Cleansing and Reliability of Smoke-Contaminated Electronics

Lennart Cider, Ph.D. Cider, Institute of Production Engineering Research, Gilteborg, Sweden.

Abstract

This work deals with problems that arise when modern surface-mounted electronics are to be reconditioned after smoke contamination.  In a fire, hydrogen chloride, which is formed when polyvinylchloride is present, is deposited on various materials.  Electronic equipment is especially sensitive since malfunction may occur after a longer or shorter time of operation due to the chloride contamination.

Earlier work has shown that through-hole electronics can be reconditioned, with good results, after deposition of up to 100µg chloride/cm² in the surrounding area.  The lower limit when cleaning is needed is often specified to 10 µg chloride/cm².

In this work, therefore, surface-mounted electronics have been exposed to smoke containing hydrogen chloride, which has contaminated the test boards with 45 to 75 µg chloride/cm².  Three different methods for cleaning smoke-contaminated electronics were investigated: manual, automatic spray, and ultrasonic.  Each method was able to clean to a contamination level lower than 1.5 µg sodium chloride equivalents/cm³.  The automatic spray method could not remove all contaminants beneath the components.  All three methods were capable of improving the surface insulation resistance to a satisfactory level.

Conformed coating can, to a large extent, protect the electronics against corrosive smoke.  However, decontamination of boards conformally coated acrylic coatings may be complicated since smoke products are partly absorbed into the conformal coating.  The large difference between chloride contamination in a fire and during the manufacture of electronic equipment is the nature of deposition.  Experiments have shown that the hydrogen chloride reacts with lead in the solder to form lead chloride.  During manufacture, chloride salts are deposited all over the test board.

Migration between conductors of different potential has not occurred.  This is due to the fact that chloride is localized on the conductors as lead chloride.  Galvanic corrosion, on the other hand, has occurred between metals within one conductor.

Introduction

Smoke-contaminated electronics are cleaned to restore the function of the equipment.  Hitherto, electronics that were cleaned were produced using the through-hole technology.  Modern electronics are predominantly surface-mounted electronics, which involve a fiber pitch and smaller cavities.  The possibility of getting undesired migration between two conductors is increased.  Hence, the need for cleaning is increased.

Electrochemical migration means that a conductor with a positive potential dissolves into positively charged metal ions, which then move in the electrical field towards the negative conductor.  The ions again turn into metal at the negative conductor, forming long, tree-shaped dendrites, which will eventually short-circuit the electronics.  Three factors are necessary for electrochemical migration:  conductors of different potential, humidity, and contamination.

A U.S. military specification, MIL-STD-2000A, for measuring cleanliness on electronics specifies that the contamination in sodium chloride equivalents (e.g., NaC1) should be lower than 1.5 µg/cm² on electronics to be conformally coated.  An equivalent amount of NaC1 means, in this case, an amount of ionic contamination having the same conductivity as NaC1 in an isopropanol/water mixture (ratio 75/25 by volume).

Cleaning smoke contaminated electronics is not currently considered until there are 10 µg/cm² of chlorides in the surroundings.  It has also been shown that through-hole electronics can be reconditioned, with good results, after deposition of up to 100 µg/cm² of chlorides in the surrounding area.  This is in accordance with research done at BELLCORE, where 600 µg/in. (600 micrograms per square inch = 93 µg/cm²) of chlorides are reported as the upper limit for reconditioning.

This work deals with the problems that arise when restoring smoke-contaminated surface-mounted electronics.  The problems are determining the state of contamination, the most effective method of removing it, and a way of ascertaining that it has been removed.

This work is a part of a series of investigations.  Earlier work reviews cleaning methods, reliability measurements, and smoke contamination.

Experimental Design and Procedure

The test board (see Figure 1) is an FR-4 laminate assembled with one 68-terminal, 50 mil pitch, J-leaded plastic chip carrier (PLCC68) on Segment 4; one leadless ceramic chip carrier (LCCC68) on Segment 3; two small outline components with 28 terminals (S028) on Segment 6; and 12 Type 2220 (CC2220) chip capacitors on Segment 5.  The test board is also equipped with comb patterns (Segment 1 and Segment 3 through 6) for measuring the surface insulation resistance (SIR), with conductor width and distance of 250 µm.

Some of the components chosen were shown in earlier experimental work to be very (PLCC68) and extremely (LCCC68 and CC2220) difficult to clean underneath due to their size and low stand-off height.  These different types of components were therefore used to evaluate the effect on the cleaning method.

The treatment and analysis of the test boards are indicated in Table 1. The right-hand side of the table shows the analysis of each board (1 to 100), and the left-hand side shows the treatment.

Table 1

Experimental Design

Table 2

Location of Each Test Board and Steel Board During Smoke Deposition

            All of the boards and components were cleaned in ultrasonic equipment in IPA/water 75/25 at 40ºC or 1 hour.  The level of cleanliness as determined by ionic contamination testing using Protonique MCM-1 unit.  All of the boards and components had a contamination under the detection limit, 0.001 µg NaC1/cm².  The test boards were screen-printed with a Rosin Mildly Activated paste (Multicore, SN62RM10BAS86), assembled, and soldered in an ERSA ERS450 1R reflow soldering equipment.

            All of the mounted boards, except 12 according to the experimental design, were first pre-cleaned in terpene (Citrikleen XPC) for 5 minutes, then in IPA/water 75/25 for 30 minutes at 40ºC in ultrasonic equipment.  This was done to establish a low and well-defined level of cleanliness before the boards were contaminated with smoke.

            Six test boards were dried at 105°C for 1 hour, then dip-coated with an acrylic conformed coating (Humiseal 1B31).  This acrylic conformal coating cures in air and is one of the most common on the Swedish market.

Table 3

Composition of the Fire Used for the Smoke Deposition

Smoke Contamination

Smoke was deposited on the test boards in a house used for controlled fires.  This house contains two rooms, called the fire room and the deposit room, which are connected by a passage.  The location of each test board was randomized.  Eight steel boards labeled A through H, were placed in order to cover the corners and interior of the deposition area.  The test boards and the steel boards were placed in a horizontal position on a rack about 5 cm above the floor.  The internal location is shown in Table 2.

The smoke was generated by a Fire burning wood, PVC cable, cellular plastic and FR-4 boards.  The composition is given in Table 3.  The fire was started by electrically heating a wire kept in alcohol.

A container with 10 liters of water was placed immediately over the fire to limit the temperature rise and to produce a humid smoke.  This composition resulted in a final chloride content on the eight steel boards in the deposit room of 45 to 75 µg/cm².  The temperature in the deposit room was monitored with a PC system during the 4 hours of exposure, as seen in Figure 2a.  A thermocouple, fitted under one of the components on one test board, registered a maximum temperature of 50°C after about 15 minutes as seen in Figure 2b.

A number of test boards were placed in a humid atmosphere immediately following the smoke deposition.  A total of 53 test boards were kept for 20 hours at 90% rh, 30°C.  These test boards were put in a dry atmosphere (<30% rh) after this treatment.  The remaining test boards were placed in a dry atmosphere immediately after the smoke deposition.

 Cleaning

A total of 66 test boards were reconditioned using three different methods after the smoke deposition.

 Method 1 is used by the companies that participated in this study.  A 5% water solution of a commercial alkaline cleaning agent Euroclean F-42 was used. 

 Table 4

Chloride Content on Metal Surfaces on Test Boards (µg/cm²) 5, 12, 19, and 26 According to the Bresle (Segments 1 and 2) and Relectronic (Segment 6)

There are five steps in this procedure.  First, the boards were sprayed with cleaning agents.  Then they were manually brushed and rinsed with tap water and deionized water.  Next, they were dried with high-pressure air.  Finally, they were dried in vacuum 40°C.  This procedure took 3 minutes, plus 1 hour for drying.

In Method 2, the test boards were cleaned in an automatic defluxing machine Contax, which was designed for post solder cleaning.  The unit uses immersed big pressure spraying.  The cleaning agent used in this study was a mixture of 75 isopropanol and 25% water.  The three-step procedure includes high-pressure spraying at 50°C for 20 minutes, rinsing with isopropanol, and drying with warm air.  The procedure took 30 minutes, including the drying.

Method 3 is an extension of Method 1.  An ultrasonic cleaning machine (40 kH 275 W, 15 liters) was used.  The cleaning agent was Euroclean F-42, the same used in Method 1.  This method involved six steps.  First, the boards were rinsed with tap water.  They were then dipped in cleaning agent and manually brushed.  Next, they were leached in cleaning solution for 10 minutes, followed by 10 minutes of ultrasonic agitation.  The boards were then rinsed with tap water and deionized water and dried with pressurized nitrogen.  Finally, they were dried at 80°C for 1 hour.  This procedure took 25 minutes, plus 1 hour for drying.

 Table 5?

Chloride Content on Metal Surfaces on Test Boards (µg/cm²) 5, 12, 19, and 26 According to Bresle (Segments 1 and 2) and Relectronics (Segment 6)

Eighteen test boards were dried at 105°C for 1 hour, and Humiseal 1B31, a conformal coating, was applied using dip coating.  This was done to investigate the effect of conformal coating on the reliability and cleanability of a smoke-contaminated board.

Results

The level of chloride contamination was investigated with two different methods, both of which are convenient for mobile conditions.  The Bresle method employs quantitative chemical titration, and the Relectronic Chloride Quick Test is a qualitative test strip showing a rough estimation in the interval 0 to 20 (µg/cm²) chlorides.

The chloride contamination was measured on the steel boards after the smoke contamination using both methods.  The Bresle test was repeated one day later on the same steel boards. The results showed an almost even distribution of chlorides.  The average is 50 (µg/cm²) chlorides, and the maximum is 75 (µg/cm²) (on Board B), as seen in Table 4. The results do not change from Day One to day Two.  The Relectronic test only shows "cleaning necessary" (>20 µg/cm² chlorides).

The chloride content was also estimated on four of the test boards.  Two of the boards, Boards 5 and 12, were reference boards that were not exposed to smoke, and two, Boards 19 and 26, were smoke-contaminated.  The Relectronic test showed no chlorides for the reference boards and more than 20 (µg/cm²) for the smoke-contaminated boards.  The Bresle method showed an incorrect level of more than 20 (µg/cm²) chlorides for the reference boards and up to 91 (µg/cm²) for the smoke-contaminated boards.  The Bresle method showed an incorrect level of more than 20 (µg/cm²) chlorides for the reference boards and up to 91 (µg/cm²) for the smoke-contaminated boards.  These results are summarized in Table 5.

 Laboratory Analyses

The chemical analysis was performed by ABB Corporate Research. The compounds were leached from the test boards with 50°C deionized water for 1 hour.  The following compounds were analyzed:

Table 6

A Selection of Chemical Compounds Analyzed on Test Boards

All values are given

• Cations:  Ag, A1, Ca, Co, Cr, Cu, Fe, K, Mg, Mo, Na, Ni, Pb, Sb, Si, Sn, Zn
• Anions: C1, Br, F, NO, SO₄², succinic acid, adipic acid, acetic acid, formic acid
• Total amount of organic carbon (OC)
• Equivalent amount of NaC

All of the anions, including chloride, are analyzed by means of ion chromatography.  May of the compounds were present in very small amounts.  Data for the most significant compounds are shown in Table 6.  Large amounts of lead were found on smoke-contaminated test Boards 15 and 21; the amounts of silicon and sulfate are higher after cleaning with Method 1 than they were before; and Method 3 shows a very low contamination level overall, whereas Method 2 has higher values.

The extraction measurements were performed in a Protonique MCM-1.  The extraction method has been evaluated by Bergendahl and Dunn.  The test board were leached in a standardized manner according to MILSTD-2000A with isopropanol water (75%/25%).  The conductivity is automatically converted into an equivalent amount of NaC1.  All of the cleaned test boards were below the military specifications (1.5 µg/cm²), as seen Table 7.  The smoke-contaminated boards showed large values.

 Table 7

Extraction Measurements on Test Boards in a Protonique MCM-1

The test boards were visually inspected with a WILD 420 microscope and a JEOL JSM-840A scanning electron microscope.  This inspection was done to find differences between cleaning methods and differences between the reference, smoke-contaminated, and cleaned boards

An example of typical smoke damage is shown in Photograph 1a.  Soot particles are found everywhere, and the solder joints have been discolored.  The component has been removed in Photograph 1b.  The conductors under the component changed from bright metal to dull grey due to the hydrogen chloride in the smoke.  One can also see that the smoke penetrated between 1.5 to 2 mm below the component.  This component, SO-28, has a relatively stand-off height of 200 µ.m.

The next sequence of photographs shows the solder mask between two solder joints.  These photographs, taken with the scanning electron microscope, are greatly enlarged.  Photograph 2a shows the reference board without soot, while Photograph 2b shows a smoke-contaminated board with soot deposits, which are seen as irregular white particles.  Cleaning Methods 1 and 3 removed all of the soot, as seen in Photographs 2c and 2e.  Method 2 only reduced the amount of soot particles, as seen in Photograph 2d.

    The conductor surfaces approximately 2 mm from the edge of the component as shown in photographs of the five test boards investigated.  Photograph 3a shows the eutectic structure of the solder; the darker area is tin-rich, and the lighter area is lead rich.  An area that reached with hydrogen chloride from the smoke is shown in Photograph 3b.  Cleaning Methods 1 and 3 removed almost all of the chloride product on the conductors, as seen Photographs 3c and 3e.  Method 2 only reduced the amount of these unwanted products, as seen in Photograph 3d.

    The FR-4 board material under the components was not contaminated with chloride products.  Using energy dispersive spectrometry, the chloride products on the conductors were found to be lead chloride.

    Visual inspection revealed that test boards left in humid conditions for a longer time had more corrosion products.  Conformally coated test boards withstood the hydrogen chloride, but soot particles were found partly buried in the coating.

    The test boards were tested for 4 weeks according to IPC-TM-650, Test Method Standard 2.6.3.3, Class 3, (85°C, 85%rh, polarization voltage 50 VDC and measurement voltage 100 VDC with reversed polarity).  Surface insulation resistance (SIR was monitored during the 4 weeks of testing for all of the comb patterns.  The purpose was to investigate the reliability of the electronics.

It was found that smoke-contaminated test boards had very low SIR values (<1 MΩ), Smoke-contaminated test boards with conformal coating (CC-1) had SIR values in the range 200 to 1200 MΩ and even higher (800 to 2300 MΩ) when a conformal coating (CC-II) as applied after the cleaning.  The undamaged reference boards had values from 1600 to 2800M, and the conformally coated reference boards (CC-1) had the highest values at 4000 MΩ.  An SIR value higher than 100 MΩ is good for most applications and necessary for high-reliability applications.  The influence of an extra 20 hours in a humid atmosphere after the smoke contamination was not evident in this SIR measurement.

Discussion

The Bresle analysis is a fast and relatively easy way of attaining a quantitative value of chloride content after a fire.    It is designed for, and works properly on, iron and steel surfaces.  The results of this experiment show that it does not work correctly on tin and lead.  The other chloride tester, Relectronic Chloride Quick Test, can give reliable results on these metals, but it is merely a qualitative method.

Analysis of the chloride compounds and location showed that lead chloride is formed on the conductors only.  The amount of chloride on other surfaces of the test board is under the detection limit.  Smoke penetrates into small cavities under surface-mounted components, and lead chloride is formed.  The penetrating depth was up to 2 mm for components with a stand-off height of 200 µm.

Electrochemical migration did not occur in this experiment since the chloride compounds were localized on the conductors as rather insoluble lead chloride.  Three conditions achieve detrimental migration:  a difference in voltage, moisture, and ionic contamination between conductors.  The third condition was not fulfilled.

Cleaning smoke-contaminated electronics with Method 1, the method currently used, produces very good final results.  The chemical analysis shows a somewhat increased amount of silicon and sulphate, compounds that come from the alkaline cleaning agent and can easily be removed by rinsing the boards more thoroughly in water.  Extraction measurements, visual inspection, and SIR measurements give a joint approval for the method.

Method 2, the automatic high-pressure spray unit, cannot remove all of the lead chloride from the conductors.  Extraction measurements give the method a pass, but visual inspection shows small spot particles and lead chloride.  However, the SIR measurements are not affected and are as good as those of the other methods.

Method 3, the method tested last with ultrasonics, shows very low final residues. The chemical analysis shows contamination levels that are almost as low as they are for a reference test board.  Extraction measurements indicate contamination levels that are well below limits for new electronic equipment.  Visual inspection and SIR measurements approve this method.

The cleaning agent is important, and the choice of isopropanol mixed with water in Method 2 may not have been the best one.  The corrosion is more severe on metal surfaces that have been left in a humid atmosphere for an additional 20 hours, and the amount of chemical contamination is slightly higher.  The other measurements do not show that this treatment affected the test boards.  This may be due to the conditions used in this investigation:  The humid atmosphere was produced in a room with extremely clean but highly humid air.  Normally, the deposition of contamination proceeds until the equipment is removed from the danger zone.  Valuable equipment must always be rescued as quickly as possible.

The findings in this experiment indicate that it may be possible to differentiate between chloride contamination from various sources such as fire, salt deposits during operation in coastal areas, or contamination from assembly.  This has been shown in a separate paper, in which the role of the relation between halide and alkali concentrations on electrochemical migration was also scrutinized.

Conclusions

All of the tested methods work when renovating smoke-contaminated electronics.  The current method, used by the renovation companies that participated in this experiment, does a very good job.  Small amounts of residue from the alkaline cleaning agent were found on the test boards, but this can be easily solved by improving the final water rinse.  The final result is remarkably good.

The automatic spray cleaning gave the worst results in this investigation.  The visual inspection reveals remaining soot particles and lead chloride. 

Using an ultrasonic cleaner can improve the final result, compared to the current method, mainly by reducing the amount of chemical contamination left on the test boards.  The Bresle method of analyzing chloride works properly on steel surfaces, but it shows incorrect results on tin/lead surfaces, which are often used in electronics.  The Relectronic test for chlorides works on both steel and tin/lead, but it has a restricted measurement interval.

Conformal coating provides good protection against corrosive smoke.  However, renovating smoke-contaminated conformally coated electronics may be difficult because soot particles partly stick in the coating, at least in acrylic coatings.

Electronic left in a humid atmosphere showed more severe corrosion damage than those that are not left in a humid atmosphere.  The surface insulation resistance was not affected and the amount of chemical contamination was only slightly affected by this prolonged treatment in a humid atmosphere.

Acknowledgements

The financial support of Brandforsk, Elsan, Larmijarist, Polygon, Sankon, Skeab, and Nutek is gratefully acknowledged.

References

1.         Olesen, S.T., "Effects of Corrosive Smoke on Electronics, "Electronik Centralen Denmark, Report ECR-145, 1984.

2.         Reagor, B. T., "Smoke Corrosivity:  Generation, Impact, Detection, and Protection," Journal of Fire Sciences, Vol. 10, No. 2 (1992), pp. 169-179.

3.            Granath, G., "Rengöring av elektronik vid till verkning och sanering," ("Cleaning of Electronics During Production and Reconditioning"), IVF-rapport GE9001, 1990.

4.            Tegehall, P-E., "The Impact of Crevices Beneath Surface-Mounted Devices on the Cleaning Efficiency," Soldering & Surface Mount Tech., No. 8 (1991), pp. 46-52.

5.         Brox, B., "Värdering av ny tillgänglig teknik: praktiskn forsook med testkort," ("Evaluation of New Available Technology: Experimental Work with Test Boards").  Technique for Clean Electronics (TRE), Report B3, IVF Göteborg, Sweden, 1990.

6.            Andersson, L-H, and Jungel, M., "Värdering av nagra metoder för bestämming av ytklorider pa brandplaster," ("Evaluation of Some Methods for the Determination of Surface Chlorides at Fire Sites"), FOA Sweden, Report C 20733-2.4, Sept. 1988.

7.            Bergendahl, C.G., and Dunn, B.D., "Evaluation of Test Equipments for the Detection of Contamination on Electronic Circuits," ESA STM-234, 1984.

8.         Dunn, B.D., and Bergendahl, C.G., "An Evaluation of the Solvent Extraction Method for the Detection of Ionic Contamination on Substrates Supporting Large Surface-Mounted Devices," Brazing & Soldering, No. 13 (Autumn 1987), pp. 20-32.  Concentrations and the Electrochemical Migration on Printed Circuit Boards," presented at the Electrochemical Society Conference on Corrosion and Reliability of Electronics Materials and Devices, Toronto, Canada, October 1992.