conductive flux environment. establish the internal corrosive depth gel- Stress Corrosion of Component of attack due to the presence of pref Figure 8 illustrates how the more Leads. Probably the most revealing erential sites of corrosion and the non- noble silver does not enter into the results to have been produced by this uniform cross-sectional areas of the corrosion reaction. This galvanic attack evaluation program are those which damaged strands. The more reactive is believed to be intensified by the indicated the low stress corrosion RA fluxes were observed to cause rath presence of active chloride ions, Cl , cracking resistance of gold-plated er extensive corrosion in locations and abietic acid present in the flux Kovar leads when stressed close to this where the copper substrate had been which form a green product (green material's yield point, then subjected exposed and only slight corrosion plague) which can be washed safely to the soldering fluxes under evalua beneath the layer of supposedly intact away during post-soldering cleaning. tion. These results are listed in the last and pore-free silver plating. A potential danger arises, however, two columns of Table 2 and, for full The worst case of wire strand corro when strong RA-type fluxes and resi details, the photomicrographs and sion produced by one of the stronger dues are entrapped under the wire captions presented in Figs. 9-21 must RA fluxes is shown in Fig. 8. It is noted insulation material by capillary action be reviewed. WELDING RESEARCH SUPPLEMENT I 299-s The extensive degree of general and later stage—lightly etched to reveal a conditions have not been widely stress corrosion tracking (SCC) shown transgranular mode of SCC propaga reported in the literature.'I"i,s Effec by these longitudinal sections is quite tion. tive protection of Kovar component surprising because close visual inspec As in the case of the example cited, leads from SCC has been achieved1" by tion of the lead surfaces does not several photographed SCC paths did firstly chemically removing the work- always indicate any sign of surface not begin at the standard scratch. It is damaged (from lead-stamping opera corrosion (e.g., see Fig. 7, flux Z1, type thought that the gold plating on the tions) surface of the lead prior to thin RMA). The photomicrographs (Figs. lead surfaces is porous and capable of gold plating. The thin plate is designed 9-21) depict longitudinal sections initiating and sustaining SCC growth to protect the lead from oxidation made through the lead mid-planes by diffusion of chloride ions from the during component manufacture, but is under the "standard scratch" and in residual flux to the crack tip. The elec finally removed by dip-coating with a any region away from the scratch trochemical mechanism of SCC in ductile pore-free finish of eutectic tin- which possesses a particularly severe alloy steels is accelerated by the appli lead solder, and any gold plate remain corrosion site. The results of applying cation of anodic currents, and a similar ing adjacent to the component-to- either as-received or boiled flux to the situation is thought to exist in the case lead glass-to-metal seal is additionally stressed Kovar leads may be compared of these gold-plated Kovar leads, protected by a silicone varnish. and, although not consistent, most anodic currents being set up between Another method" to minimize the fluxes appear to have an enhanced the porous plating and the less noble SCC failure of leads is to plate the corrosiveness once boiled. Kovar alloy. It is to be noted that the Kovar with 12.5 microns of nickel prior It should be noted that several of the corrosion products occupy a larger to gold-plating. However, even this component samples suffered lead volume than the Kovar from which process will not overcome chloride- breakages as they were being carefully they are formed; these products will assisted SCC if the nickel is mechani removed from their containers at the tend to increase the SCC propagation cally cracked during either component end of the 56-day test period. Many of rate by creating a wedging action and manufacture or subsequent lead-form these fractured items were viewed by additional stress concentrations at the ing operations.17 scanning electron microscope. With crack tip. It has been suggested that it is one typical example, the lead was The fracture of Kovar alloy as a result impossible to avoid the SCC failure of microsectioned, polished and—at a of exposure to SCC environmental Kovar leads by solely increasing the thickness of the gold finish. One report1'' states that many thick gold- plated transistor leads containing re sidual stresses induced by 90 deg angle bends, in the presence of a commer cial soldering flux and a humid envi ronment, were observed to fracture after only 23 days. The halide concentration necessary to promote SCC in Kovar has not been previously studied. In the chemical industry, where these corrodents are a serious problem, it has been ob- served111 that an extremely small con centration of 0.02% aqueous NaCI will readily cause SCC of high alloy stain less steels, including AISI 316 stainless steel. Based on the flux evaluation results presented in Table 3, an attempt has been made to index the various fluxes according to their ability to cause cor rosion. This "corrosivity index" ap pears as the summation of all the numerically classified results for gen eral surface corrosion and SCC. A comparison between this index and the ionic and halide contents of the various fluxes is presented in Figs. 22 and 23. These show some correla tion between a high corrosivity and either a high halide or high ionic con tent. There is, however, no relation ship between the SCC susceptibility of Kovar leads and either halide or ionic content of individual fluxes. In the case of one as-received R-type flux, Fig. 8—One of the strongest and hygroscopic RA-type fluxes produced severe galvanic designated A3, it is seen that a halide corrosion of the copper conductors from these stranded wires (A). The SEM photograph (B) concentration as low as 0.0011% may of one strand shows a silver "shell" whin h, when microsectioned (Cand D) reveals extensive initiate and propagate sufficient SCC wasting of copper. (The original form of silver plate is marked on C, but this was crushed by the mounting media; the polished sections have been lightly etched in ammonium peroxide to cause lead fracture (Fig. TIB). to highlight the copper grain structure.) A—optical photograph; B—SEM photograph, X350; C-microsection, X400; D—detailed photomicrograph, X900 (reduced 23% on reproduc Effect of Skin Secretions on Stressed tion) Component Leads. Spacecraft elec- 300-sl OCTOBER 1980 Q. o X o < LLI a. O Fig. 9-Flux A1. Pitting corrosion seen under Fig. 70— Flux A2. Surprisingly little corrosion Fig. 11—Flux A3. Only one site of stress scratch (A), with general stress corrosion at under scratch (A, Q, but severe lead embrit- corrosion (B) with cracks propagating to sites B, C and D tlement in one selected region (B) 75% of lead thickness I O < LLI IMPORTANT NOTE Figures 9-21 consist of tour views Z each laid out in quadrant form LU wherein the quadrants are to be identified according to the follow Q. ing notation: o A (As-received (As-received > LU flux; region flux; selected Q under scratch) I regions) O < D LU C tn (Boiled flux; (Boiled flux; region under selected scratch) regions) Fig. 13—Flux Kl. Extremely severe form of stress corrosion cracking through the lead Fig. 12— Flux A4. Only very slight pitting thickness (A-D) with some exfoliation cor a. corrosion adjacent to scratch on (A) rosion (D). In one region (D) the gold plate o is blistering due to the buildup of corrosion _J LU products > o < LU tf> LU tr z LU o _i >LU LU o Io < LU tf> LU Fig. 15—Flux K3. Leads have fallen apart due Fig. 76—Flux Ml. Pitting corrosion adjacent Fig. 14—Flux Kl. Appearance of stress corro to the presence of a fine network oi hairline to scratches (A,C) with some surface corro sion cracking similar to Fig. 16 stress corrosion crac ks sion and blistering of gold plating (D) WELDING RESEARCH SUPPLEMENT ; 301-s Ml Fig. 17—Flux M2. Severe case of stress corro Fig 19— Flux Zl. Slight intergr'anular corro Fig. 20—Flux Z2. Surface pitting corrosion sion cracking and associated blistering (B- sion in one region of stressed lead (A,B,D) and severe stress corrosion (C) O) (D—boiled flux) causes skin secretions to be deposited upon stressed component leads, was evaluated by means of two clean flat- packaged components supporting the "standard scratch." Once contami nated by perspiration, these samples were subjected to the stress and humidity environment in parallel with cleaned control samples and the pre viously' described fluxed samples. All devices were inspected at regular intervals and classified according to the degree of observed surface corro sion. The results are listed in Table 3. After 21 days, the contaminated sample had become surface-corroded to an extent less than 50"o of its lead surface area. The cleaned control sam ple remained uncorroded. At the end of the 56-day period, the contami Fig. 18-Flux M3. An extraordinary mixture nated and control samples were Fig. 2'1-Flux Z3. Surface pitting corrosion of corrosive classes, general Kovar corrosion microsectioned; these as-polished (A,B,C) and severe stress corrosion (D) and blistering due to corrosion product (A), photomicrographs appear in Figs. 24 slight intergranular and pitting corrosion and 25, respectively. Despite the (B), zero corrosion (C) and severe stress somewhat innocuous surface appear difficult to remove completely such corrosion (D) ance of the contaminated leads, the high concentrations of salt from the sections reveal highly branched stress vicinity of similar fingerprints when tronic units are generally handled by corrosion cracks beneath the stress- they are present on component sur operators under "clean room" condi raising scratch as well as in selected faces due to the porous nature of the tions. The preferred codes of practice regions beneath the gold plate. Many gold plating on Kovar leads. for component assembly by hand or of the fine cracks which penetrated An induction period prior to Kovar wave soldering to printed circuit more than 50% of the lead thickness cracking may be dependent on the boards recommend that operators occurred in regions well away from buildup and penetration of a local wear finger cots or lint-free gloves surface blisters and would not have corrosive solution somewhere along when they bend, straighten and insert been apparent from visual inspection the lead surface and SCC, in the worst component leads into pcb termination alone. case promoting lead fractures, may be areas. These practices are certainly not The component control samples observed only after a long period of universal since some operators feel showed absolutely no signs of surface time when the environmental condi restricted when wearing hand covers or stress corrosion cracking following tions are conducive to crack propaga for deiicate soldering operations or tor the 56-day exposure to temperature tion. reworking incorrect joints on high and pure water humidity. These test results emphasize the density boards. The use of bare hands Measurements have been made1'1 on need to preclude unnecessary han in these instances may be justified the salt contributed by fingerprints on dling of Kovar and probably other provided the boards are thoroughly handled printed circuit boards. Indi iron-nickel-cobalt alloy lead materials solvent-cleaned by approved meth vidual prints were found to contribute by bare hands. When circumstances 1 ods immediately after assembly. as much as 30 micrograms of NaCI as permit handling, every precaution The effect of severe handling, which measured on the DCM. It may be must be taken to ensure complete 302-sl OCTOBER 1980 © = R AS-RECEIVED TO • = RMA BOILED CONDITION • = RA A1 K3 130 KEY MANUFACTURERS' Q. Dl SCRIPT I ON OF FLUX ACTIVITY o 120 A2 _l UJ B 110 K2 > —<—• UJ 100 Q CD K1 X < 90 o QL M3 Z2 < o 80 > UJ tr C/3 M2 UJ u. 70 QL X 60 A3 UJ Z3 Q. Q 50 ©- -© o ©- ~® _l A0 Ml UJ AA > 30 1/1 Z1 X O 20 CC o LX QL o 10 < u UJ 0 (/> UJ 10 100 1000 10000 100 000 QL D.C.M. TEST RESULTS (jjg Nad /j_tI FLUX ) Fig. 22—Relationship between the ionic content of a flux and its corrosive index o0L _l UJ > ® = R • = RMA I o • = RA Q. o X o Q. o _l removal of potential corrosive films by hardware. The surprisingly strong cor for the various classes of fluxes exam UJ adequate post-assembly cleaning pro rosive nature of many commercial ined in the program are as follows: RA > cedures. fluxes precludes their use whenever 0.14 s for gold-plated and 0.5 s for there is the slightest chance that a copper wires; RMA 0.25 s for gold- X Conclusions trace of their residues may remain on plated and 1.3 s for copper wires; R o delicate electronic materials which, for 0.35 s for gold-plated and 2.0 s for QL 1. The flux survey and subsequent the new generation of ESA spacecraft, < copper wires. UJ program of testing and evaluation of may require long storage times prior to The fully activated RA-type fluxes CO specific types of commercial liquid launch and may have, in the case of generally contain a high ionic concen soldering fluxes may serve as a rough communications satellites and Space- tration and possess a high fluxing effi guide to flux selection for application lab, operating lives of 10 years. ciency (FE) whereas no particular rela during the soldering of spacecraft 2. The mean solder wetting times tionship could be established between WELDING RESEARCH SUPPLEMENT : 303-s Fig. 24—Section across handled speci Fig. 25—Section across clean control sam men—severe stress corrosion cracking. pie—no evidence oi any form of corrosion A—under scratch; B-selected region A—under scratch; B-selected region Fig. 2b—Photograph showing the deleter ious eiiect oi Class 2 top-marginal pass) and Class 3 (bottom-tail) (luxes following the copper mirror test FE and ionic concentration for the 6. Concerning gold-plated Kovar RMA and R-type fluxes. The applica component leads, these may become tion of heat was found to increase the slightly surface-corroded, but degrada 7. The leads of electronic compo ionic content of most tluxes. tion by stress corrosion cracking nents may become severely degraded (SCC), as caused by the majority of the 3. Although somewhat subjective, by the transfer of contamination, such tested fluxes, is not always visible until the standard copper mirror test and as perspiration from an operator's bare such leads fracture in two. To avoid the proposed copper sheet test pro hands. Such practices must be SCC initiation and propagation, the vided comparable results. Boiling indi avoided, and it is an ESA requirement' following ESA soldering requirements' vidual fluxes did not produce different that clean white gloves or finger cots must be enforced: results. Neither test method is consid be worn in order to avoid the form of ered to be particularly selective in • Flux and residue are removed imme catastrophic SCC depicted in Fig. 27. assessing the corrosiveness of fluxes diately after soldering. under service conditions. • Stress relief bends must be provided Acknowledgment 4. Only the "ad hoc" corrosion tests between component body and part are considered likely to shed light on termination. The authors wish to thank Mr. D.S. the component lead and wire prob • Leads must not be sharply bent; use Collins for his assistance with the me lems encountered on ESA spacecraft the minimum lead-bending require tallography and Mr. H. Smith of the projects; they reproduce the material ments. Fulmer Research Institute, England, fot characteristics and the interrelated • Leads must not be forced to lie flat the halide content and pH determina mechanisms of galvanic corrosion and during soldering, and they must be tions. stress corrosion cracking. formed accurately to prevent residual 5. Concerning silver-plated copper stresses, e.g., components must not be References wire, fluxes of the RA-type should not held into PCB plated-through holes by 1. ESA-PSS-14/QRM-08, "The Manual be used. If there is any likelihood that the spring properties of their leads; Soldering of High Reliability E lectrical Con the plating is damaged or porous, then this could provide ideal conditions for nections," 1918. certain of the RMA and R-type fluxes catastrophic SCC failure. 2. QQ-S-571, "Non-corrosive Rosin- listed in Table 3 would not be recom • Non-authorized fluxes, solvents, cored Solder Wire," 1963. 3. DTD-599A, "Non-corrosive Flux for mended, particularly when the ingress etc., must be removed from the work Soft Soldering," March 1961. of flux or contaminated cleaning solu area. 4. MIL-F-14256C, "Flux Solutions of Ros tions cannot be prevented. In case of • Also, mounting pads, conformal in or Mildly Activated Rosin," 1963. doubt, the cleaning/vacuum bake coatings, etc., should be designed to 5. Peters, S.T., and Wesling, N., "Corro remedy outlined in the literature" may limit stresses due to differential ther sion of Silver-plated Copper Conductors," eliminate corrosion. mal expansion.-" SAMPE 13th National Symposium, May - MARGINAL PASS . . . . . _l_ 0 001 0.0_|_1 01 1.0 HALIDE CONTENT OF AS RECEIVED FLUX , 7. BY WEIGHT • 27-Effect oi halide content on copper mirror test results 304-sl OCTOBER 1980 1968. tion tor Semiconductor Devices—Group B It was evident from the replies 6. Peters, S.T., "Review and Status of Red Tests." received that the majority of compa Plague Corrosion of Copper Conductors," 15. Dunn, B.D., "Product Assurance and nies do not carry out any form of Insulation/Circuits, May 1970, p. 55. Choice of Materials for Satellite Construc incoming inspection tests or chemical 7. Reich, B., "Stress Corrosion Cracking tion," Metall, August 1976, 30 (9), pp. controls on any of their liquid flux of Cold-plated Kovar Transistor Leads," Sol 711-720. purchases. In general, it was found id State Technology, April 1969, pp. 36-38. 16. Elkind, M.|., and Hughes, H.E., "Pre that company purchase orders refer 8. Studnick, W.R., and Foune, C.C, vention of Stress Corrosion Failure in Fe- "Testing for Corrosivity in Activated Liquid Ni-Co Alloy Semiconductor Leads," Bell enced only that procurement should Soldering Fluxes," The Western Electric Telephone Laboratory Report in Physics oi be against certain national specifica Engineer, |an. 1973, Vol. XVII, No. I, pp. Failure in Electronics, 5 (1967), pp. tions and that acceptance testing was 3-8. 447-495. assumed to have been performed at 9. Weirick, L.J., "A Metallurgical Analysis 17. Harboe, R., and Adams, L., "An Inves the supplier's plant prior to batch re of Stress Corrosion Cracking of Kovar Pack tigation of the CMOS Lead Corrosion Prob lease. age Leads," Solid State Technology, 1975, lem." ESTEC-Working Paper No. 1023 (con In total, 71 flux samples wore Vol. 18, No. 3, pp. 25-30. fidential). received—Table 4. The most common 10. Dunn, B.D., "Reliable loints for 18. Reich, B., "SCC of Gold-plated Kovar Spacecraft, Part 2," Electronic Production, Transistor Leads," Solid State Technology, flux products originated from British, 1978, Vol. 7, No. 4, p. 23. 1969, Vol. 12. No. 4, pp. 36-38. U.S. and German manufacturers. A few VI. Brons, )., "Evaluation of Post-solder 19. Spa'hn, H., "Performance Require were received with Dutch and Belgian Flux Removal," Welding journal, 54(12), ments tor Stainless Steels in the Chemical brand names. All samples were sub Research Suppl. Dec. 1975, pp. 444-s to 448- Process Industry," Proceedings of the Stain jected to a halide content determina s. less Steel 1977 Conference, London, p. tion according to the method de 12. Tautscher, C.)., "Printed Wiring Board 161. scribed in the literature1, here, the Cleanliness Testing," Circuit World, Vol. 4, 20. Dunn, B.D.. "The Resistance of percentage by weight of halide is cal No. 2, p. 30. Space-quality Solder Joints to Thermal Fa culated as chlorine on the weight of 13. B.S. Draft Specification No. 77/78244 tigue," Circuit World, Part I, Vol. 5, No. 4, the non-volatile portion of flux. intended to replace B.S. No. 41 1, 1954. 1979, pp. 11-17; Part 2, Vol. 6, No. 1, 1979, 14 MIL-S-19500D, "General Specifica pp. 16-27. A pH value determination was attempted by placing chemically treated pH papers in the as-received Appendix: Summary of Survey Conducted to Establish Soldering Fluxes flux for one minute; color changes of Utilized by European Manufacturers of Electronic Equipment for Space the papers were dependent on the Application hydrogen-ion content of each flux. All fluxes were acidic, having a pH of less Thirty of the major European com more samples of fresh uncontami- than 7. However, the dye in the papers panies engaged in the manufacture of nated flux as used at their plants. The did not respond strongly to all fluxes electronic hardware for ESA spacecraft samples were accompanied by com and these results are somewhat sub applications were approached. Twen plete information about batch identi jective. ty-five companies responded by for ty, manufacturer's date of purchase Each flux was subjected to a copper warding for analysis and testing one or and intended usage. mirror test. This test was performed Table 4—Summary of Various Flux Types Tested, User Companies and Laboratory Results Laboratory test results Copper mirror test Flux Country and Batch Halide content code no. of user identity of as-received pH of original 24 h at 30 C and 50% RH no. comp any of flux sample, % flux As-receiv ;d 3oiled 1 GB 4 a 0.069 5.1 (1) (1) S 11 a 0.037 4.5 (0) (0) NL 13 j 0.022 4.8 (1) CM GB 24 a 0.016 4.8 (1) (1) b 0.054 4.5 0) (1) c 0.028 4.8 (1) (1) GB 25 c 0.053 4,8 (1) (I) 2 D 9 a 0.0012 4.0 (1) (I) DK 10 a 0.0003 4.8 (0) (0) NL 13 c 0.0011 4.8 (0) (0) B 14 b 0.0064 4.5 (0) (0) NL 18 b 0.0003 3.9 (I) (I) DK 26 b < 0.0004 4.8 (0) (0) 3 D 1 a 0.03 5.1 (1) (I) GB 3 a 0.051 5.0 0) (1) DK 10 b 0.038 5.1 (0) (0) B 14 a 0.044 4.5 (0) (0) NL 18 a 0.032 5.0 (0) (0) E 19 a 0.024 5.1 (1) (I) 4"" B 5 b 0.01 4.5 (0) (0) DK 7 b 0.0095 4.8 (1) (1) NL 13 d 0.012 4.8 (0) (0) S 27 a 0.01 4.8 (1) (I) (Continued on next page) WELDING RESEARCH SUPPLEMENT 305-s using flux in the as-received condition glass slides as required by MIL-F- corrosive attack on the copper. and after boiling a standard volume of 14256C" Approximately 0.05 ml of flux The pass/fail criteria of this test are flux for one minute at approximately was placed on the copper side of the given in the footnote to Table 4. Any 200°C (392°F) in order to simulate the mirrors which were then stored in the penetration of the copper thickness soldering operation. The copper mir horizontal position inside a humidity could be readily seen by holding each rors were prepared in the laboratory by chamber at 30 ± 2°C (86 ± 3.6°F) and mirror against a light source. Such cor vacuum-depositing 5000 A of pure 50% relative humidity for 24 h. Each rosive attack was classified as a class 3 copper onto treated vapor-degreased slide was then visually examined for flux failure. Attack on, or modification Table 4—Summary of Various Flux Types Tested, User Companies and Laboratory Results (Continued) Laboratory test results Copper mirror test, Flux Country and Batch Halide content 24 h at 30°C and 50% RH code no. of jser identity of as-received pH of original no. company of flux sample, % flux As-received Boiled 5 E 6 a 0.46 4.5 (0) (0) NL 13 w 0.48 - - - B 14 c 0.48 4.5 (0) (0) D 22 a 0.46 4.5 (1) (I) 6"" DK 7 a 0.83 4.5 d) (1) S 11 b 0.28 3.9 (1) (1) NL 13 k 0.59 4.8 (2) (2) 7"" DK 7 c 0.067 3.9 (I) 0) NL 13 e 0.011 4.5 (2) (2) F 31 a 0.0009 4.8 (2) (2) 8 NL 13 t 0.42 - - - CB 15 b 0.40 4.5 (0) (0) CB 25 b 0.36 4.8 (0) (0) 9 DK 10 d 2.31 4.5 (2) (2) GB 30 a 2.32 4.8 (1) (1) 10 \l 13 q 0.0002 - - - GB 25 q 0.0018 4.5 (2) (2) 11 DK 10 e 0.70 3.9 (0) (0) GB 15 a 0.44 3.9 (2) (2) 12 B 5 q 0.0035 4.8 (2) (2) 13 D 8 < 0.0006 4.8 (2) (2) 14 D 8 5.84 3.9 (2) (2) 15 D 8 < 0.0006 4.5 (0) (0) 16 D 8 0.0009 4.2 (0) (0) 17 D 9 0.0019 4.5 (2) (2) 18 DK 10 0.88 3.9 (1) (1) 19 NL 13 0.0009 4.2 (0) (0) 20 NL 17 0.004 4.8 (1) 0) 21 NL 17 0.0028 4.5 (0) (0) 22 NL 17 0.68 4.5 (2) (2) 23 NL 18 0.012 5.1 (2) 24 I 32 a 0.078 5.0 - - F 21 a 0.048 4.5. (1) (1) 25 D 22 b 0.019 4.8 (1) d) 26 DK 26 q 12.95 2.0 (3) (3) 27 NL 13 0.006 (0) (0) 28 NL 13 0.0046 29 NL 13 0.008 30" NL 13 f 0.0029 4.5 (2) (2) 31 D 8 e < 0.0004 4.2 (2) (2) 32«b \l 13 7.93 3.4 (3) (3) 33"" NL 13 4.45 3.6 (3) (3) 34"" NL 13 0.0011 4.8 (0) (0) 35"" NL 13 0.0013 4.8 (1) (I) 36"" NL 13 0.73 4.8 (D (I) 37"" NL 13 0.75 4.8 (2) (2) 38"" NL 13 62.50 1.0 (3) (3) 39 I 32 < 0.0002 10.0 ta,The copper mirror test was performed on flux samples in the as-received condition, in the same concentration as used by the participating ESA contractors The test was repeated after each sample had been boiled for 1 mm at 200 ± 5°C to simulate the soldering operation. The pass/fail criteria are classified as follows: Class 0—Pass: no discoloration of flux or copper; class 1-Marginal pass: strong discoloration of flux, but no attack on copper, class 2-Marginal pass: slight discoloration of flux with salmon pink 'etch' on copper, class 3—Fail: excessive etching with penetration of copper Ihickness 'hlHuxes chosen for further testing 306-si OCTOBER 1980 of, the copper surface was not classi tractors and used only for the pre- physical makeup of each particular fied as a failure, but as a marginal pass tinning of "difficult" metals. solder joint configuration. (either class 1 or 2). Examples of 2. With the exception of the ex 3. Slight variations are observed in classes 2 and 3 are shown in Fig. 26. tremely corrosive fluxes, there appears the halide content and pH values of The results of the copper mirror test to be no relationship between the different batches of the same flux type z and the halide content analysis are resistance to copper mirror corrosion when purchased in different countries. UJ Q- compared in Fig. 27. The following and the individual flux halide concen This indicates that flux manufacturers O observations are made following the trations (from Fig. 27). However, this may change the chemical formulation _i results of the tests: is unlikely to be true for bi-metallic of their brand name products. Howev >UJ 1. Based on the copper mirror test corrosion or in the presence of alloys er, this has no marked effect on the results, the extremely corrosive fluxes susceptible to stress corrosion. For a copper mirror test. have a halide content in excess of 4.0% realistic assessment of the corrosive 4. Flux types as-received and after and pH values of less than 3.5. Such nature of any flux, it is therefore rec boiling produce identical copper mir fluxes are rarely employed by ESA con ommended that the corrosion test ror test results. assembly reproduce accurately the WRC Bulletin 258 May 1980 International Benchmark Project on Simplified Methods for Elevated Temperature Design and Analysis: Problem I—The Oak Ridge Pipe Ratchetting Experiment; Problem II—The Saclay Fluctuating Sodium Level Experiment by H. Kraus Problem I—The Oak Ridge Pipe Ratchetting Experiment is analyzed by general purpose finite element computer programs and by approximate analytical techniques. The methods are described and the results are compared to the experimental data. Problem II—The Saclay Fluctuating Sodium Level Experiment is analyzed by special purpose computer programs and by approximate analytical techniques. The Methods are described and the results are compared. Experimental data are not yet available. Publication of these reports was sponsored by the Subcommittee on Elevated Temperature Design of the Pressure Vessel Research Committee of the Welding Research Council. The price of WRC Bulletin 258 is $10.00 per copy, plus $3.00 for postage and handling. Orders should be sent with payment to the Welding Research Council, 345 East 47th St.. Room 801. New York, NY 10017. WRC Bulletin 259 June 1980 Analysis of the Radiographic Evaluation of PVRC Weld Specimens 155, 202, 203, and 251J by E. H. Ruescher and H. C. Graber This report is one of a series of analyses of nondestructive examination data obtained from heavy-section steel weldments with intentionally introduced flaws prepared by the PVRC Subcommittee on Nondestructive Examination of Pressure Components. The primary objective in this work area was to determine the radiographic detectability of deliberately induced flaws. A group of evaluation teams without prior knowledge of the number, type or location of the intentional discontinuities independently examined each specimen in accordance with identical instructions. The results of these examinations were used as the basis for decisions regarding the flaws. This report describes the evaluation techniques used to reduce the data from the detectability of the deliberately induced and naturally occurring flaws in the weld specimen. Publication of this report was sponsored by the Subcommittee on Nondestructive Examination of Pressure Components of the Pressure Vessel Research Committee of the Welding Research Council The price of WRC Bulletin 259 is $11.00 per copy, plus $3.00 for postage and handling. Orders should be sent with payment to the Welding Research Council, 345 East 47th St., Room 801, New York, NY 10017. WELDING RESEARCH SUPPLEMENT , 307-s