IEEE TRANSACTIONS ON DEVICE AND MATERIALS RELIABILITY, VOL. 10, NO. 1, MARCH 2010 71

The Influence of H2S Exposure onImmersion-Silver-Finished PCBsUnder Mixed-Flow Gas Testing

Shunong Zhang, Michael Osterman, Member, IEEE, Anshul Shrivastava,Rui Kang, and Michael G. Pecht, Fellow, IEEE

Abstract—This study focuses on the corrosion of immersionsilver (ImAg)-finished copper land patterns on printed circuitboards (PCBs) under H2S exposure. Eight test conditions wereexamined with varying levels of H2S, temperature, relative humid-ity, and exposure time. The results indicated both direct chemical-reaction corrosion and electrode-reaction corrosion, particularlygalvanic corrosion. H2S gas was a stronger driving force forcorrosion on ImAg-finished PCBs than SO2 gas. Temperaturehad a significant influence on ImAg-finished surface PCBs. Testsfound extensive corrosion on ImAg-finished PCBs at 40 ◦C, evenin very low humidity. On ImAg-finished surfaces, the corrosionwas nonuniform during the early period of exposure, with thecorrosion modes being mainly pitting, open mouth, and particles.The corrosion products at this early stage mainly included Cu2Oand Ag2S. As time progressed, the corrosion products mainlyincluded Ag2S, Cu2S, CuS, and CuO. These formed a passive filmon the surface. ImAg-finished PCBs are vulnerable to H2S gas;nonuniform severe corrosion was found to occur at defects on theImAg surfaces. The mixed-flow gas test produced creep corrosionon ImAg-finished PCBs using only H2S gas. Dendrite corrosionproducts growing from an edge with a solder mask were generallylonger than those growing from an edge without a solder mask.The corrosion products of these dendrites mainly included Cu2Sor CuS.

Index Terms—Corrosion, H2S environment, immersion silver(ImAg) printed circuit board (PCB), mixed flow gas (MFG) test.

I. INTRODUCTION

W ITH THE ADVENT of the restriction of hazardoussubstances legislation in Europe, which forbids the use

of lead in most electronic products, the electronic industryhas seen an increased level of interest in the use of Pb-free

Manuscript received June 15, 2009; revised August 21, 2009. First publishedSeptember 29, 2009; current version published March 5, 2010. This workwas supported in part by the Center for Advanced Life Cycle Engineering,Department of Mechanical Engineering, University of Maryland.

S. Zhang and R. Kang are with the Department of Systems Engineering ofEngineering Technology, Beijing University of Aeronautics and Astronautics,Beijing 100191, China (e-mail: zsn@buaa.edu.cn; kangrui@buaa.edu.cn).

M. Osterman and A. Shrivastava are with the Center for Advanced LifeCycle Engineering, Department of Mechanical Engineering, University ofMaryland, College Park, MD 20742 USA (e-mail: osterman@calce.umd.edu;ashrivas@calce.umd.edu).

M. G. Pecht is with the Department of Electronics Engineering, City Uni-versity of Hong Kong, Kowloon, Hong Kong, and also with the Center forAdvanced Life Cycle Engineering, Electronics Products and Systems Center,University of Maryland, College Park, MD 20742 USA (e-mail: mgpecht@cityu.edu.hk; pecht@calce.umd.edu).

Color versions of one or more of the figures in this paper are available onlineat http://ieeexplore.ieee.org.

Digital Object Identifier 10.1109/TDMR.2009.2033194

surface finishes for printed circuit boards (PCBs). Prior to theban on lead, SnPb hot-air solder level (HASL) finishes domi-nated the market. With Pb-free electronics, organic solderabilitypreservative (OSP), immersion silver (ImAg), electroless nickelimmersion gold (ENIG), or immersion tin (ImSn) finishesare now being used. With improved laminates and processes,the use of Pb-free HASL is beginning to increase. The al-ternatives to HASL, generally, have much flatter and thinnercoatings [1]–[7].

Among the Pb-free finishes, ImAg and OSP are the preferredfinishes for many applications, while ImSn and ENIG are usedfor niche applications [4]. ImAg has many desirable character-istics, including maintaining solderability or wettability for upto 12 months prior to assembly, and good conductivity. ImAgcan be applied flat and thin, and the cost of plating ImAg ishalf the price of ENIG and is comparable with ImSn finishes[6], [7]. However, ImAg has been shown to have issues in high-sulfur environments [3].

High-sulfur environments can come from many sources, forexample, design facilities, where the Kaolin (China) clays areused to model products; paper mills, where sulfur is used in thebleaching process; power plants, where geothermal sources areused to turn steam turbines; oil refineries, where sulfur gasesare produced during the processing of crude oil; and waste-treatment plants, etc. [3].

Researchers have studied the corrosion phenomenon andcorrosion mechanisms in ImAg PCBs [1]–[6]. Some studieshave also compared various kinds of Pb-free PCB finishes in thesame batch experiments [1], [2], [4], [5]. In order to simulate thefield-failure modes and assess the reliability of Pb-free PCBs,researchers have used mixed-flow gas (MFG) tests [1], [2], [4]and clay tests [5], [6] to evaluate corrosion resistance.

Cullen [1] indicates that the tarnish of ImAg is mainly avisual problem and does not pose a functional loss until thetarnish reaches deep enough to expose the underlying copper.Exposed PCB copper, when submitted to environments ofliquid water and high-sulfur content, will form soluble salts.These salts may become mobile and creep between conductors,resulting in the possibility of electrical failure. PCB surface fin-ishes have varying vulnerabilities to corrosive anions. All PCBsurface finishes are sensitive to sulfur. Creep corrosion can beproduced using a condensing-vapor test, which uses H2S pro-duced by adding 1% hydrochloric acid to 0.1 g/l sodium bisul-fide to form a “sulfur chamber.” All finishes exhibit creeping

1530-4388/$26.00 © 2010 IEEE

72 IEEE TRANSACTIONS ON DEVICE AND MATERIALS RELIABILITY, VOL. 10, NO. 1, MARCH 2010

TABLE ICORROSION ENVIRONMENT CLASSES, BASED ON

Cu2S GROWTH ON COUPONS [4]

corrosion within 24 h under this environment. Cullen alsoconducted Class III MFG testing and found that creep corrosiondid not occur until the humidity was increased above 93%,creating condensation.

Veale [2] conducted MFG tests (100 ppb H2S, 200 ppb NO2,200 ppb SO2, 20 ppb Cl2, temperature of 28 ◦C–29 ◦C, 75%RH,and test duration of 20 days) to study the corrosion resistanceof OSP, ImAg, ENIG, and ImSn. The copper reactivity wasmonitored throughout the test, and the accumulated corrosionper day was 3500 Å. From these results, Veale concluded thatnone of the tested coatings could be considered immune tofailure in 10 years under an ISA G3 environment. However,based on documented copper reactivity (350 nm/day) for anexposure of 20 days, the condition would match 10 years undera moderate-level ISA G2 and also match one year under a mild-level ISA G1 (or Battelle III). The results indicate that noneof the coatings can be considered immune from failure in aBattelle class III environment, but ImSn and OSP could beexpected to survive a Battelle class II environment. Table I com-pares the classification of Battelle and ISA environments [4].

Xu et al. [4] conducted MFG testing on PCBs with OSP,ImAg, ImSn, and ENIG finishes under 40 ◦C, 69%RH,1700 ppb of H2S, 200 ppb of NO2, 20 ppb of Cl2, and200 ppb of SO2, which represented Battelle class IV and ISAclass G2 conditions. To further simulate operating conditions,half of the samples were also biased at 10 V with a currentcontrol limit at 4 mA. The corrosion rate was monitored usingCu sample weight gain and was determined to be ∼600 nm/day,averaged over 28 samples installed at various locations insidethe MFG chamber. An IPC B25 comb structure with 1.0 oz(1 oz = 28.35 g) finished copper was used as the test vehicle.The width of the copper lines was 0.01 in and the spacingbetween adjacent lines was 0.013 in. The same test vehicle wasdistributed to several major PWB finish chemistry suppliersto be coated with OSP, ImAg, ImSn, and ENIG. Before theMFG test, the test boards were processed through two Pb-freereflow cycles. Seven different ImAg processes from severalsuppliers were tested in this experiment. The results after twodays showed that all the ImAg finishes were covered withgrayish corrosion products, mostly Cu2S. Four out of sevenImAg finishes showed fiber-assisted electrochemical migrationafter five days of MFG exposure. Creep corrosion along a fiber(fiber-assisted creep corrosion) was observed after ten days.From the test results, it was found that ImAg-finished boardswere more susceptible to corrosion than the other three finishesthat were tested. A second issue associated with ImAg wasblistering or peeling of the conductive corrosion products fromthe surfaces. After ten days of exposure in the MFG test, most of

the samples only showed minor blistering. Peeling and flakingwere only observed after 40 days of MFG exposure.

Schueller [5] first used high-sulfur clay to drive corrosion,in 2007. He selected Chavant clay (type J-525) containing30%–50% elemental sulfur and heated the clay with water tosimulate an actual environment. Creep corrosion was found ondifferent finishes, depending on the weight and heating times ofthe clays. The test methods were designed to simulate differentharsh class-GX environments.

After Schueller’s study, Zhou and Pecht [6] conducted re-search on the assessment of ImAg-finished PCBs using a clay-exposure test. Three ImAg-finished PCBs with different ImAgfinish thicknesses (0.36, 0.22, and 0.21 μm) were taken asthe test samples. After exposure in a sealed enclosure withclay, the ImAg-finish surfaces were found to be sulfurized.Creep corrosion products that are shaped like dendrites wereformed on ImAg-finished pad edges under high-sulfur andhigh-humidity environments. CuS was the main component ofthe creep corrosion, accompanied by a little bit of Ag2S. Theoccurrence probability and the length of the creep-corrosionproducts were dependent on the coverage of ImAg finish onthe copper trace and were reduced as the thickness of the ImAgfinish was increased, and the solder mask was separated fromthe pad edges.

Sulfur has clearly been shown to be a factor in the corrosionof ImAg finish. Yet, no test has been conducted in a corrosiveenvironment containing a single corrosive gas such as H2S orSO2, etc., during the analysis of ImAg as a PCB finish. Thus,the behavior of ImAg under such conditions remains largelyunknown. In our study, MFG tests using only H2S and SO2,respectively, were conducted with ImAg and other finishedPCBs. This paper, however, focuses solely on the influenceof H2S on ImAg-finished PCBs under MFG testing. Eightexposure levels are presented and compared.

II. PROCESS OF SAMPLE PREPARATION

For this study, unassembled PCBs with ImAg finish wereused as test samples. The thickness of the ImAg finish was0.23–0.36 μm, as detected at five different locations by X-rayfluorescence spectroscopy. One ImAg-finish PCB was seg-mented into many pieces. Each piece was approximately50 mm × 50 mm in size and contained various distributionsof finished through holes and pads. The test-sample preparationprocedure included the following:

1) cut ImAg surface-finished PCB board into pieces using aDremel tool;

2) rinse under running tap water;3) wash individual pieces in an ultrasonic cleaner with

deionized water;4) dry the pieces in ambient air;5) place pieces on aluminum foil and dry for an additional

2 h at 110 ◦C;6) place pieces in plastic storage bags;7) select three pieces for inspection with an environmen-

tal scanning electron microscope (E-SEM, Model: FEIQUANTA 200) and energy dispersive X-Ray spec-troscopy (EDS).

ZHANG et al.: H2S EXPOSURE ON ImAg-FINISHED PCBs UNDER MFG TESTING 73

Fig. 1. Typical ImAg-finished surface before testing.

Fig. 2. Point analysis of (a) ImAg-finished surface and (b) solder-masksurface before testing.

Fig. 1(a) shows a typical ImAg-finished PCB segment sam-ple before testing. Fig. 1(b) shows a typical through hole underE-SEM. Fig. 1(c) shows the two different kinds of edges of anImAg surface: with a solder mask and without a solder mask.Fig. 1(d) is a typical ImAg surface under high magnification.

The elements on the ImAg surface included Ag and Cu(determined by an EDS point analysis at 30 kV). The elementson the solder mask included C, O, Si, S, Ba, and Ca. Thesefindings were reconfirmed by a separate SEM (S-530) and EDS(Link ISIS) system under a high vacuum and 10 kV. Figs. 2(a)and 3 show an EDS point analysis on ImAg-finished surface,and Figs. 2(b) and 4 show a point analysis on solder-masksurface.

Bare copper samples were also prepared in order to monitorthe corrosion rate. The copper samples were prepared as persection 7.6.1 of ASTM B810-01a [11].

III. EXPERIMENTS AND RESULTS

The results of eight MFG tests using H2S gas are presentedin this paper. For each test, the test samples were suspended

Fig. 3. Point analysis on ImAg-finished surface by EDS (10 kV).

Fig. 4. Point analysis on solder-mask surface by EDS (10 kV).

TABLE IITEST MATRIX

by insulated wires in the MFG chamber [15]–[21]. Duringthe exposure, visual observations were conducted. The testspecimens were examined under E-SEM, and the elementalcomposition of each specimen was measured by EDS afterremoval from the MFG chamber. The examined conditions aresummarized in Table II.

A. Test 1

The first test condition included a temperature of 40 ± 1 ◦C,a relative humidity of 45 ± 1%RH, and an H2S concentration of150 ± 10 ppb. Upon termination after 4.5 h, corrosion productswere observed on the surfaces of the PCB pieces. Corrosionproducts appeared brick red, gray, and blue in color [Fig. 5(a)].Under E-SEM examination, some surfaces showed no obviouschange [e.g., Fig. 5(b)]; some surfaces showed many smallpitting points [e.g., Fig. 5(c)]. An EDS point analysis at 30 kV

74 IEEE TRANSACTIONS ON DEVICE AND MATERIALS RELIABILITY, VOL. 10, NO. 1, MARCH 2010

Fig. 5. 4.5-h exposure (45 ± 1%RH, 150 ± 10 ppb H2S, T = 40 ± 1 ◦C).

Fig. 6. EDS point analysis for pitting points in test 1 (30 kV).

indicated the presence of Ag, Cu, S, C, O, Br, Si, and Ba in a pit(Fig. 6). Some surfaces showed large pits [e.g., Fig. 5(d)]. AnEDS point analysis on the pit at 15 kV showed that the elementsincluded Cu, Ag, S, C, O, and Si (Fig. 7). Some substancesseemed to have migrated near the edge of a hole without asolder mask [Fig. 5(e) and (f)]. A similar phenomenon was ob-served at other holes and pads. An EDS point analysis at 15 kVon a surface edge showed the presence of the elements C, O, Si,Ag, and S (Fig. 8). Another EDS point analysis at 15 kV near

Fig. 7. EDS point analysis for a pit in test 1 (15 kV).

Fig. 8. EDS point analysis on a surface edge in test 1 (15 kV).

Fig. 9. EDS point-analysis inspection near a surface edge in test 1 (15 kV).

the surface edge showed that the elements included C, O, Si,Ag, Cu, and S (Fig. 9).

Discussion: The C, Si, and Ba that were found by EDSinspection (Fig. 6) likely came from the solder mask. Br wasfound in the pitting surface [Figs. 5(c) and 6]. Based on thegas and air path, it was speculated that the Br came fromthe water that was in the heater barrel in order to increasehumidity. However, the water was distilled water that camefrom a distilled-water instrument.

It was speculated that the corrosion mechanism on the boardwas primarily due to the porosity of the ImAg finish. At hightemperatures, some pores on the surface will expand. Ag willreact with H2S and O2 to form Ag2S and Ag2O. Cu from poresin the Ag will also react with H2S and O2 to form Cu2S, CuS,Cu2O, or CuO.

In this test, the relative humidity was about 45%. The Cuin Fig. 9 shows that there was a Cu exposure at the edge of

ZHANG et al.: H2S EXPOSURE ON ImAg-FINISHED PCBs UNDER MFG TESTING 75

Fig. 10. 4.5 h exposure (5 ± 2%RH, 150 ± 10 ppb H2S, T = 40 ± 1 ◦C).

the ImAg-finished surface, and electrochemical migration ofCu2+ or Cu+ occurred. The Ag in Fig. 9 also indicated thatelectrochemical migration of Ag+ occurred. In addition, onthe ImAg finished surface, there were one or more complexelectrode-reactions systems [22]. In particular, when Cu isexposed, Cu and Ag will form galvanic corrosion and react fastwhen they meet sulfur.

Fig. 5(b)–(d) shows that the corrosion was very nonuniformon the ImAg surface. This means that there were some defectssuch as manufacturing defects, slight scratches, or Cu exposureon the surface before the test. When these defects were exposedin the MFG chamber, the corrosion was much faster than inother places, e.g., Fig. 5(d).

B. Test 2

For the second test, the relative humidity was reduced to 5 ±2%RH while maintaining the temperature and gas concentra-tion levels from the first test. Comparing test 2 with test 1, it isdifficult to distinguish which test condition is more severe. Thecolors of the corrosion products after 4.5 h of exposure werealso brick red, gray, and blue [Fig. 10(a)]. Observing underE-SEM, some surfaces of the through-hole pads showed noobvious changes, while some surfaces of the through-hole padsshowed dark points [e.g., Fig. 10(b)]. A surface of a through-hole pad showed a little pit, which looked like an open mouth[Fig. 10(c)]. Deep pits were also found on some surfaces [e.g.,Fig. 10(d)].

After 24 h of exposure, another specimen was taken out.The corrosion products, which were brown in color after theprevious exposure, changed to gray and blue color [Fig. 11(a)].Pitting points, open mouth, and pits on ImAg-finished surfaceswere observed under E-SEM [Fig. 11(b)–(d)].

Fig. 11. 24-h exposure(5 ± 2%RH, 150 ± 10 ppb H2S, T = 40 ± 1 ◦C).

Discussion: For this test, the relative humidity was so lowthat the electrochemical reaction could have been ignored, andthe direct chemical reaction should exist on the ImAg surfacewith H2S.

Before the MFG tests using H2S gas, several tests usingonly SO2 were conducted [23]. One of the tests included atemperature of 40 ± 1 ◦C, a relative humidity of about 5% RH,a concentration of SO2 of about 400 ppb, and exposure durationof three days. One sample each was removed for inspectionafter 48 and 72 h. The samples did not show obvious changes[Fig. 12(a) and (b)]. The surface of the sample after 48 h ofexposure time showed some little dark points [Fig. 12(c)] underE-SEM. However, point analysis by EDS did not show specialelements; e.g., Figs. 12(d) and 13 show a point analysis by EDSof the surface of the 48-h ImAg sample. Comparing Fig. 10(a)with Fig. 12(a) and (b) and the results of the analysis, H2S wasshown to be a strong driving force of corrosion on ImAg PCBs.

C. Test 3

For Test 3, the following conditions were imposed: 32 ±1 ◦C, 80 ± 3%RH, and 150 ± 30 ppb. The exposure durationwas 24 h. After 4.5 h, one sample was removed for examina-tion. The removed specimen exhibited corroded surfaces withslight color changes that could be observed by unaided eyes[Fig. 14(a)]. E-SEM examination of the surfaces found thatsome surfaces of the through-hole pads showed no obviouschanges [Fig. 14(b)]. However, pits were also found [e.g.,Fig. 14(c) and (d)].

Fig. 15(a) shows an obvious color change from slight brownto slight gray and blue on the sample after 24 h. E-SEMexamination found surface pitting [Fig. 15(b)] and exposedcopper [Fig. 15(c) and (d)].

76 IEEE TRANSACTIONS ON DEVICE AND MATERIALS RELIABILITY, VOL. 10, NO. 1, MARCH 2010

Fig. 12. ImAg surfaces after 48 and 72 h of exposure time under MFG testingusing SO2 (5%RH, about 400 ppb SO2, T = 40 ± 1 ◦C). (a) 48 h. (b) 72 h.(c) 48 h. (d) 48 h.

Fig. 13. EDS point analysis on ImAg surface after 48 h of exposure timeunder MFG testing using SO2 (20 kV).

Discussion: Comparing Fig. 14(a) with Fig. 10(a), theamount of corrosion in test 3 was much less than in test 2.The influence of high temperature (40 ◦C) proved to be moresignificant than low temperature (32 ◦C) in environments witha low concentration of H2S gas, even if the humidity was ashigh as 80%RH.

Comparing Fig. 14(c) with Fig. 14(d), the pits in Fig. 14(c)seem to have expanded from pores, while the pits in Fig. 14(d)may be from scratches that occurred before the test and thatcould not be seen by the unaided eye, and the pits in Fig. 15(c)and (d) also exhibit similar characteristics.

D. Test 4

The conditions in test 4 included a temperature of 32 ± 1 ◦C,a relative humidity of 80 ± 3%RH, and an H2S concentrationof 650 ± 20 ppb. After the set exposure time of 4.5 h, corrosion

Fig. 14. 4.5-h exposure (80 ± 3%RH, 150 ± 30 ppb H2S, T = 32 ± 1 ◦C).

Fig. 15. 24-h exposure (80 ± 3%RH, 150 ± 30 ppb H2S, T = 32 ± 1 ◦C).

products, brown with a little gray, could be observed on muchof the ImAg-finished surfaces [Fig. 16(a)]. E-SEM examinationfound evidence of pitting and large areas of exposed copper[Fig. 16(c) and (d)]. However, there were still some surfacesthat were not showing any obvious changes [Fig. 18(b)].

Discussion: By comparing Fig. 16(a) with Fig. 14(a) it canbe seen that the influence of a high concentration of H2S wasmore significant in a high-humidity environment.

ZHANG et al.: H2S EXPOSURE ON ImAg-FINISHED PCBs UNDER MFG TESTING 77

Fig. 16. 4.5-h exposure (80 ± 3%RH, 650 ± 20 ppb H2S, T = 32 ± 1 ◦C).

Fig. 17. 4.5-h exposure (35 ± 3%RH, 650 ± 20 ppb H2S, T = 32 ± 1 ◦C).

E. Test 5

The test conditions for Test 5 included a temperature of32 ± 1 ◦C, a relative humidity of 35 ± 3%RH, and an H2Sconcentration of 650 ± 20 ppb. Surfaces exposed for 4.5 and24 h are shown in Figs. 17 and 18, respectively.

E-SEM examination found pits in different corrosion prod-ucts [e.g., Fig. 17(b)–(d)]. After 24 h, the remaining test speci-mens had gray corrosion products on most surfaces [Fig. 18(a)].

Fig. 18. 24-h exposure (35 ± 3%RH, 650 ± 20 ppb H2S, T = 32 ± 1 ◦C).

Fig. 19. 4.5-h exposure (75 ± 1%RH, 100 ± 10 ppb H2S, T = 32 ± 1 ◦C).

E-SEM examination found particle corrosion products [e.g.,Fig. 18(b)] and pits [e.g., Fig. 18(c) and (d)].

Discussion: By comparing Fig. 17(a) with Fig. 16(a), it canbe seen that the degree of corrosion in test 5 was much less thanin test 4 because of the decrease in humidity.

F. Test 6

For test 6, the H2S gas concentration was 100 ppb under arelative humidity of 75%RH. After 4.5 h, corrosion productswere not observable with naked eyes [Fig. 19(a)]. However,E-SEM inspection found evidence of corrosion on the finishedland [Fig. 19(b)]. A 15-kV EDS point analysis of a particleindicated the presence of Ag, S, C, O, and Si (Fig. 20).

G. Test 7

Under an H2S gas concentration of 50 ppb in test 7, no obvi-ous corrosion was observed after 4.5 h of exposure [Fig. 21(a)].A 30-kV E-SEM inspection of select areas did not reveal anyobvious changes, while some pitting [Fig. 21(b)] was observed

78 IEEE TRANSACTIONS ON DEVICE AND MATERIALS RELIABILITY, VOL. 10, NO. 1, MARCH 2010

Fig. 20. EDS point analysis for a particle in test 6 (15 kV).

Fig. 21. 4.5-h exposure (75 ± 1%RH, 50 ± 10 ppb H2S, T = 32 ± 1 ◦C).

Fig. 22. EDS point analysis on a pitting in test 7 (15 kV).

under 15-kV E-SEM inspection. EDS analysis at 15 kV did notreveal the presence of S (Fig. 22).

H. Test 8

For test 8, a longer exposure time and a higher H2S con-centration (1050 ± 50 ppb) were used, and creep corrosion wasobserved. Fig. 23 shows the samples at different exposure timesin this test. It can be seen that the color of corrosion productschanges from brown to gray with time.

After 1.5 h of testing, pits were observed on some surfacesunder E-SEM inspection (e.g., Fig. 24).

For the sample after 4.5 h of testing, particle [Fig. 25(a)]and dendrite corrosion products [Fig. 25(b)] were found on thesurface of an ImAg-finished PCB piece, and very thin dendriticcorrosion products were also found at the edge with the soldermask [Fig. 25(c)] and also without the solder mask [Fig. 25(d)].A 15-kV EDS analysis of a particle indicated the presence

Fig. 23. Samples in test 8 (75 ± 1%RH, 1050 ± 10 ppb H2S, T = 32 ±1 ◦C). (a) 1.5 h. (b) 4.5 h. (c) 24 h. (d) 48 h. (e) 72 h. (f) 96 h.

Fig. 24. Corrosion status after 1.5 h of exposure in test 8.

of Ag, Cu, S, C, O, and Si (Fig. 26). The element weightpercentage and atomic percentage (Fig. 27) showed that theatomic percent ratio of Cu and O was about 2 : 1, the atomicpercent ratio of Cu and S was about 4 : 1, the atomic percentratio of Ag and S was about 2 : 1, and the atomic percent ratioof Ag and O was about 1 : 1. Therefore, it could be speculatedthat the corrosion products were Cu2O and Ag2S, with a smallamount of Cu2S or CuS.

ZHANG et al.: H2S EXPOSURE ON ImAg-FINISHED PCBs UNDER MFG TESTING 79

Fig. 25. Corrosion status after 4.5 h of exposure in test 8.

Fig. 26. EDS analysis on a particle after 4.5 h in test 8 (15 kV).

Fig. 27. Element weight percent and atomic percent by EDS semiquantitativeanalysis.

In addition, a 15-kV EDS analysis of a dendrite on the soldermask revealed the presence of Cu, Ag, S, C, O, Si, and Ba(Fig. 28). This dendrite was mainly Cu2S or CuS, consider-ing Fig. 4 and Fig. 28 and the galvanic-cell electrochemicalcorrosion.

The dendrite corrosion products after four days are shownin Fig. 29. The ImAg-finished surfaces appeared to be coveredwith particles [Fig. 29(a)]. These particles may be Ag2S, Cu2S,CuS, or CuO that form a passive film under H2S exposure.

Fig. 28. EDS Analysis on the dendrites after 4.5 h in test 8 (15 kV).

Fig. 29. Corrosion status after 96 h exposure in test 8.

More pronounced dendrites can be observed in Fig. 29(b)and (c) compared with Fig. 25(c) and (d). The thickness ofthe dendrites increased over time. Dendrite growth was morepronounced at the interface with the solder mask. This findingmay be due to a lower amount of silver coverage at this interfaceas compared with the nonsolder-mask-defined areas. Dendriteswere also found growing from solder mask [e.g., Fig. 29(d)]which was likely caused by the copper-line exposure under thesolder mask.

In this test, eight bare copper samples were used to moni-tor the tests. The thickness gain was measured by the cross-sectional method, and the corrosion rate was ∼1670 nm/day,which was equivalent to one year under an ISA G3 level.

IV. CONCLUSION

The following summarizes the results of the study that hasbeen presented in this paper.

1) There are two mechanisms of the corrosion of ImAg-finished PCBs under H2S gas environment. One is direct

80 IEEE TRANSACTIONS ON DEVICE AND MATERIALS RELIABILITY, VOL. 10, NO. 1, MARCH 2010

chemical-reaction corrosion, while the other is electrode-reaction corrosion, particularly, galvanic corrosion.

2) H2S gas is a strong driving force for corrosion on ImAg-finished PCBs compared with SO2 gas.

3) Elevating the temperature from 32 ◦C to 40 ◦C promotedcorrosion, even in very low humidity.

4) On the ImAg-finished surfaces, the corrosion is nonuni-form for the early period of exposure; the corrosionmodes of pitting, open mouth, and particles were ob-served, with corrosion products mainly consisting ofAg2S and Cu2O. As time progressed, the corrosion prod-ucts mainly included Ag2S, Cu2S, CuS, and CuO andformed a passive film on the surface.

5) Surface defects are more vulnerable to corrosion thansurfaces without such defects.

6) Dendrite corrosion products growing from the edge with asolder mask are usually longer than those formed from anedge without a solder mask. This finding is likely due tothinner Ag coverage near the solder mask. The corrosionproducts of these dendrites mainly consisted of Cu2Sor CuS.

REFERENCES

[1] D. Cullen, “Surface tarnish and creeping corrosion on Pb-free circuitboard surface finishes,” IPC Works, 2005.

[2] R. Veale, “Reliability of PCB alternate surface finishes in a harsh indus-trial environment,” in Proc. SMTA, 2005, pp. 494–499.

[3] P. Mazurkiewicz, “Accelerated corrosion of printed circuit boards due tohigh levels of reduced sulfur gasses in industrial environments,” in Proc.32nd Int. Symp. Test. Failure Anal., Nov.12–16, Renaissance Austin Hotel,Austin, TX, 2006, pp. 469–473.

[4] C. Xu, D. Flemming, and K. Demerkin, Corrosion Resistance of PWBSurface Finishes, Apex, Los Angeles, CA, Feb. 20–22, 2007.

[5] R. Schueller, “Creep corrosion on lead-free printed circuit boards in highsulfur environments,” in Proc. SMTA Int., 2007, pp. 643–654.

[6] Y. L. Zhou and M. G. Pecht, “Assessment of immersion silver finishedcircuit board assemblies using clay tests,” in Proc. ICRMS, Chengdu,China, Jul. 20–24, 2009.

[7] W. Q. Wang, A. Choubey, M. Azarian, and M. G. Pecht, “An assessmentof immersion silver surface finish for lead-free electronics,” J. Electron.Mater., vol. 38, no. 6, pp. 815–827, Jun. 2009.

[8] P. Zhao and M. G. Pecht, “Field failure due to creep corrosion on com-ponents with palladium pre-plated leadframes,” Microelectron. Reliab.,vol. 43, no. 5, pp. 775–783, May 2003.

[9] Standard Guide for Mixed Flowing Gas (MFG) Tests for Electrical Con-tacts, ASTM Designation: B845-97, 2008.

[10] Standard Practice for Conducting Mixed Flowing Gas (MFG) Environ-ment Tests, ASTM Designation: B827-05, 2000.

[11] Standard Test Method for Calibration of Atmospheric Corrosion TestChambers by Change in Mass of Copper Samples, ASTM Designation:B810-01a, 2006.

[12] W. H. Abbott, “The development and performance characteristics of flow-ing mixed gas test environments,” IEEE Trans. Compon., Hybrids, Manuf.Technol., vol. 11, no. 1, pp. 22–35, Mar. 1988.

[13] Environmental Conditions for Process Measurement and Control System:Airborne Contaminants, ISA-S71.04-1985, 1985.

[14] Airborne Contaminants Test Methods, Nov. 2000, Telcordia GR-63-CORE Issue 2, Section 5.5.

[15] “CALCE standard operating procedures,” Mixed Flow Gas Chamber-Rev. A, Sep. 1999.

[16] CALCE Standard Operating Procedures for ESEM, [Model:Quanta200F(26A6)].

[17] Precision Calibration System Instruction Manual: Model 491 Interim,Kin-Tek Lab., Inc., La Marque, TX.

[18] Operating Instructions: Trace Source ULED Permeation Sources,Kin-Tek Lab., Inc., La Marque, TX.

[19] OptiSonde General Eastern Chilled Mirror Hygrometer User’s Manual,Gen. Eastern Instrum., Wilmington, MA, Oct. 2007.

[20] 8270_74_80_84 Mass-Flow Control Boxes Instruction Manual, MathesonGas Products, Montgomeryville, PA.

[21] ML8780 H2S Analyzer Operation Manual, Monitor Labs Ltd.,Englewood, CO, 1994.

[22] C. N. Cao, Principles of Electrochemistry of Corrosion. Beijing, China:Chem. Ind. Press, Feb. 2008.

[23] S. Zhang, A. Shrivastava, M. Osterman, M. Pecht, and R. Kang, “Theinfluence of SO2 environments on immersion silver finished PCBsby mixed flow gas testing,” in Proc. ICEPT-HDP, Beijing, China,Aug. 10–13, 2009, pp. 116–122.

Shunong Zhang received the B.S. and M.S. degreesin power engineering and Ph.D. degree in systemsengineering from Beijing University of Aeronauticsand Astronautics (BUAA), Beijing, China, in 1990,1995 and 2009, respectively.

She is a Professional Engineer with the Depart-ment of Systems Engineering of Technology En-gineering, BUAA. She has more than 15 years ofengineering experience and has implemented morethan 20 projects that concern prognostic and healthmonitoring (PHM) technologies for electronic prod-

ucts, failure analysis on electronic parts, reliability assessment and verificationtechnologies for products or systems, professional training on sixσ method-ology for manufactories, software development for airplane and aeroenginemonitoring systems, system safety of airlines, etc. In recent years, her researchinterests have focused on PHM, failure analysis and reliability physics, andreliability assessment.

Michael Osterman (M’91) received the Ph.D. de-gree from the University of Maryland, College Park,in 1991.

He has been with the University of Maryland since2002, where he is a Senior Research Scientist andthe Director of the Center for Advanced Life Cy-cle Engineering (CALCE) Electronic Products andSystem Consortium. He heads the development ofsimulation-assisted reliability assessment softwarefor CALCE and simulation approaches for estimat-ing time to failure of electronic hardware under test

and field conditions. He is one of the Principle Researchers in the CALCEeffort to develop simulation models for failure of Pb-free solders. In addition,he has lead CALCE in the study of tin whiskers and has authored severalarticles related to the tin-whisker phenomenon. Further, he has written eightbook chapters and numerous articles.

Dr. Osterman is the recipient of the Best Session Paper Award in the 41stInternational Symposium on Microelectronics, IMAPS 2008 and the BestPaper–Maurice Simpson Technical Editors Award in the Institute of Environ-mental Sciences, 2008. He is a member of ASME and SMTA.

Anshul Shrivastava received the B.S. and the M.S.degrees in mechanical engineering from the Univer-sity of Maryland, College Park, in 2004.

He is a Research Engineer with the Centerfor Advanced Life Cycle Engineering, MechanicalEngineering Department, University of Maryland.His research areas include failure analysis of elec-tronic systems, accelerated testing, and materialscharacterization.

ZHANG et al.: H2S EXPOSURE ON ImAg-FINISHED PCBs UNDER MFG TESTING 81

Rui Kang received the B.S. and M.S. degreesin automation science and electrical engineeringfrom Beijing University of Aeronautics and Astro-nautics (BUAA), Beijing, China, 1987 and 1990,respectively.

He is a Professor with the Department of SystemsEngineering of Technology Engineering, BUAA,where he is also the Chair Professor of Reliabil-ity Engineering Research Center. He is an AdjunctProfessor of Electronics Engineering with the CityUniversity of Hong Kong, Kowloon, Hong Kong. He

is also a Visiting Professor with various prestigious universities, including thefollowing: Automation School, Nanjing University of Science and Technology,Nanjing, China; Mechatronics Engineering School, University of ElectronicScience and Technology of China, Chengdu, China; and Aerospace Engineer-ing College, Nanjing University of Aeronautics and Astronautics, Nanjing.He has developed six courses and has published four books and more than100 research papers. He has advised more than 40 M.S. and Ph.D. studentsas of 2009. He is an expert in reliability in technology and industry for Chinesenational defense. His main research interests include reliability design andexperiment technology based on the Physics of Failure, prognostics and healthmanagement (PHM), and logistics support.

Prof. Kang is the Chief Editor of the Journal of Reliability Engineering. Hereceived the Chinese government award for outstanding scientific research.

Michael G. Pecht (F’92) received the M.S. degreein electrical engineering and the M.S. and Ph.D. de-grees in engineering mechanics from the Universityof Wisconsin, Madison.

He is a Visiting Professor in electronics engi-neering with the City University of Hong Kong,Kowloon, Hong Kong, and with the MechanicalEngineering Department, University of Maryland,College Park, where he is the George Dieter ChairProfessor in Mechanical Engineering, a Professor inApplied Mathematics and the founder and Director

of the Center for Advanced Life Cycle Engineering (CALCE), ElectronicsProducts and Systems Center. He has written more than 20 books on electronicproducts development, use and supply chain management, and over 400 techni-cal articles. He has been leading a research team in the area of prognostics forthe past ten years. He has consulted for over 100 major international electronicscompanies, providing expertise in strategic planning, design, test, prognostics,IP and risk assessment of electronic products, and systems.

Dr. Pecht is the Chief Editor for Microelectronics Reliability and an Asso-ciate Editor for the IEEE TRANSACTIONS ON COMPONENTS AND PACKAG-ING TECHNOLOGY. He served as Chief Editor of the IEEE TRANSACTIONS

ON RELIABILITY for eight years and was on the advisory board of IEEESPECTRUM. He is the recipient of the highest reliability honor award, theIEEE Reliability Society’s Lifetime Achievement Award in 2008, the EuropeanMicro- and Nano-Reliability Award for outstanding contributions to reliabilityresearch, the 3M Research Award for electronics packaging, and the IMAPSWilliam D. Ashman Memorial Achievement Award for his contributions inelectronics reliability analysis. He is a Professional Engineer, an ASME Fellow,and an IMAPS Fellow.