Indoor Air 2000; 10: 47–56 Copyright c Munksgaard 2000Printed in Denmark. All rights reserved

INDOOR AIRISSN 0905-6947

Hygroscopic Fine Mode Particle Deposition on ElectronicCircuits and Resulting Degradation of CircuitPerformance: An Experimental Study

ANDRES LITVAK1,2,3*, ASHOK J. GADGIL1 AND WILLIAM J. FISK1

Abstract A portion of electronic equipment failures is a conse-quence of particle deposition on electronic circuits in normal in-door environments. Deposited hygroscopic particles reduce theelectrical isolation (EI) between conductors. In laboratory experi-ments, we investigated the mechanisms, locations, and effects ofparticle deposition on electronic circuits with surface mountedchips (SMCs) and also on small television sets. One set of elec-tronics was exposed for 281 h to an unusually high concentrationof artificially-generated ammonium sulfate particles while a sec-ond set (experimental controls) was exposed to normal indoorparticles. The particle mass concentration in the high-exposurechamber was 500 times higher than normal. Television reliabilitywas observed and the changes in EI between adjacent legs ofSMCs were measured. The experiments demonstrate the stronginfluence of electrostatic forces on the locations and rates of par-ticle deposition. Although televisions did not fail after exposureto concentrated aerosols, the EI between adjacent legs of theSMCs was, in many cases, greatly diminished. Relative humidityhad a very strong influence on the magnitude of EI. A qualitativeexplanation of the mechanisms of particle deposition and circuitdegradation is proposed, including the role of fibers. Finally, apotential method to reduce particle deposition on electronic com-ponents is discussed.

Key words Electronics; Experiments; Deposition; Failures;Indoor air; Particles

Received 8 June 1998. Accepted for publication 15 February 1999.

C Indoor Air (2000)

IntroductionBackgroundAn estimated $1 billion is spent annually to repairfailed electronic circuits in U.S. telephone switching of-fices and approximately 20% of these failures (worth

1Indoor Environment Department, Lawrence Berkeley National Laboratory, Berkeley, CA 94720, USA, Tel: π1 510 486 6591, Fax: π1 510 486 6658, 2Labora-toire des Sciences de l’Habitat, DGCB-URA CNRS 1652, ENTPE, Rue Maurice Audin, 69518 Vaulx-en-Velin Dedex, France, 3Present address: CETE deLyon - DVT/HTC, 46 rue St-ThEobald - BP 128, 38081 LıIsle dıAbeau Cedex, France, Tel: π33 4 74 27 51 67, Fax: π33 4 74 27 52 52, e-mail: andres.litvak-/cetelyon.equipement.gouv.fr, *Author to whom correspondence should be addressed

US$200 million) are thought to be caused by indoor airpollutants (Weschler, personal communication;Weschler and Shields, 1991b). Given the widespreaduse of electronic equipment in the U.S. economy, thetotal U.S. cost of electronic equipment failures causedby indoor air pollution must be many times larger. Pro-tection of electronic circuits against indoor air pol-lutants represents an important issue for industrialeconomies which rely on increasingly smaller elec-tronic devices for communication and information pro-cessing. In the future, as electronic components, cir-cuits, and spacing between adjacent conductors con-tinues to get smaller,1 degradation of circuit reliabilityfrom indoor particle deposition may assume evenmore importance.

Indoor ParticlesAirborne particles commonly have a bimodal mass dis-tribution. The coarse-mode particles (diameter rangeapproximately 2 mm to 20 mm) are typically generatedas a result of the mechanical wear of materials. Theycan derive from mineral and biological sources and in-clude fibers or biological particles. Coarse mode par-ticles are usually removed from the indoor air at amoderate to high efficiency with the filters in buildingair handling systems. Fine mode particles have diam-eters smaller than approximately two micrometers andare typically generated as a result of combustion and

1For example, in some new electronic devices with ‘‘fine-pitch’’spacing, the spacing between pins is approximately 0.25 mm.Such devices may have 300 pins in one side of an integated circuitchip and more than 1,000 total pins.

Litvak, Gadgil and Fisk

photo-chemical gas-to-particle conversion. Fine-modeparticles may be generated indoors but many of theseparticles enter buildings with outdoor air. Fine-modeparticles are not efficiently removed by the filters usedmost commonly in building air handling systems.

Fine mode particles may cause greater hazard to in-door electronics than coarse particles. They are moredifficult to filter from the air and they are often hygro-scopic as discussed subsequently. Also, fine particlescan deposit equally well on vertical and horizontal sur-faces. Voltages applied to electronics can producestrong electric fields dramatically, enhancing fine par-ticle deposition (for example, it is common to find en-hanced deposits of fine-mode particles on top of powersupply lines on a circuit board).

Major mechanisms affecting coarse mode particledeposition onto electronic circuit features includegravitational settling and inertial impaction. The majorforces affecting fine mode deposition include dif-fusional deposition, and deposition caused by thermo-phoretic and electrostatic forces.

Hygroscopic particles, both coarse and fine mode,can absorb moisture from the environment and changetheir diameter (and mass), and thus their major depo-sition mechanism. Once deposited, the increase in theirconductivity with increasing water content can also in-crease their impact on circuit performance. Such hygro-scopic particles can easily represent as much as half ofthe total mass of fine particles in a building; most ofthese hygroscopic particles are ammonium sulfate salts(Sinclair et al., 1990). When the environment reaches acritical relative humidity (CRH), these salts deliquesceand become electrically conductive. Depending ontheir composition, fine particles tend to deliquesceabove 50–65% relative humidity (RH) (CRH is 81% forpure ammonium sulfate).

Contamination Effects on Electronic EquipmentIn the microelectronics manufacturing environment,sensitive electronic features on the integrated circuit(IC) chip are directly exposed to indoor air. De-posited airborne particles can substantially degradedevice yield by causing failures on exposed electroniccomponents (Cooper, 1986). Therefore, extremely lowairborne particle concentrations are maintained inmanufacturing environments (e.g., clean rooms). Fur-thermore, IC chips are encapsulated in the finalstages of the manufacturing process. Thus, whenelectronic components are operated in the workplaceenvironment, many of their critical and sensitive fea-tures are isolated from the ambient air. Howeversome features still remain directly exposed to ambi-ent air including the ‘‘legs’’ of surface mounted chips

48

(SMCs) that provide the electrical connection be-tween the SMCs and the electrical traces on the cir-cuit board. Particles can accumulate on a circuit overthe years and create bridges of agglomerated par-ticles between adjacent non-protected conductors.The reliability of contaminated electronics may notbe immediately affected when such accumulation oc-curs. For example, circuit boards contaminated withhygroscopic salts may function normally at low hu-midity (below the CRH) but fail during a high hu-midity episode (above the CRH) when the depositedhygroscopic particles permit leakage currents andelectrical shorts. Failures of circuit boards in the tele-communications industry due to hygroscopic dusthave been documented (Burnett et al., 1992; Franken-thal et al., 1993). A cost-benefit study suggests that,in many telephone switching offices, the costs ofmore efficient building filters and continuous fan op-eration are more than offset by the savings from re-duced electronic circuit failures (Weschler, 1991a).

Particle-induced degradation of electronic circuitsdepends on several factors including the layout of thecircuit, the characteristics of the particles, the locationand magnitude of deposition, and the degree of threatto circuit performance from reduced electrical isolationof adjacent conductors. Compared to the substantialbody of literature related to the prevention of depo-sition of fine particles on semiconductor surfaces dur-ing the manufacturing process, relatively few studiesof electronics circuit degradation from particle depo-sition in the workplace environment have been pub-lished.

Research is needed to better understand the mechan-isms involved in the deposition of indoor particles onelectronic components and circuit surfaces during op-eration. An improved understanding is also needed ofthe impacts of particle deposition on electronic circuitreliability. Such knowledge will assist in the develop-ment of analytical tools for predicting electronic failurerates as a function of various ventilation and particlefiltration scenarios and also facilitate the selection ofcost-effective HVAC systems capable of reducing elec-tronic failure rates induced by indoor particles.

ObjectivesThe primary objective of this work was to assess theimpact of hygroscopic particle deposition on the elec-trical functioning of exposed electronics. We aimed toexperimentally demonstrate that particle deposition onelectronic surfaces results in a decrease in electrical iso-lation and/or premature electronic equipment failures.Related objectives were to study the influence of elec-tric field strength on particle deposition and to quan-

Hygroscopic Fine Mode Particle Deposition on Electronic Circuits and Resulting Degradation of Circuit Performance

tify the influence of relative humidity on the electricalisolation of exposed electronicsHygroscopic Fine ModeParticle Deposition on Electronic Circuits and Result-ing Degradation of Circuit Performance.

MethodsOverviewIn an earlier experimental study, researchers developeda submicron particle exposure chamber where the rateof deposition throughout the chamber was acceleratedby maintaining a high velocity field within thechamber (Frankenthal et al., 1993). In contrast, we de-signed and built a sub-micron particle exposurechamber where deposition is accelerated by increasingthe particle concentration. The particle deposition rateon indoor surfaces is usually represented as a first or-der process such that the deposition rate scale linearlywith the particle concentration away from the walls, inthe ‘‘core’’ of the room (Nazaroff et al., 1993).

Experiments were conducted to study the depositionof fine-mode polydisperse particles on electronicsplaced in an exposure chamber. As a control experi-ment, we exposed identical electronic circuits to nor-mal ambient laboratory air in a second exposurechamber.

In the first experimental phase, electronic equipmentwas exposed for 281 h simultaneously in the two ex-posure chambers. The elevated particle concentrationin the experimental chamber was approximately 500times greater than the reference particle mass concen-

Fig. 1 Layout of surface trace conductors and surface mountedchip circuit. A differential DC voltage was applied to type II legsvia connectors 1 and 2

49

tration of 30 mg/m3 from a study of 38 commercialbuildings (Turk et al., 1989). Thus, particle depositionquantities in the first exposure chamber were expectedto be equivalent to a 16-year exposure to a normal in-door particle concentration. The control exposurechamber had a normal indoor particle concentration.

In the second experimental phase, the electronicequipment was visually and microscopically inspected.The performance of the televisions was assessed byvisual observation of the displays during operation fora range of relative humidities. The electrical isolationof adjacent legs of the SMCs was also measured overa range of relative humidities.

ElectronicsSeveral identical television sets (Radio Shack Portavi-sion, model number 16–121A) were purchased from alocal retailer. These were selected to represent typicalelectronic devices in office buildings such as computersand video display terminals that have integrated digi-tal electronic circuits as well as high voltage video cir-cuits with strong electrical fields. During the experi-ment, the television sets were connected to a video dis-play pattern generator (LEADER, model LCG-395A).

Standardized SMC circuits were specially manufac-tured for our research project and are shown in Figure1. They comprise a 1-cm square dummy surface-mounted chip whose 52 legs (spaced 0.65 mm center-to-center) are connected to the surface trace conductorson the circuit board. The surface trace conductors ex-tend over a 9 cm by 7 cm circuit board, and terminatein a 22-pin edge connector. Thirteen legs emerge fromeach of the four sides of the square dummy chip. Fromeach set of 13 legs, five consecutive legs were individu-ally connected to the edge connector via five individ-ual surface trace conductors. The remaining 32 legs ofthe chip were connected in a ‘‘parallel’’ configuration;

Fig. 2 Diagram of primary experimental apparatus

Litvak, Gadgil and Fisk

every alternate leg was connected to the same surfacetrace conductor, forming two sets of 16 legs in parallel.Legs were classified in two categories: type I, for theindividually connected legs and type II for the parallel-connected legs.

Exposure ChambersThe experimental system is illustrated diagrammati-cally in Figure 2 and the physical conditions duringexperiments are summarized in Table 1. Ten televisionsets and 20 SMC circuit boards were exposed to a highconcentration of particles in the experimental exposurechamber (EC1). Fourteen circuit boards were orientedvertically and six were horizontal (with the chips fa-cing up). Among the vertically-oriented circuit boards,type II legs on five were electrically grounded, legs onanother five were at a 10 VDC-differential voltage (lowvoltage) and legs on the remaining four were at a 240

Table 1 Experimental parameters during the Phase 1 exposure test

EC1Temperature: TminΩ22.7æC

TmaxΩ28.7æCTaverageΩ25.7 æC (Standard deviation: 1.2æC)

Relative Humidity: RHminΩ25%RHMAXΩ51%

PHYSICAL PARAMETERSEC1Temperature: TminΩ22.7æC

TmaxΩ28.7æCTaverageΩ25.7 æC (Standard deviation: 1.2æC)

Relative Humidity: RHminΩ25 %RHmaxΩ51%RHaverageΩ42% (Standard deviation: 6.3%)

Aerosol concentration: 15,300 mg/cm3

Time of exposure: 281 hours

EC2Temperature: TminΩ22.2æC

TmaxΩ27.9æCTaverageΩ24.1æC (Standard deviation: 0.9æC)

Relative Humidity: RHminΩ10%RHmaxΩ48%RHaverageΩ23% (Standard deviation: 7%)

Time of exposure: 281 hours

Ambient laboratory airTemperature: TminΩ21.3æC

TmaxΩ25.9æCTaverageΩ23.2 æC (Standard deviation: 0.6æC)

Relative Humidity: RHminΩ10%RHmaxΩ53%RHaverageΩ26% (Standard deviation: 8%)

ELECTRONIC PARAMETERSEC1Circuits 1–4: VoltageΩ240 VDC OrientationΩverticalCircuits 5–9: VoltageΩ10 VDC OrientationΩverticalCircuits 10–14: VoltageΩGROUND OrientationΩverticalCircuits 15–20: VoltageΩGROUND OrientationΩhorizontalTelevision sets 1–10.

EC2Circuits A-B: VoltageΩGROUND OrientationΩverticalTelevision sets A-B-C-D.

50

VDC-differential voltage (high voltage). All legs on thehorizontally oriented circuit boards and all type I legs,regardless of board orientation, were electricallygrounded.

As a control, four television sets and two verticallyoriented SMC circuit boards with all legs groundedwere exposed to normal laboratory air in a second ex-posure chamber (EC2).

The dimensions of EC1 and EC2 were respectively0.43 m¿0.66 m¿1.24 m and 0.6 m¿0.44 m¿0.76 m.Both chambers had aluminum walls, floor and ceiling.One wall in EC1 was of plexiglass, allowing visual ob-servation of the television screen displays. EC2 had afront that was open to the laboratory air.

EC1 was sealed carefully and maintained slightlydepressurized to prevent the laboratory air and EC2from being contaminated by the generated particles.The relative humidity in EC1 was controlled by injec-

Hygroscopic Fine Mode Particle Deposition on Electronic Circuits and Resulting Degradation of Circuit Performance

tion of dry or humid clean air into the chamber. Hu-midified air was created by pumping the air throughbubblers.

Temperature and relative humidity sensors werelocated in EC2 and in the laboratory. In order toavoid contamination from the high concentration ofparticles, temperature and humidity sensors for EC1were located in a 3.8-liter aluminum box suppliedwith filtered air from EC1. The solid-state tempera-ture sensors (AD-590-KH, Newark Electronics) werecalibrated at multiple temperatures using a precisionmercury thermometer in a temperature-controlledwater bath. The relative humidity sensors (GeneralEastern Model RH-3) were calibrated at five referencerelative humidities created with saturated salt solu-tions. The estimated uncertainties in the temperatureand relative humidity measurements were ∫0.3æCand ∫3% RH, respectively.

In EC1, an internal mixing fan was installed adjacentto the location where the stream of concentrated par-ticles entered the chamber. The mixing fan forced airperpendicular to the incoming particle stream. A testwas performed, using a sulfur hexafluoride (SF6) tracergas, to check the mixing within EC1. SF6 was injectedinto the inlet air stream that normally contained a highconcentration of particles and SF6 concentrations weremeasured versus time at three locations inside EC1.The SF6 concentrations varied less than 1.8% from theaverage concentration.

In order to minimize the temperature increase inEC1 due to the heat dissipation from the televisionsets, three external fans directed laboratory air at ahigh velocity toward the exterior surfaces of EC1. Withthese fans operating, the temperature in EC1 was ap-proximately 2.5æC higher than the surrounding roomtemperature. The spatial variation in temperaturewithin EC1 was approximately 2æC.

Particle GenerationParticles were generated by atomizing an aqueoussolution of ammonium sulfate (10 g Lª1) with 1% ofuranin2 (sodium salt of fluorescein) using a constantoutput atomizer (TSI Model 3076) and compressed fil-tered dry air. The resulting particle stream was driedby passing it through two diffusion dryers mounted inseries. The particle stream passed through the 25.3-litersettling chamber where large particles were removedby settling. The particles were neutralized with a Kr-85 source, which brought the particle charge distri-

2The uranin was added to permit quantification of the amount ofparticle deposition on the circuits. These results will be providedin a subsequent paper.

51

bution to a Boltzmann equilibrium. Finally, the particlestream entered EC1 at 3 liters per minute, through a1.25 cm ID copper tubing.

Particle MeasurementsDue to the limitations of the particle instrumentation,the high concentration of particles in EC1 was not ana-lyzed directly. Instead, a volume of particle-laden airfrom EC1 was mixed with a larger volume of HEPA-filtered air in a dilution chamber. Samples from the di-lution chamber were then analyzed with the particleinstrumentation.

Particle size distributions in the range (0.1–2 mm)were measured in eight size bins with an optical par-ticle counter (OPC) [PMS Model LASAIR 1003]. A dif-ferential mobility analyzer (DMA) consisting in anelectrostatic classifier (TSI Model 3071) coupled to anultrafine condensation particle counter (TSI Model3025) measured the particle size distribution in therange (0.01–0.34 mm). The generated polydisperse par-ticle had a mass median diameter of 0.48 mm, with ageometric standard deviation of 1.8. The particle massconcentration in EC1 was found to be approximately15 mg/m3.

Electronics Reliability TestsAll SMC circuit surfaces were cleaned with a methanolbath and cotton swabs and dried with compressedHEPA-filtered air before the controlled exposure toparticles. For reference, the electrical isolation betweenthe legs of the clean SMCs was measured as a functionof relative humidity. Electrical isolation was then meas-ured after the controlled exposure of the SMCs to par-ticles. Before these post-exposure measurements ofelectrical isolation (phase 2), the exposed surface traceconductors on the circuit boards were cleaned onemore time, since the location of interest on circuitswere the legs of the chips and the gaps between theselegs. Thus, the measurements of electrical isolationwere not affected by particle deposition between andon the electrical traces of the circuit boards (in modernelectronics, these traces on circuit boards are normallyprotected from indoor air contaminants). Electrical iso-lation measurements took place with the SMCs locatedinside EC1 at a controlled humidity.

For the electrical isolation measurements, a 200 VDCdifferential voltage was applied between the 2 ter-minals connecting the adjacent legs of type II on 12exposed SMCs. All were vertically oriented circuits:four at high voltage (circuits 1 to 4), four at low voltage(circuits 5 to 8) and four grounded (circuits 10 to 13).A multimeter electrometer (Keithley Model 619) meas-ured the leakage current through the edge connector

Litvak, Gadgil and Fisk

and, therefore, between the adjacent legs of the chip.The equivalent resistance between the parallel legs wascalculated, as the quotient between the differentialvoltage and the leakage current. For every circuit,measurements were performed under different stablerelative humidities from 30% RH to 90% RH (measure-ments at lower RH preceded measurements at higherRH). For circuit 1, a series of measurements of EI wassubsequently made at approximately 60% RH, afterdeliquescence of ammonium sulfate aggregates on thelegs of the SMC.

The maximum electrical resistance detectable withthe multimeter and 200 volt power supply was ap-proximately 28.5 TW. To check multimeter perform-ance, repeated measurements were made with severalprecision resistors in the range of 108 to 1013 W. Thesesame measurements were performed using anothercalibrated pico-ammeter (Keithley Model 674). The dif-ferences in measurements of electrical resistance wereless than 3%. Measurement precision was also assessedusing a SMC circuit. In successive measurements be-tween type II legs of a clean SMC as a function of rela-tive humidity, the relative standard deviation of themeasured resistance was ∂20%.

Television set reliability was assessed by visual ob-

Fig. 3 Ammonium sulfate accumulation between adjacent legsof SMC circuits. Particle aggregates (white in color) depositedpreferentially between legs of SMCs exposed to a differentialvoltage

52

servations of screen displays while televisions receivedthe signal from the video display pattern generator.

Electronic circuit features were observed with an op-tical microscope (Micromaster EL). The microscopehad a total magnification of 40.

ResultsObservationsTelevision screen displays were periodically monitoredduring the 281-h exposure test. Relative humidity was25–52% during the test. Subsequently, television setswere exposed in EC1 for 20 h, without high concen-trations of particles, to a relative humidity varyingfrom 70 to 90% RH. No malfunctions were observedeven with relative humidity above ammonium sulfateCRH (81% RH).

Exposed television sets were inspected. Particle ag-gregates were observed both inside and outside the de-vices on various surfaces such as the screen displaysand interior electronics. Particle deposition was spa-tially very uneven. During the approximately 12-dayexperiment, dendritic accumulations of particles up toa few millimeters in length were observed to growfrom the televisions’ external surfaces. Particle aggre-gates formed several bridges (a few millimeters long)between the television sets and the rack supportingthem. Particle accumulations appeared to be influ-enced by the electrostatic field created by the tele-visions; for example, the aggregates were perpendicu-lar to the television walls when the televisions were on

Fig. 4 Particle deposition pattern between legs of SMCs exposedto high and low differential voltage before and after deli-quescence. (a) and (b) show the same legs of SMC Number 5.(c) and (d) show the same legs of SMC Number 1

Hygroscopic Fine Mode Particle Deposition on Electronic Circuits and Resulting Degradation of Circuit Performance

and directed downwards, when the televisions wereturned off.

The legs of SMCs were visually inspected. Thehighest levels of particle accumulation were observedbetween adjacent type II legs with a differential voltage(see Figure 3). Locations of high particle deposition in-cluded the sharp edges and especially at the sharpbends of the legs of the SMCs. However, particle ac-cumulation was also observed between legs that weregrounded (legs of type I).

Microscopic examinations revealed the presence ofagglomerations of ammonium sulfate particles that oc-casionally bridged adjacent legs of SMCs. Bridges ofparticles were spatially non-uniform. They were notexclusively located between adjacent legs exposed to ahigh differential voltage – we observed uneven particleaccumulation between legs of both types, even ongrounded legs. However, the accumulation of de-posited mass was dramatically enhanced on legs ex-posed to a high differential voltage. Deposition pat-

Fig. 5 Electrical isolation as a function of relative humidity, showing exponential relationship. On each plot, a unique data point stylerepresents a unique circuit. Dark data points are used for exposed circuits

53

terns between the high voltage legs were bulky,whereas bridges of deposited particles were thinner forcircuits with conductors at low voltage or ground.Moreover, we could observe the presence of fiberswithin some of the finer agglomerations of particles.Fibers could not be observed within heavier accumu-lations of particles, possibly because the deposited par-ticles obscured the fibers. In the absence of fibers, fineragglomerations of particles were preferentially locatedat the sharp bends of the legs including bridges be-tween adjacent bends in legs.

Figure 4 shows particle deposition patterns betweenlegs of chips at various relative humidities. It wasclearly observed that for higher relative humidities(above 81% RH), ammonium sulfate particle aggre-gates between adjacent legs of chips disappeared afterdeliquescence, whereas non-hygroscopic fibers of un-known composition persisted.

The surfaces of the circuit boards were also inspec-ted and particle accumulation was observed at two dis-

Litvak, Gadgil and Fisk

Fig. 6 Range of electrical isolation at approximately 60% RH onclean and exposed circuits for the three differential voltages

tinctive recurrent locations on the circuit boards with ahigh differential voltage: a) between the pair of surfacetraces (on the circuit board) exposed to a high differen-tial voltage; and b) on the thin edges of the circuitboard, in the vicinity of the conducting traces exposedto high voltage.

Electrical Isolation MeasurementsEI between adjacent legs of SMCs was measured beforeand after the circuits’ exposure to the highly concen-trated particle mixture. Similarly, EI was measured be-fore and after the control circuits were exposed to nor-mal laboratory air in EC2. The circuits are respectivelyreferred to as ‘‘clean’’, ‘‘exposed’’ and ‘‘control’’. Asshown on Figure 5, the EI between adjacent legs oftype II showed a large (approximately exponential) de-crease with increasing relative humidity. The EI forSMC circuits exposed to a high differential voltage de-creased by approximately 3 orders of magnitude (fromapproximately 1012 W to 109W) when RH increasedfrom 35 to 65% RH.

The EI for circuits 1 to 4 (vertically oriented withhigh voltage) decreased by several orders of magni-tude after exposure to the high concentration of par-ticles. For these circuits, at 60% RH, the ratio be-tween EI on exposed circuits to EI on clean circuitsvaried from 1:600 to 1:5,000. On the other hand, forcircuits 5 to 8 (vertically oriented, low voltage) and11 to 13 (vertically oriented, grounded), this ratiowas smaller than 1:20, at 60% RH. (Circuit 10 had anelevated ratio of 1:1,000 due to the high EI of theclean circuits.) For the control circuits, the post-ex-posure values of EI were essentially unchanged fromthe pre-exposure values.

Figure 6 shows the range of the measured EI onclean and exposed circuits for the three differentialvoltages at a relative humidity of 60% RH. Low voltage

54

and grounded circuits exposed to high concentrationsof particles have EIs in the same order of magnitudecompared to clean circuits.3 Circuits exposed to a highdifferential voltage have EIs that are about a factor of1000 lower (from TW to GW) after particle exposure.

The influence of deliquescence of the deposited par-ticles on EI was studied. EI, at approximately 60% RH,between adjacent legs of SMC circuit 1 were measuredtwice – once before deliquescence of deposited am-monium sulfate and once after obtaining deliquescenceby exposure to RH above 80%. EI was 20% lower afterdeliquescence, which is not a significant change giventhe measurement precision.

At 90% RH, EI on circuit 3 was initially approximately3.2 MW. However, the EI on circuit 3 was not stable at90% RH and dramatically decreased with time. Al-though EI continued decreasing, we stopped the electri-cal measurements at 8.93 kW to prevent equipment dam-age. Circuits 1 and 2 did not exhibit this instability at90% RH. Subsequent microscopic observations of thesurface of SMC circuit 3 revealed evidence of burnslinking inter-trace conductors at the location where thelegs of the chip extend to the surface of the circuit board.

DiscussionTests with the television sets did not demonstrate thatexposure to high concentrations of particles causedpremature failures. The absence of television set fail-ures was surprising. Two potential explanations follow.First, deliquescence of deposited particles in the tele-vision sets during experiments with high humiditymay have been prevented by the elevated (but un-measured) air temperatures inside the television sets,relative to air temperatures surrounding the tele-visions. Due to the higher temperatures inside the tele-visions, the relative humidity inside televisions mayhave been lower than the critical relative humidity ofammonium sulfate. Second, the spacing between elec-tronic conductors in the television electronics waslarger than that in the most modern electronics.

Experimental results showed that the electrical iso-lation of the SMC circuits was, in many cases, substan-tially decreased after particles deposited on the cir-cuits. Also, electrical burns occurred on one SMC ex-posed to a high concentration of particles and to a highhumidity.

The particle accumulation on the SMC circuits andthe television sets was spatially very uneven. The loca-tions of deposited particles indicated very clearly that

3The EI range for clean grounded circuits on Figure 6 is affectedby the unusually high EI of Circuit 10 prior to particle exposure.

Hygroscopic Fine Mode Particle Deposition on Electronic Circuits and Resulting Degradation of Circuit Performance

electrostatic forces have a strong influence on particledeposition rates and locations.

The experiments indicate that particle accumulationon SMC circuits and episodes of high RH are two inde-pendent processes that can together cause considerabledrops of EI between adjacent conductors of SMC cir-cuits. EI decreased approximately exponentially withincreasing RH, in some cases from 1012W at 30% RHto 106W at 90% RH. These results corroborate previousfindings where EI of circuits exposed to various dustswas measured as a function of relative humidity (Bur-nett et al., 1992; Frankenthal et al., 1993). The EI rangesin electronics taken from field settings, measured at asimilar relative humidity, are comparable to our re-sults.

In our experiments, only one SMC circuit showedan extreme decrease of EI at high RH (circuit 3). Thisdecrease resulted in surface trace conductor deterio-ration on the circuit-board (where the legs of the SMCconnect to the surface trace conductors), whereas par-ticle accumulation occurred primarily between the legsof the chips, i.e., above the circuit-board. The previousresearch also found that the EI decreased dramaticallyas RH increased and that the decrease in EI was morepronounced as the RH reached the CRHs of dusts (Bur-nett et al., 1992).

The failure threshold for digital and analog circuitsas dust contamination becomes more conductive wasassessed previously (Burnett et al., 1992). Based on amodel, complete functional failure was predicted to oc-cur when the parasitic resistance between adjacentleads of conductors drops to approximately 1 MW.Based on this criterion, some of the EI decreases in ourstudy are significant enough to affect electronics re-liability.

Ammonium sulfate particles bridging between ad-jacent legs of the SMCs was frequently observed inconjunction with the presence of fibers. It appears thatthe fibers between adjacent legs of SMC circuits pro-vide a substrate for bridging by particles. In addition,we observed that particle accumulations bridging ad-jacent legs systematically disappeared after deli-quescence of the hygroscopic matter, whereas non-hy-groscopic fibers still persisted (as shown on Figure 4).Hence, by providing fine particles with a substrate fordeposition and a pathway between adjacent conduc-tors, fiber deposition appears to be a potentially im-portant factor for the reliability of electronic equip-ment.

In the light of these observations, we suggest thefollowing understanding of the contamination pro-cess. First, hygroscopic particles accumulated be-tween adjacent legs of the SMC to form a bridge,

55

sometimes depositing on fibers already present be-tween legs, under the influence of the electric field.Then, when the RH reached the CRH the depositedmatter deliquesced and flowed over the legs (due tosurface tension and/or gravity), occasionally reachingthe circuit-board surface. Finally, during high hu-midity episodes, deliquesced matter sometimesformed a new bridge between surface-trace adjacentconductors at their junction with the legs of theSMC. This theory is consistent with: (a) the low EImeasured on circuit 1 at 60% RH, after deli-quescence, even in the absence of visible bridge be-tween adjacent legs; and (b) the degradation of thecircuit board observed for circuit 3.

The source of the fibers observed on the SMC boardsis uncertain. Presumably some fibers deposited fromlaboratory air before the SMCs and televisions weresealed in the exposure chambers. Also, some fibersmay have been from the cotton swabs used to clean theSMC circuits before they were installed in the exposurechambers. Since fibers and large particles are presentin normal indoor air, fiber and large particle depositionon circuits would be a normal occurrence. The SMCshad a very limited period of exposure to normal air-borne fibers (a few days) but the exposed SMCs andtelevisions had the equivalent to ∂16 years of exposureto fine particles.

Efficient air filters have been used to reduce elec-tronics failure rates caused by indoor air pollution(Weschler, 1991a; Weschler and Shields, 1991b). Ourexperimental results point to another option – amodification of the geometry of the legs of the SMCcircuits. We found that particles tended to depositpreferentially at the sharp bends in the legs of theSMCs presumably because of the strong electric fieldcreated by these sharp bends. Smoothing or elimin-ating the bends in the legs, i.e., increasing their ra-dius of curvature, would reduce the particle depo-sition at these locations. Hence, smoothing or elimin-ating the sharp bends in the legs of SMCs could bean innovative method to reduce failure rates in elec-tronics.

Summary and ConclusionsWe have shown that particle deposition results in sig-nificant drops in EI between adjacent legs of chips ofelectronic components, although we did not prove thatdeposition of hygroscopic particles causes prematurefailure of a typical electronic device (television). More-over, we found that EI decreased by several orders ofmagnitude with increased relative humidity below theCRH of the particles deposited on the circuit board.

Litvak, Gadgil and Fisk

Fine particle accumulation on electronic circuit featureswas found to be spatially very uneven. Based on thelocations of heavy particle deposition, it is evident thatelectrostatic forces substantially enhance particle depo-sition at certain locations within circuits.

This research suggests a possible scenario for hygro-scopic fine mode particle contamination of electroniccomponents, that reveals the importance of fibers dur-ing the deposition mechanism. Hence, future experi-mental research would be more realistic if the air sur-rounding the electronic equipment contains both fineparticles and coarser particles and especially fibers.Lastly, this paper suggests that the failure rates of elec-tronic equipment may be decreased by smoothing oreliminating the sharp bends in the legs of surfacemounted chips.

AcknowledgementsThe authors would like to thank Al Kanzaki and VinceHoney of the LBNL electronics shop for extensive tech-nical advice and for design and fabrication of the sur-face mounted chip circuits. The valuable input fromPatrick Depecker of the Institut National des SciencesAppliques de Lyon and the technical reviews of a draftof this document by Charlie Weschler, Tracy Thatcher,and F. Remi Carrie are also greatly appreciated. Thiswork was supported by the Assistant Secretary of En-ergy Efficiency and Renewable Energy, Office of Build-ing Technology, State, and Community Programs, Of-

56

fice of Building Systems of the U.S. Department of En-ergy (DOE) under contract No. DE-AC03–76SF00098and by the National School of Public Works (ENTPE),Lyon, France.

ReferencesBurnett, W.H. Sandroff, F.S. and D’Egidio, S.M. (1992) ‘‘Cir-

cuit failure due to fine mode particulate air pollution’’. In:Proceedings of the 18th International Symposium for Testing andFailure Analysis, Los Angeles, CA, pp. 329–333.

Cooper, D.W. (1986) ‘‘Particulate contamination and microe-lectronics manufacturing: an introduction’’, Aerosol Scienceand Technology, 5, 287–299.

Frankenthal, R.P., Siconolfi, D.J. and Sinclair, J.D. (1993) ‘‘Ac-clerated life testing of electronic devices by atmosphericparticles: why and how’’, Journal of the ElectrochemicalSociety, 140, 3129–3134.

Sinclair, J.D. Psota-Kelty, L.A. Weschler, C.J. and Shields, H.C.(1990) ‘‘Measurement and modeling of airborne concen-trations and indoor surface accumulation rates of ionicsubstances at Neenah, Wisconsin’’, Atmospheric Environ-ment, 24A, 627–638.

Turk, B.H., Grimsrud, D.T., Brown, J.T., Geisling-Sobotka,K.L., Harrison, J. and Prill, R.J. (1989) ‘‘Commercial build-ings ventilation rates and particle concentrations’’, ASH-RAE Transactions, 95, 422–433.

Nazaroff, W.W. Gadgil, A.J. and Weschler, C.J. (1993) ‘‘Cri-tique of the use of deposition velocity in modeling indoorair quality’’. In: Nagada, N.L. (ed) Modeling of Indoor AirQuality and Exposure, American Society for Testing and Ma-terials, ST 1205. Philadelphia: ASTM, 1993. pp. 81–104.

Weschler, C.J. (1991a) ‘‘Predictions of benefits and costs de-rived from improving indoor air quality in telephoneswitching offices’’, Indoor Air, 1, 65–78.

Weschler, C.J. and Shields, H. C. (1991b) ‘‘The impact of venti-lation and indoor air quality on electronic equipment’’,ASHRAE Transactions, 97, 455–463.