References

2-1. J. M. Crowley, Fundamentals ofApplied Electrostatics, Krieger/Wiley, 1991. 2-2. J. A. Cross, Electrostatics: Principles, Problems and Applications, AlP/Hilger, 1987. 2-3. E. WhitWker, A History ofthe Theories ofAether and Electricity, Volumes I and n, Dover, 1990. 2-4. W. A. Harrison, Electronic Structure and the Properties of Solids, Dover, 1989. 2-5. (ESD) Protection Test Handbook, Second Edition, KeyTek Instrument Corporation, 1986, page 7. 2-6. Mll..-HDBK-263A, paragraph 3.24, Table mof Appendix A. 2-7. Understanding Solid-State Electronics, Second Edition, Radio Shack (Texas Instruments, Inc.), 1972, page 5. 2-8. Electrostatic Discharge Training Manual, NAVSEA SE 003-AA-TRN-OIO, Naval Sea Sys• tems Command, 1980, pages 31-41. 2-9. S. Wong, "A Piezoelectric Crystal Electrostatic Discharge Generator," Conference Proceedings of the EMCIESD International, Anaheim, CA, April 12-15, 1994. 2-10. N. Jonassen, "Do Gases Charge?" Proceedings of the 16th Annual EOSIESD Symposium, Las Vegas, NV, September 27-29,1994. 3-1. 0.1. McAteer, Electrostatic Discharge Control, McGraw-Hill, 1989. 3-2. DoD-HDBK-263, ESD Control Handbook for Protection of Electrical and Electronic Parts, Assemblies. and Equipment (Excluding Electrically Initiated Explosive Devices), Table IV, pages 9-11. 3-3. J. M. Kolyer, R. Rushworth, and W. E. Anderson, "ESD Control in an Automated Process," EOS/ESD Symposium Proceedings, EOS-9, 1987, pages 41-50. 3-4. J. M. Kolyer, D. E. Watson, and W. E. Anderson, "Controlling Voltage on Personnel," EOS/ESD Symposium Proceedings, EOS-ll, 1989, pages 23-31. 3-5. 1. M. Kolyer and D. E. Watson, ESD from A to Z: Electrostatic Discharge Control for Electron- ics, Van Nostrand Reinhold, 1990, page 40. 3-6. Ibid., pages 11, 160, and 161. 3-7. Ibid., pages 193-201. 3-8. T. Dangelmayer, ESD Program Management, Van Nostrand Reinhold, 1990. 3-9. Kolyer and Watson, op. cit., pages 3-27, 39-56. 4-1. Comments by J. M. Kolyer in "ShOUld Testing Be Done Beyond the Lab?" edited by Donald Ford, EOSIESD Technology Magazine, August/September 1987, pages 15 and 22. 4-2. Electrostatic Discharge (ESD) Protection Test Handbook, KeyTek Instrument Corp., 1983, pages 10, 12, 13, 16, and 18. 4-3. W. Simmons and P. Adamosky, "Meeting the New 883 Test Method of IC Static Discharge Testing," Evaluation Engineering Magazine, July 1988, page 82. 4-4. D. E. Frank, "ESD Considerations for Electronic Manufacturing," presented to American Society of Manufacturing Engineers Westec Conference, Los Angeles, CA, March 21-24, 1983 (Douglas 7324). 4-5. R. Moss, "Exploding the Humidity Half-Truth and Other Dangerous Myths," EOSIESD Technology Magazine, April 1987, page 10.

163 164 ESD FROM A TO Z: ELECTROSTATIC DISCHARGE CONTROL FOR ELECTRONICS

4-6. P. S. Neelakantaswamy and R. I. Turkrnan, "ESD Failures of Board-Mounted Devices," Elec• tronic Packaging and Production Magazine, February 1987, page 132. 4-7. D. M. Yenni and 1. R. Huntsman, "The Deficiencies in Military Specification M1L-B-8170S: Considerations and a Simple Model for Static Protection," presented at the Reliability Analysis Center EOS/ESD Symposium, Denver, CO, 1979. 4-8. J. O. Lonborg, "Static Survey Meters," EOSIESD Symposium Proceedings, EOS-S, 1983, page 63. 4-9. S. S. Sullivan and D. D. Underwood, "The Automobile Environment: Its Effects on the Human Body ESD Model," EOSIESD Symposium Proceedings, EOS-7, 1985, page 103. 4-10. D. L. Lin, M. S. Strauss, and T. L. Welsher, "Big Problem Uncovered: Zapper Data Differs," EOSI ESD Technology Magazine, August/September 1987, page 9. 4-11. R. E. McAteer, G. H. Lucas, and A. McDonald, "A Pragmatic Approach to ESD Problem Solv• ing in the Manufacturing Environment, a History," EOSIESD Symposium Proceedings, EOS-3, 1981, page 34. 4-12. "Nuclear Air Ionizers Recalled by NRC," Evaluation Engineering Magazine, March 1988, page 65. 4-13. W. R. Van Pelt, "Polonium-21O Contamination," Chemical and Engineering News, April 11, 1988, page 4. 4-14. N.10nassen, "The Physics of Electrostatics," distributed at the Sixth Annual EOS/ESD Sympo• sium, Philadelphia, PA, 1984. 4-IS. G. Baumgartner, "Electrostatic Measurement for Process Control," EOSIESD Symposium Pro• ceedings, EOS-6, 1984, page 2S. 4-16. "Latent ESD Failures: a Reality," Evaluation Engineering Magazine, April 1982, page 80. 4-17. G. T. Dangelmayer, "ESO-How Often Does It Happen?" EOSIESD Symposium Proceedings, EOS-S, 1983, page I. 4-18. B. Rodgers and W. Tan, "Attacking ESD Where It Lives," Circuits Assembly Magazine, 1une I99S, page 40. 4-19. B. N. Stevens, "Determining the Surface Resistivity ofESD Protective Cellular Packaging Ma• terials," EOSIESD Symposium Proceedings, EOS-8, 1986, page 136. 5-1. T. S. Speakman, "A Model for the Failure of Bipolar Silicon Integrated Circuits Subjected to Electrostatic Discharge," International Reliability Physics Symposium Proceedings, 1974. 5-2. O. J. McAteer, Electrostatic Discharge Control, McGraw-Hill, 1989, page 173. S-3. D. C. Anderson, "New Approach to Handling Charged Devices Without Causing ESD Dam• age," Technical Record of the Expo '92 International Conference on Electromagnetic Compat• ibility, sponsored by EMC Technology Magazine, Reston, VA, May 18-22, 1992, page 174. 6-1. GIDEP Alert H6-A-83-02, "Materials, Plastic, Antistatic," December 27, 1983. (Antistats from contaminated NASA instrument mirrors.) 6-2. G. C. Holmes, P. J. Hubb, and R. L.10hnson, "An Experimental Study of the ESD Screening Effectiveness of Antistatic Bags," EOSIESD Symposium Proceedings, EOS-6, 1984, page 78. 6-3. S. A. Halperin, "Selecting the Proper Protective : Part 11," EOSIESD Technology Magazine, October/November 1988, page IS. 6-4. MIL-HDBK-773, "Electrostatic Discharge Protective Packaging," 1 April 1988. 7-1. N.l. Safeer and 1. R. Mileham, "A Material Evaluation Program for Decorative Static Control Table Top Laminates," EOSIESD Symposium Proceedings, EOS-6, 1984, page 85. 7-2. 1. R. Huntsman and D. M. Yenni, "Charge Drainage vs. Voltage Suppression by Static Control Table Tops," Evaluation Engineering Magazine, March 1982. 7-3. E. H. Russell, "Safely Grounding Static-Control Work Surfaces," EOSIESD Technology Maga• zine,1une 1987, page 10. 7-4. R. Kallman, "Comments on Coping," Evaluation Engineering Magazine, August 1988, page 108. 8-1. "Arm Hair Pin-Pointed as New Hazard," Evaluation Engineering Magazine, May 1984, page 70. REFERENCES 165

8-2. Scott's Standard Methods of Chemical Analysis, 6th Edition, Volume I, page 334. 8-3. NAV SEA SE 003-AA-TRN-0IO, "Electrostatic Discharge Training Manual," pages 35 and 36. 8-4. R. D. Anderson, "Alert Error Corrected: How Sweet It Is!" EOSIESD Technology Magazine, OctoberlNovember 1987, page 8. 8-5. "Designing for Compliance: Immunity to ESD," Application Note 106, Special Supplement, Compliance Engineering Magazine, 1991. 9-1. N. B. Fuqua and R. C. Walker, "ESD Controls Study, Final Report," prepared for NASA by the Reliability Analysis Center, Rome Air Development Center, September 1981, page 30. 10-1. S. A. Halperin, "Anything But Static: The EOSIESD Association Takes Charge," Compliance Engineering Magazine, Winter 1991, pages 13-28. 10-2. S. Weitz, "New Trends in ESDTest Methods," EMC Test and Design Magazine, February 1993, pages 22-26. 10-3. 1. C. Hoigaard, "ISO 9000 Promotes Automated Continuous Monitoring," Evaluation Engi• neering Magazine, February 1995, pages 105-108. 12-1. K.1essen and 1. Barto, "Static Control Team Concept-Implementation of ESD Protection in Manufacturing," Evaluation Engineering Magazine, November/December 1983, page 94. 12-2. 1. R. Giuliano, "SD Program Nets Large Financial Gains," Evaluation Engineering Magazine, December 1986, page 26. 15-1. "New Plastics Harvest Adds Variety," Machine Design Magazine, 1uly 21,1988, page 12. 15-2. A. H. Keough, "Antistatic Resin Composition," U.S. Patent 4,623,594 (November 18, 1986). 15-3. P. O'Shea, "Totes/Bins/packaging Go Environmental," Evaluation Engineering Magazine, March 1995, pages 92-95. 15-4. D. Cronin, "Cro-Bar: A New Technique for ESD Protection," EMC Test and Design Magazine, 1anuary 1993. 15-5. "MIL-HDBK-773: Sighs of Relief," EOSIESD Technology Magazine, OctoberlNovember 1988, page 7. Appendix

Some of the following have been condensed to highlight conclusions and save space, but the majority have been reproduced in their entirety. Paper No. 1 was reprinted with permission of lIT Research Institute/Reliability Analysis Center. Paper No.5 was reprinted with permission of Evaluation Engineering Magazine. Paper No.7 was reprinted with permission of EOSIESD Technology Magazine. Papers No.2, 3, 4, 6, 8, 9, and 10 were reprinted with permission of the EOS/ESD Association. Papers No. 11-20 were reprinted with permission as noted in each case. For a quick overview, leaf through and read the abstracts.

167 PaperNo. 1

Presented at the 3rd Annual Electrical Overstress Electrostatic Discharge Symposium, Las Vegas, Nevada, September 22-24,1981, sponsored by lIT Research Institute, EOSIESD Symnposium Proceedings, EOS-3, 1981, page 75.

SELECTION OF PACKAGING MATERIALS FOR ELECTROSTATIC DISCHARGE-SENSITIVE (ESDS) ITEMS

John M. Kolyer and William E. Anderson

Rockwell International Corporation Autonetics Strategic Systems Division Electronics Operations 3370 Miraloma Avenue Anaheim, CA 92803

Note: The following is a condensed version of this paper. See the Symposium Proceedings for the complete text.

Abstract

Seven materials were evaluated. Properties such as transparency, puncture resistance, and heat seal strength were measured. Also, ESDS devices in bags were subjected to the exter• nal static field of a model "human finger" charged to 25,000 volts as well as to the extreme condition of a high-voltage continuous discharge. The assumption was that if the packaged devices survived ESD overstressing with no measurable damage they would not be subtly damaged under actual handling/transit conditions. Such damage could reduce lifetime and reliability. For the highest confidence in protection from ESD, at an acceptable cost and with military specification coverage, a double-bagging system was adopted: antistatic poly• ethylene, which is MIL-B-81705, Type II, as the intimate wrap with a foil bag, MIL-B- 81705, Type I, as the exterior "Faraday ." Another foil bag, proposed as MIL-B-81705, Type III, is effective without a Type II inner bag. A brief review is given of the context of static control plant procedures in which the protective bags are used.

Introduction

The selection of packaging materials for ESDS items is difficult because many products are on the market and the claims made by rival manufacturers are often questionable and even contradictory. The early products included antistatic or "pink poly," which is MIL-B- 81705, Type II; conductive (carbon-loaded) polyethylene; and a laminate incorporating alu• minum foil, which is MIL-B-81705, Type I. Later, variations on "pink poly" and foil bags

169 170 APPENDIX were developed; these are "blue poly" and a foil laminate bag proposed as MIL-B-81705, Type III, and so abbreviated in quotes as "Type III." Also, a new concept in "static-protec• tive" bags appeared: film with an exterior of vacuum-deposited nickel, protected by a thin lacquer coat, and an antistatic polyethylene lining; this construction has the advantage of partial transparency. The proponents of the latter bag made the industry aware of the need for "Faraday cage" protection from external static fields/discharges and the insufficiency of "pink poly" alone for worst-case handling conditions. An aluminum• coated, partially-transparent bag is now competitive with the original, nickel-coated ver• sion. In 1980, Department of Defense documents were issued, DOD-STD-1686 and DOD-HDBK-263, which emphasized that the interior of a bag must be incapable of triboelectric charging, as true of "pink poly," while the exterior should be a conductive "Faraday cage" when external static fields might be encountered. Our evaluation included all seven materials mentioned above. The objective was high confidence in protection from external fields/discharges, at an acceptable cost, even for items of the greatest ESD sensitivity. The bags were for general use: they were not for clean room applications, handling of open devices, or the packaging of items sensitive to trace contamination. Functional properties such as heat sealability were checked, and packaged MOSFETs were exposed to external static overstresses to provide assurance of freedom from damage by realistic stresses. The chosen material(s), of course, had to be cost-effective.

Experimental Methods

Figures I through 4 show some of the experimental methods. Figures 5 and 6 are a typical example of ESD damage.

Materials Tested

The seven materials are described in Table I. All were received as samples in the form of bags. "Pink poly" is MIL-B-81705, Type II; by "MIL-B-81705" we are referring, through• out this paper, to the current Revision B. Two leading brands of carbon-loaded polyethylene were tested; the volume resistivities, which were calculated from surface resistivity mea• surements with bar-clampled samples at low voltage, were 235 and 875 ohm-em. Using bar-clamped samples and a megohmeter at 10 volts, surface resistivities were in the ex• pected range for the antistatic lining of the nickel-coated material, 2 x 1010 ohms/square, and the aluminum-coated material, 7 x 1011 ohms/square. "Pink poly" and "blue poly" gave, respectively, 8 x 1010 and I x 1011 ohms/square. The black lining of the Type I bag gave an apparent 8 x 107 ohms/square, but edge effects were possible. "Type III" has a "pink poly" lining for which 2 x 1011 ohms/square was found. Surface resistivity measurements are neither closely reproducible nor a critical test for antistatic properties; static bleed-off time (Federal Standard 101, Method 4046) is better for single materials but not for the Type I laminate, whose foil layer drains the charge in terms of perceived voltage.

Conclusions on Bag Materials

Table I summarizes ratings for the materials in our various tests, notes military specifica• tion coverage, and gives approximate prices. Blank spaces mean that no testing was done. PAPER NO.1 171

CAPACITOR

NONCONDUCTING IURFACE

Fig. 1. Charged "finger" test.

-11.aoo OR -.aoo V co. 1. 5. OR • SECONDS)

...... NONCOND_ .... UCTING SURFACE

Fig. 2. Tesla coil test. 172 APPENDIX

....ITIIIIMCI

, .. XI., INCH ICY MIlD HORN ANTINNA

• • KW PEAK lOWER • ,. GIGAHERTZ .,0SECONOS

Fig. 3. Radar exposure.

• MlL.sTD-I1OC, METHOD 114, PARAGRAPH 4.1.12.2 13 HOURS'

Fig. 4. Vehicular bounce test.

If an ESD shielding rating of "good" and a degree of transparency are required, the metal-coated bags seem adequate. The aluminum version was more transparent than the nickel version and performed as well in our other tests. However, potential users should conduct their own evaluations. In our case, "excellent" ESD shielding was desired. No sacrifice in confidence in ESD protection was made in order to gain partial transparency. This position left only two candi• dates: 'JYpe I and "'JYpe m." These foil bags, incidentally, not only give maximum ESD protection but have superior sturdiness (abrasion and puncture resistance) and are excellent moisture barriers. Since 'JYpe I has a questionably antistatic lining, as discussed above, only PAPER NO.1 173

Fig. 5. Punch-through of gate oxide induced by "finger" (25,000 volts, 167 pF) with MOSFET in black bag (1200><).

Fig. 6. Punch-through of gate oxide, detail from previous slide (22,OOOX).

''Type III" with its "pink poly" lining remained as a single-bagging candidate. ''Type III" is not yet covered by a military specification, but approval is said to be pending. Our MX missile module assembly facility, for which a bagging system was being se• lected, required cushioning material around assembled modules. The choice was "pink poly" bubble-pack with skins of Richmond RCAS-1200 (the only material thoroughly tested by us for antistatic properties). Since this material is not a "Faraday cage," double-bagging was necessary; a Type I bag was chosen as the ESD shield. Type I was preferred to ''Type III" because the former has military specification coverage and the latter's "safe" lining was not required. Also, Type I is less expensive (Table 1). When cushioning is unnecessary, a MIL-B-81705 system comprises a "pink poly" (Type II) inner bag and a Type I outer bag; both materials are bought from suppliers on QPL- 81705-6 (issued December 30, 1980). The inner "pink poly" bag provides high at-work• station transparency; an operator can withdraw the pink bag and easily see its contents. If 174 APPENDIX

Table 1. Summary ofT.ata

WA1ER HEAT ANIl- ESO 11lANS- VAPOll ABRASION SI!AL I'UNC'IUlB STATIC MIL MA11!RIAL SHJl!LO PAIU!NCY PEDI. RESIST. S1UNG11I RESIST. PROP. SPEC COST-

Pink poly P B P B B P G Yes 13 Blue poly P G P B B P G No 18 Nickel- G P P P P·B G P No 30 coated A1uminum- G P P P B G P No 28 coated Black F X P B P B No 19 Type I B X B B G B P Yes 21 "Type IIr' B x B B G B G No 29

-Cenu per 8 x 10 in. bag. Iota of 2000. S181

Code: E = Excellent G= Good F=Fair P= Poor x= None the latter feature is not considered important, the "Type DI" bag is effective but does not yet have military specification coverage. The above benefits of double-bagging with Type II inside Type I are bought at a cost penalty of 13-21% over single-bagging with meta1-coated or '7ype III" bags. Using the prices in Table I, a Type I bag at 21¢ and a Type II bag at 13¢ together cost 34¢, which is 4, 5, or 6¢ more than the nickel-coated. ''Type III," or aluminum-coated bags, respectively. The above selections are based on present knowledge, but there are unanswered ques• tions concerning bag materials. For example, how permanent are their ESD shielding or antistatic properties? Accelerated aging tests combined with real-time exposures should give reassuring, or disturbing, answers. Also, new materials may be expected to appear on the market. The "bag of the future" may have most or all of the following properties: low cost, >75% transparency, high ESD shielding effectiveness (like Type I or "Type III", per• manent antistatic and ESD shielding properties, no dependence on a minimum relative hu• midity, no propensity to contaminate packaged items. minimum sloughing (with no possible shedding of conductive particles), excellent moisture barrier properties, good heat-sealability, and, last but not least, cost-effectiveness. PaperNo. 2

Presented at the 5th Annual Electrical OverstresslElectrostatic Discharge Symposium, Las Vegas, Nevada, September 27-29,1983, Sponsored by I1T Research Institute. EOSIESD Symposium Proceedings, EOS-5, 1983, page 87.

PERMANENCE OF THE ANTISTATIC PROPERTY OF COMMERCIAL ANTISTATIC BAGS AND TOTE

lohn M. Kolyer and William E. Anderson

Rockwell International Corporation Autonetics Strategic Systems Division Electronics Operations 3370 Miraloma Avenue Anaheim, CA 92803

Note: The following is a condensed version of this paper. See the Symposium Proceedings for the complete text.

Abstract

The of antistatic materials has been questioned. Therefore, accelerated life tests were conducted with several commercial bag materials. One criterion for retention of the antistatic property was ability of the bag lining to triboelectrically charge NEMA FR-4 -reinforced epoxy circuit board material. It was found that light rubbing, as occurs on withdrawing boards from bags in in-plant handling, gave the same result-a low and pre• sumably tolerable charge-whether or not antistat was present. Contamination caused higher charging. Thus, shelf life was judged to be unlimited, by this test, if bags are stored closed and clean. Another criterion was the charge on the bag itself. The static field is suppressed by foil or metallization, but surface conductivity is required to control the field for antistatic polyethylene alone (MIL-B-81705, Type II). For a leading brand of the latter, incorporating a low-volatility antistat, the shelf life was estimated as at least 8 years if bags are used at above 20 percent relative humidity. Heavy or repeated rubbing, as may be caused by vibra• tion in shipment, is another matter. Transfer of antistat from bag lining to surface of the repeatedly rubbed item seems necessary to prevent high charging. Again, a shelf life of over 8 years was assigned to the tested brand of MIL-B-81705, Type II, but data were insuffi• cient for a prediction for foil or metallized bags. Antistatic polyethylene tote boxes were also studied. Used, worn boxes can accept a triboelectric charge. Such boxes might be periodically treated with topical antistat solution. Conductive boxes are an alternative but have potential problems such as sloughing of conductive particles or triboelectric charging of ESD-sensitive items being added or removed.

175 176 APPENDIX

Introduction

The pennanence of antistatic plastic packaging materials is in doubt because large increases in surface resitivity during a I-year shelf life of films or foam sheets have been reported.' Increased susceptibility to triboelectric charging by polystyrene foam also was found for aged antistatic materials. The indication is that the antistat on the surface of the plastic volatizes into the air or is removed by contact with absorbent materials such as cardboard. Therefore, minimization of exposure to moving air or paper products was recommended.' In view of the above, our purpose was to establish the pennanence of certain antistatic materials, in the fonn of bags, which we use or might use. Of particular interest was the propensity of bag materials to triboelectrically charge a module on insertion or withdrawal. Therefore, a static charging test was devised in which a bare glass• reinforced epoxy board was pulled in a reproducible manner from aged bags and the charges on the board measured. Also, surface resistivities of aged bags were noted. Accelerated aging was achieved by exposing bags, both closed and propped open, to 160°F in a forced• draft oven. Bags were also exposed to blotting paper. Tote boxes were also of concern because used, worn, antistatic polyethylene boxes found in the plant were observed to take charges of up to a few kilovolts when stroked with a nylon cloth. Whether a given charge was really a practical danger was judged by dropping highly ESD-sensitive devices [metal oxide semiconductor field effect transistors (MOSFETs)] into a nonantistatic tote charged to various levels and electrically testing the devices for degradation.

Bag Materials Tested

Nine bag materials were tested (Table 1). Five of these had been characterized by us 2 years earlier.2 All samples except the foam were received in the fonn of approximately S x 10 in. bags; the foam was stapled into pouches. The abbreviations in Table I, e.g., "black" or "grid," are for convenience. The "Type I" material previously characterized2 was the Brand B material in Table 1. "Pink poly" was MIL-B-SI705, Type II, Qualified Products List (QPL) material, specifi• cally RCAS-1200 from Richmond Corporation. The "black" material was the brand for which a volume resistivity of 235 ohm-cm had been found.2This bag, of course, is conduc• tive rather than antistatic.

Experimental Methods

Figures I and 2 show the principal experimental methods.

Conclusions

Shelf Life of Antistatic Polyethylene Bags 1. In tenns of charging a popular circuit board material, under the specific conditions of our test, shelf life is unlimited if the bags are kept closed and clean. Antistatic polyethylene PAPER NO.2 1n

Table 1.

Type I \0 MIL-B-SI705, Type I (vapor-deposited aluminum + (brands A and B) spun-bonded polyethylene + aluminum foil + carbon-loaded polyethylene) Pink Poly 6 MIL-B-S1705, Type n (antistatic polyethylene) Foil + Antistatic Poly S Antistatic spun-bonded polyethylene + aluminum foil + antistatic polyethylene Nickel-Coated 2.5 100 Angstroms nickel + 1 mil polyester + 1.5 mils antistatic polyethylene Stainless-Steel-Coated 3 Polyester + stainless steel + antistatic polyethylene Black 4 Carbon-loaded polyethylene (conductive) Grid 4 Antistatic nylon + barrier film + conductive ink grid + barrier film + antistatic copolymer Antistatic Foam 250 Layered, antistat-treated foam

3/4-IN. THICK WOOD(o.IILI'~ ~

1/2·IN. THICK •• ~ ANTISTATIC FOAM (0.031 LI'~ .. '. ", ... ANTISTATIC lAG. Ix 10 IN. ~ FR ... EPOXY CIRCUIT IOARD (IARE. NO COI'I'ERI. e x 12 x O.OIM IN. (1.5 IN. EXTENDED INTO IAGI "

TA8LESWITH MElAMINE. FORMALDEHYDE lAMINATE SURFACE CARRIAGE (CARRIAGE PLUS CIRCUIT BOARD· 1.1 LII

~8~'ilE' WOOD AND FOAM WERE LIV lAG WITH EPOXY BOARD INSIDE. BOARD WAS HELD IY FINGER PRESSURE ON lAG AT POINT X. THEN FINGER PRESSURE WAS REDUCED TO LET WEIGHT FALL. AND CHARGE ON BOARD WAS MEASURED.

Fig. I. Static charging apparatus. 178 APPENDIX

LEADS TO MEGOHMMITE"

Y1:= .. ~- tl1t4N. ALUMINUM lUIt'ACE OF lAG IIlOUNOED EDGI' NOTE: A U·LI. WEIGHT WAI PLACED ON EACH OF THE ELECTIIODU

Fig. 2. Surface resistivity electrodes.

has a long record of successful use. Also our tote box tests suggest that limited static fields may be tolerable. Therefore, it is presumed that the relatively low triboelectric charges detected on the circuit board would not endanger attached devices. In contrast, the high charges observed when the board was stroked with carbon-loaded polyethylene might well be damaging. However, voltage suppression would limit the field on a multilayer board with internal planes of copper. 2. In tenns of surface resistivity, shelf life at usual relative humidities (over 20 percent) is conservatively estimated as 8 years for 6-mil "pink poly" bags when kept closed. This relatively good pennanence is probably due to the low vapor pressure of the antistat, which is extruded into the polymer and gradually bleeds to the surface to fonn a "sweat layer" with atmospheric moisture. See Fig. 3 and Table 2. 3. Bags should be stored closed to exclude contamination such as dust, kept out of contact with paper or other absorbent materials, and, of course, never washed with water or organic solvents. 4. Two extreme situations for using bags are: (I) light, unrepeated rubbing by the pack• aged item against the bag lining (in-plant handling) and (2) heavy or repeated rubbing due to vibration (shipment). Antistat depletion is irrelevant for light rubbing, in tenns of our board-charging test, but may be critical for heavy or repeated rubbing. In the latter case, the key factor may be neither resistivity nor lubricity but antistat transfer from bag to packaged item. The role of relative humidity is not clear. Note that the lining of MIL-B-81705, type I, which is without antistat, has been reported to develop damaging charges on MOSFETs in a vibration situation.3 Also note that a surface layer of peanut oil, instead of antistat, can prevent triboelectric charging of DIPs by polyethylene under heavy-rubbing conditions.4 5. Antistatic polyethylene bags without foil or metallization to control the static field can develop significant charges on themselves at some combination of antistat depletion and low moisture content of the "sweat layer." Here, relative humidity could be a critical factor. Thus our prediction of a long shelf life for "pink poly" (MIL-B-81705, Type II) is limited to use at >20 percent relative humidity. PAPER NO.2 179

1I&0Il10" ATUVRI

tOl.L...----,!'---~2---,l-3--- ..!----:!.f----:!. TIME. MONTHS

Fig. 3. Surface resistivity versus time for aged 6-mil ''pink poly" bags.

ESD Hazard Caused by Static Charges on Tote Boxes 1. Used, worn antistatic polyethylene tote boxes can develop static chuges of a few kilovolts when stroked with a nylon cloth. The MOSFET tests indicated that up to 4 tv could be tolerated. but weaker fields could cause ESD damage when "antennas," e.g., cUcuit lines, are present (see Paper No.3). 2. It is recommended that antistatic tote boxes be treated periodically with topical anti• stat solution. This is an inexpensive procedure in terms of antistat cost but mayor may not be practical in view of handling and scheduling complexities.

Table 2. Estimated Shelf LIves for Antistatic Bags (Used at >20 Percent Relative Humidity).

ESTIMA11!D SHELF LIFE (CLOSED, CLEAN)

EXTREMBUSE MIL-S-817OS CONDmON TYPE n FOIL OR METALLIZED Light rubbing ofFR-4 epoxy/glass 8 years Unlimited on insertion/withdrawBl (inplant use) Heavy/repeated rubbing, eg, due to 8 years Not estimated (rate and effect of vibration (shipment) antistat depletion are unknown) 180 APPENDIX

3. Conductive tote boxes have no permanence problem but may triboelectrically charge devices being removed from them, slough conductive particles, or cause ESD damage by rapid discharges as could happen with any conductor. 4. Each used must select a tote box material based on his unique parts-handling situation.

References

I. G. O. Head, "Drastic Losses of Conductivity in Antistatic Plastics," Reliability Analysis Center EOSIESD Symposium Proceedings, pp. 120-123, Orlando, FL, 1982. 2. J. M. Kolyer and W. E. Anderson, "Selection of Packaging Materials for Electrostatic Dis• charge-Sensitive Items," Reliability Analysis Center EOSIESD Symposium Proceedings, pp. 75-84, Las Vegas, NY, 1981. 3. D. M. Yenni, Jr., and J. R. Huntsman, "The Deficiencies in Military Specification MIL-B-81705: Considerations and a Simple Model for Static Protection," presented at the Reliability Analysis Center EOSIESD Symposium, Denver, CO, 1979. 4. J. R. Huntsman and D. M. Yenni, Jr., "Test Methods for Static Control Products," Reliability Analysis Center EOSIESD Symposium Proceedings, pp. 94-109, Orlando, FL, 1982. PaperNo. 3

Presented at the 6th Annual Electrical OverstresslElectrostatic Discharge Symposium, Philadelphia, Pennsylvania, October 2-4, 1984, Sponsored by EOSIESD Association and lIT Research Institute. EOSIESD Symposium Proceedings, EOS-6, 1984, page 7.

HAZARDS OF STATIC CHARGES AND FIELDS AT THE WORK STATION

John M. Kolyer, William E. Anderson, and Donald E. Watson

Rockwell International Corporation Autonetics Strategic Systems Division Electronics Operations 3370 Miraloma Avenue Anaheim, CA 92803

Abstract

Tests were conducted to determine under what practical conditions electrostatic discharge• sensitive (ESDS) items can be damaged at the work station by static charges and fields. Various materials and equipment were evaluated for possible hazards (e.g., static fields created by electrical air ionizers) which must be avoided by careful operating techniques. The data led to important conclusions and guidelines for an ESD control program in accor• dance with DoD-STD-1686. Examples are given of equipment, materials, and techniques which complement one another, and basic rules for electronic assemblers are suggested.

Introduction

A charged surface always creates a static field (E-field). However, in discussing "hazards of charges" we ignore the field and are concerned with discharges between conductors, for ex• ample from a person's finger to the copper line on a circuit-board module. In discussing "hazards of fields" we are concemed with ESD damage caused by induction. The effects of RF (radio-frequency radiation) or of magnetic fields (H-fields) are not considered in this paper. Following a description of the test methods used, we evaluate charge and field hazards, including those from electrical equipment, and discuss some methods of coping with them, Finally we briefly describe a synthesis of materials/equipment and operator techniques for handling ESDS items in accordance with DoD-STD-1686.

Test Methods

Static Meter Measurements Measurements were made with a Simco Electrostatic Locator, Type SS-2 (Simco Co., Inc.), calibrated with a 12-inch-square aluminum sheet charged to 1,000-10,000 volts by

181 182 APPENDIX a current-limiting high-voltage dc power supply (TARI Central Control Module, Static Control Services). The charges reported in this paper are apparent charges; for example, a 4-inch-square aluminum sheet charged to 10,000 volts showed an apparent charge of about 5,000 volts. The apparent charge is a measure of the strength of the field, which can damage ESDS items by induction.

Surface Resistivity

The electrodes have been illustrated. I A 5-lb weight rested on each electrode, and strips of soft carbon-filled polyolefin were placed under the blades to improve contact with static-dissipa• tive or antistatic surfaces. With this arrangement, a steel surface read lOS ohms/square, so lOS was subtracted as a correction factor. Readings were made with a Beckman Model L-1O megohmmeter at 500 volts unless otherwise noted. Multimeters were used for lower voltages.

MOSFET Damage Tests The N-channel metal oxide semiconductor field effect transistor (MOSFET) used in our tests was a Motorola 2N4351 with the shorting bar removed. The metal TO-72 type package had four leads (source, gate, drain, and substrate-case). This device contains no input/output protective diodes or circuitry. Using a Tektronix 576 curve tracer, gate-source threshold voltages (VOS(11I) were read at Vns = 10 volts and In = 10 ~A in accordance with MIL-STD-750B, Method 3403. The operator was, of course, grounded with a wrist strap when making measurements. For many experiments a "MOSFET board" was used; see Fig. 1. The substrate-case lead of a MOSFET was clipped to the "antenna" (circuit line) touched by the operator, while the gate lead was clipped to the circuit line with the projecting lead. The choice of substrate• case and gate leads, rather than another combination of leads, maximized the oxide layer area exposed to an ESD. Each circuit line was 4 inches long, and the lead projected 1.7 inches beyond the end of the board. The circuit-line antennas increased the likelihood of

COPPER CIRCUIT LINE

6 x 9 x 0.06 INCH PR-4 GLASS-EPOXY LAMINATE

Fig. 1. MOSFET board. PAPER NO.3 183

ESD damage to the MOSFETs by static fields. 2 Similar printed circuit boards with MOSFETs and circuit lines for antennas have been used in other investigations3•4 but for different purposes. Our board, held by the operator as shown in Fig. I, was intended as a worst-case simulation of ESDS modules being handled at the work station. MOSFETs were often shorted, as noted in the data, but they were also considered damaged if the current-voltage curve changed and/or VOS(1lI) shifted by more than 0.1 volt (VOS(1ll) read• ings for undamaged MOSFETs were reproducible to ±O.02 volt). Since latent ESD failures are a reality,S a VOS(1ll) shift of only 0.1 volt is considered significant. Damage is reported as a fraction; for example, "2/5" means that two MOSFETs were damaged out of five tested.

Room Ionization System The power supply was mentioned above. Two emitters (Static Control Services), each with four corona-discharge points and a set of plastic reflectors, were positioned 3 ft apart on the ceiling of an open-fronted booth 95 inches high, 71 inches wide, and 47 inches deep. This booth contained a bench 29 inches high, 60 inches wide, and 30 inches deep. The distance from the emitter points to the work surface was 59 inches. One emitter was positive and the other negative.

Shielding/Discharge Test This test for evaluating packaging materials was conducted with apparatus in accordance with published reports.6•7 In brief, a capacitive sensor, comprising two 0.75-inch-diameter aluminum disks (0.06 inch thick) separated by 0.5 inch of acrylic plastic, was placed inside a 3 x 3 inch pouch of packaging material clamped between flat aluminum electrodes. See Fig. 2. In accordance with a published procedure,8 a pulse of 1200 volts was applied from a 200 pF capacitor through a 400 kilohm resistor to the upper electrode, the lower electrode being grounded. The pulse picked up by the capacitive sensor was delivered to a storage oscilloscope. Results were conveniently expressed as "percent attenuation;" for example, if the pulse recorded by the sensor was 300 volts, the percent attenuation was (1200 - 3(0) (100)/1200 = 75. One edge of the pouch under test was always continous so that the mate• rial had electrical continuity from side to side.

Modified Shielding/Discharge Test The capacitive sensor described above was provided with 4-inch leads ending in clips. MOSFETs, described above, were clipped with the substrate-case lead connected to the upper capacitor plate and the gate lead connected to the lower plate. The sensor, including the leads and the MOSFET, was placed inside an 8 x 10 inch pouch of packaging material resting on a grounded aluminum plate, and the charged probe of an Electro-Metrics Model EDS-200 discharge unit (Model D-25, 150 ohms, 150 pF) was touched to the upper surface of the pouch (see Fig. 2). To ensure a worst-case condition, the packaging material was pulled tight against the sensor by taping down the edges of the pouch. The probe of the discharge unit was intended to represent a statically-charged person's finger touching a bag containing ESDS items and lying on a conductive surface. In a few cases an actual finger was used. 184 APPENDIX

SHIELDIMG/DISCHAIGI TEST

Fig. 2. Diagram of bag-testing procedures.

Hazards of Charges at the Work Station

Bench Tops In discussing bench tops we are concerned not with the top holding a charge but with its ability to drain charges from objects placed on it. Static-dissipative tops (surface resistivity 1()5-109ohms/square) are favorably discussed in DoD-HDBK-263 and have been suggested as ideal.9 Antistatic tops may drain charges too slowly, as will be seen below, while conductive tops are an electrical safety hazard and might damage charged ESDS items by discharging them too rapidly.9.10 (See also DoD• HDBK-263 and the discussion below.) A top may have a conductive sublayer without being satisfactory, because without suffi• cient surface conductivity the charge will be merely voltage-suppressed and not drained." In our tests, conductive or antistatic tote boxes charged to ±1O,OOO volts drained incom• pletely (e.g., 8000 volts remained on an antistatic box) after resting 1 minute on an antistatic top (lOll ohms/square) but drained completely «50 volts remained) on a static-dissipative top (109 ohms/square). One successful construction for a static-dissipative top is, in our experience, a high-pressure laminate with linen fabric in the surface layer and conductive carbon paper (grounded through 1 megohm) as the sublayer.12 The sol vent resistance of this laminate is reported to be very good,I2 and in our tests 50 daily rinses with methyl ethyl ketone had no effect on surface resistivity or appearance. Static-dissipative tops are safer than conductive tops for operators, but can static-dissipative tops also be safer for devices as suggested above? To answer this question, tests were con• ducted with the MOSFET board (Fig. 1). An operator was charged to various levels by PAPER NO.3 185 contacting the dc power supply, and a second after releasing the supply he touched the lead of the MOSFET board to either a static-dissipative bench top or a conductive work surface represented by an aluminum sheet. Both surfaces were grounded through 1 megohm. Results are given in Table 1. Both surfaces allowed damage to occur, but the static dissi• pative surface was less destructive. That a more resistive surface can be less damaging will be seen again in Table 2. In conclusion, an antistatic top will not itself hold a charge, but it drains charges too slowly from objects placed on it. The best choice seems a static-dissipative top with a bur• ied conductive layer and a securely mounted grounding lug.

Tote Boxes Tote box selection is a controversial subject. Antistatic tote boxes can lose their "sweat layer" and should be periodically treated with an antistat solution. \ Carbon-loaded plastic boxes, on the other hand, are permanently conductive but can slough conductive carbon particles. If sloughing is not considered a problem, one question remains: Are carbon-loaded boxes too conductive? The fear is that a charge on a conductive box might "zap" an ESDS item, whereas under the same conditions the charge would bleed off harmlessly from an antistatic box. Or, conversely, a charged ESDS item might discharge too rapidly to the "sparking surface" of a conductive box. 10,\3 We attempted to settle this question using the MOSFET board (Fig. 1). Either the opera• tor or the tote box was charged, and the lead of the MOSFET board was touched to the box. Two boxes were tested: an antistatic box (surface resistivity 2 x 109 ohms/square at 48 percent relative humidity and 72°F) and a black conductive tote box (2.2 x 10" ohms/square at 32 volts). Both boxes were injection-molded by the same manufacturer and had the same dimensions (approx. 7 x 10 x 3 inches deep, with a 0.08-inch wall). Results are given in Tables 2 and 3, which also include an aluminum sheet (12 x 12 x 0.06 inch) as a reference. Table 2 also includes a static-shielding bag, discussed below. When the operator was charged, the box or sheet lay on a static-dissipative bench top grounded through 1 megohm; when the box or sheet was charged, it stood on insulating stand-offs (plastic breakers) and the

Table 1. MOSFET Damage Caused by Charged Operator Touching Lead of MOSFET Board (Fig. 1) to Work Surfaces Grounded through 1 Megohm.

MOSFET DAMAGE

STATIC-DISSIPATIVE SURFACE ALUMINUM VOLTAGE (8 x 108 OHMS/SQUARE) SHEET

-100 0/5 1/5 -200 0/5 3/5 (2 shorted) -300 1/5 (l shorted) 3/5 (l shorted) -400 0/5 2/5 (0 shorted) -500 2/5 (l shorted) 3/5 (2 shorted) -1000 2/5 (1 shorted) 5/5 (4 shorted) Totals: 5/30 (3 shorted) 17/30 (9 shorted) 186 APPENDIX

Table 2. MOSFET Damage Caused by Charged Operator Touching Lead of MOSFET Board (Fig. 1) to Objects on Static-Dissipative Bench Top (8 x 10' Ohms/square) Grounded through 1 Megohm.

MOSFET DAMAGE

PAJmALLY• TRANSPARENT BAG (II x 14 INCH) ANTISTATIC CONDUCTIVE ALUMINUM WITH EXTERNAL VOLTAGE TOTE BOX TOTE BOX SHEET METALLIZATION

-300 0/3 0/3 1/3 -500 0/3 0/3 2/3 (l shorted) -1,000 0/3 2/3 (l shorted) 2/3 (l shorted) 2/3 (0 shorted) -2,000 0/3 3/3 (2 shorted) 3/3 (l shorted) -5,000 0/3 2/3 (l shorted) -10,000 0/3 3/3 (2 shorted) Totals: 0/18 10/18 (6 shorted)

operator was grounded through 1 megohm with a wrist strap. The tests were conducted at 68 percent relative humidity and n°F. The result was that the antistatic tote box was clearly less damaging than the conductive box when the operator was charged but only slightly less damaging when the box was charged. An explanation is that when the box was charged (Table 3) induction became the dominant factor; that is, MOSFETs were damaged by capacitive coupling when the lead of the MOSFET board penetrated the field of the charged box. On the other hand, when the operator was charged (Table 2) the controlling factor was the resistance of the surface being touched by the lead. The conclusion is that antistatic boxes are less of a discharge hazard than conductive boxes if handling techniques are poor. With good operator discipline, conductive boxes are

Table 3. MOSFET Damage Caused by Grounded Operator Touching Lead of MOSFET Board (Fig. 1) to Charged Objects.

MOSFET DAMAGE

ANTISTATIC CONDUCTIVE ALUMINUM VOLTAGE TOTE BOX TOTE BOX SHEET

-300 0/3 0/3 2/3 (l shorted) -1,000 0/3 2/3 1/3 (l shorted) (l shorted) -2,000 2/3 3/3 3/3 (2 shorted) (3 shorted) (3 shorted) -3,000 2/3 2/3 3/3 (l shorted) (l shorted) (3 shorted) Totals: 4/12 7/12 9/12 (3 shorted) (5 shorted) (8 shorted) PAPER NO.3 187 safe and have the advantage of not relying on a fugitive antistatic additive; however, the user must assure himself that sloughing of conductive particles is not a problem. In our assembly operations, modules are placed in Faraday-cage (foil-containing lami• nate) bags, and then the bags are placed in tote boxes. When a tote box is received at a work station, the operator removes the bag and sets the tote box aside before removing the mod• ule. Thus, the tote box is not a threat no matter what its electrical properties. Even ordinary plastic tote boxes could be used in this manner, but good practice calls for selection of conductive or antistatic materials in an ESD control program. The blanket rule is that ordi• nary plastics must be excluded from work stations.

Nonconductive Plastics When a polyethylene film (0.004 inch) at -13,000 volts (apparent charge, as are all charges reported in this paper) lay flat on grounded aluminum foil, the apparent charge was reduced to -300 volts because of voltage suppression, and MOSFET damage by a grounded opera• tor touching the MOSFET board lead to the plastic surface was 0/3. Similarly, for a vinyl sheet (0.058 inch) at -8000 volts the apparent charge fell to -300 volts and MOSFET dam• age was 0/5. The nonconductive plastics were unable to deliver their high charges, and the suppressed fields were too low to cause damage. The conclusion is that fields, rather than charges per se, are the hazard in this case. Thus the MOSFET damage reported in Table 4 when charged nonconductors were touched would presumably have been caused by a close approach without contact.

Table 4. MOSFET Damage Caused by Fields from Various Charged Objects.

APPARENT CHARGE, DISTANCE, MOSFET OBJECT VOLTS INCHES DAMAGE

Butyrate Tote +7000 2 3/3 Box (3 shorted) 6 1/3 (0 shorted) Vinyl Tote Box on +500 o (touched) 1.3 Bench (0 shorted) Vinyl Chair +500 o (touched) 1.5 (1 shorted) Plastic Identification -2000 3 1/4 Badge (0 shorted) Butyrate Screwdriver +1700 o (touched) 2/3 Handle (0 shorted) Polyester Lab Coat +1300 o (touched) 2/5 (0 shorted) Polyethylene Film -10,000 2/2 (1 shorted) Aluminum Sheet -3300 4 3/3 (1 shorted) Painted Metal +100 o (touched) 0/3 188 APPENDIX

Operators An operator's skin should always be grounded. However, a grounded operator can cause ESD damage, as seen in Tables 3 and 4, and is only the lesser of two evils versus an un• grounded operator. Good handling techniques are needed to protect ESDS items from grounded operators when charges on the items are unavoidable. For example, a module might be triboelectrically charged by spraying with a conformal coating. Then, if a grounded operator were to touch a contact a damaging discharge could occur as explained in the literature. 10.14 A solution to this problem in terms of handling techniques is for the operator to be conscious of the hazard and avoid touching leads or contacts of ESDS items. A possibility is that an operator's fingers might be reduced from conductive to static• dissipative or antistatic surfaces by his wearing appropriate gloves, possibly cotton. This idea is especially applicable to static-producing assembly or rework operations such as grit blasting.

Hazards of Fields at the Work Station

Field Strength versus Distance Table 4 gives typical data obtained when a grounded operator waved the MOSFET board (Fig. 1) in the vicinity of various charged objects. Rapid motion, however, seemed unneces• sary; when the lead of the MOSFET board was brought within 1 inch of a plastic surface at -5000 volts at the slow speed of 5 inches/minute, and then withdrawn at the same speed, MOSFET damage was 1/2. Presumably a potential difference was slowly built up by ca• pacitive coupling until the oxide layer of the device was ruptured. Incidentally, it should be noted that a charge of -3000 volts on the MOSFET board itself, with both the antennas and the operator grounded, caused 2 of 3 MOSFETs to be damaged (I shorted) by installation and removal from the board. Damage of components during insertion into highly charged printed circuit boards has been reported. IS Figure 3 shows MOSFET damage levels caused by a grounded operator (wearing a wrist strap with a I-megohm resistor) waving the MOSFET board in static fields. The circles are for various plastic surfaces, and the triangles are for an isolated 12 x 12 x ~ inch aluminum sheet charged to various negative voltages by the dc power supply. The black data points represent MOSFET failures (at least 1/3), and the white points represent no failures (0/3). Assuming a point source, field strength is directly proportional to voltage on the surface and inversely proportional to the square of the distance from the surface. Therefore, V plot• ted against d2, where V is voltage and d is distance from the surface, will give a straight line whose slope represents field strength. The same result is given by plotting rv versus d, which was done for convenience. Of course, the source of the field is an area rather than a point, and the data are not precise enough or numerous enough to establish an exact thresh• old field strength to cause damage. However, the line in Fig. 3 illustrates the principle. The equation for this line (except near the origin) is rv = Cd, or d = MC, where C is a constant (1.8 in this case) and d is the minimum safe distance in inches from a surface at apparent voltage V. Fo~example, if the charge were 3000 volts, the calculated minimum safe distance would be 3000/1.8 = 30 inches. Since no damage occurs below 50 volts, the line does not pass through the origin, and the threshold for damage by nonconductors was 300-500 volts. PAPER NO.3 189

0 l 7000 V sooo

JIAZAIDOUS SAFE

~ .oil. S 3300 • i. Q ...... ;0. -- 1 . - r.:I ~ 2000 -..., ... / ~ 1700 i 1300 ..... v ~ 1000 / .A ~ :; 500 ... ~ ~ 300 .. ~NONCONDUcroRS 100 / ... 50 "/

o 10 20 30 40 so 60 DISTANCE, INCHES

Fig. 3. MOSFET damage caused by static fields.

This estimate of minimum hazardous field strength is, of course, worst-case because the long antennas of the MOSFET board served to "gather" static fields. At the other extreme was a "MOSFET module" in which the antennas (insulated wires) were sandwiched be• tween aluminum ground planes while the projecting lead shown in Fig. 1 was eliminated. At 1 foot from a surface at -7000 volts, the "MOSFET module" gave MOSFET damage 0/5 versus 4/5 (two shorted) for the MOSFET board. However, the "MOSFET module" did allow damage (2/3, 0 shorted) at 2 inches from a surface at -7000 volts. Therefore, the constant C for the "MOSFET module" is more than 7 but less than 42. Real-life module assembly situations will lie between the two extremes represented by the MOSFET board and module.

Air Ionizers A few ionized-air blowers and nozzles (guns) were evaluated for field hazards by waving the MOSFET board (Fig. 1) at various distances from their tips while they were running. The operator was grounded through 1 megohm as usual. Table 5 gives the data. A safe working distance from the blowers or the pulsed dc ionizer was 10 inches. Inci• dentally, the pulsed de ionizer had the advantage of not blowing air. An airstream is often 190 APPENDIX

Table 5. MOSFET Damage Caused by Fields from Air-Ionizing Blowers and Nozzles.

DISTANCE FROM MOSFET BOARD, MOSFET EQUIPMENT INCHES DAMAGE

Blower Alternating Current, Corona Discharge 'JYpe: Model A (5000 volts) 2 2/3 (l shorted) 4 0/5 6 0/5 Model B (5000 volts) 4 4/10 (2 shorted) 10 0/10 Ionizer, Pulsed Direct Current, Corona Discharge 10 0/10 Type, 10,000 volts (Balanced) Ionized Air Nozzles, Corona Discharge Type: Model A (4000 volts) 2/5 (0 shorted) 2 0/5 Model B (4000 volts) 3/5 (1 shorted) 2 1/3 (0 shorted) 3 0/5 Model C (7000 volts) 1 0/5 Ionized Air Nozzle, Nuclear Type 0/5

objectionable; for example, it causes undesirable cooling during soldering. However, this fanless ionizer was as effective as the blowers in rapidity of neutralizing positive or nega• tive charges on plastics in tests at a range of 3 feet. The nozzles were operated at 30 Ib/in.2 gage air pressure. Model C was less damaging than the others; in this model, the corona-discharge point was mounted "piggyback" out of the airstream. The nuclear type caused no damage because it has no electrical field. Another advantage of the nuclear type is its compact size. However, its cost may exceed that of electrical nozzles over a period of years because of an annual leasing fee, and the poisonous radioactive material (polonium-210) presents a remote safely hazard. For very close work, on the basis of field hazard as well as bulk, the nuclear-type nozzle is preferred. The conclusion is that electrical air ionizers do not present a field hazard if the work is kept 10 inches or more away from them. However, when nozzles must be held very close to ESDS items, as in cleaning a module by blowing off particles, the nuclear type is required.

Room Ionization System Reasons for selecting room ionization as opposed to local ionization-or for using ioniza• tion at all-are beyond the scope of this paper. We are concerned here only with certain aspects of a specific dc room ionization system. PAPER NO.3 191

The equipment which we tested is meant to be operated with the positive and negative emitters at similar voltages so that there is little net field strength or ion imbalance at the workbench level 5 feet or more beneath the emitters. However, if the system were acciden• tally unbalanced would an electrically isolated conductor on the bench become sufficiently charged to damage ESOS items touched to it? To answer this question, tests were conducted in the booth described above under "Test Methods" using only one of the two emitters. The bench surface was static-dissipative as described under "Bench Tops," above. An aluminum sheet (12 x 12 x ~ inch) was isolated from the bench surface by two different thicknesses of nonconductive plastic (bubble-wrap) as listed in Table 6. Thus, a capacitor was produced by the aluminum sheet and the carbon• paper sublayer of the bench; when the two were separated by ~ inch. for example, the calculated capacitance of the aluminum sheet was 129 pF. The voltage on the aluminum sheet was measured with an Electrostatic Field Meter Model 970 from Static Control Ser• vices; in this case the field meter usually employed was not sensitive enough. Table 6 gives the data. The system is normally operated at about 8200 volts; a slight increase, to perhaps 9000 volts, may be required at low humidity. Therefore. Table 6 indicates that even if all the emitters of one polarity should fail completely there will not be dangerous charging of isolated conductors lying on the bench. In conclusion, worst-case testing showed no hazard. Having gained confidence that the room ionization system would not be part of the ESO problem, we tested its efficacy at neutralizing charges on plastics. Table 7 shows that neu• tralization of charges at the bench level was reasonably rapid for a variety of materials; the relative humidity during this test was 56 percent at 74°F. Even with the system somewhat unbalanced (15,000 positive ions/ml, 22,000 negative ions/ml), results were similar. To check the effect of humidity. the booth was sealed and the air inside it was partially dried

Table 6. MOSFET Damage Caused by an Unbalanced DC Room Ionization System.

DISTANCE FROM OBSERVED ALUMINUM VOLTAGE ON SHEET TO ALUMINUM MOSFET EMITTER VOLTAGE- BENCIITOP. IN. SHEET DAMAGE

+17.000 4 +240 2/3 (1 shorted) As required to charge aluminum sheet 4 +120 1/3 (0 shorted) As required to charge aluminum sheet 4 +80 1.5 (0 shorted) As required to charge aluminum sheet 1/4 +120 2/5 (l shorted) As required to charge aluminum sheet 1/4 +80 0/5 As required to charge aluminum sheet 1/4 +50 1/10 (0 shorted) +8200 1/4 <10 0/10 -8200 1/4 <10 0/10

*Only one emitter was operating. 192 APPENDIX

Table 7. Charge Decay Rates with or without Room Ionization.

RESIDUAL TIME, MINlITES, TO VOLTAGE AT DECAY TO SOO 5 MINUTES INITIAL VOLTS wrrn wrrnSYSTEM MATERIAL CHARGE, VOLTS SYSTEM ON' OFF"' Polystyrene Foam +5000 1.5 +1200 Polyethylene Film -500 1.8 -4800 Vinyl Sheet -5000 2.5 -800 Carbon-Loaded Polyethylene +5000 2.5 +3200 (Conductive) Aluminum Foil +5000 1.8 +3300

'15,000 positive ions/ml, 13,000 negative ions/ml. *"<100 positive ions/ml, 500 negative ions/mI. with a desiccant. At 10-20 percent relative humidity, the decay rates of the materials in Table 7 remained rapid. Another question was the effect of position of charged objects under the emitters; it was feared, for example, that a positively charged material under a positive emitter might re• ceive relatively few negative ions and so exhibit a slow decay rate. Therefore a test was run with a positively charged plastic and a negatively charged plastic at 51 percent relative humidity and 73°F. Table 8 shows that charge decay rates remained far more rapid at any position with the system on than when it was off. Tests at commercial installations of this and a competitive system gave better uniformity, presumably because a grid of emitters covered the ceiling instead of only two emitters being confined to a booth. Of course, per• fect uniformity of decay rate across the bench is unnecessary.

Table 8. Effect of Position Under Room-Ionization Emitters on Charge Decay Rates for Plastics.

PERCENT OF INITIAL CHARGE INITIAL POSmONON CHARGE, BENCH IO 20 MATERIAL VOLTS SURFACE MINUTES MINUTES MINUTES Polyethylene -1700 Under Positive 3 Bubble-Wrap Emitter Under Negative 2 Emitter Between Emitters IS 9

Butyrate +15,000 Under Positive 17 11 3 Tote Box Lid Emitter Under Negative 3

It was desired to compare the effectiveness of the room ionization system with that of a typical bench-model ionized-air blower (as in Table 5). Polyethylene bubble-wrap triboelectrically charged to -10,000 volts was found to lose half its charge after a I-minute exposure (at 60 percent relative humidity and 74°F) either to the emitters at ±8200 volts at bench level or to the blower at a distance of 5 feet. In conclusion, the room ionization system tested did not present a field or charge hazard at the bench level and was indeed effective in removing standing charges on nonconductors such as plastic packaging materials. Fig. 4 (for 49 percent relative humidity and 73°F), is a typical example of the utility of room ionization in removing a stubborn charge on plastic; similar curves appear in a published report. 16 Note that the electrical field of a charged item attracts ions to cause neutralization. Therefore, a thin sheet of plastic which lies flat on a workbench and has a largely collapsed field will be neutralized only slowly and incom• pletely. The bubble-wrap in our test was i inch thick and was folded over on itself so that the measured surface was elevated about I inch.

Assembly Operations Many static-producing manufacturing operations are listed in DoD-HDBK-263 and a Navy training manual,I7 but only two will be discussed here: peeling masking tape from a roll and grit-blasting a module to remove the coating from components for rework. A conductive masking tape would have virtually no static field because of charge drain• age and voltage suppression. Our tests showed that a commercial aluminum foil/fiber tape had sufficient conformability and tear resistance for masking, but its acrylic re• leased incompletely after oven-baking of the masked parts at 150°F. Therefore, the only deficiency of this tape might be overcome by substituting a silicone adhesive, with good release, for the unsatisfactory acrylic. A Micro Blaster (Comco, Inc.) was found to create a charge of +700 volts on FR-4 circuit board material when sodium bicarbonate "grit" was sprayed against the surface using 10 Ib/in.2 gage air pressure. By adding a "piggyback" corona-discharge ionizer (Simco PIN

-18 II) ~16 0 I~~ WITHOUT ROOM IONIZATION ="'-14 0 --...... ~-12 0 I~ ~ \ a-100 1 ---. 0 , ----- i =: 0 \ ~ -4 0 ~WITH 100M IONIZATION -2 0 \. I' 10 20 30 40 so 60 TIME. MINUTES

Fig. 4. Charge decay rates for polyethlene bubble-wrap. 194 APPENDIX

4100034) operating at 7000 volts, with the point 0.6 inch from the spray nozzle, the charge developed was reduced to only +60 volts. This charge is well below the danger level for nonconductors (Fig. 2).

Operators A properly grounded operator has no charge on his skin. However, arm hair has been re• ported to develop up to 900 volts, even at 50 percent relative humidity, when skirt sleeves are rolled up; at least one major corporation considers this a serious static problem. 18 Note that in our worst-case tests as little as 500 volts (on vinyl plastic, Table 4) was sufficient to damage a MOSFET. Obviously, head hair which dangles and might touch ESDS items is also a hazard. A synthetic-fiber smock can carry damaging fields, as seen in Table 4, and DoD-HDBK- 263 makes the point that clothing should never touch ESDS items. An "antistatic" smock is at least a partial solution, but it should not contain stainless-steel fibers which might fall onto circuit boards and cause shorts. 19 The operator's chair, if nonconductive, is also a hazard (Table 4). The general approach to the problem of unavoidable static fields on an operator's hair or clothing-or on anything in the work station- is for him or her not to bring ESDS items near charged surfaces unnecessarily. In other words, do not look for trouble. In addition, head hair can be tied back, smock sleeves rolled up (with the proviso that ESDS items are kept away from hair on the forearms), etc.

ESD-Protective Equipment and Materials

Our discussion will be limited to two items: wrist straps and bags.

Wrist Straps Various designs were tested. Bead chains gave intermittent contact but succeeded in pro• tecting MOSFETs when the operator charged himself by shuffling his feet on a carpet and touched the lead of the MOSFET board to ground. The main objection to bead chains is their tendency to ride up over sleeves and lose contact with the skin. Any design which presses grounded metal snugly against the skin seems adequate; this includes expanding stainless-steel watchband types or designs in which a metal "wristwatch" element is held against the skin by an elastic band. However, wrist straps which depend on the conductivity of carbon-loaded plastic are sUSpect,20 and a conductive fabric band shed steel fibers up to 0.09 inch long in our tests.

Faraday-Cage Bags It is well recognized that ESD-protective bags must guard against ( I) internal triboelectric charging and (2) external static fields/discharges. 9 The bags in Tables 9 and 10 all provide some shielding against external fields. In terms of the pulse attenuation test (Table 9), 93 percent attenuation must be sufficient because the foil-containing laminate bag is an excellent Faraday cage. 21 PAPER NO.3 195

Table 9. Shleldlng/Dlscharge Test.

PERCENT PULSE BA01,21 A'ITENUAll0N

Antistatic Polyethylene 75 Carbon-Loaded Polyolefin (Conductive) 97 With Conductive Ink Grid, A 78 With Conductive Ink Grid, B 50 Partially Transparent with External Metallization 99,8 Same, but Handled (Metallization Cracked on Crease) 59 Partially Transparent with "Buried" Metallization, A 39 Partially Transparent with "Buried" Metallization, B 87 Laminate Containing Aluminum Foil, with Antistatic 93 Polyethylene Lining Antistatic Polyethylene Bags Inside of and Outside of Carbon-Loaded 92 Polyolefin Bag Antistatic Polyethylene Bag Outside of Partially Transparent Bag with 93 External Metallization

The high values given by a partially transparent, externally metallized bag (99.7 percent) and a volume-conductive bag (97 percent) are artifacts, in a way, because conductive mate• rial "shorts out" the upper and lower plates in the test; wrapping these bags in antistatic polyethylene lowered their pulse attenuation to the 92-93 percent range. By this test it would seem that carbon-loaded polyolefin is as good a Faraday cage as aluminum foil, but such is not the case; discharges from a person's finger can damage ESDS items inside conductive plastic bags.3,21 A more significant test is the device-in-bag procedure of Table 10. The conditions were certainly worst-case because a very low resistance (150 ohms) was used; a resistance of

Table 10. Modified Shleldlng/Dlscharge Test.

MOSFET BAG (AS IN TABLE 9) VOLTAGE DAMAGE

Carbon-Loaded Polyethylene (Conductive) +5000 3/3 (2 shorted) With Conductive Ink Grid, A +5000 1/3 (l shorted) Partially Transparent with External Metallization +5000 3/3 (3 shorted) Partially Transparent with "Buried" Metallization, A +5000 1/3 (1 shorted) Laminated Containing Aluminum Foil, with Antistatic +5000 0.5 Polyethylene Lining +10,000 0.5 Partially Transparent with External Metallization ** -8000 (est)'" 2.3 (0 shorted) Same, but Handled**· -8000 (est)· 2/3 (2 shorted) Partially Transparent with "Buried" Metallization, A -8000 (est)· 3/3 (0 shorted)

*The EDS-200 probe was replaced by the fmger of it person triboelectrically charged by walking on a carpet at 24% relative humidity and 73°F. **Surface resistivity between sides = 256 ohms/square at 0.2 volt. ***Surface resistivity between sides = 2 x 10" ohms/square at I 00 volts. 196 APPENDIX

1500 ohms is specified in the human ESD model of DoD-HDBK-263, and a realistic esti• mate for human resistance is said to be 350 kilohms.6 However, the foil-containing laminate bag survived even this test. MOSFETs in a partially transparent, externally metallized bag were damaged by a dis• charge from the finger of a statically charged person (Table 10), and the spark burned off a small area of metal. Published data6 seem to disagree, but conditions differed somewhat from ours; for one thing, the capacitive probe had no leads as ours did. Also, the bag may not have been pulled tightly over the upper plate of the probe as it was on our test. As expected, MOSFET damage was worse (Table 10) when the externally metallized bag had a low-conductivity crease caused by cracking of the metallization by handling. Also note that externally metallized bags, being good conductors, can participate in discharges at the work station (Table 2). The conclusion, which we had reached earlier,21 is that only foil-containing laminate bags provide absolute protection against worst-case fields and discharges. A suitable com• mercial product is a three layer laminate with aluminum foil (0.00035 inch thick) sand• wiched between antistat-treated spun-bonded polyethylene on the outside and antistatic polyethylene on the inside. Note that Mll..-B-81705, Type I, is a foil-containing laminate but has an unsatisfactory liner which does not prevent internal triboelectric charging.4•9

Equipment/Material Selection and Handling Techniques

One school of thought recommends exclusion of conductive materials 13 while another, the "conductive approach," visualizes most objects in the work station as conductive, grounded, and therefore at zero potential. 22 Either approach, or a mixture of the two approaches, will be effective with proper maintenance of materials (e.g., periodic treatment of antistatic surfaces with topical antistat) and handling techniques which accommodate the limitations of the equipment/materials. Table 11 illustrates the principle of complementary materials and techniques. The fact is that ESD-protective materials must be used correctly to be effective. 23 In other words, operators are more important than equipment. For example, a grounded con• ductive chair would be more dangerous to ESDS items then an ordinary chair if a careless operator without a wrist strap caused discharges by brushing device leads against the uphol• stery. Conversely, a very skilled operator can protect ESDS items with only the bare essen• tials of equipment, including a wrist strap, because he knows where the dangers lie. Therefore, operator training is essential for ESD control programs as prescribed in DoD-STD-1686. A wide range of equipment/materials is described in DoD-HDBK-263, and it is up to the user to integrate these with operator disciplines in a unified approach. In our program, handling techniques accommodate a mixture of conductive, static-dissipative, and antistatic materials. Some basic rules are: (1) Never touch ESDS device leads with the fingers unnecessarily, and keep ESDS items away from all surfaces except the bench top and the assembly to which the item is being attached. (2) Keep ESDS items in Faraday , such as foil-containing laminate bags, when away from the static• safe work station. (3) Clear the immediate work area of all but essential objects. A tote box might fail to drain completely and bear a charge-but if kept to one side it will be harmless. (4) Be sure to ground conductors, such as soldering iron tips, which touch leads or contacts and can discharge directly into ESDS devices. (5) Recognize the limitations of supplementary PAPER NO.3 197

Table 11. Complementary EqulpmentlMaterlals and Techniques.

ESDHAZARD EQUIPMENT/MATERIAL TECHNIQUES

Bench top Static-dissipative laminate Avoid unnecessarily touching any surface, even this, with leads of ESOS items. Ordinary plastics Ionization equipment, Identify and exclude if possible; topical antistat treat with antistat; use local or room ionization. Chair If possible, use conductive Don't closely approach with chair on grounded ESOS items. conductive surface. Tote box Conductive Be aware of possible sloughing. Tote box Antistatic Periodically treat with antistat solution. Tote box Either type Ensure grounding when on work surface; avoid touching with ESOS items; place ESOS items in Faraday cage, such as foil-containing laminate bag, before adding to box; keep box to one side when working on ESO items. Operator's skin Wrist strap Always use strap and check its resistance often. Operator's arm hair and None Tie head hair back; don't head hair closely approach hair with ESOS items. Operator's clothing Antistatic smocks Don't closely approach with ESOS items. Unavoidable static fields Field meter Avoid in accordance with (eg, from electrical Figure 3. equipment) Static fields and Foil-containing laminate Never remove ESOS items discharges away bags or other Faraday from Faraday cage except from work station cages at static-safe work station.

protective measures such as humidification or ionization and do not be overly dependent on them. Humidification merely lessens ESD hazards, and ionized air requires time to neutralize charges or cannot neutralize them at all insofar as the static field is collapsed by proximity to a ground plane. (6) Check wrist straps frequently. Various resistance testers are on the market. (7) Check with a field meter for static fields in the work station. For example, a nearby window might generate a high enough charge from dry, dusty wind blowing over it to create a dangerous field in the work area. If fields are unavoidable, apply the equation d = #11.8 as explained in the text above. (8) Be conscious that any assembly operation, such as peeling tape from a roll, creates static charges. Deal with these as required; for example, pull masking tape off the roll slowly and "wash" it in ionized air before applying it to contacts on a circuit board. (9) Remember that you, the operator, are a primary hazard even though grounded because you can participate in discharges. (10) Be scrupulous in excluding common plastics, which may be personal items such as sandwich bags, from the 198 APPENDIX work station. As a last resort, ordinary plastics which definitely cannot be excluded must be treated with topical antis tat. In general, equipment and materials must be selected carefully from the gamut of prod• ucts available, and then handling techniques must be adapted to the products chosen.

Conclusions

1. Tests confirmed that static-dissipative bench tops are preferred to conductive sur• faces not only for personnel safety reasons but because a slower discharge rate can reduce damage to ESDS items. 2. A charged conductive tote box is only slightly more hazardous than a charged anti• static box when an ESDS item enters the static field of the box. But when boxes rest on a grounded static-dissipative work surface, a conductive box participates much more readily in harmful discharges than does an antistatic box. Both types of boxes have their pros and cons, and either is suitable if used properly. 3. Statically charged nonconductive plastics can present a field hazard when the appar• ent charge is as little as 500 volts. However, a charge of as much as 13,000 volts is not deliverable to an ESDS item by contact (discharge) when the field is suppressed to 300 volts apparent charge by proximity to a ground plane. 4. Under worst-case conditions, using a MOSFET with long "antennas," a rough equa• tion for the minimum safe distance d (in inches) from a surface charged to voltage V is: d = .[V/I.8. When the antennas are shielded, as they often are in practice on circuit-board modules, the constant in this equation rises from 1.8 to at least 7. 5. Based on the above data, basic rules are: (1) Ground all conductors so that none can have a charge of over 50 volts. (2) Check with a field meter and allow no more than 300 volts apparent charge on nonconductors such as ordinary plastics. (3) If static fields are unavoidable, keep ESDS items d inches away from charged surfaces according to the worst• case equation d = .[V/I.8. 6. Electrical air-ionizing blowers or nozzles were found not to present a field hazard if ESDS items are kept at least 10 inches away from the corona-discharge points. For close work, nuclear-type air nozzles are indicated. 7. A pulsed dc benchtop ionizer, with no fan, was found to be as effective as conven• tional ac blowers. The lack of an airstream is advantageous. (Note: Slow-pulsed ionizers later proved hazardous; see "Ionization" in Chapter 2.) 8. A room ionization system with dc emitters proved to be effective-about equal to a typical bench-model blower at 5 feet- and relatively uniform regardless of the position of charged objects on the bench surface under the emitters. At worst-case conditions (emitter of only one polarity operating), an isolated conductor at bench level (capacitance 129 pF) was not charged sufficiently to damage MOSFETs. 9. The possibility of a conductive masking tape of foillfiber construction is suggested. The aluminum foil prevents a static field by charge drainage and voltage suppression. 10. A corona-discharge point mounted above a small-scale grit blaster reduced the charge developed on circuit-board material from 700 volts to the harmless level of 60 volts. II. A bead-chain wrist strap is effective but tends to ride up over sleeves. Snug-fitting straps in which metal contacts the skin are recommended. 12. In out tests, MOSFETs were damaged in partially transparent metallized bags, under worst-case conditions, when the bags were touched by an operator charged to an estimated PAPER NO.3 199

8000 volts by walking on a carpet. In contrast, MOSFETs in a foil-containing laminate bag were unaffected by a lO,OOO-volt discharge from a simulator with human-level capacitance but three orders of magnitude less resistance than a person. Only foil-containing laminate bags are recommended for Faraday-cage protection against worst-case fields and discharges. 13. A synthesis of complementary equipment/materials and handling techniques is necessary for an effective ESD control program in accordance with Dod-HDBK-1686. Examples are given. 14. Operators are more important than materials. Some basic operator disciplines are, in brief: (I) keep ESDS items away from all surfaces except the bench top and the assembly to which the item is being attached, (2) use Faraday-cage bags/, (3) clear the imme• diate work area of all but essential objects, (4) beware of charged conductors, (5) do not be overly dependent on supplementary protection such as humidification or ionization, (6) check wrist straps frequently, (7) check for static fields with a meter, (8) beware of static charges caused by assembly operations, (9) remember that a grounded operator can partici• pate in harmful discharges, and (10) rigorously exclude common plastics.

(Note: Later research modified some of the above conclusions, as seen in the text of this book. For example, we became more cautious about ionization.)

References

I. I. M. Kolyer and W. E. Anderson, "Pennanence of the Antistatic Property of Commercial Anti• static Bags and Tote Boxes," Reliability Analysis Center EOSIESD Symposium Proceedings, pp. 87-94, Las Vegas, NY, 1983. 2. B. A. Unger, "Electrostatic Discharge Failures of Semiconductor Devices," Reliability Physics Symposium, April 1981. 3. I. R. Huntsman, D. M. Yenni, Jr., and O. E. Mueller, "Fundamental Requirements for Static Protective Containers," presented at 1980 NEPCON/West Conference, Anaheim, CA. 4. D. M. Yenni, Ir., and I. R. Huntsman, "The Deficiencies in Military Specification MIL-B-81705: Considerations and a Simple Model for Static Protection," presented at the Reliability Analysis Center EOS/ESD Symposium, Denver, CO, 1979. 5. "Latent ESD Failures: A Reality," Evaluation Engineering Magazine, p. 80, April 1982. 6. I. R. Huntsman and D. M. Yenni, Ir., "Test Methods for Static Control Products," Reliability Analysis Center EOSIESD Symposium Proceedings, pp. 94-109, Orlando, FL, 1982. 7. Electronics Industry Association Interim Standard No.5, January 1983. 8. "Electromagnetic Shielding Effectiveness," technical infonnation sheet from the Bemis Co., Inc., ESD Protective Materials Dept., undated, received 1983. 9. N. B. Fuqua and R. C. Walker, "ESD Controls Study, Final Report," Reliability Analysis Center No. 01115-30-2, September 1981. 10. B. Unger, R. Chemelli, P. Bossard, and M. Hudock, "Evaluation of Integrated Circuit Shipping Tubes," Reliability Analysis Center EOSIESD Symposium Proceedings, pp. 57-64, Las Vegas, NV, 1981. II. J. R. Huntsman and D. M. Yenni, Jr., "Charge Drainage vs. Voltage Suppression by Static Con• trol Table Tops," Evaluation Engineering Magazine, March 1982. 12. I. R. Mileham and N. I. Safeer, "Selection of Static Eliminating Decorative Table Top Mats or Laminates," presented at 1984 NEPCON/West Conference, Anaheim, CA. The material we tested is No.6 in the tables of this reference. 13. D. C. Anderson, "ESD Control: To Prevent the Spark that Kills," Evaluation Engineering Maga• zine, pp. 120-131, July 1984. 200 APPENDIX

14. R. G. Chemelli, B. A. Unger, and P. R. Bossard, "ESD by Static Induction," Reliability Analysis Center EOSIESD Symposium Proceedings, pp. 29-35, Las Vegas, NY, 1983. 15. J. E. Berry, "Static Control in Bare Board Testing," Electronic Packaging and Production, pp. 161-162, August 1981. 16. C. F. Mykkanen and D. R. Blinde, "The Room Ionization System: An Alternative to 40 Percent RH," Evaluation Engineering Magazine, pp. 76-88, September 1983. 17. "Electrostatic Discharge Training Manual," NAVSEA SE 003-AA-TRN-0IO, Published by Di• rection of Commander, Naval Sea Systems Command. 18. "Arm Hair Pin-Pointed as New ESD Hazard," Evaluation Engineering Magazine, p. 70, May 1984. 19. GIDEP (Government-Industry Exchange Program) Alert No. D5-A-84-OI, May 21,1984. 20. GIDEP (Government-Industry Exchange Program) Alert No. MX-A-82-02, March 21, 1983. 21. J. M. Kolyer and W. E. Anderson, " Selection of Packaging Materials for Electrostatic Dis• charge-Sensitive Items," Reliability Analysis Center EOSIESD Symposium Proceedings, pp. 75-84, Las Vegas, NY, 1981. 22. R. Euker, "ESD in Integrated Circuit Assembly," Static Digest, published by Static Control Systems Division, 3M Co., July 1983. 23. G. E. Hansel, "The Production Operator: Weak Link or Warrior in the ESD Battle?" Reliability Analysis Center EOSIESD Symposium Proceedings, pp. 12-16, Las Vegas, NY, 1983. PaperNo. 4

Presented at the 7th Annual Electrical Overstress/Electrostatic Discharge Symposium, Minneapolis, Minnesota, September 10-12,1985, Sponsored by EOS/ESD Association and ITI Research Institute, EOSIESD Symposium Proceedings, EOS-7, 1985, page 111.

PERFORATED FOIL BAGS: PARTIAL TRANSPARENCY AND EXCELLENT ESD PROTECflON

John M. Kolyer and William E. Anderson

Rockwell International Corporation Autonetics Strategic Systems Division Electronics Operations 3370 Miraloma Avenue Anaheim, CA 92803-4192

Note: The following is a condensed version of this paper. See the Symnposium Proceedings for the complete text.

Abstract

Commercial foil-containing laminate bags provide "Faraday-cage" protection for electro• static discharge (ESD)-sensitive items but are opaque. However, foil may be perforated to give see-through capability if the metal is thick enough and the stamped holes have out• wardly directed burrs which may act as "lightning rods" to intercept discharges. A wire screen serves the same purpose, with its high points presumably receiving the discharges. This principle is applicable to various other containers, e.g., tote boxes. Thus, partial trans• parency need not mean a sacrifice in ESD protection.

Introduction

Commercial "Faraday-cage" bags are intended to protect against electrostatic fields and discharges. A secondary goal is protection against EMI (electromagnetic interference), spe• cifically at 1-10 GHz per MIL-B-81705, Type I. These bags fall into four classes: (1) foil-containing laminates which give good ESD and EMI protection but are completely opaque, (2) metallized bags in which the metallized layer is thin enough to allow partial transparency with a considerable sacrifice of ESD and EMI protection (also, external met• allization can crack on handling, thus lowering shielding effectiveness), (3) bags with a grid of conductive ink that are fairly transparent but give poor ESD protection, and (4) metal• fiber-containing bags that are also fairly transparent but give poor ESD protection. Therefore, our objective was a new bag laminate which combines the "best of both worlds" to a considerable extent: see-through capability along with good ESD protection and EMI protection in the 1-10 GHz (radar) range.

201 202 APPENDIX

Materials Tested

Nine commercial ESD-protective bag materials were tested (Table 1): antistatic poly, carbon-loaded, ink grid, buried metallization A, external nickel, copper fibers, buried met• allization B, buried metallization C, and foil laminate. Several of these products have been described and characterized by us. 1.2,3 The other six materials in Table 1 were fabricated in the laboratory from antistatic polyethylene (6 mils) and either aluminum foil, perforated aluminum foil, or metallic screens.

Experimental Methods

The test conditions are listed in Table 2 and diagrammed in Figs. 1,2, and 3.

Table 1. Bag Materials Tested.

THICKNESS TRANSPARENCY • DESIGNATION MILS % DESCRIPTION

Antistatic Poly 6 87 Antistatic Polyethylene (MIL-B-81705, Type ll) Carbon-Loaded 4 0 Conductive, Carbon-Loaded Polyethylene Ink Grid 4 68 Laminate with Buried Grid of Conductive Ink and Antistatic Liner Buried Metal- 3.5 74 Polyester + Vapor-Deposited Metal + Antistatic lization A Polyethylene Liner External Nickel 2.5 35 Vapor Deposited Nickel + Polyester + Antistatic Polyethylene Liner Copper Fibers 3.7 78 Fine Copper Wires Dispersed in Polyethylene Buried Metal- 3.3 60 Resinous Protective Coating + Vapor-Deposited lization B Aluminum + Carrier Film + Antistatic Polyethylene Liner 2-mil Perforated 12 38 Perforated Foil Encapsulated in Antistatic Foil Polyethylene Buried Metal- 3 60 Polyester + Vapor-Deposited Aluminum + lization C Antistatic Polyethylene Liner 5-mil Perforated 15 38 Perforated Foil (See Text) Encapsulated in Foil Antistatic Polyethylene Foil Laminate 10 0 Antistic Spun-Bonded Polyethylene + 0.35-mil Aluminum Foil + Antistatic Polyethylene Liner 1.4-mil Foil 7.4 0 1.4-mil Aluminum Foil + 6-mil Antistatic Polyethylene Liner 5.6-mil Foil 11.6 0 Four layers of 1.4-mil Aluminum Foil + 6-mil Antistatic Polyethylene Liner Aluminum Screen 23 59 Aluminum Screen (See Text) Sandwiched Be- tween Sheets of 6-mil Antistatic Polyethylene Copper Screen 35 41 Copper Screen (See Text) Sandwiched Between Sheets of 6-mil Antistatic Polyethylene Table 2. ShieldingIDischarge Test Conditions.

TEST NUMBER 2 3 4 5 6 7 8

Capacitive No No Yes Yes Yes Yes Yes Yes Sensor Volts +25,000 About 35,000 +5000 -6000 +10,000 About 35,000 +24,000 About 35,000 to-8000 Resistance, 0 Tesla Coil 150 Human 150 Tesla Coil 150 Tesla Coil ohms Capacitance, 167 Tesla Coil 150 Human 185 TeslaCoil 185 Tesla Coil picofarads Time, seconds (Discharge) 30 (Discharge) (Discharge) (Discharge) 30 (Discharge) 30 Surface Nonconductive Nonconductive Grounded Grounded Grounded Nonconductive Grounded Grounded Under Bag a~ ~ ....

fA 204 APPENDIX

ROHCOlIDUCTlVE SURFACE

Fig. I. Tests number 1 and 2 (Table 2).

us ISTOI.

PR>BI (TISTS 3.5.7)

Fig. 2. Tests number 3, 4, 5, 7, and 8 (Table 2).

HONCONDUCTlVE SURFACE

Fig. 3. Test number 6 (Table 2). PAPER NO.4 205

Table 3. Shielding/Discharge Data Summary.

Bag Material

Antlatatlc Poly F

Carbon·Loaded F Ink Grid (F) F Burled Metallization A (F) f F

External Nickel F F

Copper Fiber. P Burled Metallization B P F f

2·mll Perforated P (F) f F foil

Burled Metallization C P P (F) F F F F 5-mll Perforated p p p p , F foil

Foil laminate p p p p (F) f 1.... mll foil p (F) 5.8·mll Foil p P Aluminum Screen p p

Copper Screen p p

Rating

Lagend: P = pealed (no MOSFETI damaged) (F) = barely failed /115(081') F .. failed (1115(18) or fa lure ratio> 115)

Results

The data are summarized in Table 3. Note the excellent results given by screens. After publication of this paper a screen bag was commercialized; the aluminum screen was sand• wiched between layers of antistatic polyethylene. Perforated foil was not as effective as screen and just as expensive. so it was never commercialized.

Conclusions

1. Metallized. carbon-loaded. or ink grid bags have been reported not to withstand a direct static charge above 25 kV. Our own tests indicate a much lower threshold for these materials. Note that personnel walking on a carpet at 1~20 percent relative humidity gen• erate typically 35 kV. 2. Two independent groups of investigators have recommended (EOS/ESD Sympo• sium. 1984) that highly ESD-sensitive items should always be protected by metal-foil bags. 206 APPENDIX

3. We have now found that metal foil can be perforated and still shield against ESD and radar. Wire screen is even better. Outwardly directed burrs on the foil, or prominences on the screen, may act as "lightning rods" to intercept discharges. 4. The effectiveness of perforated foil or screen laminates as a moisture vapor barrier would depend on the permeability of the used. 5. Preliminary designs with performed foil or wire screen would cost twice as much as a commercial foil laminate. However, added expense might be justified in protecting valu• able, high-reliability electronics.

References 1. J. M. Kolyer and W. E. Anderson, "Selection of Packaging Materials for Electrostatic Discharge• Sensitive Items," Reliability Analysis Center EOSIESD Symposium Proceedings, pp. 75-84, Las Vegas, NV, 1981. 2. J. M. Kolyer and W. E. Anderson, "Permanence of the Antistatic Property of Commercial Anti• static Bags and Toite Boxes," Reliability Analysis Center EOSIESD Symposium Proceedings, pp. 87-94, Las Vegas, NV, 1983. 3. J. M. Kolyer. W. E. Anderson. and D. E. Watson. "Hazards of Static Charges and Fields at the Work Station." Reliability Analysis Center EOSIESD Symposium Proceedings. pp. 7-19. Phila• delphia. PA, 1984. PaperNo. 5

Presented at the Electrical Overstress Exhibition, Anaheim, California, January 21,1986. Proceedings of the Technical Program, 1986 Electrical Overstress Exhibition, page 34.

COST-EFFECTIVE METIIODS OF TESTING/MONITORING WRIST STRAPS

John M. Kolyer and Donald E. Watson

Rockwell International Corporation Autonetics Strategic Systems Division Electronics Operations 3370 Miraloma Avenue Anaheim, CA 92803

Abstract

An ungrounded operator is extremely dangerous to ESD (electrostatic discharge)-sensitive devices as illustrated by a test at high relative humidity (54 percent) in which the operator damaged transistors by rolling his chair. An ESD-protective smock did not help to protect devices. A maximum resistance from operator to ground of 10 megohms is recommended, and this value was exceeded in 3 of 36 tests under manufacturing conditions. Because of flexing of the cords and other abuse, wrist straps have a limited lifetime, but they are too expensive to scrap in a replacement cycle and instead must be monitored. Periodic monitor• ing is unsatisfactory because ESD-sensitive items will be exposed to an ungrounded opera• tor for some period of time, even if only an hour, when a strap fails. Then material review action might be required on all hardware handled since the last successful test of that strap. Therefore, continuous monitors may be "the wave of the future." Commercial continuous monitors work on the capacitance principle, but another method is a continuity check across the operator's skin, up and down a two-conductor cord, and in and out two separate ground connections. Prototype monitors of this two-conductor design are being tested in a produc• tion environment. High-cost items, e.g. $1000 modules, especially deserve continuous monitoring.

Introduction

This paper is presented in the form of slides shown at the Electrical Overstress Exposition (EOE) in Anaheim, CA, January 21,1986. Accompanying each slide is additional discus• sion and clarification as required. Wrist straps have a limited lifetime I because of frequent flexing and other abuse. Since a time-based replacement cycle would scrap many good straps and be wasteful, monitoring is necessary. This monitoring is preferably continuous, because the failure of a wrist strap in a

207 208 APPENDIX periodic check raises the questions: "How long ago did this strap fail?" and "Have products been damaged in the meantime?" As shown in the following outline, the danger of a failed strap was demonstrated, simple methods for testing resistance were devised, straps in an assembly plant were checked, and means of monitoring were investigated.

OUTLINE

• DAMAGE CAUSED BY LACK OF WRIST STRAP • RESISTANCE VS CHARGE ON OPERATOR • ALLOWABLE RESISTANCE, OPERATOR TO GROUND • RESISTANCE TEST METHODS • THROUGH STRAP • OPERATOR TO GROUND • EQUIPMENT • MEGOHMMETER • ZAPFLASH • RESISTANCE TESTING DATA • FACTORS IN RESISTANCE, OPERATOR TO GROUND • MONITORING • PERIODIC VS CONTINUOUS • CONTINUOUS DESIGNS • CAPACITANCE TYPE • TWO-CONDUCTOR TYPE • ADVANTAGES OF TWQ..CONDUCTOR TYPE • CONCLUSIONS

The charge on the operator was read by means of a Trek Model 512 field meter aimed at an electrically isolated I-ft2 aluminum plate to which the operator was attached by a wrist strap. The temperature was nOE The transistor was a Motorola 2N4351 MOSFET. The testing procedure and damage criteria have been described.2 PAPER NO.5 209

DAMAGE CAUSED BY LACK OF WRIST STRAP

• GROUNDED OPERATOR IS LESSER OF TWO EVILS VS UNGROUNDED OPERATOR (KOLYER ET AL, EOSIESD SYMPOSIUM 1984)

• UNGROUNDED OPERATOR IN ESC-PROTECTIVE SMOCK, 54% RELATIVE HUMIDITY

• ROLLED CHAIR: 500 VOLTS, TRANSISTORS DAMAGED

• SHUFFLED FEET: 1000 VOLTS, TRANSISTORS DAMAGED

A charged operator is a very great danger because he can damage ESD-sensitive devices by direct discharge or "injection" rather than indirectly by induction from a static field. The published data3 were obtained with an operator holding a probe attached to an elec• trometer and an oscilloscope. The Trek meter and l_ft2 plate were used as described under the preceding slide. Since 10 volts was our limit of detection. the result at I megohm agrees with that obtained by the more sophisticated method: a I-megohm resistor in the wrist strap allows negligible voltage on the operator.

RESISTANCE VS CHARGE ON OPERATOR

RESISTANCE, OPERATOR PEAK VOLTAGE TO GROUND, MEGOHMS ON OPERATOR REFERENCE <2 10 11 50 30 100 80 INFINITE (NO STRAP) 1550 1 <10 2 INFINITE (NO STRAP) 1100 2

1. HUNTSMAN AND YENNI, EOSIESD SYMPOSIUM 1982 2. OUR DATA 210 APPENDIX

ALLOWABLE RESISTANCE, OPERATOR TO GROUND

• FOR DEVIDCES SENSITIVE TO 50 VOLTS MIN: 50 MEGOHMS

• FOR DEVICES SENSITIVE TO 20 VOLTS MIN: 10 MEGOHMS

• CONSERVATIVE LIMIT FOR ALL DEVICES: 10 MEGOHMS

These thresholds are based on the data of the preceding slide. A charge of 11 volts on the operator is hannless to most devices at the present time; in the future this allowable resistance might have to be lowered. A I-megohm resistor is presently standard for wrist straps. As an added precaution, a static-limiting floor finish (UL288, Hanson-Loran Chemical Company) was used. The chair-rolling or foot-shuffling operator in the slide before last would have generated little charge with UL288. In the test, a conventional acrylic finish was on the tile floor. A Beckman L-lO megohmeter at 10 volts was used. For quick tests, a Zapflash (see two slides below) is suitable.

RESISTANCE TEST METHOD, THROUGH STRAP ALLIGATOR CUPS PAPER NO.5 211

RESISTANCE TEST METHOD, OPERATOR TO GROUND

The same megohmeter at 10 volts was used. The metal can (tin-coated steel) is a conve• nient, hand-filling probe. The Zapflash was tested with various resistance levels. At 1.0 megohm the light (indicat• ing continuity) was very bright. At 3.0 megohms the light was relatively bright, but at 5.2 megohms it was dim and at 6.2 megohms it was barely visible.

RESISTANCE TEST METHODS, EQUIPMENT

• MEGOHMMETER AT 10 VOLTS, EG. BECKMAN MODEL L-10

• ZAPFLASH (ANDERSON EFFECTS). HIGH·RESISTANCE CONTINUITY TESTER. LIGHTS AT ABOUT 6 MEGOHMS MAX. GOOD FOR QUICK CHECKS

Various combinations of wrist straps and operators gave a total of 36 tests. These mea• surements were made in a manufacturing situation. One operator had spliced the cord of his bead chain, thus increasing its resistance from 1.0 to 1.8 megohms; this is not a serious increase, but it might have been. Another operator was pleased with her loose-fitting expan• sion band; resistance from her skin to ground was 30 megohms for this reason. 212 APPENDIX

RESISTANCE TEST DATA

RES~TANCE,MEGOHMS TYPE NUMBER NUMBER THROUGH OPERATOR TO STRAP SAMPLES OPERATORS STRAP GROUND EXPANSION 5 7 1.00-1.05 1.0-9.0 BAND (1) EXPANSION 5 8 1.00-1.05 1.05-30 BAND (2) EXPANSION 2 1.0 2.0,4.3 BAND (3) CLOTH BAND 3 4 2.7-110 1.5-65 (CONDUCTIVE) BEAD CHAIN 4 8 1.0-1.8· 1.5-40

·SPLICED CORD GAVE 1.8

These examples illustrate the need for operator discipline, especially in the absence of con• tinuous monitoring of the straps. The three out-of-spec cases occurred in a program of periodic (weekly) checks and illus• trate the desirability of continuous monitoring. In a periodic check, the operator may slide the band up his arm to tighten it or make some other adjustment to pass the test. An analogy is annual inspection of automobile exhaust emissions; once the driver has passed inspec• tion, he might readjust his engine to exceed the allowable level of smog-producing gases.

FACTORS IN RESISTANCE, OPERATOR TO GROUND

• OVERSIZED EXPANSION BAND

• INCREASED RESISTANCE OF CLOTH BAND (SOILING)

• STRETCHING OF CLOTH BAND, CAUSING LOOSENESS

• SOILING OF BEAD CHAIN

• DRYSKIN

• IN 36 TESTS, RESISTANCE EXCEEDED 10 MEGOHMS IN 3 CASES: EXPANSION BAND (LOOSE FIT), CONDUCnVE CLOTH BAND, BEAD CHAIN PAPER NO.5 213

PERIODIC VS CONTINUOUS MONITORING

• PERIODIC

• ADVANTAGE: LOW EQUIPMENT COST

• DISADVANTAGE: ESD-SENSmVE DEVICES ENDANGERED FOR SOME nME INTERVAL, EVEN IF ONLY AN HOUR

• CONnNUOUS

• ADVANTAGES: DEVICES ALWAYS PROTECTED, WRIST STRAP QUALITY SECONDARY, LESS OPERATOR DISCIPLINE NEEDED

• DISADVANTAGE: RELAnVE HIGH EQUIPMENT COST (BUT COST OF LABOR FOR PERIODIC MONITORING MUST BE SUBTRACTED)

CONTINUOUS MONITOR, CAPACITANCE TYPE

• CHECKS CAPACITANCE, EG, 150 pF

• ALARM FLASHES AT ALLOWABLE MAX OPERATOR-TO-GROUND RESISTANCE, EG, 3.5 MEGOHMS TOTAL

• USES CONVENnONAL WRIST STRAP

• COMMERCIALLY AVAILABLE FROM MORE THAN ONE SUPPLIER

Commercial products include the Simco Wrist Strap Monitor Model M50A, the Ground Gard from Static Prevention, Incorporated, and the GM-ICT from Westek. Charleswater Products has a developmental product to be introduced early in 1986. 214 APPENDIX

Note that in the absence of continuous monitoring, all hardware handled since the last successful test of a failed strap might be subject to a material review action.

CONTINUOUS MONITOR, CAPACITANCE TYPE: DIAGRAM

1·MEGOHM RESISTOR

This is a very simplified representation of the Simco Wrist Sgrap Monitor Model M50A.

CONTINUOUS MONITOR, TWO-CONDUCTOR TYPE

• CHECKS LOOP ACROSS SKIN, INTO FIRST GROUND CONNECTION, AND OUT SECOND GROUND CONNECTION

• ALARM FLASHES AT 20 MEGOHMS, OPERATOR·TO-GROUND

• SPECIAL SPLIT·CONDUCTOR WRIST STRAP NEEDED

• PROTOTYPES TO BE TESTED IN PLANT PAPER NO.5 215

The product under evaluation is from Semtronics Corporation. Westek is developing their own model.

CONTINUOUS MONITOR, TWO-CONDUCTOR TYPE: DIAGRAM

", T ...... ------_/

This is a very simplified representation of the Semtronics prototype. The commercial system, Part No. EN435S, will comprise a Sentinel monitor, a dual coil cord, and a wrist strap.

ADVANTAGES OF TWO-CONDUCTOR TYPE CONTINUOUS MONITOR

• FAILURE PROBABLY IN ONE CONDUCTOR, NOT BOTH, SO ESD-SENSITIVE ITEMS NEVER JEOPARDIZED

• CHECKS GROUND CONNECTIONS AS WELL AS OPERATOR

• NOT FOOLED BY HIGH-CAPACITANCE OBJECT TOUCHING STRAP

• WHEN BOTH CONDUCTORS ARE UNPLUGGED FROM BOX, MECHANICAL SWITCH SHUTS OFF ALARM, EG, WHEN OPERATOR TAKES BREAK 216 APPENDIX

At least one maker of the capacitance-type monitor expects to add a feature to check ground connections. Also, the problem of the strap being deceived by a high-capacitance object may possibly be overcome. With these improvements, the capacitance-type monitor would more nearly equal the two-conductor type, so that a choice between the two might largely depend on cost. Cost effectiveness is not easy to calculate at this point, since the costs of periodic inspec• tion and of hardware damaged due to lack of continuous monitoring must enter the equa• tion. In general, periodic checks may be adequate for low-cost products, but continuous monitoring is indicated for expensive assemblies. After all, most of us consider the wrist strap our first line of defense, and if it fails we have dropped our guard, however briefly, in the absence of continuous monitoring.

CONCLUSIONS

• WRIST STRAP FAILURE CAN EASILY CAUSE DAMAGE OF ESD-SENSmVE DEVICES

• RESISTANCE TEST METHODS ARE DESCRIBED

• 3 FAILURES (RESISTANCE OVER 10 MEGOHMS) IN 36 CHECKS OF STRAPS IN PLANT

• LOOSE BANDS AND DRY SKIN RAISE RESISTANCE

• CONTINUOUS MONITORING PREFERRED

• TWO-CONDUCTOR MONITOR PREFERRED OVER CAPACITANCE TYPE

• TWO-CONDUCTOR MONITOR PROTOTYPES TO BE TESTED IN PLANT

• HIGH-COST ITEMS, EG, $1000 MODULES, DESERVE CONTINUOUS MONITORING PAPER NO.5 217

References 1. A. P. Hahl, "A Wrist Strap Life Test Program," Reliability Analysis Center EOSIESD Sympo• sium Proceedings, pp. 94-96, Philadelphia, PA, 1984. 2. J. M. Kolyer, W. E. Anderson, and D. E. Watson, "Hazards of Static Charges and Fields at the Work Station," Reliability Analysis Center EOSIESD Symposium Proceedings, pp. 7-19, Phila• delphia, PA, 1984. 3. J. R. Huntsman and D. M. Yenni, Jr., ''Test Methods for Static Control Products," Reliability Analysis Center EOSIESD Symposium Proceedings, pp. 94-109, Orlando, FL, 1982. PaperNo. 6

Presented at the 8th Annual Electrical Overstress/Electrostatic Discharge Symposium. Las Vegas. Nevada. September 23-25. 1986. Sponsored by EOS/ESD Association and aT Research Institute. EOSt ESD Symposium Proceedings. EOS-8. 1986. page Ill.

METHODOLOGY FOR EVALUATION OF STATIC-LIMITING FLOOR FINISHES

John M. Kolyer and Dale M. Cullop

Rockwell International Corporation Autonetics Strategic Systems Division Electronics Operations 3370 Miraloma Avenue Anaheim. CA 92803

Hanson Loran Chemical Co .• Inc. 6700 Caballero Blvd. Buena Park. CA 90620

Note: The following is a condensed version of this paper. See the Symnposium Proceedings for the complete test.

Summary

Methodology is described for evaluating floor finishes which limit static buildup on per• sonnel wearing ordinary shoes and drain charges from personnel wearing conductive foot• wear or heel grounders. Procedures include triboelectric charging of various synthetic shoe sole materials and an accelerated scrubbing test to predict durability. The objective is a low• charging. easily maintained. and cost-effective product.

Introduction

Several commercial ESD-control floor finishes were evaluated in mid-1985. Some were unsatisfactory in limiting static generation. while others had questionable floor finish proterties such as scrub resistance; in general. non-ESD properties seemed compromised by antistat addition. All the products were judged too expensive to be cost-effective. These deficiencies prompted our study. Our objective was a low-charging. easily maintained. and cost-effective acrylic coating. and the search involved development of methodology for evaluating commercial finishes and perfecting new and better ones. The emphasis was on simple. relatively inexpensive equipment and realistic performance tests.

218 PAPER NO.6 219

We prefer the tenn "static-limiting" because this property, not conductivity per se, is what is desired if ordinary shoes are worn. "Static dissipative" and "antistatic" refer to surface resistivity ranges, and surface resistivity correlates imperfectly with triboelectric charging of shoe soles, as will be seen. "Static-preventing" would be a misnomer or false claim because all antistatic materials penn it some charging, especially of certain "problem" synthetics. When conductive footwear or heel grounders are prescribed, low surface resistivity is the key ESD property, and tests for triboelectric charging are superfluous. However, our methodology remains pertinent because it predicts durability in tenns of surface resistivity as well as triboelectric charging propensity.

Floor Finishes Tested

The nine commercial products tested represented most of these offered in the U.S. in mid- 1985. These were acrylic, modified with wax in some cases, with a minimum solids content of 15-24.5 percent and pH values in the 8-10 range. Note that deficiencies in these prod• ucts may have been more or less corrected after the time of our testing, because floor finish fonnulations can be changed at a moment's notice. Also, several new products have ap• peared. Hence this paper is concerned with general methodology rather than evaluation of specific products. The four experimental products tested prepresent stages in the evaluation of a new com• mercial product which was a wax-modified acrylic with a nominal solids content of 20 percent and a pH of 9.7. Grounded aluminum (alloy 6061, temper T6) was included in the tests as a reference material to demonstrate that a conductive floor surface can charge synthetic shoe soles. Thus a reduced surface resistivity is necessary but not sufficient.

Experimental Methods

The various experimental methods are shown in Fig. I through 7. Fig. 8 shows that the actual voltage in the walk test can deviate considerably from the meter reading because of inertia of the needle. To be conservative, we multiply meter readings by a factor of 2.

Cost Effectiveness

Cost effectiveness is a vital "property" and must be considered in selecting a floor finish. In conjunction with ordinary ungrounded shoes, the use of a static-limiting floor finish, like humidification and room ionization, is a supplementary procedure, as opposed to pri• mary ESD control methods such as the use of wrist straps. Supplementary procedures are intended to create a relatively benign background or "safety net" for imperfect primary methods. A perfect acrobat doesn't need a safety net, and a perfect operator who always wears his wrist strap doesn't need a static-limiting floor finish, but in reality the reduction of background hazards is worth some expense. The question is: how much expense? 220 APPENDIX

lxI-INCH TILE SAMPLE (CAPSULE HOLDS SIX SAMPLES, ONE INSIDE EACH FACE)

Fig. 1. Heel mark test.

ADJUSTABLE SCALE

RUIIER RING

STRING (TO WINDLASS)

Fig. 2. Topaka slip test.

CROSS SECTION

Fig. 3. Gloss test. PAPER NO.6 221

CABLE (TO RECIPROCATING MECHANISM)

Fig. 4. Detergent or water scrub test.

Fig. 5. Roller test.

Table I compares estimated costs of using a static-limiting floor finish derived from experimental product No.4 versus a standard nonantistatic fmish of similar chemical nature (acrylic and polyethylenic polymers). It is interesting that the chemicals account for only a small fraction of the annual cost of floor maintenance: 4.2 percent for the conventional finish and 6.5 percent for the static-limiting finish. For 10,000 ft2 of floor space, the pre• mium for using the static-limiting finish is $235/year, which seems a low cost for any sig• nificant additional EDS protection. Note that the special finish could help justify saving mishandled items in material review actions; for example, if an operator touched an ESD-sensitive item without wearing his wrist strap, the argument might be made that the probable charge on his skin was below the sensitivity level of the item based on walk-test monitoring data. 222 APPENDIX

, ...-c..':f .... \ ':.:.~.... ' !'.' FARADAY ...... "':' ..": " CUP .~.. . . - -. ... ,.'. :.':> GROUNDED STEEL PAl L STRING (TO WINDLASS)

Fig. 6. Drag test.

AUJillrul SHEET (1 "2)

STAND-OFF INSULATOR (POLYETHYLENE BEAKER WITH SLIT FOR ALlJUNUM SHEET)

Fig. 7. Walk test. PAPER NO.6 223

0

1000 0

.... 0 ~ i 0 = -i 0 00 f 0 t 500 cP vi ...-' i

0

VOLTS, ACTUAL

Fig. 8. Apparent versus actual voltage in walk test.

Table 1. Estimated Annual Costs of Using Conventional versus Static-Limiting Acrylic Floor Finish.

COST, S/IO,OOO Fr2

CONVENTIONAL FINISH, STATIC-LIMITING FINISH, $ IO/GALLON SI5/GAlLON

Finish (8 Coats @ 5 Gallons/Coat) 400 600 Neutral Cleaner (30 Gallons) 165 N/A Antistatic Neutral Cleaner N/A 300 (30 Gallons) Maintenance* (5 h/day@ $IO/h, 13,000 13,000 5-DayWeek)

Totals 13,565 13,900

Premium o 235

*Scrubbing/recoating, restoring, damp-mopping, dust-mopping. and possibly spray-buffmg. These operations are done piecemeal and intennittently, as required. 224 APPENDIX

An advantage of the use of static-limiting floor finish with ordinary shoes as a supple• mentary procedure is that no disturbing or even visible change is made in the work environ• ment. In contrast, humidification can cause discomfort due to "mugginess," and room ionization has been known to create anxiety about its imagined electrical effects on people. However, thorough testing must assure that voltages are in fact sufficiently controlled with the great variety of shoe soles which personnel may wear. For example, certain vinyl com• pounds may be "problem" materials, as indicated by high charges in both the walk test and laboratory tests. Static-limiting floor finishes may be applied to conductive tile without destroying its grounding capability. For example, two coats of the product derived from experimental product No.4 increased the surface resistivity of a popular brand of conductive tile from 1 x 107 to 3 X 107 ohms/square and increased the resistance to ground (through a 5-lb, 2.5-inch-diameter NFPA 56A electrode) from 7 x 106 to 3 X 107 ohms at 10 volts. These threefold or fourfold increases do not prevent rapid electrical drainage from conductive footwear, and the finish provides wear-resistance and an attractive gloss.

Conclusions

1. Nine products commercially available in mid-1985 were evaluated to illustrate our methodology and found to be generally unsatisfactory in performance and too high in cost. Resistance to scrubbing and foot traffic seemed compromised by antistat addition. How• ever, formulations are easily changed, and today's versions may be improved. 2. The products tested were not literally "zero-charge." They limited triboelectric charg• ing of shoe soles but did not stop it. As of May 1986, some suppliers felt that ordinary, ungrounded shoes could be worn without charging enough to damage ESD-sensitive de• vices, but others felt that conductive footwear or heel grounders were needed to drain charges including these created by movement of clothing. 3. The "acid test" of a static-limiting floor finish is the measurement of charges on people who walk on it. When ordinary shoes are worn without conductive straps, surface resistivity correlates imperfectly with charging and can be misleading. 4. The ever-changing shoe sole market presently includes a variety of synthetics such as polyurethane, vinyl, ethylene-vinyl acetate, and styrene-butadiene rubber (SBR). These materials vary in susceptibility to triboelectric charging; vinyls are high and polyurethanes tend to be low (hence they are often used, with additives, in ESD-control footwear). Also, different synthetics may respond in varied and unexpected ways to different floor finishes. Therefore, an averaging effect is desired, and screening of candidate finishes should in• clude several synthetics, not just Neolite (an SBR compound) as in AATCC Test Method 134-1979. 5. Compared to synthetics, leather soles are a minor problem. They hold only 15 per• cent of the market, are usually used only on expensive shoes, and in general triboelectrically charge much less than synthetics. 6. In laboratory tests, cylinders of various synthetics were rolled down a ramp of coated tile, or pieces cut from actual shoe soles were dragged across the surface; in both cases the specimens fell into a Faraday cup to measure the charge. The results roughly agreed with voltages observed on people with random shoe soles when a variety of synthetics and sev• eral people were measured to get average effects. PAPER NO.6 225

7. Based on these ESD tests, methodology is proposed for developing/evaluating new products in the laboratory. Basic floor finish properties such as slip, gloss, and scrub resis• tance are included. The test equipment is simple and easy to use. 8. Correlation of laboratory tests with a floor test in a busy supermarket was good for floor finish properties: durable finishes were distinguished from those subject to hazing and scuffing. However, ESD properties deteriorated in all cases due to spray-buffing under the wrong conditions ( not thick enough). It is recommended that spray-buffing of static-limiting floor finishes be done with caution. A thin restorer coat keeps most of the luster while reviving ESD properties. 9. With proper maintenance, including mopping with a solution of antistatic cleaner, a good static-limiting floor finish should control the charging of walking personnel to 200 volts maximum (peak voltage of spikes) for most, but not necessarily all, synthetic soles at 50 percent relative humidity and 72°F. An experimental finish was successful at 33 percent relative humidity and 72°F in limited testing. 10. Our tests were done in the vicinity of 50 percent relative humidity, which proved acceptable for comparative purposes (screening of candidate finishes), but final on-the• floor testing must involve the lowest relative humidity to be met in practice. Literature data suggest a threefold rise in charging of rubber soles when the relative humidity drops from 50 to 15 percent at 72°F, but the increase could be much more. 11. Static-limiting floor finishes lower triboelectric charges on rolling equipment as well as walking people. For example, annoying charges on supermarket shopping carts with polyurethane wheels were reduced to an innocuous level. 12. The use of static-limiting floor finish with ordinary, ungrounded shoes is only a supple• mentary procedure which provides a "safety net" for primary procedures such as the use of wrist or ankle straps. Therefore, the added expense of using the finish must be moderate for this backup procedure to be cost-effective. An example is given in which switching from a conventional to a durable, easily maintained static-limiting finish costs a premium of only $235/year/lO,OOO ft2 of floor. On conductive tile, the finish permits grounding while pro• viding wear-resistance and gloss. Since some sort of finish must be used, why not a static• limiting or dissipative one at little added expense?

Note: This paper appeared as an article in EOSIESD Technology Magazine, April 1987, page 20:

Getting ESD Floor-Finish Tests Rolling Toward Standarization . ... By John M. Kolyer, Rockwell International and .-...... Dale M. Cullop, Hanson loran Chemical Co. ~,1~::=; These tests are good candidates for a standard method of ~'!:...... ___ evaluating the static·limiting properties of floor finishes. PaperNo. 7

This paper is reprinted. by permission. from EOSIESD Technology. October/November 1987.

TOTE BOX MATERIAL: HOW GOOD IS IT?

1. M. Kolyer. W. E. Anderson and D. E. Watson

Rockwell International Corp.• Electronic Operations. Anaheim.CA

Until recently. all tote boxes for ESD control were made of polyolefin and either topically applied antistat. extruded-in antistat. or extruded-in graphitic (conductive) carbon. Each of these materials has its limitations. Topically applied antistat is only an expedient because the antistat wears off after some undetermined period of use. Also. a wall of plain polyolefin. even with an antistatic surface. provides little Faraday-cage shielding protection from external static fields or discharges (Table I). Extruded-in antistat provides a longer life than topically applied antistat because it continues to bleed to the surface for some time. creating a weakly conductive sweat layer from atmospheric moisture. However. handling. heat. contact with paper products. or exposure to solvents will eventually deplete all the antistat (Ref 1). Again. shielding is poor. Extruded-in conductive carbon offers the advantage of permanence. but it has several problems. A carbon-loaded polyolefin tote-box wall is conductive enough to endanger people; a current of over 100 rnA, which is usually fatal (DOD-HDBK-263). can be carried at 110 V (Table 2). If the conductive box itself is charged it is more dangerous to devices than antistatic or nonconductive boxes (Table 3 and Ref 2). A high-conductivity surface is also dangerous to devices (Ref 3). especially if the opera• tor should be charged and touches a sensitive lead to the box (Table 4). However. the wall of a carbon-loaded polyolefin box is not conductive enough to be a good Faraday cage; note that the volume resistivity of carbon-loaded polyolefin is on the order of 102 Q-cm versus 1~ Q-cm for aluminum foil. Refs 3 and 4 agree that highly ESD• sensitive components should always be protected by metal-foil bags. not carbon-loaded bags or bags with thin. see-through metallization. This conclusion is applicable to other containers such as boxes. Furthermore. carbon-loaded polyolefin imparts high triboelectric charges to nonconductors stroked on it (Table 5 and Ref 1) and sloughs conductive particles that could fall into open microelectronic devices and cause shorts.

Looking at New Materials

The basic defect of conventional boxes is that they are insufficient Faraday cages. So. two multilayer designs with permanent ESD properties have been investigated: fiberboard-foil

226 PAPER NO.7 227

Table 1. Shlelding/Discharge Test.

CAPACITANCE RESISTANCE MOSFET TOTE BOX TYPE (pF) (0) v DAMAGE· Antistatic or carbon- Human Human -8,000 3/3 (OS) loaded polyolefin (0.140 in. wall) Same as above 97 1,500 +5,000 2/2 (2S) Same, but with air gap"'''' 97 1,500 +15,000 2/2 (2S) Carbon-loaded poJyo- 97 1,500 +5,000 0/3 lefin as above, but lined with 0.0003-in. aluminum foil Corshield containing one Human Human -8,000 0/10 layer of 0.00025-in. Tesla coil Tesla coil 35,000 1/5 (IS) aluminum foil Corshield folded to give Tesla coil Tesla coil 35,000 0/5 two layers of the foil Vinyl (0.011 in.) on Tesla coil Teslacoil 35,000 0/5 20-gauge aluminum sheet (0.0375 in.) Sandwich of aluminum Tesla coil Tesla coil 35,000 0/5 screen sandwiched between O.06O-in. sheets of "Forbon" hard vulcanized fiber (NVF Co., Div.)

*For example, 2/5 (IS) would mean that five MOSFETs were tested, two were damaged and one of those damaged was shorted. **I-in. gap between each electrode and inner surface of box.

and vinyl-metal sheet. Multiple layers are necessary because a homogeneous wall is not capable of providing both a safe antistatic or nonconductive surface and Faraday-cage protection. Commercially available fiberboard-foil construction ("Corshield" by Conductive Con• tainers Inc.) consists of aluminum foil sandwiched between layers of fiberboard. The fiber• board has a naturally antistatic surface but is covered with an antistatic coating to seal in

Table 2. Current Carried by Tote Boxes.

RESISTANCE (0) CURRENT (mA)

TOTE BOX TYPE (OOV 200V IIOV 220V

Antistatic 2x H)9 2xl()9 5 x 10-5 1 x 10"4 Carbon-loaded polyolefin (950 n at 1.5 V) 120 230 Corshield 5 x 10' 4x 10' 2 x 10"4 5 x 10"4 Vinyl-aluminum sheet >10 12 2x 10" <10-7 1 x 1()-6 228 APPENDIX

Table 3. Damage to MOSFETs by Grounded Operator Touching Charged Tote Box.

VOLTAGE OF CHARGED roTE BOX roTA!. MOSFETs roTE BOX TYPE +SOO +1000 -8000 DAMAGED

Antistatic 0/1 0/5 1/5 1/15 Carbon-loaded polyolefin 1/5 2/5 3/5 6/15 Corshield 1/5 0/5 1/5 2/15 Vinyl-aluminum sheet 0/5 0/5 1/5 1/15

Table 4. Damage to MOSFETs by Charged Operator

roTE BOX TYPE MOSFET DAMAGE

Antistatic 0/5 Carbon-loaded polyolefin 3/5 (15) Corshield 0/5 Vinyl-aluminum sheet 0/5 Bare aluminum sheet 4/5 (35)

Table 5. Trlboelectrlc Charging Data.

V ON COUPON nCONCOUPON

roTE BOX TYPE FR-4 ACRYlIC AI. FR-4 ACRYlIC AI.

Antistatic 70 700 0 0.6 6.4 0.0 Carbon-loaded polyolefin 400 800 0 3.8 8.5 0.0 Corshield 200 200 0 1.0 2.9 0.1 Vinyl-aluminum sheet 200 200 800 3.3 2.6 2.5

sulfur-containing impurities that might tarnish silver-plated leads. The foil provides su• perior Faraday-cage protection. For example, a discharge from a key held by a person charged to 8000 V caused no damage to field effect transistors (MOSFETs), whereas an impractically heavy wall (0.140 in.) of carbon-loaded polyolefin allowed transistor dam• age (Table 1). Even the Tesla coil test was passed when the Corshield box was folded to give two layers of foil (Table 1). In previous work (Ref 5), only constructions with heavy foil or metal screen passed this test. Note that the Tesla-coil test is the worst case in electrical stress and positioning of the device in the box. However, this test is not the worst case in statistical significance because only five parts are tested for a "pass" rating. Furthermore, our Tesla• coil test does not use worst-case acceptance criteria because subtle damage may not be seen by the curve tracer, and the MOSFETs used are sensitive to 100 to 200 V whereas some new devices may be affected by only 20 volts. If a cost-effective material can pass this test, we would use that material. PAPER NO.7 229

Fiberboard-foil costs less than other materials. Also, it can be stored easily as flat sheets and then folded into boxes when needed. Its only major defect is its limited durability, but heavier fiberboard will probably prove sturdy enough for most applications. The vinyl-metal sheet design should satisfy the market niche requiring extreme durabil• ity. The metal, either steel or aluminum, provides high structural strength and is coated on both sides with tough vinyl, e.g., 0.010 in. thick, by either or powder-coating. The resulting nonconductive surface is safe for people. Also, in a contrived charge-device model test (Table 6), the nonconductive surface was even safer for devices than an antistatic surface. The relatively heavy metal wall, 0.0375-in. aluminum (20 guage), is a virtually impreg• nable Faraday cage and suppresses the voltage of a static charge, no matter how much the surface may be stroked, so that the box never has an appreciable E field when the metal is grounded via bare metal feet on the bottom. In our test, the effectiveness of draining off charges onto an antistatic bench-top was better for a vinyl-metal sheet box than for a carbon-loaded polyolefin box (Table 7). However, good contact with the bench surface, aided by flatness of the bottom of the box, can be critical. The slightly flexible Corshield box benefited from being conformable and lying flat, whereas the rigid antistatic or carbon-loaded polyolefin boxes were slightly "dished" (concave) so that only edges or corners made contact. The nonconductive vinyl surface's only defect is that it can triboelectrically charge con• ductors, whereas an antistatic or conductive surface cannot (Table 5). However, charging of nonconductive surfaces, e.g., conform ally coated circuit-board modules, seems a more im• portant issue, and nonconductive vinyl was less of an offender than carbon-loaded polyolefin (Table 5).

Test Methods

Below are the various test methods that were used to derive the results shown in each of the tables accompanying this article.

Shielding/Discharge Test (Results in Table 1) An electrode 1.5 in. square was taped against the inside surface of one wall of the tote box and connected to the substrate-case lead of a Motorola 2N4351 MOSFET, and a similar elec• trode was taped against the inner surface of the bottom of the box and connected to the gate

Table 6. Charged-Device Model Test

TOTE BOX TYPE MOSFET DAMAGE

Antistatic 2/5 (IS) Carbon-loaded polyolefin 2/5 (OS) Corshield 1/5 (IS) Vinyl-aluminum sheet 0/5 Bare aluminum sheet 2/5 (IS) 230 APPENDIX

Table 7. Drain Time Test.

v (NEGATIVE)

'J'O're BOX TYPE AFI'ER I SEC AFI'ER 5 SEC

Antistatic 8000 8000 Carbon-loaded polyolefin 5000 900 Corshield 2500 600 Vinyl-aluminum sheet with bolt heads for feet 900 600

lead. Then a discharge was made to the outer wall of the box over the electrode. This dis• charge was from a charged person holding a key, from a capacitor connected to a resistor and a steel probe, or from a Tesla coil operated for 30 sec. The box sat on a grounded plate during the test.

Current Carried by Tote Boxes Test (Results in Table 2) NFPA 56A electrodes were placed 1 in. apart on the box surface, and the resistance was read with a Beckman Model L-IO megohm meter.

Damage to MOSFETs by Grounded Operator Touching Charged Tote Box Test (Results in Table 3) A grounded operator held the substrate-case lead of a MOSFET and touched the gate lead to the charged tote box resting on a nonconductive plastic stand-off.

Damage to MOSFETs by Charged Operator Test (Results in Table 4) An operator charged to +1000 volts held a MOSFET (as in Table l) by the substrate-case lead and touched the gate lead to the grounded tote box.

Triboelectric Charging Test (Results in Table 5) 1.5-in.-square coupons of uncoated aluminum or FR-4 epoxy circuit-board laminate, un• coated or coated with acrylic conformal coating, were shaken in the tote box for 30 sec and then dropped into a Faraday cup or measured with a static field meter. All charges were positive. PAPER NO.7 231

Charged-Device Model Test (Results in Table 6) The capacitor (1300 picofarads) representing a charged circuit board was FR-4 epoxy lami• nate, 0.096 x II x 15 in., copper-plated on both sides with I-in. unplated borders. The substrate-case lead of a MOSFET (as in Table I) was connected to the lower side of the capacitor, which was suspended by nonconductive twine. Then the upper plate was charged to + 1000 volts, and the gate lead of the MOSFET was touched to the grounded tote box being tested.

Drain Time Test (Results in Table 7) The tote box, suspended by nonconductive twine, was charged to -8000 volts, placed on a melamine-fonnaldehyde laminate table top (1010 il/sq.) for either 1 or 5 sec, and lifted again; then the field on the box was measured with a static meter.

Summary

Table 8 summarizes our evaluations. Together, the two new multilayer boxes should satisfy all in-plant handling needs for an ESD-control program. These boxes are able to afford secure Faraday-cage protection for even the most sensitive items when electrically continu• ous are being used.

TableS.

TOlE BOX TYPE

CARBON- CARDBOARD- VINYL- LOADED ALUMINUM METAL CHARACTERISTIC ANTISTATIC POLYOLEFIN FOIL SHEET ESD shielding Poor Poor Excellent Excellent Drain time Fair Good Very good Excellent Triboelectric charging Nonconductors Low High Low Low Conductors None None Almost Some none Danger to devices Box charged Low Moderate Low Low Operator charged Low High Low Low Danger to people Low High Low Very low Permanence (ESD) Poor-fair Excellent Excellent Ex.cellent Durability (physical) Good Good Fair Good Cost Moderate High Low High 232 APPENDIX

Cost advantage depends on many factors, most notably the number of units produced and the fabrication method. In general, a vinyl-aluminum sheet box would be competitive with an injection-molded carbon-loaded polyolefin box. A vinyl-steel sheet box, though cheaper than aluminum, would be almost three times heavier in the same gauge. In contrast with the above choices, the fiberboard-foil box is inherently inexpensive (Table 8). Last of all, fiberboard boxes (Corshield) are commercially available at this time: how• ever, vinyl-metal sheet designs are still in the prototype stage. Another interesting multilayer design is one which utilizes an aluminum screen or foil sandwiched between layers of hard vulcanized fiber (Table 1). The vulcanized fiber mate• rial is naturally antistatic, even at low humidity (5 x 1011 O/sq after seven weeks' storage at nop and 12% RH). As in the case of the vinyl-metal sheet box, the commercial success of this design would depend upon the development of a practicable fabrication method, but both appear to be excellent alternatives for future ESD control.

References 1. J. M. Kolyer and W. E. Anderson, "Pennanence of the Antistatic Property of Commercial Anti• static Bags and Tote Boxes," Reliability Analysis Center EOSI.ESD Symposium Proceedings, EOS-5, Las Vegas, NY (1983): 87-94. 2. J. M. Kolyer, W. E. Anderson and O. E. Watson, "Hazards of Static Charges and Fields at the Work Station," Reliability Analysis Center EOSIESD Symposium Proceedings, EOS-6, Philadlephia, PA (1984): 7-19. 3. R. O. Enoch and R. N. Shaw, "An Experimental Validation of the Field-Induced ESO Model," Reliability Analysis Center EOSIESD Symposium Proceedings, EOS-8, Las Vegas, NV (1986): 224-231. 4. G. C. Holmes, P. J. Huff and R. L. Johnson, "An Experimental Study of the ESO Screening Effectiveness of Antistatic Bags," Reliability Analysis Center EOSIESD Symposium Proceed• ings, EOS-6, Philadelphia, PA (1984): 78-84. 5. J. M. Kolyer and W. E. Anderson, "Perforated Foil Bags: Partial Transparency and Excellent ESO Protection," Reliability Analysis Center EOSIESD Symposium Proceedings, EOS-7, Min• neapolis, MN (1985): 111-117. PaperNo. 8

Presented at the 9th Annual Electrical OverstresslElectrostatic Discharge Symposium, Orlando, Florida, September 29-October I, 1987, Sponsored by EOSIESD Association and ITI Research Institute, EOSIESD Symposium Proceedings, EOS-9, 1987, page 41.

ELECTROSTATIC DISCHARGE (ESD) CONTROL IN AN AUTOMATED PROCESS

John M. Kolyer, Ronald Rushworth, and William E. Anderson

Rockwell International Corporation Autonetics Electronic Systems Sensors and Aircraft Systems Division 3370 Miraloma Avenue Anaheim, CA 92803

Note: The following is a condensed version of this paper. See the Symposium Proceedings for the complete text.

Summary

In the robot-based, computer-controlled installation of devices such as OP-AMPs and CMOS on printed circuit boards, an exaggeratedly sensitive model assembly served to locate ESD hazards and later monitor the process for their absence. This "coupon" approach applies uniquely to automation with its freedom from unpredictable operator error.

Introduction

The objectives were (l) to reveal ESD hazards in an automated, robot-based manufacturing process for mounting components on printed circuit boards (PCBs), (2) to remove or con• trol these hazards, and (3) to monitor the process for continued ESD safety. The process and the ESD-sensitive components of concern will be described. Then ESD damage mechanisms, hazards, and remedies will be reviewed, and possible hazards will be related to process steps. Next, the "coupon principle" of using an exaggeratedly sensitive model assembly to locate and evaluate ESD hazards will be explained and illustrated with specific coupon designs. Finally, test data obtained with coupons will be discussed. Haz• ards were eliminated to give an ESD-safe process which could be checked periodically by means of fresh coupons.

233 234 APPENDIX

Description of Automated Process

In 1986, Rockwell International at Anaheim, CA, opened a new Automated Manufacturing Cell (AMC) for placing and reflow-soldering surface-mounted components on PCBs to build modules for guidance systems. The AMC (Fig. 1) is a highly automated, computer integrated, robot-based manufacturing system comprising several work stations arranged in a loop. An Automated Guided Vehicle moves around the cell to shuttle parts between sta• tions. Materials are delivered to the AMC from an Automated Material System (AMS). With reference to Fig. I, the assembly steps are: (1) removal of parts in tote boxes from an automated carousel, (2) fluxing, tinning, cleaning, lead-forming, and kitting, (3) dis• pensing of dots of solder paste at the rate of two per second, (4) dispensing of adhesive and placement of flatpack, radial, and axial components at the rate of one per 18 seconds, (5) curing of adhesive and baking out of volatiles from the solder paste, (6) vacuum oven bak• ing and vapor-phase soldering, (7 and 8) cleaning to remove flux and other contamination as well as solder balls, (9) manual installation of over-sized or heat-sensitive components and also rework, and (10) inspection using sonic digitizers. Units 7 and 8 are connected by a conveyor belt. Work instructions and parts information are stored in a central, fault-tolerant computer system. At each station, a laser bar code scanner is used to enter data. Necessary manufac• turing data appear on cathode ray tubes (CRTs). This paperless system allows easy product changeovers, and several products can be built simultaneously. The precision afforded by automation reduces component scrap so that subassembly costs are minimized and quality is maximized. The relative humidity in the AMC room is controlled above a minimum of 30 percent (at nOF). This reduces triboelectric charging but of course does not prevent it.

Coupon Principle

An automated process, being free of unpredictable operator error, uniquely lends itself to a "coupon" approach in which an exaggeratedly ESO-sensitive model assembly is passed through the process as a coupon to reveal ESO hazards. Various coupon designs used in our tests all contained a 2N4351 MOSFET. The assumption was that if this MOSFET showed no measurable damage, in terms of gate-source threshold voltage and shape of the curve on a curve-tracer, there would be no subtle damage (not detected in quality control tests) to devices on a real module. Undetected damage could lead to latent failure and hence shorter lifetime and reduced reliability. Latent failures are known to occur with CMOS devices. A coupon is shown in Fig. 2, and a close-up of the MOSFETs on this coupon is shown in Fig. 3.

Experimental Methods

The unedited paper describes how the coupons were tested before and after passing through various process steps. ESO hazards revealed by the coupons were removed, and fmally coupons were passed without damage through the entire process and the AMC was declared ESO-safe.

~ ~

;II ;II

~ ~

:u :u

011 011

S! S!

7' 7'

AMC AMC 7"

CD CD

® ®

cell. cell.

manufacturing manufacturing

@ @

Automated Automated

1. 1.

Fig. Fig.

o o

o o

PASTE PASTE

STATION STATION

\ \

APPLlCAnON APPLlCAnON

SOLDER SOLDER

WORK WORK

PHASE PHASE VAPOR VAPOR

SOLDERING SOLDERING

--

.:0-

STAnON STAnON

ASSEMBLY ASSEMBLY

WORK WORK

ROBonc ROBonc

6 6

O O

4 4 O O 238 APPENDIX

Fig. 2. Coupon.

Summary and Conclusions

1. Model assemblies (coupons) which were more ESD-sensitive than real PCB assem• blies proved effective in locating ESD hazards in an automated assembly process. This coupon approach was feasible because automation is free from unpredictable operator error. 2. Relevant ESD damage mechanisms were direct injection (01), the field induced model (FIM), and damage by the static (8) field from the triboelectrically charged board itself (FFB). The most sensitive coupons were damaged by 500 V by DI, 30,000 V at 12 inches by FIM, and 700 V by FFB. 3. The FFB mechanism threatened coupons consisting of bare-board laminates with mounted MOSFETs, but fields on the actual PCBs were minimized by voltage suppression by metallic heat rails and ground planes.

Fig. 3. Close-up of encapsulated MOSFET (simulating a resistor) and plugged-in MOFSET. PAPER NO.8 237

4. Some proposed damage mechanisms never materialized. Triboelectric charged by hot, blowing air was experimentally discredited, and flowing electronic heat-transfer liquid in the vapor-phase soldering process did not damage the most ESD-sensitive coupons. 5. One coupon design was a simulated resistor for handling by robots in the component preparation and adhesive-bonding steps, while other designs were simulated PCB assem• blies for passing through the rest of the process. Designs differed in ESD-sensitivity, and when a more sensitive design was damaged, a less sensitive, more realistic design was tried because all the designs were far more ESD-sensitive than real PCB assemblies. 6. ESD damage to coupons led to corrective actions. These included eliminating syn• thetic brushes, removing supplier-installed grounding wires mixed with brush bristles, and making sure all machinery was grounded. 7. The test program culminated in successful passage of five replicate coupons through the entire process. Monitoring would be done in the same way. 8. In conclusion, the empirical coupon approach is recommended for revealing ESD hazards in automated processes and assuring safety after hazards are removed. 9. However, the coupon must be demonstrably more sensitive than actual components to all pertinent ESD damage mechanisms. Introduction of a new component, or a change in the process, could necessitate coupon redesign. 10. Finally, periodic checks with coupons are necessary because continued ESD safety is not guaranteed. For example, a ground connection might fail and machinery parts might be charged by electrical leakage.

Note: This paper was presented as an article in EOSIESD Technology Magazine, August/September 1988, page 10. A good summary was included:

Controlling ESD with Coupons by John M. Kolyer, a_lei llllhworth and WIlliam E. ~, Rockwell International Corp., Autonetics Electronic Systems, ~_~...,. Sensors and Aircraft Systems Div. A comprehensive and detailed overview of Rockwell International's use of coupons to detect and manage EOSjESD hazards in an automated assembly facility. Human unpredictability can cause many static problems, but humans also monitor and control ESD. Here's how Rockwell International uses coupons to detect and manage EOSjESD hazards in an automated assembly facility where few humans need go.

Etll,,,,'s Note: AI,b," IIUIO"""• bly 11M to IISSU" 111111 qUllllty de• etlllSsembly ,limllUl'" "","y of ,II, ,p/It 0 mucII "'uc,tI IIu""," _k p,06/,ms CIIUS,tI by ,II, mowmelll fDlU, Rockwtll 1"'HlUltlllllGl IIltd 11M mlS/obs of IIumllll bellllS. til, _tllods d"cribetl below. TII,y ,II,n nmllllll ,II, probltm of tI"ecl• "" btu,d 011 ,II, coupOll prlllcipl" illl 11M cOII,ro/lill, ESD I" ,,., lib• w"",11I 0 ,ubtu_bly _ Itllll- se/lCt of IIumllll supI/'V/II011. III II tlw '0 ESD tllllll ,1101' '0 be _ limltttl WilY, ESD COIItroi becomlS fllCtll"d Itllnds III f", til, IlIltH till,.. l1li t1u,cilt III nmolt Stllli/ll. III, qUIIlijiClltlOllIlM "9UIIlijiclltlOll III ",tI" '0 spo, IIn11S of ESD vul• of til, f«lIity. ",,,,bill,y tlurilll lIu'omlltttlllssem- PaperNo. 9

Presented at the 10th Annual Electrical Ovestress/ Electrostatic Discharge Symposium, Anahein, California, September 27-29, 1988 Sponsored by EOS/ESD Association and rrr Research Institute. EOSI ESD Symposium Proceedings, EOS-IO, 1988, page 99.

CORROSION AND CONTAMINATION BY ANTISTATIC ADDITIVES IN PLASTIC FILMS

John M. Kolyer and Jack D. Guttenplan

Rockwell International Corporation Autonetics Electronic Systems Sensors and Aircraft Systems Division 3370 Miraloma Avenue Anaheim, CA 92803

Abstract

One commercial brand of MIL-B-81705, Type n film contained organic acid and caused corrosion of solder-coated device leads on circuitry. However, solderability was unaffected in accelerated tests. Even acid-free antistats can stresscrack polycarbonate, fog instrument mirrors, weaken adhesive bonds, and discolor epoxy paint. These problems are reduced by a new generation of Type n films.

Introduction

Contamination of surfaces by antistats from antistatic films can cause (l) corrosion of sol• der (when the antistat contains organic acid), (2) stresscracking of polycarbonate plastic, (3) fogging of instrument mirrors, (4) loss of strength of adhesive bonds, and (5) discolora• tion of epoxy-polyamide paint Other adverse effects, e.g., contamination of gyroscope balls, have been discussed in the literature. I

Corrosion of Solder

In early 1987 a Government-Industry Data Exchange Program (GIDEP) Alert2 described corrosion of solder-coated leads by traces of n-octanoic acid in one commercial brand of "pink poly" antistatic film (MIL-B-81705, Type II). Therefore, we ran tests to determine the amount of acid present and the nature of corrosion caused by it In our tests, extraction of a suspect (acid-contaminated) bag with refluxing ethanol gave acidity corresponding to 514 ppm by weight n-octanoic (caprylic) acid, in agreement with 400-650 ppm found by methylene chloride extraction by other investigators.3 The acidity

238 PAPER NO.9 239 of a sample of neat (undiluted) antistat from the manufacturer of the suspect film corre• sponded to 8.9 wt. % n-octanoic acid, and other investigators3 found up to 15 wt. %. From our acidity data, the wt. % of antistat in the film = (514) (100)/89,000 = 0.6, which is roughly the level the manufacturer said had been added to the polyethylene by blending and extrusion. Our corrosion testing is summarized in Table I. Degradation of solderability has been suggested2.3 but was not found in our tests. When painted with neat acid-contaminated anti• stat, without aging, leads remained wettable by solder, and strength of solder joints was normal. In long-term tests, circuit-board specimens and resistors with freshly tinned copper leads were stored in suspect bags at elevated temperature and humidity or immersed in moist antistat or moist n-octanoic acid at 120°F. The conclusion was that corrosion of pack• aged items does occur but is only superficial (dull, gray tarnish) because so little acid reaches the solder surface. There was no effect on solderability except in the drastic test with neat n• octanoic acid where enough corrosion products built up to cause dewetting of solder. When the acid content was very low, e.g., 0.05 wt. % as n-octanoic acid in the neat antistat, antistats of the diethanolamide type ("proposed antistat" in Table I) were noncorrosive, and the cosmetic problem of tarnishing disappeared. Most items already stored in acid-contaminated packaging were left as is, but new items were packed in acid-free containers. Panicky repackaging that could deform leads or cause electrostatic discharge (ESD) damage was avoided.

Stresscracking of Polycarbonate

Polycarbonate, which is notably subject to stresscracking, was attacked by all antistats tested, though to different degrees. Stresscrack agents for polycarbonate include commonplace liquids such as ethylene glycol, heptane, and com oil in the literature4 and, in our tests, pink liquid hand soap, No. 30 motor oil, and RMA (rosin, mildly activated) solder flux. There• fore, it was not surprising that liquid antistats, which have considerable solvent power, were also stresscrack agents. At 2.4% tensile strain on the surface of ~-inch-thick bent strips of polycarbonate, ethoxylated tertiary amine or diethanolamide-type antistats caused cracks to propagate completely through the plastic within 24 hours. At 0.3% strain, the antistats still caused catastrophic cracking, but the time was much longer: 2 to 7 weeks. When conventional MIL-B-81705, Type II films or new-generation "dry" films (dis• cussed below) were rubbed on the surface of bent polycarbonate strips at 2.4% strain, the new films caused significantly less cracking than did the "old." The explanation is that the new films have minimal liquid antistat on their surface. Thus they may be safe when the stress level on the polycarbonate surface is very low, but the cautious recommendation is to avoid contacting polycarbonate items with any plastic that contains a liquid antistat.

Fogging of Instrument Mirrors

Fogging of instrument mirrors by outgassed antis tat during storage was observed and was simulated by suspending an aluminum plate inside an antistatic bag in an oven at 160°F. After one week, the plate was rinsed with acetone, and the solution was analyzed by infra• red spectroscopy. It was found that about 20% of the original antis tat on the inner surface of 240 APPENDIX

Table 1. Corrosion Study.

GIDEP ALERT E9·A·86·02 ISSUED JANUARY 23,1987 • RESULTS

• ONE COMMERCIAL BRAND OF PINK POLY AN11STATIC FU.M - ACCELERATED TESTS (MIL-B-81105, TYPE 0) CONTAINS TRACES OF N-ocTANOIC • HEAT ANTISTAT (SUSPECT BRAND) CAUSED DULLJNO/ ACID GRAYING, NO EFFECT ON SOLDERABILITY • PURE n-oCTANOIC ACID FORMED HEAVY CORROSION • n-ocTANOIC ACID REACTS WITH SOLDER, FORMING PRODUCTS, SOME EFFECT ON SOLDERABILITY TIN/LEAD SALTS • FfIR ANALYSIS - CARBOXYLIC DERIVATIVES (ORGANIC • SOLDERABILITY CAN BE REDUCED ACID SALTS) OF TIN AND LEAD • PROPOSED NEAT ANTISTAT (ESSENTIALLY FREE Of ORGANIC ACID) HAD NO EFFECT ON SOLDER OR TESTING PROGRAM SOLDERABILITY

- LONG-TERM EXPOSURE TESTS • 8O'l'o R.H. AT 100"F • DESIGNED TO ANSWER FOLWWlNG QUESTIONS: - SEALED AND UNSEALED BAGS - IF OPERIJIONAL PARrS OR MODULES ARE STORED IN - 3-MONTHS EXPOSURE nus MATI!RIAL, WR.L THERE BE PHYSICAL OR • AMBIENT HUMIDITY AND TEMPERIJURE I!LI!CJ1t1CAL DAMACB, I.B., CORROSION, IMPAIRED -SEALED AND UNSEALED BAGS SOLDBRABn.JTY, CONDUCTIVE PATHS? -I-YEAR EXPOSURE

- IF THERE IS DAMAOE, WHAT DISPOSmON SHOULD BE - LONG-TERM STORAGE TESTS MADE OF PARrS AND MODULES PRESENTLY STORED OR • SUSPECT BRAND OF BAGS CONTAINS LOW LEVEL, PREVIOUSLY EXPOSED? TYPICALLY SOD PPM, n-oCTANOIC ACID • OTHER BRANDS ESSENTIALLY FREE Of ORGANIC ACID • CORROSION EfFECTS IN SUSPECT BAOS LIMITED TO • TEST SPECIMENS DULLING/GRAYINO; NO EFFECT ON SOLDERABILITY - PCB SPECIMENS, SOLDER-COATED CIRCUITRY BOTH SIDES • NO CORROSION/NG EFFECT ON SOLDERABILITY IN OTHER COMMERCIAL BRANDS - PCB MATERIAL COUPONS, COPPER-CLAD BOTH SIDES, SOLDER-DIPPED OR SOLDER-PLATED AND Rl!A.OWl!D

- RESISTORS WITH COPPER LEADS, FRESHLY TINNED APPEARANCE OF PCB TEST SPECIMENS AFTER ACCELERATED IMMERSION TESTS • ACCELERATED EXPOSURE TESTS

- PARTIAL IMMERSION AT 120"1' FOR 7 DAYS • MOIST N-oct'ANOIC ACID • IMMERSION IN PURE n-ocTANOlC ACID • MOIST NEAT ANTISTAT (FROM SUSPECT BRAND) SATURATED WITH WATER-HEAVY CORROSION • MOIST PROPOSED NEAT ANTISTAT PROOUCT ACCUMULATION IN LIQUID

- HUMIDITY TEST (MIL-STD-202, METHOD 106) • IMMERSION IN SUSPECT BRANO HEAT • PAINT ON THIN FILM ANTJSTAT - SLIGHT GRAYING IN LIQUID - D-OCTANOIC ACID • IMMERSION IN PRGPOSED NEAT -NEAT ANTISTAT (FROM SUSPECT BRAND) ANTJSTAT - NO EFFECT - PROPOSED NEAT ANTISTAT • CYCLE RM. TEMP. TO UO·F AT 90-98% R. H., TWICE DAILY FOR \0 DAYS ACTION TAKEN

• EVALUATION METHODS • PURCHASE OF SUSPECT BRAND SUSPENDED - PCB SPECIMENS • ADD SOLDER PASTE AND VAPOR PHASE SOLDER • PARTS AND MODULES PRESENTLY STORED: LEFT • VISUALLY DETERMINE DEGREE OF SOLDER WETTING "AS IS" (SOLDERABILITY NOT AFFECTED; REPACKAGING COULD CAUSE ESD DAMAGE, LEAD - SOLDER-COATED COUPONS DAMAGE, CONTAMINATION, ETC.) • SGLDER DIP • VISUALLY DETERMINE DEGREE OF SOLDER WETTING • PARTS AND MODULES PREVIOUSLY EXPOSED; LEFT - RESISTOR LEADS "AS IS" (NO PROBLEM OR FAILURE MODE HAS • MEASURE SOLDERABILITY ON MENISCOORAPH BEEN ATTRIBUTED TO THIS EFFECT) PAPER NO.9 241 the bag (as detennined by rinsing a fresh bag) had migrated as vapor to contaminate the plate. The same level of transfer is estimated to occur in 6 months at room temperature, based on a Cox chart which relates vapor pressure of organic compounds to temperature. The solution to the fogging problem obviously is to use ESD-protective packaging mate• rial free of fugitive, surface-seeking additives. Another example of the undesirability of transferable antistats is the packaging of items such as resistors whose surface must main• tain high electrical resistivity. In this instance the weakly conductive "sweat layer" that is desired on the package surface becomes a liability on the surface of the stored item.

Loss of Strength of Adhesive Bonds

Intuition suggests that contamination of surfaces by "greasy" antistat would greatly reduce the strength of adhesive bonds. An experiment was conducted using aluminum (alloy 2024 T3) treated with Forest Products Laboratories etch solution (30 parts by weight deionized water, 10 parts concentrated sulfuric acid, and 1 part sodium dichromate) at 150°F for 20 minutes followed by rinsing with deionized water and drying at 150°F for 20 minutes. Lap• shear specimens were prepared using a 2-mil (0.OO2-inch) bond line of Scotch weld EC2216 (3M Co.), which is a filled epoxy-polyamide adhesive. These bonded aluminum specimens were cured at 155°F for 15 hours and tensile-tested by a standard method.5 Before bonding, the surfaces had been (1) left clean, (2) contaminated with a slight amount of antistat by rubbing with an antistatic bag, or (3) liberally smeared with neat antistat. Bond strength data (averages of four detenninations) are given in Table 2. Failures were cohesive (within the adhesive) except as noted in the table. As shown by the relative bond strengths, slight contamination by rubbing with the bags had a negligible effect, and gross contamination by ethoxylated tertiary amine or diethanolamide antistat has surprisingly little effect on bonding with one frequently used structural adhesive. How• ever, other might behave differently and would have to be tested. Our result merely shows that fugitive antistats do not necessarily cause severe bonding problems.

Table 2. Adhesive Bond Strength Data

AV.BOND STANDARD RELATIVE CONTAMINANT STRENGTH, PSI DEV.,PSI BOND STRENGTH None 2588 199 1.00 Antistat from rubbing with suspect bag 2468 357 0.95 Antistat from rubbing with new- generation "dry" bag 2583 100 1.00 Ethoxylated tertiary amine antistat 1954 438 0.76* Suspect amide antistat containing 8.9% n-octanoic acid 2196 132 0.85**

"About 10% adhesive failure (bare aluminum exposed). ""Trace, perhaps 0.1 %, adhesive failure. 242 APPENDIX

Discoloration of Epoxy-Polyamide Paint

Instrument panels painted with gray epoxy-polyamide paint (MIL-C-22750, Type I) were discolored by prolonged contact with the acid-contaminated brand of MIL-B-S1705, Type II film. The film had been pressed tightly against the surface, and wrinkles in the film corresponded to a pattern of whitish streaks or marbling. Abrasive rubbing compounds failed to remove these blemishes, and waxes did not fully restore gloss. Repainting was necessary. When drops of liquid were left on a painted surface for 24 hours at 120°F, n-octanoic acid caused the most attack (swelling and softening of the paint), followed by the acid• contaminated antistat, a low-acid diethanolamide antistat, and an ethoxylated tertiary amine antistat. The explanation for this order of attack is that the acid was the best paint solvent, the amide was intermediate, and the amine was the poorest solvent. At room temperature, the acid-contaminated film caused staining in I day when a pres• sure of 5 psi was maintained but no staining in the absence of pressure. A sample of new• generation pink antistatic bubble-wrap, which had minimal antistat on the surface, caused no staining even after 6 months at 5 psi. Attack on paint probably is rarely noticed because the film is not usually tightly pressed against the bagged item as chanced to happen in our case. Our temporary solution was to use MIL-B-SI705, Type I, which of course is antistat-free.

Summary and Conclusions

1. One commercial brand of MIL-B-SI705, Type II film was contaminated with n• octanoic acid and caused corrosion of solder, as revealed in a GIDEP Alert. However, our corrosion tests indicated that the problem was only cosmetic because solderability was unaffected; in fact, the acid resembles flux, which is also an organic acid. Therefore, pan• icky repackaging was avoided in our operations. The issue is now past, because the offend• ing product has been removed from the market. 2. Antistat-containing films are suitable for most purposes and have a long history of success in ESD control, but they must be used with discretion because the fugitive, surface• seeking additive can stresscrack polycarbonate, fog instrument mirrors, and discolor epoxy paint. Adhesive bond strength also may be reduced, but the effect was less than expected in tests with a commonly used epoxy adhesive. 3. Liquid antistats have considerable solvent power and can even attack rubber and wire insulation, but in practice the damaging effects are limited by the small amount of antistat transferred by rubbing or volatilization. For example, n-octanoic acid in sufficient quantity will destroy solder completely, but a minimal amount of the acid contained in rubbed-off antistat merely dUlled solder surfaces. 4. Antistat contamination problems are reduced, but not eliminated, by a new genera• tion of "dry" MIL-B-SI705, Type II films with a minimal amount of "acid-free, amine• free" amide antistat on the surface. By "acid-free" is meant negligible acidity, and "amine-free" refers to a low amine content, e.g., 4-6 wt. % diethanolamine. Amines are not corrosive to solder under conditions of use, and ethoxylated tertiary amine has long been used as an antis tat. PAPER NO.9 243

5. Assuming that their permanence6 proves adequate, the new-generation bags and bubble-wrap will be a great impovement over the old "wet" or "greasy" products but are not a panacea because the difference is quantitative, not qualitative. Packaging materials free of liquid antistat still must be used in critical applications such as shipment or storage of poly• carbonate items. Potential users should run tests to find the most cost-effective packaging system.

Acknowledgment

The assistance of R. Rushworth, D. R. Violette, A. E. Carmellini, and T. J. Hester in per• forming chemical analyses and corrosion tests is greatly appreciated.

References I. M. K. Bernett, H. Ravner, and D. C. Weber, "Electroactive Polymers as Alternate ESD Protec• tive Materials," EODI ESD Symposium Proceedings, pages 115-119, orlando, Fl., 1982. 2. GIDEP Alert No. E9-A-86-02, on "Materials, Plastic, Antistatic," issued January 23, 1987. 3. J. Anderson, R. Denton, and M. Smith, "Antistatic Polyethylene Package Corrosion," EOSI ESD Symposium Proceedings, pages 36-40, Orlando, Fl., 1987. 4. Modern Plastics Encyclopedia, 1978-79, pages 528-529. 5. "Standard Test Method for Strength Properties of Adhesives in Shear by Tension Loading (Metal• to-Metal)," ASTM D 1002-72 (reapproved 1983). 6. J. M. Kolyer and W. E. Anderson, "Permanence of the Antistatic Property of Commercial Anti• static Bags and Tote Boxes," EOSIESD Symposium Proceedings, pages 87-94, Las Vegas, NY, 1983. PaperNo. 10

Presented at the 11 th Annual Electrical Overstress/Electrostatic Discharge Symposium, New Orleans, Louisiana, September 26-28,1989, Sponsored by EOS/ESD Association and m Research Institute. EOSIESD Symposium Proceedings, EOS-II, 1989, page 23.

CONTROLLING VOLTAGE ON PERSONNEL

John M. Kolyer, Donald E. Watson, and William E. Anderson

Rockwell International Corporation Autonetics ICBM Systems Division Electronics Operations 3370 Miraloma Avenue Anaheim, California 92803

and Dale M. Cullop

Hanson Loran Chemical Co., Inc. 6700 Caballero Blvd. Buena Park, California 90260

Note: The following is a condensed version of this paper. See the Symposium Proceedings for the complete text.

Abstract

Controlling voltage on personnel is a vital and demanding requirement in ESD protection. New data show that the allowable resistance to ground must be lower than was thought. Furthermore, wrist-strap systems often fail, so monitoring must be continuous; resistive• type monitors are best. A cost-effective program is described.

Introduction

Controlling voltage on personnel has long been recognized as of primary importance in preventing electrostatic discharge (ESD) damage to electronics. The reason for concern is that operators' fingers often touch leads or contacts of ESD-sensitive (ESDS) devices, and a flow of current from person to device can destroy or "wound" it by the direct injection (01) mechanism. A "wound," passing undetected in routine tests, may result in latent fail• ure. Thus, improperly grounded people are the principal ESD hazard or at least prime sources of ESD for destroying parts. A resistance of about I megohm is generally agreed upon for current-limiting resistors to protect operators wearing wrist straps. However, questions remain: (I) What should be the

244 PAPER NO. 10 245 maximum allowable resistance to ground (ARTG) for protection of ESDS items as opposed to protection of operators? (2) How can the triboelectric charging of operators be mini• mized to make the resistance to ground less critical? (3) Should this resistance be checked at intervals or continuously? (4) If a continuous wrist-strap monitor is used, which design is best? (5) How can skin-voltage control and monitoring be made part of a cost-effective program? We have attempted to answer these questions based on experience as well as experimental data.

Experimental Methods

An operator sat at a workbench and charged himself either by stroking a sheet of Aclar film on the bench surface or by shuffling his feet. The resulting data are shown in Fig. I. Fig. I represents worst-case charging conditions, but surveys of operator grounding in manufac• turing areas showed many bad cords and instances of high resistance to ground because of operators' dry skin. Therefore, we recommend continuous monitoring with units like the one illustrated in Fig. 2 and 3.

A Cost-Effective Program Including Skin-Voltage Control

We emphasize basic, proven methods of ESD control and deemphasize methods which are of questionable value and may even be part of the problem. The major elements in our suggested program for ESDS items in Class I of MIL-STD-1686A are:

1. Operator Disciplines A sufficiently skilled operator could work with almost no special equipment, and, con• versely, the most expensive appliances won't prevent damage by an ignorant or careless operator. Remember that a grounded operator is only the lesser of two evils versus an un• grounded operator because a charged device could be damaged by rapid discharge to the grounded operator's finger (the Charged Device Model or CDM). To maintain their skills, operators must be retrained periodically.

2. Use of Test Equipment to Aid Operator Judgment An ESD-educated operator needs access to basic test equipment including a field meter, a continuity checker such as a Zapflash, and a surface resistivity meter. Several operators might share these instruments.

3. Skin Voltage Control As described in this paper, the ARTG of 10 megohms is met using continous wrist-strap monitors and antistatic lotion. l~ri------' I

o • VOlTME IY PYT-SOO • • VOLTMEIY TlEK 512 I ( ) • MOSFETIWWiE '1 2 I / /

III (2/10) 1(515)" / 'OO o (115) ~ , i •••••••••••••••• • ••••••••••••• , ••••••••••••••• " •••••••• V (0/5) I (0/5) I' I / I / 1 • / / /

~\I//---r4./ SHOES, STATJC• ,r/ FLOOR " /. LlMITJIIi FINISH / /

100 1000 MEGOIfG

Fig. 1. Plot of triboelectric charging data. PAPER NO. 10 247

CONTINUOUS MONITOR, TWO-CONDUCTOR TYPE: DIAGRAM

ALARM \ \' "', .. ,,

'" ~~T ...... ------,-"

Fig. 2.

4. Exclusion of Nonconductors from Workstations This precaution must be rigidly enforced, with frequent checks. A field meter and surface resistivity meter are essential. Nonconductive surfaces which can't be eliminated may be treated with topical antistat, necessarily chloride-free in some applications, and so labeled. Bench tops should have static-dissipative surfaces and a buried ground plane to suppress voltage.

5. Grounding of All Conductors This is easily checked with a Zapflash. If a conductor, such as a microscope eyepiece hous• ing, cannot be conveniently grounded, the operator must satisfy himself that it never be• comes dangerously charged. The best policy is to ground even small-area conductors.

6. Control of Electrical Fields Paper No.3 gives safe distances for loo-volt-sensitive devices from surfaces at various apparent voltages as determined by a field meter. The operator should run frequent checks on his static-safe workstation to locate charges. Then, if these can't be eliminated, he will work a safe distance from them. 248 APPENDIX

Fig. 3. Two-conductor continuous monitor with strap.

7. Minimization of Triboelectric Charging of Operators Static-limiting floor finish is very cost-effective as seen in Paper No.6. Choose a brand based on its ability to limit tribroelectric charging of operators wearing shoes with various types of soles. We recommend a "walk test" with a dozen of your personnel: the voltage on them is measured with a field meter or with a Voyager PVT-300 Personnel Voltage Tester. Surface resistivity of the finish should be at most 10" ohms/square, but note that surface resistivity does not correlate with triboelectric charging for materials in general and corre• lates imperfectly for floor fmishes. No matter what finish is used, operators should not fidget or shuffle their feet unnecessarily; there is no sense in "looking for trouble." PAPER NO. 10 249

8. Use of Ionizers Be cautious with room ionization as mentioned above. Cleanrooms, by the way, are a special case; an ionization system, correctly engineered, may reduce the number of airborne particles. Use local ionizers only for a clearly defined purpose, such as to control charge buildup in processes like grit-blasting. Never use ionization as a vaguely conceived "safety net."

9. Humidification In very dry areas, humidification is desirable because it makes antistatic materials with "sweat layers" function better and it reduces triboelectric charging. But don't let it build false conficdence, and beware of corrosion or other problems as mentioned above. As with ionization, the use of humidity control must be judicious.

10. Static-Safe Packaging We recommend "Faraday-cage" packaging with aluminum foil as shielding, because inde• pendent studies agree that foil is the only adequate protection from external fields and dis• charges for highly ESD-sensitive items. When away from a static-safe workstation, an ESDS item must be in a Faraday-cage container such as a tote box for in-plant handling or a sturdy, suitably cushioned Faraday-cage package for shipment.

11. Elimination of Unnecessary Equipment Our recommended program is cost-effective because it emphasizes operator skills coupled with basic, clearly useful equipment. Money is saved by eliminating marginally valuable appurtenances such as special smocks, footwear, chairs, or floors. "Buy the basics, forget the frills." Besides being a waste of capital, some frills may be counterproductive by instill• ing false confidence so that operator disciplines are relaxed.

12. Program Administration Of course, good record-keeping and day-to-day management are essential. Operators should be encouraged to keep records, e.g., of hazards such as E fields that they're en• countered, and to make suggestions. Incidents of hardware mishandling should be dis• cussed with the group. Operators will monitor their individual workstations by using test equipment as recommended above. With alert, knowledgeable operators and an emphasis on the basics, ESD control becomes a simple matter except for silicon wafer processing, paint-spraying, or other processes which involve triboelectric charging beyond the operator's control. Static-limitation must be designed into such processes, e.g., by using conductive solvents or blowing ionized air over surfaces. Automated processing is, of course, out of the operator's hands by definition and requires special approaches such as the use of ESDS coupons to locate hazards. In large companies, a Program Coordinator should be appointed to lead a working group, conduct monthly meetings, brief upper management, etc. 250 APPENDIX

Summary and Conclusions

1. An operator, grounded through various resistances, triboelectrically charged himself by shuffling his feet or by stroking a plastic film (representing a garment sleeve) on a work• bench surface. Devices sensitive to 80-100 V were damaged by being touched by the op• erator at a little over 70 megohms resistance to ground in the foot-shuffling tests and a little over 7 megohms in the plastic film tests. An explanation for the difference involves rapidity of voltage rise, which is greater for the thinner dielectric (film versus shoe soles). 2. Considering such factors as the methodology of the tests, the effect of relative hu• midity, and the fact that women's shoe soles tend to be thinner than men's (as used in the test), we recommend an allowable resistance to ground (ARTG) of 10 megohms. This is the maximum, and a suitable minimum for most work is 1 megohm, so as to limit current to well under 1.0 mAo When high voltages are being handled, ground fault circuit interruptors will guard against the remote possibility of an electrically overstressed resistor being car• bonized so that it falls below a safe resistance. 3. To minimize triboelectric charging, operators shouldn't fidget or shuffle their feet unnecessarily. Also, a static-limiting floor finish is effective; for example, the voltage on an ungrounded, foot-shuffling operator was reduced from 3300 V with an ordinary finish to 180 V with the special finish. With the special finish, the ARTG could have been increased by a factor of about 8, because the charging of the operator was so low even when he was ungrounded. However, the ARTG must be set assuming normal floor surfaces. 4. The static-limiting floor finish remained effective, though less so, after severe scuff• ing and scraping to simulate wear. We advise selecting a durable brand of finish which is easily maintained by standard procedures. 5. One-conductor wrist straps show poor reliability. In fact, checks on wrist straps in electronic assembly areas disclosed not only a disturbing rate of cord failure but many instances of excessive resistance to ground caused by loose wrist bands or operators' dry skin. Clearly, monitoring is essential. 6. This monitoring should be continuous because ESD damage can occur in nanosec• onds and an ESDS item can be "zapped" during the interval, however brief, between peri• odic checks. All hardware handled since the last successful check of a failed strap could be subject to a material review action and might even have to be scrapped! Furthermore, cords sometimes fail in an intermittent way which could go undetected in periodic checks but be revealed in continuous monitoring. We highly recommend these monitors, but some users have been troubled by features such as the audible alarm. Potential buyers should run trials in their own assembly areas. 7. Our preferred monitor design is the two-conductor resistive type which checks ground connections as well as the operator. Also, the two conductors provide redundancy (if one fails, the other keeps working), and the monitor is not fooled by a high-capacitance object touching the strap. The property of interest-resistance-is measured directly. Antistatic lotion is often required for dry skin, and the monitor gives notice when the lotion must be renewed. 8. This paper sketches a cost-effective program, including skin-voltage control, for pro• tecting items in sensitivity class 1 of MIL-STD-1686A. Elements of the program include operator disciplines, test equipment for operators' use, continuous monitoring of skin volt• age, exclusion of nonconductors, grounding of conductors, control of E fields, minimiza• tion of triboelectric charging of operators, judicious use of ionization and humidification, static-safe packaging, cost-saving by eliminating unnecessary equipment ("Buy the basics, forget the frills"), and astute program administration. Paper No. 11

This paper is reprinted, by permission, from EOSIESD Technology, October/November 1990, pages 18-23.

ESD-CONTROL MYTHS, OLD AND NEW

John M. Kolyer and Donald E. Watson

Rockwell Intemational Corporation Defense Electronics Autonetics Strategic Systems Division San Bernardino, California

By "myth" we mean a more or less widely held belief that sounds credible but proves false when examined critically. Many such beliefs have clouded ESD-control standards. First, we'll discuss some old myths, then some new ones. Among old ESD-control myths, now put to rest in most people's minds by long repeti• tion of the truth, are the following: Wrist straps render human skin harmless. The truth is that a grounded operator versus an ungrounded operator is only the lesser of two evils because of the Charged Device Model (CDM). A charged device may discharge to the grounded operator's fin• gers and be damaged. If the operator is ungrounded and chances to be at the same poten• tial as the device, no damage will occur. So grounding, in this unlikely case, would be an "evil." Components are safe once they're mounted on circuit boards. Circuit lines may act as antennas to increase the likelihood of ESD damage and discharges caused by the Field Induced Model (FIM). Also, more than one device may be damaged by Direct Injection (01) and electrical overstress (EOS) when a single discharge occurs to a branched circuit line. Carbon-loaded plastic provides a Faraday cage. Carbon-loaded plastic, which was the earliest ESD-protective material (in the sense of not holding triboelectric charges) is not conductive enough to shunt a direct discharge safely away from an ESD-sensitive item inside a bag. Instead, the discharge acts through the wall of the bag and can cause damage if the device leads are in a vulnerable position. High humidity solves all ESD handling problems. Triboelectric charging is merely reduced by high humidity, not eliminated. Operator disciplines cannot be relaxed when humidification is provided. Humidification is expensive and may introduce problems such as corrosion or operator discomfort. Ionization allows the use of plain, nonconductive plastics. This statement is untrue because nonconductive plastics can become triboelectrically charged and cause ESD dam• age by the FIM before the ions have time to neutralize the charge. Ionization is a substitute for grounding of personnel (by wrist straps, for example). This egregious idea was actually promoted by an equipment supplier several years ago. The fact is that operators can become triboelectrically charged (by shuffling their feet, for example) and damage ESD-sensitive items with 01 by touching them despite a shower of ions that neutralize their skin relatively slowly.

251 252 APPENDIX

Devices with protective networks are not ESD-sensitive. Protective networks raise the threshold for ESO damage to a limited extent, such as from 100 V to 2,000 V by the Human Body Model (HBM), not infinitely. In other words, the sensitivity class in accor• dance with MIL-STD-1686A may be raised (from Class 1 to Class 2 in the previous ex• ample), but the device will remain ESO-sensitive. In our system of ESO control, the precautions wouldn't change because we treat all ESO-sensitive classes the same as a mat• ter of convenience; maintaining "islands" of different disciplines within a plant is impractical. So much for old myths. Unfortunately, a second generation of myths has arisen, created partly by vendors' biases. We'll debunk several of these new, less obvious, and more insidi• ous untruths. The following discussion includes our own biases, of course, but these are based on data and not influenced by sales concerns. Readers are welcome to contact us c/o ESDlEOS Technology on points of disagreement. Background Ionization Is needed when highly ESD-sensitive Items are being handled. This statement suggests a cure-all or panacean view of ionization, but ioniza• tion is not a cure-all. In fact, ionization for no valid, well-defined reason can be useless and actually become part of an ESO problem by creating unbalanced charges on ungrounded conductors. The possible occurrence of such charging has been noted for an electrically unbalanced AC ionizer (Ref. 1). Hence, ionizer manufacturers have shown much concern about balancing which in many models is now automatic. In contrast to humidification, which reduces triboelectric charges by inhibiting them in the first place, ionization can neutralize a charge only after it has formed. Thus, ionization is a process of correction rather than prevention. However, it does have its place in ESO control when used judiciously. Local ionization helps control triboelectric-charging pro• cesses, such as peeling tape from a roll, and room ionization does control standing charges on common plastics and garments. Also, room ionization may have an incidental air-clean• ing effect in cleanrooms, but potential users should run tests or at least review published data to prove that vendors' claims of reduced particles are true. While malfunctioning electrical ionizers could charge isolated conductors as we men• tioned earlier, nuclear ionizers have a possible problem of polonium 210 contamination. One brand of nuclear ionizers was recalled by the Nuclear Regulatory Commission in Feb• ruary 1988. Actually, this danger is remote. In terms of cost, nuclear and electrical ionizers may be similar; you must make your own calculations. In sum, ionization is not a harmless vitamin pill but strong medicine that has possible dangerous side effects and should be used only when risks are carefully controlled and outweighed by benefits. The resistance-to-ground for personnel may be 100 mn or more because charges drain otT even at very high resistance as shown by steady-state tests. The fallacy of this statement is that a steady-state test is unrealistic. The voltage spikes on seated operators being triboelectrically charged under real conditions, such as shuffling feet or brushing a garment sleeve over the work surface, must be measured. We did these measurements and found that MOSFETs were damaged at only about 8 mn resistance-to-ground when an operator stroked a plastic film, representing a sleeve, on a workbench under conditions where the individual generated -3,800 V when ungrounded (Ref. 2). The implications of such data are serious, and we recommend an allowable resistance-to-ground of 10 mn maximum (with the usual 1 mn minimum for safety). The maximum resistance should be controlled with a continuous wrist-strap monitor (see the next myth). Occasional checks of wrist straps are sufficient. Logically, this statement is inde• fensible because almost everyone agrees that 01 from operators' fingers is a major, or PAPER NO. 11 253 even the major, ESD-damage mechanism, and it is well known that this damage occurs in a matter of nanoseconds. Therefore, we can't drop our guard for even a fraction of a second-monitoring of skin resistance-to-ground must be continuous. Even if wrist straps were checked every hour (the real interval is usually daily or even weekly), when a failed strap is discovered, all the ESD-sensitive items handled by the wearer during the last hour are suspect. The only reasons continuous monitoring is illogically avoided are its per• ceived cost and trouble. But, in reality, this monitoring is practical. More and more con• tinuous monitors are being used in industry, especially on government programs, and we have found certain models to be quite cost-effective. We prefer the two-conductor resis• tive rather than the capacitance type. Note that dryness of operators' skin may raise the resistance-ta-ground far above 10 mO, even with a perfect wrist strap and good skin contact. The monitor also reminds operators to apply antistatic lotion, usually once or twice during a shift. All ESD damage is caused by a spark. As a bare, unqualified statement, this is mean• ingless. The first problem is semantic: What is a "spark"? Is it a visible, audible discharge? If so, an ungrounded operator's finger at 1000 V will damage a MOSFET, for example, by touching a lead even if there is no spark visible or audible. Even if a spark is defined as "any discharge causing damage by 01," electrical fields can act through space to destroy MOSFET's by the FIM with no spark at all. In the latter case, the spark occurs inside the device, but this definition of spark as "an internal discharge" results in a useless truism that all ESD damage is caused by discharges inside devices, which gives us no guidance in preventing these internal discharges. We want to know what happens outside a device to cause the "spark" inside. Therefore, a more productive state• ment is: "All ESD damage is caused by fields (FIM) or discharges (01, CDM)." The only useful idea in the "spark theory" is that a rapid discharge may cause damage under conditions where a slower discharge won't (see the next myth). Conductive surfaces are dangerous; antistatic surfaces are safe. We object to this generalization, though it contains more than a grain of truth. MIL-HDK-773 strongly favors antistatic surfaces as being safer for ESD-sensitive items, and our own data sup• port this. For human safety, conductive surfaces are obviously undesirable because they can carry a lethal current (over I milliampere). However, even static-dissipative surfaces can cause ESD damage to sensitive devices such as MOSFET's by the CDM, and avoid• ing conductive surfaces completely is impracticable because tweezers, soldering iron tips, and operators' fingers are ubiquitous in workstations. Our approach, which we consider a prudent compromise, is to use a mix of antistatic and conductive surfaces, with the latter being minimized. If handling procedures are correctly developed and followed (admittedly this is a big "if'), conductors will be grounded and ESD-sensitive items will be uncharged so the CDM can't happen with either a conductive or an antistatic surface. In other words, we're avoid• ing the trouble of covering fingertips with antistatic cots and so on by relying on operator skills to prevent 01 and CDM, which can still happen with antistatic surfaces, though to a much lesser extent. A relatively (not completely) "forgiving" work environment in terms of discharges, with all antistatic or static-dissipative surfaces, seems unfeasible to us, and damage by fields (by the FIM) could still happen. Antistatic surfaces (of DIP tubes, for example) do not triboelectrically charge pack• aged ESD-sensitive items by rubbing against them. In 1983, we were surprised to find that antistatic polyethylene (MIL-B-81705, Type II) charged epoxy-glass circuit-board material, and researchers at another major corporation were skeptical at first but verified 254 APPENDIX our results (Ref. 3). The truth is that the contamination effect (antistat transferring to the rubbed surface so that liquid separates from liquid with minimum charge generation) greatly reduces triboelectrification but does not prevent it. An antistatic material whose surface layer of antistat has been removed by volatilization or wear will not benefit from the con• tamination effect nor will permanently antistatic materials with dry surfaces. Furthermore, packaging materials that depend on a fugitive antistat will hold a charge when the antistat is sufficiently depleted. Triboelectric charging is quite material-specific, so tests should be run with the particu• lar packaged item and packaging material (both fresh and aged) in question. Highly conductive surfaces (of metals, for example) do not participate in triboelec• tric charging and are safe for ESD·sensltive items. Nothing could be further from the truth! Conductors only seem not to be triboelectrically charged because when they are grounded, the charge runs off almost instantaneously. An isolated sheet of aluminum or steel takes a charge of equal magnitude and opposite polarity to the charge taken by a plastic sheet stroked against it. In fact, grounded metals and carbon-loaded, conductive plastics are effective in charging nonconductors and causing damage by the COM, which can occur during the charging process with no discharge when, for example, an intense, uneven field is produced on a circuit board with mounted devices (we have called this the field-from-board or FFB damage mechanism). For example, epoxy-glass circuit-board lami• nate is readily charged by sliding on stainless-steel shelves in an oven, and people are charged by walking with insulated, rubber-soled shoes on a conductive floor. Ironically, such a floor, which was designed for use with conductive footwear, actually can make matters worse with ordinary shoes. Conductive flooring is always beneflcial for ESD control. We've just mentioned that a conductive floor may be the opposite of beneficial when ordinary, nonconductive shoes are worn. We say "may be" because shoe sole materials vary greatly in charging propensity. Rubber and certain vinyl compositions tend to give especially high charges. Another problem with conductive floors is safety. In one of our tests, operators with wet, leather-soled shoes (as they might have been ifthey'd come in out of a rainstorm) showed a resistance-to-ground low enough for a dangerous current of over I milliampere to pass through their bodies if they touched a 110 V hot lead. Ground-fault interrupters may be needed when such situations could occur. The point is that special footwear must be used with conductive floors, and the safety issue must be considered. Only then are conductive floors beneficial. Conductive chairs and special smocks are essential In a rigorous ESD·control program and allow some operator disciplines to be relaxed. Conductive chairs are possibly dangerous to both people and devices, insofar as a lethal current might be carried or the COM might be encouraged by the conductive surface, but we do not reject them on that basis alone. Our main reason not to use conductive chairs is that they are unnecessary; ordinary vinyl-upholstered chairs are not a hazard according to our measurements of electrical-field strength in the work area on the bench. Though vinyl does exhibit static charges, the fields from them are sufficiently attenuated by distance and the shielding effect of the operator's body to be harmless at the bench surface where ESO-sensitive items are being handled. This setup assumes that operators will not bring the items near the upholstery, as one of our operator disciplines specifically forbids. Thus, we are depending on operator skill to avoid the cost of expensive, foolproof equipment, but operators are always the critical factor no matter how much money is thrown at the problem of ESO control. PAPER NO. 11 255

As for smocks, the shielding effect of fabric with widely spaced conductive threads is dubious and, in any event, isn't needed with tight-fitting shirts or blouses with short or rolled-up sleeves. Proximity to the skin suppresses fields on the cloth, and distance to the operator's hands, in the normal working position, further weakens fields to the point of being harmless. You can make measurements with a hand-held field meter to support or refute this conclusion in your own assembly area. Another consideration is cost. Smocks are good uniforms for raising ESD awareness and impressing visitors, but they are rather expensive to collect, clean, and reissue. No equipment justifies relaxing operator disciplines. Operators are the "captains of the ship," and ineptitude on their part can sink the most lavishly capitalized ESD-control pro• gram. Forget expensive frills and invest in operator training. The hair hazard is eliminated by area ionization. Here's another case of expecting too much from ionization. As stated earlier, area (room) or any other kind of ionization should be used only for clearly defined purposes, and controlling static charges on hair is not one of them. Long hair will constantly recharge as it swings around, and the resulting field could cause damage by the FIM before air ions neutralize the charge. The best course is to require hair to be tied back so static voltages on it are suppressed by nearness to the skin. Use a hand-held meter to verify that the field in the vicinity of ESD-sensitive items is harmlessly weak, even with the operator's head bowed for close examination. We use the "Charge-Distance Rule," an empirical relation between the apparent voltage on a surface and the safe distance from that surface (Ref. 4). The equation is d = ~~:, where d is the distance in inches and V is the apparent charge in volts measured vith a hand-held meter. For example when V is 1,000 V, d is 18 in. Ordinary floors in ESD-controlled areas must always be resurfaced or specially treated. This assertion seems reasonable at first, but concrete floors, especially new ones, often contain enough moisture to be weakly conductive. The presence of salts in the con• crete also may help. Sealers may be soapy and afford low triboelectric charging of ordinary shoe soles, as does the thin layer of antistat on floors coated with a static-limiting floor finish. Therefore, concrete floors should be checked for surface resistivity if conductive footwear is used, or triboelectric charging of people (which correlates imperfectly with surface resistivity) should be checked if ordinary shoes are worn. Soiled. uncoated vinyl-asbestos tile, though unattractive in appearance, may also impart relatively low charges to people walking on it. In general, evaluate the floor before you consider altering it. If nothing else, baseline data will show the improvement obtained and help, in retrospect, to justify the cost of resurfacing (with conductive tile, for example) or treating (such as with static-limiting floor finish). All cellulosic materials change from antistatic to nonconductive at moderately low relative humidity (15 to 20%). We believed this myth until our testing disclosed a coated, plasticized (regenerated cellulose) that remains antistatic down to about 4% relative humidity at room temperature. (For further information, see EOS/ ESD Technology, OctoberlNovember 1989, p. 9.) Antistatic masking tapes are ESD-safe. As with many myths, the fallacv of this one lies in oversimplification. Yes, antistatic masking tapes are safe in the sense that they will not show a field because incipient charges drain to the worker's hand, while any field from the nonconductive adhesive layer is suppressed by the weakly conductive tape layer. How• ever, when the tape is pulled from a nonconductive surface (for example, epoxy-glass circuit-board laminate), that surface may be charged enough to damage mounted devices by 256 APPENDIX the FFB mechanism. The point is that charging of the substrate, as well as the tape itself, must be considered before generalizing that the tape is ESD-safe. Short drain times of charges from objects, such as tote boxes, deposited on work surfaces is of major importance and justifies expensive laminates and grounding schemes. Our experience and judgment indicate that situations where short drain time is critical are rare. A commonly cited scenario in support of short drain time is a grounded operator reaching into a charged tote box at rest on a bench and touching a sensitive device lead. Damage occurs by the CDM because the body of the device is charged by induction along with the tote box. However, such an incident is unlikely because the grounded opera• tor usually will touch the box and discharge it before reaching in. In fact, reaching in and touching a device lead without touching the box wall might be a difficult trick. We're satis• fied with a reasonably short drain time without taking extreme measures, such as using a conductive surface, which also has dangers, to minimize the time. Drain time has always been a secondary consideration, not a controlling factor, in our selection of work surfaces and grounding techniques. Overkill protection Is necessarily expensive. This myth may stem from the idea that you get what you pay for. However, overkill can come cheap. An example is a stone house built near a quarry; the stone will last for thousands of years but is cost-competitive with wood, which decays in a lifetime. Similarly, aluminum foil is a far more effective ESD shield than see-through, lOO-angstrom-thick metallization but is similar in cost. A laminate bag consisting of wire screen sandwiched between layers of antistatic poly• ethylene is indeed overkill, resisting even 35,000 V from a Tesla coil, but this packaging at a few dollars for a 10 x 12 in. bag is not really expensive when a $100.000 module for a vital aerospace or defense application must be protected. Continuous wrist-strap monitor• ing. discussed earlier, also may seem like overkill-though logically it is a necessity-but costly failures or material review actions are avoided. In fact, you can't afford not to have it. Latent ESD failures are commonplace. Perhaps we should call this one a "sus• pected myth" because no one is sure. A consensus among leading device experts, whom we questioned two years ago, was that latent failures are real but rare. In 1988, British Telecom researchers stressed components to 90% of their HBM threshold and then sub• jected them to accelerated testing to look for premature (latent) failures; however, they found none (Ref. 5). Of course, this negative result doesn't rule out latent failures under all conditions; any negative generalization is difficult if not impossible to prove. Intuition supports the latent failure concept when we picture stressed devices as "walking wounded" because wounded soldiers indeed may die early, but devices are not human nor wounded in the sense that flesh can be. Intuition often uses false premises and is wrong-though it could be correct by accident in this instance. The point is that skepticism is in order until strong evidence for latent failures accumulates. But even the possibility of a few latent failures in critical equipment, such as missile guidance systems or life-supporting medical appliances, is reason enough for a rigorous ESD-control program. Myths, old and new, are oversimplifications or bad generalizations that tend to lull be• lievers into a false sense of security while they rely on a product or approach that has fatal shortcomings or is needlessly expensive. For example, someone concerned only with "sparks" would ignore the FIM and FFB, and someone relying on ionization as a cure-all could have ESD-sensitive items damaged by poorly trained operators. In stating the truth as we see it, our object has been to dispel irrational fears, roused by some vendors' scare tactics, as well as foolish complacency. We have suggested when to spend and when to save because throwing money at the problem in the wrong places (for expensive, unnecessary equipment) and PAPER NO. 11 257 misguided thrift (refusal to buy continuous wrist-strap monitors) are both counterproduc• tive. Buy the basics, forget the frills. Our opinions are often controversial but supported by realistic tests as well as experience and common sense. They form the basis of a flexible, cost-effective system of ESD control that has proven itself over years of practice (Ref. 6).

References 1. GIDEP Alert H7-A-85-02, 1985. 2. Kolyer,1. M., D. E. Watson, and W. E. Anderson. "Controlling Voltage on Personnel," EOSIESD Symposium Proceedings, EOS-ll, 1989, p. 23. 3. Private communication with I. R. Huntsman and D. M. Yenni, 3M, 1983. 4. Kolyer, I. M., W. E. Anderson, and D. E. Watson. "Hazards of Static Charges and Fields at the Workstation," EOSIESD Symposium Proceedings, EOS-6, 1984. 5. Woodhouse, I., and K. D. Lowe. "ESD Latency: A Failure Analysis Investigation," EOSIESD Symposium Proceedings, EOS-IO, 1988. p. 47. 6. Kolyer, I. M., and D. E. Watson. ESD From A to Z: Electrostatic Discharge Controlfor Elec• tronics. Van Nostrand Reinhold, New York, NY, 1990. Paper No. 12

This paper is reprinted, by permission, from Evaluation Engineering, September 1991, pages 115-119.

ESD TESTING OF SILICON WAFERS

John Kolyer and Donald Watson

Rockwell International Corporation Defense Electronics Autonetics Strategic Systems Division San Bernardino, California

As a typical ESD-sensitive device, the Motorola 2N435 I Metal Oxide Semiconductor Field• Effect Transistor (MOSFET) has made an excellent "white rat" for our research. 1 However. this destructive testing has drawbacks: the cost of the MOSFETs as well as annoying delays when more pieces must be ordered. Now we have found that hundreds or even thousands of tests can be run on a single silicon wafer. The wafer is P-type silicon coated with 1,000 A of oxide. With either MOSFETs or wafers, a voltage differential, which may be momentary, punctures the oxide in a micro• scopic dielectric strength test. The wafer is 4" in diameter and 0.020" thick and has a resistivity of 5 to 60 n'cm, tvpicallv 15 to 25 n·cm. The oxide thickness is 800 to 1,200 A, nominally 1,000 A. This type of wafer serves to monitor sodium contamination in furnaces where the oxide is grown, so the upper surface holds a hundred sputtered aluminum dots for making electrical mea• surements to detect sodium. To make electrical contact with the underlying semiconductive silicon, one side of the wafer was broken off and solder drops (labeled "s" in Fig. I) were applied 2" apart on the exposed edge and pressed by spring-loaded contacts with terminals A and B. Resistance between A and B was I x lOs n at 10 V. The probe, electrically isolated with Teflon™, was held by a pivoted wooden arm which could be adjusted for reach or radius as well as rotated. The procedure was to start at maxi• mum radius and sweep the wafer at successively smaller radii. At first, an attempt was made to utilize the aluminum dots, but the best procedure, which provides an almost unlimited number of test points, was to avoid the dots and probe directly on the oxide. The steel probe, 0.112" in diameter with the tip polished to a radius of O.OOS", was springloaded to press the oxide surface with a force of 0.6 lb (Fig. 2). Resistance across the oxide layer was determined with a Beckman L-12 Megohmmeter with shielded leads (RG-58IU coaxial cables) at 10 V. The low or negative megohmmeter lead was connected to terminal T (which was connected to probe terminal P) while the high or positive megohmmeter lead was connected to A. Undamaged oxide read approximately 5 x 1012 n, with the reading influenced by capacitance of the wafer apparatus. Damaged oxide almost always read less than 5 x 10 10 n, usually in the lOS to 109 range but sometimes lower. Slight degradation, e.g., 5 x 1011 n, rarely occurred. Intermittents occasionally oc• curred, as with damaged MOSFETs.

258 PAPER NO. 12 259

Fig. 1. Wafer test setup.

CURRENT FLOW ~

PROBE ESD I EVENT r

.. : ....\.,: ...... : ...... • ~-- P - TYPE SILICON

s s (solder) (solder)

Fig. 2. Diagram of silicon wafer test. 260 APPENDIX

For comparison tests, MOSFETs were similarly measured with the low megohmmeter lead to the substrate-case lead of the MOSFET and the high megohmmeter lead to the gate lead of the MOSFET. Fresh MOSFETs read over 1013 0, and degraded MOSFETs usually read less than 1010 0, but subtle damage, e.g., 2 x 1012 0, occasionally occurred. The arbi• trary criterion for failure was a resistance of less than 1 x 1012 O. This is more rigorous than our former definition of failure as a shift of more than 0.1 V in gate-source threshold volt• age, which allowed considerable damage to the gate oxide. 2 The difference between the wafer and a MOSFET gate was leakage current. For ex• ample, when the voltage across the gate of a MOSFET was slowly raised, the resistance fell to 2 x 10 11 0 at 102 V (0.5 nA current) and 1 x 1010 0 at 110 V (11 nA) before failure occurred and the resistance fell several orders of magnitude. In a similar test, the wafer oxide began to leak (2 x 1012 0,0.05 nA) just before failure at 107 V. In MOSFETs, perhaps the major leakage path is over the surface of the chip to the nearby source and drain metallization. This leakage is significant in Charged Device Model (COM) tests involving charge drainage from a pin to a surface. However, in testing where the rise time is fast, gate leakage does not protect the MOSFET and it is successfully modeled by the wafer. The voltage sensitivity of the wafer apparatus was checked with a modification of the Human Body Model (HBM). 3 Terminal A was grounded, and a 1,500-0 resistor connected to a charged l00-pF capacitor was touched to terminal T. MOSFETs were tested by simi• larly touching the gate lead with the substrate-case lead grounded. Both wafer and MOSFETs began to fail at about 80 or 100 V (Table 1) as expected for 800 to 1,000 A of silicon oxide. The following four examples illustrate that the wafer method correlates with the MOSFET device tests. The examples deal with:

• Safe distance of ESO-sensitive devices from charged surfaces. • Maximum allowable resistance to ground for operators. • Hazard of small, charged, conductive objects on the workbench. • Superiority of foil-containing ESD-protective packaging material.

Example I. To demonstrate the Field-Induced Model (AM) for ESO damage, a I ftl aluminum plate on a nonconductive standoff was connected to terminal P, and terminal A was grounded. AcIar™ film (9" x 12") triboelectrically charged to -15,000 V was slowly moved toward the plate to induce a charge. In repeated tests, when the film was approxi• mately 40" from the plate, the oxide on the wafer failed as noted by a burst of RF heard from an AM radio tuned between stations. This result is consistent with our Charge-Distance

Table 1. Voltage Sensitivity Test.

FAILURE RATE (FAlLEDrrnsTED) VOLTAGE PROBE ON WAFER 2N4351 MOSFET

--60 0/5 0/5 -80 115 0/5 -100 1/5 115 -120 2/5 2/5 -140 SIS 4/5 -160 SIS SIS PAPER NO. 12 261

(CD) Rule, which specifies a conservative safe working distance of 68" from surfaces charged to 15,000 V.2 Example 2. In tests for allowable resistance to ground (ARTG) for personnel, an un• grounded person with a capacitance of approximately 200 pF stroked the Aclar™ film on a carpet. The film was quickly lifted, which charged the person to -4,000 V as read by a field meter on an electrically isolated aluminum plate connected to the individual's skin. The charging process was repeated with the person grounded by a wrist strap with a I-MQ resistor and a finger pressed against T (A was grounded). The failure rate (number failed/number tested) was 0/10. However, when the resistance in the strap was increased to 15 MO, the failure rate was 6/10. This result is consistent with previous data using MOSFETs.4 The explanation is that the voltage surge or spike exceeded 100 V and damaged the oxide when the resistance to ground was too high; the charge on the person could not be drained fast enough. In retrospect, the test could have used a wafer instead of MOSFETs to demon• strate this principle and set the maximum ARTG at 10 MO. Example 3. The setup in Fig. 1 modeled a circuit-board module with the lead from T to P representing circuitry leading to the oxide and the wafer representing circuitry beyond the oxide. The capacitance on the probe side was determined by charging T to -1,000 V with the probe resting on Teflon and reading the charge with a Keithley Model 617 Pro• grammable Electrometer. The charge was 12 nCo Therefore, C = QIV = 12 pF. By this method, a ceramic capacitor labeled 100 pF held 103 nCo On the wafer side, C was 34 pF. A discharge to T represented the so-called floating model in which an ungrounded ESO-sen• sitive item is touched by a charged surface.5 The question was: Are low-capacitance con• ductive objects, such as small screwdrivers, at comparatively low voltages really a hazard? To obtain an answer, the objects in Table 2 were touched to a power supply lead, and then to T. Alternatively, the probe was rested on Teflon, and the gate lead of a MOSFET was connected to P by means of a jumper while the substrate-case lead was connected via a jumper to B. The jumpers raised the capacitance from 12 to 14 pF on the probe side and from 34 to 36 pF on the wafer side. The voltages in Table 2 are at the moment of contact. For example, the ceramic capacitor was touched to the power supply lead at -1,000 V but lost 50 V, by leakage or by attracting air ions, during the one-second transfer to T. The data in Table 2 show the usual statistical scatter, but the wafer and MOSFET tests agree that small conductive objects are indeed hazardous at about 500 V/5 pF or 3,000 V/l pF, even with considerable capacitance (12 pF) in circuitry leading to the oxide layer. The inverse situation is the COM in which charged DIPs with a capacitance of only 1 to 3 pF are damaged by discharge from a lead to a workbench or other surface.s

Table 2. Floating Model Test.

CAPACITANCE, pF, BYELECI'ROMETER FAILURE RATE (FAILEOImSTED) CAPACnuR METIiOD VOLTAGE PROBE ON WAFER 2N43S1 MOSFET

Ceramic, rated 3.3 -480 2/5 0/5 at 3.3 pF -950 3/5 4/5 Small screwdriver with 3" 4.5 -450 3/5 1/5 blade and plastic handle, -900 5/5 2/5 held in grounded hand I" nail embedded in rubber, 0.9 -1,000 0/5 0/5 held in grounded hand -3,000 3/5 4/5 262 APPENDIX

Example 4. A capacitive probe was constructed to two O.75"-diameter brass disks, 0.009" thick, separated by 0.5" of nonconductive plastic. The inner conductors of l' lengths of RG-58/U coaxial cable were soldered to the disks and the cable shields were grounded. The probe was placed inside an ESD-protective bag which was taped to a grounded alumi• num plate so that the bag material was stretched tightly over the upper disk. The lead from the upper disk was connected to T while the lead from the lower disk was connected to A. A 200 pF capacitor charged to -3,800 V was touched to the bag above the upper disk, and then the probe was disconnected and the wafer oxide was tested as usual. Table 3 shows the results, which agree with the MOSFET data in similar testS. 2•6 The antistatic polyethylene and black conductive plastic bags were not sufficiently conductive to prevent a damaging electrical pulse between the disks. The nickel-coated bag was super• ficially conductive, but the spark discharge removed the metallization, resulting in a bare spot 0.01" wide. Only the foil-containing bag succeeded in preventing a momentary poten• tial of 100 V or more across the oxide.

Table 3. Packaging Test.

FAILURE RATE (FAu.EDtrESTEDl BAG BY PROBE ON WAFER

Antistatic polyethylene (MIL-B-81705, Type ll), 0.006" wall 5/5 Black conductive plastic, 0.004" wall 5/5 Metallized polyethylene with 100 Aexternal nickel 5/5 Laminate containing 0.00025" aluminum foil 0/5

Summary

In conclusion, the wafer test duplicates results with actual devices when the rise time is fast. Data are somewhat scattered, just as with MOSFETs. The practical object is to obtain a failure rate of 0/5 in a worst-case test to establish safe working conditions and to select ESD-protective packaging.6 A MOSFET test has been recommended for packaging specifi• cations, but the method of Example 4 would do as well.·

References 1. J. M. Kolyer and O. E. Watson, ESD from A to Z: Electrostatic Discharge Control for Electron• ics, Van Nostrand Reinhold, 1990, pp. 21 and 96. 2. J. M. Kolyer, W. E. Anderson, and O. E. Watson, "Hazards of Static Charges and Fields at the Work Station," EOSIESD Symposium Proceedings, 1984, p. 7. 3. OoO-HDBK-263, May 2,1980, p. 24. 4. J. M. Kolyer, O. E. Watson, and W. E. Anderson, "Controlling Voltage on Personnel," EOSIESD Symposium Proceedings, 1989, p. 23. 5. O. J. McAteer, Electrostatic Discharge Control, McGraw-Hili, 1989, pp. 176 and 189. 6. J. M. Kolyer and W. E. Anderson, "Perforated Foil Bags: Partial Transparency and Excellent ESO Protection," EOSIESD Symposium Proceedings, 1985, p. 111. PaperNo. 13

This paper is reprinted, by permission, from Evaluation Engineering, October 1991, pages 11 0-117.

CDM AND WORK SURFACE SELECTION

John M. Kolyer and Donald E. Watson

Rockwell International Corporation Defense Electronics Autonetics Strategic Systems Division San Bernardino, California

Work surfaces for ESD control must be conductive enough to allow charges to drain from conductive objects, yet resistive enough to restrict the current flow to a safe level in terms of the Charged Device Model (CDM).I An unsafe level of current flow has been equated with sparking.2•3 A study was undertaken to determine the best compromise between drain time and CDM protection based on device damage data, not mere intuition. Using a Beckman L-12 Megohmmeter set at 10 V with the high or positive lead con• nected to the grounding lug of l-ft2 panels or mats, or clamped to the edge of bag samples, the resistance to ground was measured in two ways:

• With the low or negative lead of the megohmmeter connected to a dual in-line package (DIP) square-tipped lead 0.015" wide and 0.009" thick (terminal D in Fig. I) pressing the surface (S) with a force of 0.008 lb. • With the low lead of the megohmmeter connected to a National Fire Protection Asso• ciation (NFPA) 99 electrode (5 lb., 2.5" diameter, aluminum foil over rubber) resting on the surface.

Eleven surfaces were tested with the following results (resistance in ohms through D is given in parenthesis):

• Aluminum alloy 6061-T4 sheet (approximately 10 0, but varied from 3 to 1,100 0). • Carbon-loaded conductive polyolefin bag (l x l()!l 0). • Hard laminate with conductive fibers (5 x 1()!l 0). • Bag with external nickel (2 x 101 0). • Rubber mat (I x 109 0). • Special vinyl mat (6 x 109 0). • Three hard laminates (2 x 1010 0, 5 X 10 10 0, 7 X 1010 0). • Antistatic polyethylene (MlL-B-81705, Type II) (1 x 1011 0). • Painted steel as used for office furniture (5 x 1012 0).

Except for the aluminum and the nickel-coated bag, the NFPA 99 values are plotted against the D values as shown in Fig. 2.

263 264 APPENDIX

""--.11', \ If.\

\ \ IJ2 I

CODE: D • DIP LEAD SOLDERED TO WIRE w B • TEFLON BLOCK ON WHICH D RESTS S • SURFACE UNDER TEST T • REMOTE TERMINAL FOR PROBE P • PROBE TERMINAL W • WAFER s • SOLDER A,B • WAFER TERMINALS J1, J2 • JUMPERS G • GATE LEAD OF MOSFET SC • SUBSTRATE-CASE LEAD OF MOSFET

Fig, 1. CDM apparatus.

The CDM was modeled using the setup of Fig. 1 as explained in Ref. 4. To represent a charged device, the switch was closed, and the powered (negative) lead from a calibrated power supply (Power Designs. Inc., Model 3KlOB) was connected through an 8-MO resis• tor to terminal A before turning on the power. With the power supply turned on before connecting it to terminal A, the 1,000 A oxide layer on the wafer and a MOSFET with its leads shunted were damaged. This illustrates the fact that shunting does not necessarily provide ESD protection. The MOSFET is tested by moving the probe to a Teflon™ surface and using the jumpers as shown in Fig. 1. With the assembly charged, the power supply lead was disconnected from terminal A, the switch opened, and block B (0.25" thick) removed to let D fall to the grounded surface

S under test. As shown schematically in Fig. 3, capacitor C2 (representing the capacitance of the wiring from terminal T to P and the probe itself when contacting the wafer) then drained into the test surface through resistance R2, while C. (representing the capacitance from the oxide layer through terminal A, the switch, the wire w and to terminal D) was insulated by the wafer oxide or by the gate oxide of the MOSFET. The l4-pFcapacitanceofC. is worst• case for a large DIP; ordinary DIPs are 1-3 pF at the work surface.s The RC time constant for C2 was as expected, given measurement uncertainties, e.g., approximately 3 s for anti• static polyethylene vs 1.7 s calculated, with the fall in voltage at terminal A being moni• tored with a Trek Model 512 Field Meter. After 5 s, the switch was closed and terminal T was touched with a grounded I50-MO resistor to discharge the setup before measuring the resistance of the wafer oxide or MOSFET PAPER NO. 13 265 gate with the megohmmeter. Failure was judged by a significant resistance drop as ex• plained in Ref. 4.

The wafer oxide failed when the voltage at C2 decreased by more than 100 V, indicating that the dielectric strength of the oxide was exceeded. Even with the resistance through D 12 (Rz in Fig. 3) at 4 X 10 n, the wafer oxide finally failed at 20 s when drainage of CI was allowed to continue. At the arbitrary drainage time of 5 s, only painted steel was safe among the surfaces tested. In contrast to the wafer oxide, MOSFETs were partially protected by leakage current, as explained in Ref. 4. The failure rate (number failed/number tested) decreased with increas•

ing resistance R2• The rate did not approach zero until approximately 1010 n at 500 V/14 pF, 1011 nat 500 V/214 pF, or 1012 nat 1,000 V/14 pF as shown in Fig. 4. The failure rate reflected no sudden "sparking point" as the resistance decreased; rather, there was a smooth rise in the probability of failure. Without leakage current, MOSFETs would be damaged even by antistatic polyethylene at 500 V/14 pF, as is the wafer oxide. But, in fact, as the resistance through D increases, the leakage rate exceeds the drainage rate; therefore, the MOSFETs are not damaged. By shunting the wafer with a lOS n resistor between terminals P and A, wafer failure began abruptly below lOS n as expected. Thus, the shunted wafer can serve for testing purposes as an idealized 50%-failure-rate MOSFET as shown in Fig. 4. The conclusion drawn from Fig. 4 is that the optimum resistance through D is in the 1010 to 1011 n range, corresponding to an NFPA 99 measurement of 107 to lOS n for the data points in Fig. 2. The rubber mat with a resistance through D of 1 x 109 n (l x 10«' n by NFPA 99) was judged too conductive (approximately 40% MOSFET failure), while the

~ 10 f M 81 ~ u.z 1) i ~ C!) 6 g § III 5 Jj I a: '0 4 J 3 4 5 6 7 8 9 10 11 12 13 Log of Resistance to Ground through 0, Ohms

Fig. 2. Correlation of resistance to ground data. 266 APPENDIX

12pF lor Probe on Warer. 14 pF lor MOSFET 34pF lor Probe on Warer, 36 pf lor MOSFET 105 otmls (rrom A to B. Figure 1) Resistance to Ground through 0 (Figure 1) Total Resistance rrom Olide 10 Ground

Fig. 3. Schematic for CDM apparatus (Fig. 1) with switch open and D resting on S. vinyl mat at6 x I()'I n was borderline, The hard laminate at 5 x 0' 1 was clearly unsuitable, but fortunately most hard laminates for ESD control give 1010 to 1011 n through D as desired. Incidentally, the CDM danger of conductive bags lying on the work surface is apparent from Fig. 4. Though the preferred hard laminates are safe for MOSFETs at 500 V, damage does occur at 1,000 V, especially at higher capacitance levels, as shown in Fig. 4 and reported previ. ously.6.7 Therefore, even these surfaces must be considered possibly hazardous todevices in Class I ofMIL·STD·1686A. If drain time is not an issue, a nonconductive surface with an underlying conductive layer for voltage suppression is best.' Skin resistance varies greatly but was on the order of 107 n through 0, corresponding to a MOSFET failure rate of 60%. 'Therefore, a grounded operator is the lesser of two evils vs an ungrounded operator.' This COM hazard can be alleviated by wearing antistatic gloves or finger cots or avoided altogether by wearing nonconductive latex gloves or cots.' How· ever, the wearer of nonconductive gloves or cots must remember that charged objects can· not be discharged by touching as they nonnally are, and even small charged objects at relatively low voltage, e.g., a screwdriver at 5 pF and 500 V, can cause ESO damage by direct injection or the floating model.' PAPER NO. 13 267

Tests also were run with 1% tolerance, 1/20-W thin-film resistors (Class 1 of MIL-STD-

1686A) shunted between terminals P and A in Fig. l. At 3,000 V and C1 = 214 pF, a hard laminate (5 X 1010 0 through D) caused less than 0.01 % resistance shifts for 10-0, 100-0, lO-kO, or 3OO-ill resistors. However, the carbon-loaded conductive bag shifted the resis• tance of the 3OO-k 0 resistor by -0.05%, and the aluminum surface caused shifts in all the resistors to a maximum of -0.38%. The conclusion is that the CDM is not a major hazard for thin-film resistors with I % or even 0.1 % tolerance when the work surface is a hard laminate with 107 to lOS 0 resistance to ground by NFPA 99. ICs or discrete devices usually will not exceed 500 V when near the work surface be• cause of voltage suppression. In fact, the inputs of most ICs now are protected by zener diodes to at least 1,000 V. Therefore. the major concern may be for Class I devices mounted on circuit boards that have relatively high capacitance and also high voltage because a large board tilted with respect to the work surface escapes effective voltage suppression. Accord• ingly modules should be checked with a field meter and, if necessary, neutralized with

5/5 ////////////////////////////////////~"""""""'~

Shunted Wafer. ~ '~ 500orl000V. ~ \.; 12pF / ~ ~ \.; / ~ 4/5 ~ Unshunted Wafer. ~ • ~ 500 or l000V. ~ ~ 12pF ~ 500V ~ \.; / ~ ~ \.; / ~ 3/5 () ~ () ~ ~\ \.; ____ ~%..:ai~re______/'~ __(OOOV ~ ,~ ~ ~ ~ 215 MOSFETpata • .\\~ () ~ • 500V.14 pF \ ~ () 1000V. 14 pF \ ~ :-.; D 500V. 214 pF '. , 1/5 ~~. \r500V. \~ 214pF · ··

2 3 4 5 6 7 8 9 10 11 12 13 Log of Resistance to Ground through D, Ohms

Fig. 4. MOSFET and wafer failure rates. 268 APPENDIX

ionized air before a sensitive tenninal or edge contact touches the bench surface or the operator's fingers. In conclusion, while supporting the popularity of work-surface laminates with an NFPA 99 resistance to ground of 101 to lOS 0, the data make the important point that these surfaces are only relatively CDM-safe. They are nonsparking, but discharges can cause ESD dam• age without an evident spark. For example, in many cases, D fell to surface S and damaged MOSFETs with no RF heard from an AM radio tuned between stations. Therefore, operator disciplines must assure that the ability of the surface to drain charged objects does not become a liability by damaging voltage-sensitive devices.

References

1. Carlton, D., "Selecting an ESD Workstation," Evaluation Engineering, January 1991, p. 94. 2. "New Test Proposed for Checking ESD Safety of Materials," Compliance Engineering, Fall 1990, Vol. VII, Issue 5, p. 77. 3. "Electrostatic Discharge Protective Packaging," MIL-HDBK-773, April I, 1988. 4. Kolyer, J. M. and Watson, D. E., "ESD Testing With Silicon Wafers," Evaluation Engineering, September 1991, p. liS. 5. McAteer, O. J., Electrostatic Discharge Control, McGraw-Hill, 1989, p.176. 6. Kolyer, J. M., Anderson, W. E., and Watson, D. E., "Hazards of Static Charges and Fields at the Work Station," EOSIESD Symposium Proceedings, 1984, p. 7. 7. Kolyer, J. M., Anderson, W. E., and Watson, D. E., "Tote Box Material: How Good Is It?" EOSIESD Technology, OctoberlNovember 1987, p. 13. 8. Kolyer, J. M. and Watson, D. E., ESD from A to Z: Electrostatic Discharge Control for Electronics, Van Nostrand Reinhold, 1990, p. 121. 9. Kolyer, J. M. and Watson, D. E., ESD from A to Z: Electrostatic Discharge Control for Electronics, Van Nostrand Reinhold, 1990, p. 46. PaperNo. 14

This paper is reprinted, by permission, from EOSIESD Technology, February/March 1992, pages 27-28.

IS YOUR WORK SURFACE CDM SAFE?

John M. Kolyer

Rockwell International Corporation Defense Electronics Autonetics Strategic Systems Division San Bernardino, California

The charged device model (CDM), in which charge on the circuitry of an ESD-sensitive device flows from a contact, lead, or terminal to a benchtop or other surface, has become the focus of increasing concern. The flow has been found to create sufficient voltage differ• ential or current within the device to puncture gate oxide, degrade thin resistor films, or cause other overstress damage. I Because of this, MIL-HDBK-773 warns against conduc• tive surfaces. Obviously, a highly conductive surface, such as metal, is unsafe by the defmition of the CDM as a discharge-caused failure model. Conversely, a nonconductive surface is assumed to be safe because discharge is avoided. However, some surface conductivity is desired to drain static charges from objects such as tote boxes. A tradeoff is needed; there must be a surface-to-ground resistance high enough to suppress the CDM while not overextending the drain time. One study found that the resistance to ground for work surfaces was correlated with the CDM failure rates of MOSFETs sensitive to lOOV. (These MOSFETs are more easily dam• aged than most, but not all, devices).2 The conclusion was that the resistance through a lead that contacts the surface should be above 1010 ohms, because the MOSFETs were protected by a 1010 ohm leakage path. When the resistance through the lead was 109 ohms, for ex• ample, leakage around the gate oxide of the MOSFET was not fast enough to prevent a damaging voltage differential of lOOV. As seen in Fig. I, a conductive rubber mat and a laminate with conductive fibers were unsafe, while painted steel was safe but nonconductive. Between these extremes are the best surfaces: hard laminates resistive enough to prevent CDM damage yet sufficiently conductive to drain charged objects. How can you evaluate your own work surface? Here are three methods:

1. The conventional method measures resistance to ground with a 5-lb., 2.5-inch-diameter NFPA 99 electrode connected to the negative (low) lead of a megohmmeter. The positive (high) lead is connected to the grounding lug of the work surface. At 10 V the resistance should be 10'_108 ohms as seen in Fig. l. Note that results may be misleading for unhomogeneous surfaces because the large electrode cannot detect small conductive spots or fibers that a device lead might find. For example, a laminate with conductive fibers on

269 270 APPENDIX

4 5 6 7 8

Log of Resistance to Ground through NFPA 99 Electrode, Ohms

80

Brlghl (I mAl Z" '# c: oj ~ iii :> 60 CC 2- I!! ~ S ~ .r:: I- .~ Dim W LL CO (I l1li) lJ) ,Q 0 :> CO 40 ::E gc: Z Very Dim (0.3l1li) Dark (O.lI'A) 20

Log of Resistance to Ground through DIP Lead, Ohms

2 3 4 5 6 7 8 9 10 11 12

LAMINATE SKIN CONDUCTIVE HARD PAINTED WITH RUBBER LAMINATES STEEL CONDUCTIVE MAT FIBERS• • • • •

Fig. 1. MOSFET failure rate by CDM versus resistance to ground and neon bulb brightness. the surface gave a resistance through a device lead of 5 x 10' ohms with the lead resting on a fiber vs. 5 x 1011 ohms between fibers. 2. The method discussed in Ref. 2 measures resistance at 10 V through a dual in-line package (DIP) lead (0.015 x 0.009 in. in cross section) pressing the surface with a force of 0.008 lb. The lead is shifted from place to place to identify the lowest resistance, which should be above 1010 ohms. (See Fig. 1.) 3. A neon bulb test was evaluated using the DIP lead with the specified force, because pressure on soft surfaces (e.g, a rubber mat) greaty affects resistanceY An N-2 neon bulb with maximum striking voltage of 90 V and design current of 0.6 rnA was in series between the DIP lead and the negative lead of a megohmmeter at 500 V, while the positive megohmmeter lead was connected to the grounding snap or conductive layer of the material under test. The bulb became very dim at a little above 109 ohms through the DIP lead, corresponding to 0.3 microampere. PAPER NO. 14 211

The test did not detect COM-hazardous surfaces at the low end of the MOSFET failure rate curve. However, the bulb did provide a visual demonstration of the worst COM-offenders by glowing brightly and flashing on contact. With unhomogeneous, ungrounded surfaces not amenable to a resistance measurement, a flash of the bulb would suggest a COM hazard. In conclusion, COM-safe laminates or other COM-safe materials, such as soft mats, can be identified by a resistance measurement. Measurements can be taken using the standard NFPA 99 electrode or a small-area probe that simulates a device lead and detects conduc• tive inhomogeneities.

References

1. McAteer, O. I., Electrostatic Discharge Control, McGraw-Hill, 1989, p. 173. 2. Kolyer, 1. M. and Watson, D. E., "The Charged Device Model and Work Surface Selection," Evaluation Engineering, October 1991, pp. 110-\17. 3. "New Test Proposed for Checking ESD Safety of Materials," Compliance Engineering, Fall 1990, p. 77. 4. Anderson, D. C., "A Simple Approach to ESD Damage Prevention," EMC Technology, March/April 1991, p. 38. PaperNo. 15

This paper is reprinted. by permission. from EMC Test and Design. September 1993. pages 28-31.

REALISTIC TESTING OF ESD MATERIALS

John M. Kolyer

Rockwell International Corporation Defense Electronics Autonetics Strategic Systems Division San Bernardino. California

Data-sheet material properties such as surface resistivity. shielding effectiveness. and tri• boelectric charging propensity are measured under arbitrary. fixed conditions and cannot predict performance under the varied. dynamic conditions of use. Thsts that realistically simulate electrostatic discharge (ESD) damage mechanisms are needed for selecting ESD• protective materials with confidence. This article briefly describes each of the common ESD damage mechanisms~2,3 and gives examples of relevant tests.

Human-Body Model (HBM)

In the HBM. a charged person touches a lead or terminal of an ESD-sensitive device and causes a damaging flow of electrons to or from the device. The current may cause damage by its energy (current-sensitive devices. e.g .• thin-film resistors). or it may produce a differ• ential voltage that exceeds the dielectric strength of a thin oxide layer [voltage-sensitive devices. e.g .• MOSFETs (metal-oxide semiconductor field-effect transistors)). The HBM mechanism is a major. if not tlu! major. cause of ESD damage. so the skin voltage of personnel handling ESD-sensitive items must be strictly controlled by ground• ing. Wrist straps are the usual means. and the skin-to-ground resistance should be limited to a maximum of 10 Mel to prevent harmful surges of voltage when operators triboelectrica1ly charge themselves by motions such as foot shuffling.l Recently. a cordless wrist strap of novel design appeared on the market. The principle of operation appears to be corona discharge from fibrous material in the strap rather than the draining of charges to ground. A sample strap was tested in two ways. First. a person wearing the cordless strap aimed a static (field) meter at a grounded steel surface while stroking his rubber shoe soles on a nylon carpet. When he lifted one foot to reduce his capacitance the meter read +2 kV and his skin voltage was -2 kV. The same result occurred without the wrist strap. demonstrating that the strap was ineffective under the conditions of the test. In the second test. the strap-wearer sat in a chair on the same carpet and shuffled his feet without lifting them. Then he touched the cap of a 3M Static Event Detector™ (SED) with the case grounded. The SED is diagrammed in Fig. 1. (The shunt with the neon bulb pro• tects the SED from voltages above 5 kV that could damage it.) A general discussion of the

272 PAPER NO. 15 273

NE-2 SED Bulb

Fig. 1. Protected SED.

use of the SED in ESD process control has been published.4 An advantage of the SED is ease of use, because a liquid crystal display (LCD) changes color when the monitor is tripped, and no measurements or readings are needed. The SED used in this test had a capacitance of 150 pF and was tripped by voltage surges of 92 V or more, so it represented a MOSFET with a damage threshold of approximately 100 V. In repeated trials, the seated strap-wearer tripped the SED, demonstrating that MOSFETs would have been damaged by being handled under analogous conditions (gate lead touched, substrate-case lead grounded). The conclusion was that the HBM hazard was not significantly reduced by the cordless wrist strap. A cord to ground, including a current• limiting resistor to protect the wearer from shock, remains a necessity. To reduce the chance of HBM damage, the skin voltage of people wearing ordinary shoes can be lowered by special floor materials. This partial control of skin voltage is a "safety net" to protect ESD-sensitive items if they are accidentally touched by ungrounded personnel. Also, there is less chance of fingers throwing sparks on dry days and upsetting equipment in computer rooms. A practical test for candidate flooring is measuring the skin voltage on twelve randomly selected people wearing their ordinary shoes and walking nor• mally. The walker holds a probe connected to an electrically isolated plate monitored by a static meter. 2 A static-limiting floor finish with a microscopically thin surface layer of antistat reduces the skin voltage of a person on a tile floor to approximately 10 percent of that found for conventional acrylic finishes or "waxes.''2 An alternative to this special finish, which must be carefully applied and maintained, is conductive carpet. Two types of such carpet were evaluated with the results shown in Table 1. Most of the shoe soles, and all the heels, hap• pened to be rubber rather than leather. The relative humidity was 40-42 percent at 75°F. As seen in Table 1, charging varies greatly among people in this test, so data must be averaged. Also, the brand of acrylic finish, as well as the condition of the tile, including soil• ing, affects the results. Carpet 1, which contained a proportion of conductive fibers, reduced people's skin voltage to 42 percent of the level for the tile with conventional floor finish, while Carpet 2, in which all the fibers were conductive, reduced the voltage to 25 percent. These carpets did not perform as well as static-limiting floor fmish, but they may be suitable for replacing high-charging conventional carpet in computer rooms or office areas.

Machine Model (MM)

In the MM a damaging current flows between a metallic object, such as a tool, and an ESD• sensitive device. The MM is simulated by tests in which the resistance between a charged capacitor and a probe touched to a device (or sensor representing a device) is nearly zero, 274 APPENDIX

Table 1. Walk Test Data.

SKIN VOLTAGE, kV

TILE WITH REGULAR TILE WITH REGULAR PERSON ACRYLIC FINISH (WAX) ACRYLIC FINISH (WAX) NO. CARPET 1 (CONTROL FOR CARPET I) CARPET 2 (CONTROL FOR CARPET 2)

I +0.5 -2.2 ...{}.OO5 ...{}.02 2 +0.5 ...{}.8 +0.1 "'{}.4 3 +0.8 ...{}.7 +0.2 "'{}.5 4 +0.25 -2.5 ...{}.OO5 ...{}.15 5 +0.45 ...{}.9 +0.15 "'{}.3 6 +0.8 "'{}.5 +0.1 ...{}.9 7 +0.8 ...{}.8 +0.05 "'{}.5 8 +0.5 ...{}.7 ...{}.01 "'{}.2 9 +0.4 -1.7 +0.02 ...{}.15 10 0 ...{}.3 +0.4 -1.0 11 +0.05 ...{}.6 +0.1 "'{}.5 12 +0.01 ...{}.5 +0.1 "'{}.I5 Average +0.42 -1.0 +0.10 ...{}.40 whereas HBM tests involve a resistance of 150-1500 n to represent that of a person. Since the MM is the more severe condition, it is appropriate for worst-case tests of packaging materials. A convenient method for testing bag materials is to place a protected SED (Fig. I) inside the bag, which rests on a grounded aluminum plate, and then to touch a charged probe to the bag above the cap of the SED. A probe with a diameter of 0.025 inch represents a sharp hand-held tooLS More severe tests6 use a rounded probe with a diameter of 0.25 inch to represent a blunt tool. With the 0.2S-inch probe at 10 kV, the equivalence of the SED to an oxidized wafer7 failing at an average of 117 V was demonstrated as shown in Table 2, and the wafer test (Fig. 2) in turn correlated with MOSFET testsY As seen in Table 2, only packaging films with an opaque metallic layer passed the severe lO-kV, blunt-tool test with no failures in five trials. Such materials are rated "excellent." Materials failing the blunt-tool test at 10 kV but passing at 6 kV are rated "good"; these include MIL-B-S170SC Type III (see-through film laminate with vapor-deposited alumi• num protected by O.S-mil polyester film). Materials passing only the sharp-tool testS at 9 or 10 kV are rated "fair," and materials failing all three tests are "poor." "Poor" materials include see-through film laminates with the metal "out" (protected only by a thin organic coating rather than polyester film). The spark thrown by a finger on a dry day will create a hole in the metallization and trip the SED inside metal-out bags, even though these bags pass the EIA 541 (Electronic Industries Association) shielding test. The problem with the EIAtest is that it uses large, flat electrodes at relatively low voltage (I kV) and does not include real-life spark discharges. This misleading test gives a good rating to volume-conductive, carbon-loaded bags which perform poorly in our tests as well as in tests recently conducted in Germany using the EIA 541 network with a krypton switch.s A realistic approach demands that if spark discharges are a threat, they must be accounted for in the test method. Another example of the MM is discharge between a charged, isolated conductor and an ESD-sensitive device. Isolated conductors may attain considerable voltages by PAPER NO. 15 275

Table 2. Data for Packaging Materials Tested at 10 kV.

FAILURE RATE (FAILEDfTESTED)

MATERIAL 1lllCKNESS, MILS OXIDIZED WAFER SED

Antistatic polyethylene 6 4/5 5/5 (MIL-B-81705C Type n) Black, conductive 4 5/5 5/5 polyethylene Foil laminate 9 0/5 0/5 (MIL-B-81705B Type I) Foil laminate, antistatic 9 0/5 0/5 outside, antistatic polyethylene inside Metal-in, see-through 3 4/5 4/5 aluminum (MIL-B-81705C Type m) Metal-out, see-through 3.5 5/5 5/5 copper Aluminum foil 1.3 0/5 0/5 Metal-in, opaque aluminum 4 0/5 0/5 (MIL-B-81705C Type I) MIL-B-81705C Type m 198 total 2/5 3/5 lined with antistatic foam Corrugated cardboard 130 5/5 5/5 space-charging in the vicinity of air ionizers, so tests were run with a 20-pF plate placed at various distances from ceiling-mounted. pulsed DC emitters. A potential relative to ground of 250 V was required for the 20-pF plate to trip the SED (Fig. 1) with case grounded. This result is consistent with experience7 for low-capacitance charged objects. When the plate touched the cap of the SED. the capacitance rose so that the original 250 V fell to approximatelv the 92-V threshold voltage for tripping the sensor.

Packaging Material Discharge

Insulating Pivot Arm

Oxide

Conductive P-Silicon Adhesive

Fig. 2. Wafer test apparatus. 276 APPENDIX

For 250 V on the plate, cycle time was plotted against emitter-to-plate distance as shown in Fig. 3. The normal operating range for emitter voltage and working distance lay in the safe area of the graph and the conclusion was that space charging of 20-pF isolated conduc• tors would not threaten devices, such as MOSFETs, mimicked by the SED.

Charged Device Model (COM)

In the CDM, an ESD-sensitive device (discrete or mounted on a circuit board) becomes charged. For example, a dual in-line package (DIP) may be triboelectrically charged by sliding inside a DIP . Then a lead or terminal discharges to a conductive surface. The question arises: How conductive is "conductive" to cause the CDM discharge? Attempts have been made to find an answer.9•1O One method of testing surfaces for CDM safety uses the SED as shown in Fig. 4. The double-pole switch is closed, the power supply is turned on to charge the apparatus to I kV, the switch is opened, and then the probe is touched to the surface under test. Capacitances

C 1 and C2 are 5-10 pF to represent circuit lines connected to a board-mounted MOSFET as represented by the SED.

6

oI 5 Danger Zone (250 V or More on Plate) at 22 kV on Emitters

.I.... 3 ...GI ~ (No Danger 2 Zone at 10- /0 fIj 12 kV on Emitters) Nonnal Operating

1 ,/ R~

1 2 3 4 5

Distance from Emitters to Plate, feet

Fig. 3. Pulsed air-ionizer data. PAPER NO. 15 277

Probe

Fig. 4. CDM test schematic.

7 The RC time constant for draining C2 at 8 pF is 8 x 10- second (rapid discharge) for probe-to-surface R = 1()5 ohms and 0.8 second (slow bleed-off) for R = 1011 ohms. In prin• ciple, this constant could be calculated. However. in practice, R and C usually cannot be measured with confidence, so an empirical test is necessary. In general, materials with a surface resistivity of at least 109 ohms/square are CDM-safe at I kV and C, =C 2 =5-10 pF. But keep in mind that surface resistivity readings vary with the manner of measurement, including electrode configuration. Curves of surface resistivity versus relative humidity (RH) can be derived by confining samples with saturated chemical solutions that control the relative humidity in the air above them. II As seen in Fig. 5, the resistivity of many materials varies greatly with RH, so the same material may be CDM-safe at low humidity but unsafe at high humidity. At approxi• mately 50 percent RH, vulcanized fiber was well above 109 ohms/square and passed the above-mentioned CDM test with the SED; this material was selected for tweezers used to handle ESD-sensitive dice. Similarly, a recently developed antistatic nylon for probe tools for assembling the same dice passed the CDM test at 50 percent RH. A permanently anti• static vinyl plastic, suitable for tote boxes, also passed. In contrast, a carbon-impregnated cardboard packaging material had a surface resistivity less than lOS ohms/square at 50 per• cent RH and failed the CDM test.

Field-Induced Model (FIM)

In the FIM, an electrical field (E-field) induces a charge on an ESD-sensitive device to cause failure. The SED is unsuited for FIM simulations because it is tripped by voltage surges rather than by overvoltage gradually accumulated as a device is moved into an E-field. The oxidized wafer or actual MOSFETs must be used. In the case of the FIM, or any ESD damage mechanism, actual devices or an SED may be needed to sense ESD events because the risetime, e.g., 100 picoseconds, may be too brief for any oscilloscope to measure. '2 278 APPENDIX

1013 ...... ~ .... _ ...... ~ ......

v • Antistatic Vinyl F • Vulcanized Fiber N • Antistatic Nylon C • Carbon-Illpregnated Cardboard

105 ..~ __.. ____ .. ~ .. ~ __~~ o 10 20 30 40 50 60 70 80 90 100 RH,%

Fig. 5. Surface resistivity vs. relative humidity.

At assembly benches or shipping/receiving stations, the source of a damaging E-field can be a nonconductive tool or container. Therefore, nonconductors should not be used in handling ESD-sensitive items. Materials for this purpose must be sufficiently conductive for charges to drain to ground, via a grounded operator's hand or a static-dissipative work surface, within a few seconds. As a rule of thumb, the surface resistivity should be below 10'2 ohms/square for a material to be antistatic and not retain charges, and curves like those in Fig. 5 show the cut-off RH at which 10'2 ohms/square is reached. Vulcanized fiber re• mains antistatic down to approximately 3 percent RH and is judged to be ESD-safe under normal working conditions. The other three materials in Fig. 5 remain antistatic even at 0 percentRH. Also relating to the FIM is the propensity of a work surface or packaging material to charge nonconductors rubbed against it. For example, a circuit board might be triboelectrically charged by sliding against packaging material, and then the E-field on the board could damage ESD-sensitive devices being installed or already installed. Triboelectric charging propensity relates to the CDM as well as the FIM. For example, minimal charging of the plastic cases of DIPs sliding out of a DIP tube is desired to reduce the CDM hazard. PAPER NO. 15 279

Table 3. Trlboelectrlc Charging Data.

VOLTAGE ON COUPON, kV'

FR-4 GF POLYIMIDE OVERALL MATERIAL READINGS/AVG. READINGS/AVG. READINGS/AVG. AVG. RATING

Antistatic +1 +1 +3/+2 +1 +1 +2 /+1 0-1 0/-0.3 Medium Charging vinyl Cardboard +1 +1 +1/ +1 +0.4+2 +3 /+2 -1-1-1/-1 I Medium Charging Conventional +4 +4 +4/ +4 +4 +2 +2 / +3 +2 +3 +1/ +2 3 Medium Charging vinyl Black conduc- +7 + I +3 / +4 +18 +11 +18/+16 +2+1 +1/+1 7 High Charging tive plastic Antistatic 0 0 0 / 0 o 0 0 / 0 0 0 0/ 0 o Low Charging polyethylene

• Results for three tests and the average.

Experience shows that standard triboelectric charging tests with quartz and Teflon™ are insufficient, and specific materials of concern must be evaluated. For example, a perma• nently antistatic vinyl plastic, suitable for tote boxes. was tested (Table 3) using circuit• board laminates (FR-4, GF, and poly imide) with the FlM in mind. Triboelectric charging is notoriously erratic and unreproducible, so these tests were run in triplicate. Because of the scatter of the data, the best that can be done in such testing is to make the rough classifica• tions shown. Black conductive plastic is well known to produce high charges, whereas antistatic polyethylene (MIL-B-SI705C, Type II) imparts very low charge because antistat invisibly rubs off onto the surface of the other material. The antistatic vinyl, which had a dry surface with no antistat layer, was a medium charger -like cardboard or conventional vinyl.

Summary

Novel tests are appropriate when standard tests are insufficiently specific and realistic. The object is to simulate worst-case ESD-hazard scenarios so that materials for tools, contain• ers, work surfaces. and flooring can be selected with maximum confidence.

References

1. O. I. McAteer, Electrostatic Discharge Control, McGraw-Hili, 1989. 2. I. M. Kolyer and D. E. Watson. ESD from A to Z: Electrostatic Discharge Control for Electronics, Van Nostrand Reinhold, 1990. 3. I. M. Kolyer. "Fundamentals of ESD Control," Technical Record of Expo '91, International Conference on Electromagnetic Compatibility, EMC Technology Magazine, Reston, VA May 18-20,1992, pages 154-161. 4. I. D. Campbell, "ESD Process Control and Measurements," EMC Test and Design, September/October 1992, page 63. 5. I. M. Kolyer and D. E. Watson, "Packaging for High-Voltage Discharge Protection," Evaluation Engineering, March 1992, pages 96-100. 280 APPENDIX

6. I. M. Kolyer, "Toward an Ideal ESO-Protective Package", Proceedings of the 1994 EMC/ESO International Conference, Anaheim, CA, April 12-19, 1994. 7. I. M. Kolyer and O. E. Watson, "ESO Testing of Silicon Wafers," Evaluation Engineering, September 1991, pages 115-119. 8. I. B. Brinton, "Test Questions Non-Metallized Bags." EOSIESD Technology, Ianuary/February 1992. pages 12-13. 9. I. M. Kolyer and O. E. Watson, "COM and Work Surface Selection," Evaluation Engineering, October 1991, pages 110-117. 10. I. M. Kolyer., ''Testing Surfaces for ESO Safety," Evaluation Engineering, November 1994, pages S-36-5-40. 11. 1. M. Kolyer and R. Rushworth, "Humidity and Temperature Effects on Surface Resistivity," Evaluation Engineering, October 1999, pages 106-110. 12. The Pulsed EMf Handbook, Second Edition, KeyTek Instrument Corp., Wilmington, MA, 1991, page 23. Paper No. 16

This paper is reprinted, by pennission, from Evauluation Engineering, November 1994, pages S-36-S-40.

TESTING SURFACES FOR ESD SAFETY

John M. Kolyer

Rockwell International Corporation Defense Electronics Autonetics Strategic Systems Division San Bernardino, California

Whether we want to admit it or not, we are our own worst enemies when it comes to elec• tronic equipment. Human beings are the major source of direct injection (01), the most straightforward and common cause of ESD damage. Keeping ESD in check requires a good control program and even better testing proce• dures. Today, several practical test methods simulate 01, the ESD damage mechanism that includes the human body model and machine model, the charged device model (CDM), and the field-induced model (FlM).'·2 The culprit in all these models is a conductive surface that participates in rapid dis• charges to or from an ESD-sensitive (ESDS) device or circuitry leading to the device. The conductive approach to ESD control, in which ESDS items are handled or approached only with highly conductive grounded objects, encourages rapid discharges and must be less safe than the rival static dissipative-approach.2 The desirability of slow discharge is re• flected in MIL-HDBK-773, which cautions against damage by spark discharges.3 Where do you draw the line on conductivity? Resistance measurements vary with factors such as electrode design, applied voltage, and relative humidity (RH), and results are diffi• cult to relate to discharge rates and ESD damage. The carbon-loaded film in Table 1 is a well-known sparker and, presumably, is too conductive. The laminate has a resistance to ground in the 107 to 108 n range, previously judged to be CDM-safe for hard laminates.4 But what about materials such as skin or a rubber mat? Tests simulating damage models are necessary. A DI or CDM test has been proposed in which a neon bulb connected to a 500-V power supply (PS) flashes when contacting con• ductive surfaces.' This method is subjective and does not necessarily correlate with device damage.6 A CDM test uses actual devices (MOSFETs) for credibility, and this test now has been expanded to represent DI and the FlM.6 The expense of inventorying MOSFETs has been avoided by replacing them with a simu• lated voltage-sensitive device, the Static Event Detector™ (SED) from 3M. For our testing purposes, a MOSFET (2N4351) stressed between the gate lead (G) and the substrate-case lead (SC) is essentially a silicon-oxide layer, with a dielectric strength of approximately 100 V, bypassed by a high-resistance leakage path. This path protects MOSFETs during CDM discharges to high-resistance surfaces.4

281 282 APPENDIX

Table 1. Materlala.

RESISTANCE (OHMS)

TOOROUND BBTWEEN SURFACE 11IROUOH NFPA99 RESISnvrrY, 11IROUOHD NFPA99 EU!C'I1tODES OHMS/SQUARB, (FlO. I) EU!C'I1tODB I IN. APART NO. DESCRIPI10N AT 100 V ATIOV AT IOV AT 100 V Antistatic 1 x 1()9 2 x 1()9 9 x 10' 8xlO' polyethylene (5 x l()9at (8 x 1010 (1 X 1010 (4 x 1()9 (MIL-B-81705, 35%RH) at 35%RH) at 35% RH) at 35% RH) Type ll), 6 mils 2 Carbon-loaded 3 x 10' 1 x 10' 3 x 1()4 3xl()4 polyethylene film, @IV @IV @IV @IV 4 mils 3 Conventional 3xl()9 1 x 1010 2xl()9 lxl()9 corrugated cardboard, 110 mils 4 Hard benchtop 7 x 1()9 15 x 1010 4x 107 3 X 107 laminate with buried conductive layer 5 Rubber workbench 3 x 107 3xlO' 1 x 1()6 2xl()6 mat with buried conductive layer 6 Bare finger 4x 10' 3xl()6 N/A N/A at IOV 7 Carbon-loaded 1 x 10' 8 x 1()6 N/A N/A plastic cot on at 10V finger 8 Static-dissipative 4 x 10' 3 x 1()9* N/A N/A plastic cot on at 10V finger

• Measured through steel plate on wrist.

To cause damage, the rise time of voltage need not be rapid, because the oxide must rupture whenever the voltage differential across it exceeds approximately 100 V. MOSFETs subjected to the very slow rise time of 10 V/s failed at an average of 117 V-only slightly above the 80 V to 100 V found at very fast rise times. A MOSFET with SC grounded will fail when an antenna attached to G is moved slowly into an electrical field. 2 In the same situation, the SED is tripped only by a very intense field. creating a differen• tial voltage far above 100 V, because normal sensor response requires voltage surges. When rise times are fast, as they are in most ESD events, the MOSFET and the SED give equiva• lent results. An SED is visibly tripped at the moment of the ESD event. On the other hand, MOSFETs can give false indications of failure by being damaged at other points in the test process. PAPER NO. 16 283

The SED used for these tests had a capacitance of 150 pF and was tripped at 92 V. SED and MOSFET failure rates were identical in a CDM test, and similar agreement has been found in packaging tests with spark discharges (Table 2). As shown in Fig. 1, the sensor (SED, MOSFET, thin-film resistor, or neon bulb) was held between alligator clips AI and A2• The SED cap was connected by a fine wire (26-gauge) for flexibility. Terminal Ts was connected to one of two probes: a dual in-line package lead (the D probe) or an edge contact (EC) comprising a 0.25-in.-wide copper circuit line on glass• filled epoxy board material mounted on a Teflon™ block (the EC probe).4

Table 2. Test Data. c,

SIMULATED CONDmON MATERIAL 4 pF 204 pF 1028 pF

Charged DIP falls to I at 35% RH, 3, 4 0/10 0/10 0/10 grounded surface (MOSFETs: 0/5) (Probe D) (CDM) 8 0/10 3/10 10/10 (See Reference 4) 2/10 9/10 10/10 5 10/10 10/10 10/10 (MOSFETs: 5/5) 2,6,7 10/10 10/10 10/10 Fingers grasp edge I at 35% RH 0/10 0/10 0/10 contact of board (wrapped on finger) (Probe EC) (01, FIM, 8 4/10 9/10 10/10 CDM) 6, 7 10/10 10/10 10/10 Bag or box falls I in. I (6 x 10 in. bag) 0/10 1/10 2/10 onto edge contact of board (Probe EC) (CDM, 01) 3 (2 x 5 x 10 in. 0/10 2/10 2/10 box) 2 (6 x 10 in. bag) 10/10 10/10 10/10 Charged DIP falls to 2 (1.6 in! 1.4 pF) 0/10 2/10 2/10 isolated conductor on 0.25 in. Teflon (Probe D) (CDM) 2 (6.3 in,2 5.6 pF) 10/10 10/10 10/10 on 0.25-in. Teflon Charged circuit board Material C, = 9414 pF with lOO-ohm (O.1K) or 300 kilohm Steel 0.1 K, ~.055%; 300K, ~.039% (300K) 1/20-watt thin- film resistor touched by 2 0.1 K,~.OOI%; 300K,tO.001% pointed, grounded tool (Probe D) (CDM) 5 0.1 K, to.005%; 300K, ~.OOI %

Material C, = 839pF N-2 neon bulb as sensor C, =4pF (Probe D) (CDM) 2 Slight flash in Aash dim light 5 No flash in Slight flash in dim light dim light

Note: Failure rates for the SED are expressed as fraction of failed/tested. Resistance shifts (highest of three tests) are in percent. 284 APPENDtX

MOSFn

I: SED ~"

.... - _~-0 -= =-_--- ~ ~ "'"

•

Fig. 1. Test apparatus.

Tenninals T T1, T, and T. were claw clips. Tl and Tl were grounded. With AI and Al " empty, capacilaJ'lCe CI measured between Tl and a metal plate beneath the apparatus was 4 pF. Similarly, capacitance Cl measured between T. and the plate was 8 pF with !he 0 probe installed and to pF with the EC probe installed.

Capacitances CI and C2 represent those of circuit lines on a board. A similar measure• ment between the plate and Ees on aerospace computer boards of various designs gave from 3 pF to 31 pF. To raiseC a capacilorsomelimes was held byT, and T , if desired, C l 1 could also be l'3ised by a capacitor" held by Tl and T•. Testing was done at 1000 V. The SED sometimes was tripped by the surge when the PS was turned on with switch Sclosed. The MOSFETs were always damaged, so the voltage was increased slowly; for example. at 100 VIs. After 10 s at 1000 V, S was opened and block B was pulled aside to let D fall to the surface. The SED was tripped by a I'3pid discharge through D. For the carbon-loaded film, with 10' n resistance through D (Table t), the 8-pF capacitorC2 dl'3ined with an RC time constant of 8 x 10.. 1 s, the time constant for the laminate (.5 x 1010 n through D) was 0.4 s. These times represent a I'3pid discharge vs a relatively slow bleed-off. In theory, RC might be estimated in all cases. But, in pl'3ctice, R and C may be difficult or almost impossible to measure, so an empirical test is more pl'3ctical and convincing than a calculation. Table 2 gives examples of simulated real-life handling conditions for ESDS items The RH was approximately 50% except for tests with antistatic polyethylene at 35% RH. With PAPER NO. 16 285 the caveat that generalization beyond the conditions of the tests is unjustified, conclusions drawn from Table 2 are:

o The rule of thumb that a resistance to ground of 101 to lOS Q is desired for COM safety of hard benchtop laminates is supported by the data.4 This rule does not extend to other materials, such as antistatic polyethylene. o The most meaningful resistance measurement was through O. In the first test in Table 10 9 2, 10 Q to ground gave no failures; 10 Q gave several failures, especially at high CI; and lOS to lOS Q gave all failures. o Elevated RH reduces the general ESO problem but makes certain materials become discharge hazards. Antistatic polyethylene was safe at 35% RH but not at 50% RH. o ESO-protective finger cots should be in the antistatic range (109 to 1011 Q/sq) rather than static-dissipative (lOS to lOS Q/sq).

o Even cardboard, with its high resistance through 0, gave failures at high CI when a box fell onto an EC. This result emphasizes that discharge safety depends on the mode of contact. o Isolated (ungrounded) conductors such as the carbon-loaded film are COM threats even at low capacitance. The converse situation (conductor at 1000 V, device uncharged) gave similar results in past work. A I pF conductor was harmless, a 3 pF conductor was damaging.1

o Even with a very conductive surface and an unrealistically high CI, the resistance of thin-film resistors shifted only slightly in a COM simulation.4 o An N-2 neon bulb as a sensor gave a subjective result rather than a clear pass/fail.

Incidentally, conductive surfaces can be hazardous to personnel handling high voltage. The carbon-loaded polyethylene film could carry a calculated current of 33 rnA at 1000 V between NFPA 99 electrodes I inch apart. An AC current of 21 rnA to 40 rnA causes mus• cular inhibition.8 In conclusion, ESO-protective materials high in the antistatic range (1010 to 1011 Q/sq) may be necessary for worst-case 01, COM, or FIM hazards. However, a rigorous antistatic approach requires finger cots and restricts the normal use of conductive tools. Tests, such as those in Table 2, are recommended to indicate real needs and avoid unnecessary and im• practical precautions.

References 1. McAteer, 0.1., Electrostatic Discharge Control, McGrawHilI, 1989, pp. 173-188 2. Kolyer,l. M., and Watson, O. E. ESD from A to Z: Electrostatic Discharge Controlfor Electronics, Van Nostrand Reinhold. 1990, pp. 5, 6, 9,157,160,169, and 205. 3. MIL-HDBK-773, April 1988.4.3.1. 4. Kolyer,1. M., and Watson, O. E., "COM and Work Surface Selection," Evaluation Engineering, October 1991, p. 110. 5. "New Test Proposed for Checking ESO Safety of Materials," Compliance Engineering, Fall 1990, p. 77. 6. Kolyer,l. M., "Is Your Work Surface COM Safe?" EOSIESD Technologv, February/March 1992, p.27. 7. Kolyer,l. M., and Watson. O. E., "ESO Testing of Silicon Wafers," Evaluation Engineering, September 1991, p. \15. 8. OoO-HOBK-263, May 1980,7.3.1.3. PaperNo. 17

This paper is reprinted, by pennission, from Evaluation Engineering, October 1990, pages 106-110.

HUMIDITY AND TEMPERATURE EFFECfS ON SURFACE RESISTIVITY

John Kolyer and Ronald Rushworth

Rockwell International Corporation Defense Electronics Autonetics Strategic Systems Division San Bernardino, California

This investigation evaluated antistatic materials used in ESD control for electronics. The objectives were to define curves for surface resistivity (a measure of the antistatic property) vs relative humidity, detennine the temperature effect, and set an effective lower limit for controlled relative humidity in assembly areas. Twelve materials were studied: four commercial antistatic films, including a polyethyl• ene film with a radiation-cured coating, three cellulosic (cellophane) films, two kinds of paper (present in static-controlled areas in the fonn of documents such as work instruc• tions), a detergent (representing topical antistats), a static-limiting floor finish, and leather (representing leather shoe soles). With reference to Fig. A, antistatic polyethylene I met MIL-B-81705, Type II, and con• tained an amide antistat; antistatic polyethylene 2 contained an ethoxylated tertiary amine antistat; the antistatic nylon also contained an ethoxylated tertiary amine antistat; and the paper was 20-lb bond made by the bleached kraft process. Also, the filter paper was Whatman No.1 made from cotton linters; the detergent was Joy (Proctor & Gamble); the floor finish was a high-solids acrylic; and the leather was split, vegetable-tanned, and unfinished. To facilitate resistance measurements, both sides of a nonconductive film (cellulose acetate) were dipped in the detergent and floor finish. Strips of film or sheet were held between clamps to make a square, as in ASTM D 257, and sealed inside with saturated reagent-grade salt solutions (MgCI2, NaN03, KCI, KN03, K2SO.) to control the relative humidity (see Table 1).1-3. A drying agent (anhydrous, calcium sulfate with indicating blue color) afforded 0% RH. The jars were stabilized at three fixed temperatures, 41 to 46°F, 75 to 80°F, and 108°F, until the samples reached equilibrium moisture content and the resistance readings became constant. Although no acidic or basic vapors that might affect resistance were found, confir• matory detenninations were made at 75 to 80°F with an alternate set of salt solutions (KF,

K2C03, NaHSO., NaN03, NH4HlO.). The apparatus is shown in Figs. 1-6. Figure 1 is an individual . Figure 2 shows a leather sample in the clamps. The jars were placed in a refrigerator (Fig. 3) or a low-heat oven (Fig. 4). Figure 5 shows the interior of the oven with a light bulb (bottom of picture) as the thennostatted heater, then the fan, then the samples. In Fig. 6, a jar is tested under ambient conditions.

286 PAPER NO. 17 287

Code: o 75· 80 ° F, alternate solutions

----- 41· 46°F -0- -- -0- 108 0 F

10' ...... 10' ...... _ ...._

AntJetatlo PoIyeIIIyIene 1 Antlltetlo PoIyethy..... 2

o

10' ...... 10' ...... o 10 20 30 40 50 eo 70 eo 90 100 o 10 20 30 40 50 eo 70 eo 90 100 .... % .... %

10' ...... 10' ......

Polyethylene with RId.- Curld COetIng

10·' ...... "- ...... ~ ~ 10·' ...... "---- ...... o 10 20 30 40 50 eo 70 eo 90 100 o 10 20 30 40 50 eo 70 eo 90 100 .... % .... %

Fig. A. Plots of resistance data 288 APPENDIX

Code: -,--...... 75 - 80° F ° 75 - 80° F, alternate solutions ----- 41· 46°F -0- ---0- 1080F

10' ...~~~-!-~~~~~~ 10· ...... - ...... ~ ...... o 10 211 30 040 50 eo 70 eo 110 100 o 10 211 30 040 50 eo 70 AH,% AH,%

10' ...... -. ....

10' ...... - ...... -!o~~~ 10' ...... o 10 211 30 040 50 eo 70 eo 110 100 o 10 211 30 040 50 eo 70 eo 110 100 AH,% AH,%

Fig. A. continued. PAPER NO. 17 289

Code: --_ 75-80o F o 75 - 80 ° F. alternate solutions -a-_ 41 - 460F -0- ---0- 1 08 ° F

107 ...... -1" .... ""'1' ...... 1"""1

Detergent

10·' ...&..&..&...... 10' ...~ ...... ~ ...... o 10 20 30 40 110 eo 70 eo 80 100 o 10 20 30 40 110 eo 70 eo 80 100 ""'''' ""''''

10·' ...~~ ...... ~~~~~1;.I 10' ...&..&..&...... o 10 20 30 40 110 eo 70 eo 80 100 o 10 20 30 40 110 eo 70 eo 80 100 ""'''' ""'''' Fig. A. continued. 290 APPENDIX

Table 1. Relative Humidity over Saturated Aqueous Solutions.

TEMPERA11JRE RELATIVE SOLUTE of HUMIDITY,'"

MgCI2·6Hp 43 35 75 33 108 32

NaN0J 43 78 75 75 108 70 KCI 43 88 75 85 108 82

KNOJ 43 95 75 94 108 89

K2S04 43 98 75 97 108 96 KF·2Hp 75-80 27

K2COJ·2H2O 75-80 43 NaHSO.. HP 75-80 52 NH4~P04 75-80 93

As seen in Fig. A, agreement of results with the standard vs alternate solutions generally was good. Incidentally, sulfuric acid solutions, sometimes used to control relative humidity, did evolve acidic vapor and were rejected for that reason. The resistance across the films was measured at 100 V with a Beckman L-12 megohmmeter with the cables shielded to prevent RF interference. Time of electrification was 5 s. Runs were made in triplicate with the averages plotted in Fig. A. Standard deviations generally were 5 to 60% of the mean. No data were rejected. The clamps contacted both sides of the specimens, so R in Fig. A is the same as surface resistiv• ity (in M Q/sq) only for the volume-conductive materials (paper, leather, and cellophane films). For the other materials, the surface resistivity is twice the value of R in Fig. A. The data provided several conclusions:

• Surface resistivity is controlled by two factors. First is the moisture content which, in tum, is controlled by the relative humidity; second is the temperature. A similar tempera• ture effect has been reported for charge decay time,4 which would be expected to corre• late with surface resistivity for materials such as antistatic polyethylene. At constant relative humidity, the resistance of the antistatic materials decreased exponentially with temperature, as occurs with dielectrics' and semiconductor.6 Metals show the opposite effect. • The reality of the temperature effect is proved by the dry-air data. At high relative humidity, the temperature effect becomes small or nil as the materials move from the semiconductive toward the conductive state. • Because of the significant thermal effect, the temperature should be given, along with the relative humidity, in reporting surface resistivity data. PAPER NO. 17 291

• Surface resistivity commonly is assumed to vary exponentially with relative humid• ity, S but the plots of log resistance vs relative humidity (Fig. A) often were curves depart• ing far from linearity. The cellulosic materials (paper and cellophane) and leather constitute one family of curves; all are fairly close to straight lines of approximately the same slope.

2

3 4

5 6

Figs. 1-6. Resistance apparatus. 292 APPENDIX

The antistatic and polyethylene with radiation-cured coating constitute another family of fairly similar curves, while the curves of antistatic nylon are somewhat similar but shifted much upward. The detergent and static-limiting floor finish form an• other family, with curves rapidly rising, especially for the floor finish, as 0% RH is approached. Therefore, the floor finish is very sensitive to slight changes in relative humidiy at the low end. • Antistatic polyethylene is sometimes assumed not to function in dry air, but the two brands tested were effective. The surface resistiviy remained less than 1012 n,tsq so that charges could drain quickly from the surface. • The graphs provide cutoff relative humidities below which materials lose their anti• static propery (that is, the surface resistiviy exceeds 1012 n,tsq). At room temperature, the cutoffs are 0% RH (no cutoff) for polyethylene with either of two extruded-in antistats or an antistat-impregnated, radiation-cured coating; 0% (no cutoff) for a detergent "sweat layer"; 3 to 4% for a plasticized cellophane film with or without a coating; 9% for a static-limiting floor finish; 18% for leather or an unplasticized cellophane film; 23 to 26% for paper; and 50% for antistatic nylon . • These data support the requirement for a minimum controlled relative humidity of 30% to keep most antistatic materials functional. This agrees with experience. So-called antistatic nylon is an exception, but this film is really not antistatic in practical terms and has been the subject of a GIDEP Alert7 for that reason.

The plasticized, coated cellulosic film Ecostat,8 contains moisture and a hygroscopic softener or plasticizer. No antis tat is concentrated at the surface where it can be removed by rubbing. After vigorous rubbing with this film, nonconductive surfaces remained nonconductive. However, they were made antistatic by being rubbed with, or merely pressed against, the antistatic polyethylenes or the polyethylene with the radiation-cured coating. In contrast to other materials, Ecostat remained antistatic after brief rinsing with water. Soaking at 81°F for 24 h extracted the plasticizer so the film reverted to unplasticized cello• phane (Fig. 7) and the room-temperature cutoff relative humidity rose from 4 to 18%. In preliminary static decay tests (FED-STD-101, Method 4046) at 8 to 10% RH and 75°F, Ecostat showed a decay time of 5 to 11 s. It was noncorrosive to copper or Sn62 solder when exposed in contact with these metals at 100% RH and 75 to 80°F for one year or 120°F for one month. This film appears to be suitable for a noncontaminating intimate wrap; e.g., inside MIL• B-81705, Type I, to meet the intent ofMIL-HDBK-773. It also is biodegradable. PAPER NO. 17 293

CI Cellophane, pla.t• • Cellophane, plaat. • coatad o Cellophane, pla.t.. aner extraction e Cellophane, plast•• coated, after extraction

1~pp~-r~--..~~--p-~-r ..

10' • ~

10'

10' ....~~~ .. ~ .. --6.-- ..... o 10 20 30 «) 50 60 70 80 90 100 RH,%

Fig. 7. Effect of water extraction on cellophane films.

References 1. International Critical Tables, Vol. I, p. 67. 2. Stokes and Robinson, Industrial and Engineering Chemistry, Vol. 41,1949, p. 2013. 3. Handbook o/Chemistry, edited by N. A. Lange, McGraw-Hill, Revised 10th Edition, 1967, p. 1432. 4. Evaluation Engineering, April 1982, p. 74. 5. ASTM D 257-78, Appendix XI. 6. Encyclopaedia Britannica, 15th Edition, Vol. 18, 1989, p. 241. 7. GlDEP Alert C6-A-87-09, November 12, 1987. 8. Evaluation Engineering, March 1990, p. 96. PaperNo. 18

This paper is reprinted. by permission. from Evaluation Engineering. March 1992. pages 96-100.

PACKAGING FOR HIGH-VOLTAGE DISCHARGE PROTECTION

John M. Kolver and Donald E. Watson

Rockwell International Corporation Defense Electronics Autonetics Strategic Systems Division San Bernardino. California

The EIA-541 electrostatic shielding test for packaging materials uses the relatively low voltage of 1.000 V on a large flat electrode. This may simulate fields or low-voltage dis• charges. but possible damage from ESD includes high-voltage discharges--often as visible sparks-from people or conductive objects. This article describes a test method for simulat• ing such discharges and outlines packaging strategies to resist them. The test is conducted at -9 k V. the minus sign meaning that electrons flow from a charged capacitor toward the package. This is not a worst-case human voltage; the machine-model resistance of nearly zero is worst-case. and the ESD sensitivity of 100 V is nearly worst• case. The test conditions constitute a reasonable requirement. passed by Types I and III of MIL-B-SI705. Figure I represents an idealized. nearly zero-resistance person with 200-pF capacitance discharging from a sharp-tipped. hand-held tool after being triboelectrified to -9 kV (for example. by walking on a carpet at 20% RH). The material under test was pulled tight over the capacitive probe. Then the tip of the O.025"-diameter capacitor lead was touched to the material for 1 s. The vertical shield prevented the E field of the capacitor lead from inducing false voltages in the circuitry. The sensor was an oxide-coated silicon wafer modeling a MOSFET gate. I A pulse of 100 V or greater across the disks of the capacitive probe punctured the oxide under a steel probe pressed against the surface. The resistance of the oxide measured through the steel probe was over 10 12 n initially and usually less than 109 n when the oxide had been damaged. so the failure criterion was the upper limit of detection of 2 x 109 n bv a digital multimeter. The tests were run at approximately 50% RH. Wafer-oxide failure rates for the 14 commercial materials in Table 1 are consistent with MOSFET data in past work.2 Results of the tests indicate that a conductive layer was re• quired for passing the test (failure rate = 0/5). Foil bags (Materials 4 and 5) are prescribed in MIL-HDBK-773 and have been recommended for highly ESD-sensitive items and worst• case discharges.2•3 It was also concluded that semitransparent vapor-deposited metallization is effective only when buried under a layer of plastic (Material 6). The screen bag (Material 8) is very discharge-protective2 and relatively transparent. but expensive. The S-mil skin of Material 14 essentially is plain paper on the outside but carbon• impregnated on the inside. This resembles the successful shielding combination of Material 2

294 PAPER NO. 18 295

Capachor (200 pF)

C • Capachive Probe whh Brass Discs

Fig. 1. Apparatus.

Table 1. Commercial Materials.

FAILURE RAlE NO. MATERIAL CONDUcnVE LAYER (FAlLURES/l'ESTS) Antistatic polyethylene (pink poly). None 4/5 6 mils (Mll.-B-81705.TYpe U) 2 Cellophane. 1.4 mils None 4/5 3 Black conductive. 4 mils Carbon-loaded polyethylene 5/5 4 Foil laminate. antistatic TYvek outside. Aluminum foil. 0.25 mil 0/5 antistatic polyethylene inside 5 Foil laminate. aluminized Aluminum foil. 0.25 mil 0/5 outside. carbon-loaded polyolefin inside (Mll.-B-81705.TYpe I) 6 Metallized with 0.5-mil polyester cover Aluminum. partially 0/5 layer. antistatic polyethylene inside transparent (Mll.-B-81705.TYpe m. Class 1) 7 Metallized with 0.16" Aluminum. 150 A 0/5 cover, biopolymer inside 8 Screen sandwiched between Metallic window screen 0/5 Mll.-B-81705, TYpeU 9 Externally metallized Aluminum, partially 5/5 (thin organic coating) transparent 10 Externally metallized Copper, partially 4/5 (thin organic coating) transparent 11 Externally metallized Nickel, partially 5/5 (thin organic coating) transparent 12 Antistatic foam, 0.25" None 4/5 13 Corrugated cardboard, 0.09" None 4/5 14 Corrugated cardboard, 0.16" Carbon impregnated on 0/5 inside of skins 15 Air space None 0/5 (3") 4/5 (1S') 296 APPENDIX

Table 2. Material Combinations.

FAILURE RATE OUTER MATERIAL INNER MATERIAL (FAILURESfJ'ESTS)

1 0/5 12 1/5 2 12 5/5 lor2 3,9 or 10 0/5 1 11 2/5 2 11 0/5 3 or 11 5/5 9 3/5 10 4/5 12 1/5 12 2,3,9, 10, 11 or 12 0/5 3,9,10 or 11 12 0/5 3 or 10 2 5/5 9 or 11 2 4/5 lor 13 15 (0.5") 0/5

outside and Material 3 inside in Table 2. The 0.14" gap (largely air) inside Material 14 adds to the protective effect. The conductive layer should be buried (Materials 4, 6, 7, 8 and 14) rather than exposed (Materials 3, 5, 9, 10 and 11). This not only improves shielding but also discourages ESD damage by the charged device model.4 Sparks did not correlate with ESD failure. Material 3 failed with a visible arc, but Mate• rial5 passed; Materials 1,2,12 and 13 failed without an arc. The arc to Materials 9,10 and 11 may have contributed to failure by creating bare spots approximately 0.01" to 0.02" diameter in the metallization. (Fig.2). Choices are limited for single materials pressed against the ESD-sensitive item, but op• tions multiply for combinations including double bagging and air gaps (Table 2). For ex• ample, a pink poly bag (Material 1) or an externally, metallized bag (Materials 9 or 10) is ineffective alone, but when combined correctly with pink poly on the outside-not the inside-the metallization is buried and protected from spark damage. Cellophane (Material 2) also is an effective cover layer. Materials 3, 9, 10 or 11 succeed when combined with foam (Material 12). Table 3, derived from Table 2, provides examples of six construction types with the principles of their success: Faraday-cage shielding, a dielectric barrier of relatively heavy plastic, and protective spacing. Even inexpensive, unmetallized materials are adequate with air gaps, which need be only OS'. Commercially recyclable and biodegradable containers use construction types 5 and 6 Other commercial packaging utilizes construction type 4, in which a transparent bag is inflated with the least expensive material-air. Common packaging materials in the right configurations can prevent the damage of high• voltage discharge. The techniques of buried metallization and protective spacing offer many possibilities for cost-effective package design. PAPER NO. 18 297

Fig. 2. Bare spot 0.017 inch in diameter created on surface of Material 9 by spark discharge.

Table 3. Package Constructions.

NO. CONSTRUCTION MATERIAL NOS. (TABLE I) PRINCIPLE Single bags or thin-walled 4,5,6,7,8 Faraday-cage effect boxes in which ESD-sensitive (buried metal- leads, terminals or contacts lization dissipates press the inside surface discharge) 2 Double-bagging systems lor 2 over 3, 9, or 10; 2 over 11; Faraday-cage effect with conductive layer 12 over 3, 9, 10, or 11 3 Double-bagging systems Two layers of I, totaling 12 mils; Dielectric barrier without conductive layer two layers of 12, totaling OS' blocks discharge 4 Soft packages with air gap 1 or 2 with gap (OS' minimum) Space weakens provided by antistatic bubble-wrap; E field transparent film bag inflated with air 5 Rigid packages with 14 alone, or 13 with metallic layer, Faraday-cage conductive layer e.g., aluminum foil, bured under effect a film such as Material 2 6 Rigid packages with air gap 13 with a protective gap (OS' Space weakens minimum) provided by antistatic E field bubble-wrap or spacers and parti- tions; may have double-pane view- ing window of I or 2 with OS' gap 298 APPENDIX

References

I. Kolyer, J. M., and Watson, D. E., "ESD Testing of Silicon Wafers," Evaluation Engineering, September 1991, pp. 115-119. 2. Kolyer, J. M., and Watson, D. E., ESD from A to Z: Electrostatic Discharge Control for Electronics, Van Nostrand Reinhold, 1990, pp. 173, 175-181. 3. Holmes, O. C., Huff, P. J., and Johnson, R. L., "An Experimental Study of the Screening Effectiveness of Antistatic Bags," Reliabiliy Analysis Center EOSIESD Symposium Proceed• ings, Philadelphia, PA, 1984, pp. 78-84. 4. Kolyer, J. M., and Watson, D. E., "The Charged Device Model and Work Surface Selection," Evaluation Engineering, October 1991, pp. 110-117. Paper No. 19

This paper is reprinted, by pennission, from Evaluation Engineering, September 1992, pages 94-100.

HIDDEN CHARGES ON ESD-PROTECTIVE PACKAGING

John M. Kolyer and Donald E. Watson

Rockwell International Corporation Defense Electronics Autonetics Strategic Systems Division San Bernardino, California

A static charge is a static (immobile) excess or deficit of electrons on a surface. This charge emanates an electrical field (E field) usually easily detected by a static meter. However, the E field is weak and difficult to detect when suppressed by proximity to a nearby conductor as in packaging laminates with nonconductive outer layers sandwiching a conductive or static-dissipative interJayer. The hard-to-find charges on the laminates are called cryptocharges (meaning hidden or secret). A cryptocharge, being immobile, cannot flow into a device lead or tenninal to cause ESD damage, and the associated E field generally is too weak to cause damage by the field• induced model (FIM). Buried-metal ESD-shielding laminates can be antistat-depleted and nonconductive without being an FIM threat. Cryptocharges, however, can be brought out of hiding by induction. A conductor brought within an E field is polarized by induction. If the outer end or side of the conductor is momentarily grounded before the conductor is removed from the field, the conductor will have an opposite charge to that of the field. This is compound induction. I Although cryptocharge fields are weak, when a metal plate is pressed to the charged surface, momentarily grounded, and then lifted from the surface, potentials of up to several kilovolts appear on the plate. This effect is contact compound induction (CCI). To study CCI, several laminated and unlaminated films or sheets were tested (Table I). Material I was a pack with the film stretched over a 3.8" x 3.8" x 0.3" cardboard frame with 2.3" x 2.3" windows. Materials 2, 3 and 5 were bags stretched over cardboard in the same configuration as Material I. Material 4 was a 4" x 6" x 1.1" box. Samples were electrically cryptocharged for I min with the window area, or the side of the box, resting on a charged 2" x 2" x 0.03" aluminum plate. A 0.2-lb weight on the sample facilitated contact. See Fig. I. Alternatively, samples were tribocharged by stroking them on a nylon carpet at 70-75°F and 55-65% relative humidity (RH). CCI was accomplished by pressing a 2" x 2" x 0.03" aluminum induction plate with a Teflon™ handle against the cryptocharged surface with a force of 0.5 lb for 5 s. Then the plate was grounded momentarily by being touched with a grounded person's finger before being lifted from the surface (Fig. 2). The voltage on the plate was read with a Trek Model 512 Field Meter calibrated with a known -7 kV on the same plate.

299 300 APPENDIX

Table 1. Materials.

SURFACE RESlmVITY, OHMS!SQUARE@ 100 V AND 65% RH (75°F) MAn:RIAL TOTAL THICK- SURFACE NO. DESCRIYI10N NESS, MIL LAYERS INn:RLAYER 0un:R INNER Clean-skin 1.4 Polyethylene Static- >1013 >1013 shrink film 2 dissipative, nonmetallic 2 Mil-B-81705, 3.1 Polyester Aluminum 3 x 1010 2 x 1()9 Type III, outside, metalliza- Class I polyethylene tion,approx. inside 100 A, semitrans- parent 2W Same. but outer surface rinsed with water for 15 s >1013 2 x 1()9 3 Similar to 3.6 Polyester Aluminum 3xl()9 5 x 1()9 Material 2 outside, metalliza- but opaque polyethylene tion,opaque inside 3W Same. but outer surface rinsed with water for 15 s >1013 5 X 109 4 Paperboard 16 Paperboard Aluminum I x 1010 3 x lOS laminate outside, metalliza- transparent tion, 150 A biopolymer inside 5 Polyethylene 2.5 None None >1013 >1013 film 6 Polychloro- 7 None None >1013 >1013 trifluoro- ethylene (Aclar) film

+OR- 7kV POWER SUPPLY

Fig. 1. Deposition of cryptocharge. PAPER NO. 19 301

Fig. 2. Contact compound induction (CCI).

Table 2 shows that antistatic surfaces were unfavorable for cryptocharging. Material 2 lost its charge relatively rapidly while Materials 3 and 4 held no charge. Thus, cryptocharges would seem to reside on the outer surface, not inside the interlayer. Cryptocharges also were localized in the window touching the electrified plate (Fig. 1), whereas a charge within the interlayer should have spread throughout the sample. This led to the conclusion that the role of the interlayer is suppressing voltage, not holding charges. Like other static charges, cryptocharges decayed exponentially because the rate of decay (-liVldt) was directly proportional to the diminishing potential or voltage (V) that drives

Table 2. CCI Data.

CCI DECAY CONSTANTS (SEE TEXT) CHARGING CCI INITIAL DECAY TIME FROM MA1ERIAL NO. METIlOD VOLTAGE (VO>, kV B,DAr' HVTO I kV

Electrical (-) +5 0.176 0.970 9.1 days Electrical (+) -5 0.196 0.964 8.2 days Tribo (+) -7 0.223 0.953 7.2 days 2 Electrical (-) +7 93.6 0.872 25 min 2w Electrical (-) +7 0.421 0.982 3.8 days Tribo (-) +9 0.657 0.982 2.5 days 3 Electrical (-) 0 3W Electrical (-) +5 0.0871 0.996 18.5 days Tribo (+) -4 0.0946 0.960 17.0 days 4 Electrical (-) 0 5 Tribo (+) -3 5.71 0.999 6.8h 6 Tribo (-) -5* 0.316 0.963 5.1 days

• Field measured directly. not by CCI. 302 APPENDIX drainage of charge from the surface and, by means of the E field, attracts oppositely charged air ions. Then BV=-dVidt where B is a constant. Integrating the expression yields:

InV = InVo-Bt where V is voltage in kV at time t in days, In is the natural logarithm, and Vo is the initial voltage in kV. The values of B in Table 2 were obtained by exponential regression analysis of several data points, r being the correlation coefficient. The decay time in days from 5 kV to 1 kV is calculated by dividing In5 by B. The half-life of any cryptocharge is In2 divided by B. The effect of RH on the decay rate is uncertain. Decreasing humidity lowers the rate of charge drainage over surfaces but may slightly increase the mobility and neutralizing effec• tiveness of air ions.2 In these tests, the temperature was 75-80°F and the RH was 55-65% as measured by psychrometry. The concentration of air ions (conductivity of the air) was un• known, so the decay rates in Table 2 are only comparative. Unlike static charges that are not voltage-suppressed, cryptocharges resisted neutraliza• tion by ionized air from a blower because the E fields were very low (100 V to 300 V on the surface as sensed by the field meter). Thus, cryptocharges are not only hard to detect but also hard to remove. CCI must be distinguished from tribocharging. Uncharged Material 6 strongly tribocharged the induction plate with a positive voltage by separation, but uncharged Materials 1,2,3,4, and 5 gave negligible voltages by separation. Also, CCI for Material I was not inhibited by cementing Material 1 to the face of the induction plate, whereas tribocharging would have been minimized. In all cases, momentary grounding was necessary for CCI but would not have been for tribocharging. With Materials I, 3W, and 5, the polarity of the plate would have been posi• tive for tribocharging, as it was for a negative cryptocharge; but the plate was negative for a positive cryptocharge (Table 2). Incidentally, weak CCI charges on the plate occurred without momentary grounding and might be explained by bleed-off into the air. The CCI process apparently involved negligible electron flow between the sample and the induction plate because covering the plate with volume-nonconductive film (Material 1) did not inhibit CCI. Electron exchange, as opposed to induction, also would have made the plate the same polarity as the cryptocharge instead of the opposite polarity. Discharges from the induction plate, with a capacitance of approximately 1 pF when held by the handle, damaged wafer oxide.3 The oxidized wafer was the kind from which IC chips are cut and represents the gate of a MOSFET. Failure occurred when the discharge created a voltage differential across the oxide in excess of its dielectric strength of approxi• mately 100 V. See Figs. 3 and 4. The inverse situation is the charged device model (COM) in which a charged DIP with 1-3 pF capacitance discharges to ground with resulting ESD damage. I In terms of the l-pF plate, the last column of Table 2 gives the dangerous period for an initial CCI voltage of 5 kV. This was approximately 3 days for Material 2W, a week for Material I, and 2.5 weeks for Material 3W. The persistence of the cryptocharge on Material 3W might be due to effective voltage suppression by the relatively thick metallic interlayer or to very low conductivity of the PAPER NO. 19 303

Fig. 3. Wafer oxide test.

!!IS

3/S

2IS

115

VOLTAGE,kV

Fig. 4. Wafer oxide damage from charged induction plate. 304 APPENDIX washed polyester surface. Antistat depletion by volatilization, absorption by contact with paper products, or accidental rinsing could transform Material 3 to 3W in the workplace. In summary, cryptocharges can be deposited electrically or triboelectrically, are some• times very persistent and, under the contrived conditions of CCI testing, can emerge from hiding to induce MOSFET-damaging voltages on a conductor pressed to the surface. But are cryptocharges ever deposited under realistic conditions? And is CCI an ESD hazard or just a curious phenomenon? Electrical cryptocharging is possible when packages are handled. A 7-kV discharge from a 200-pF capacitor or the fmger of an ungrounded person was found to charge Material 1 to several kilovolts as detected by CCI, and ungrounded personnel are easily charged to 7 kV by walking on a carpet at 25% RH. As for tribocharging, Materials I, 2W and 3W attained 4-9 kV in rather humid air (fable 2) and would charge much higher under dry conditions. Assuming that a cryptocharge is present, one ESD-hazard scenario resembles field-induced double jeopardy in which compound induction and possible FIM damage are followed by CDM damage. I As a hypothetical example, a dual in-line package (DIP) or circuit-board module is placed in a clamshell pack with a negative cryptocharge on the interlayer. A few days later, a grounded operator opens the pack and touches a lead of the DIP or an edge contact of the board, accomplishing CCI and possibly causing FIM damage. Then the operator removes the positively charged DIP or module from the clamshell pack and again touches a lead or edge contact to cause a rapid discharge with resulting CDM damage. Such a sequence of events may be improbable but is not impossible. Is cryptocharging really a threat? To answer a question with a question: Is the tribocharging propensity of packaging really a threat? Or are carbon-loaded bags or tote boxes whose conductive surfaces promote the CDM by rapid discharge really a threat as MIL-HDBK- 773 suggests? No one knows. So every hazard, albeit remote, must be considered in a rigorous ESD• control program that fights the invisible to prevent the unmeasurable (subtle damage caus• ing premature failure). Since cryptocharges cannot be ignored, laminates such as Materials 1,2, and 3 must be exceptions to the rule (made with the FIM in mind) that static charges of up to 300 V measured by field meter are tolerable on nonconductors.4 Suspect packaging in use should be checked by CCI and a risk assessment made and precautions proposed. For relatively insensitive or low-value items, special procedures for removal from the package without touching leads or contacts might suffice. For items that are highly ESD-sensitive, expensive, or critical in defense or life-support systems, alternate packaging might be prudent. A cost-effective choice for avoiding cryptocharges or any form of stored energy is a volume-conductive shielding laminate such as Material 4. This material never becomes antistat-depleted, and charges cannot be retained because they bleed not only across the surface but directly through the layers.

References

1. McAteer, O. 1., Electrostatic Discharge Control, McGraw-Hili, 1989, pp. 58,176 and 179. 2. Jonassen, N., ''The Physics of Electrostatics," distributed at the 1984 EOS/ESD Symposium. 3. Kolyer, J. M., and Watson, D. E., "ESD Testing of Silicon Wafers," Evaluation Engineering, September 1991, pp. 115-119. 4. Kolyer, J. M., and Watson, D. E., ESD from A to Z: Electrostatic Discharge Control for Elec• tronics, Van Nostrand Reinhold, 1990, pp. 160, 161, 172 and 210-215. PaperNo. 20

This paper is reprinted, by pennission, from the Proceedings of the 1994 EMCIESD 1nternational Conference, held in Anaheim, CA, April 12-19, 1994.

TOWARD AN IDEAL ESD-PROTECTIVE PACKAGE

John M. Kolyer

Rockwell International Corporation Autonetics Electronic Systems Division Anaheim, California

Abstract

An ideal electrostatic discharge (ESD )-protective package would shield against worst-case external discharges and be permanently static-dissipative and volume-conductive. Also, it would have non-ESD properties such as recyclabi/ity, biodegradability, and flame-resis• tance. This paper discusses state-of-the-art packaging and proposes superior designs ofthe future that may be achieved, without major breakthroughs, by consolidation and extension ofpresent techniques. Basic to this development will be realistic ESD tests, several ofwhich are described with examples of data.

Introduction

The title of this paper says "toward" because an ideal product is, by definition, impossible. It would have to include conflicting properties as in "biodegradable, heat-sealable, low-cost, transparent, stainless steel." Furthennore, it would have to be universally suitable for count• less applications with specialized requirements. However, of paper, plastics, and metals can combine diverse properties and approach an ideal package for specific purposes. This paper discusses key electrostatic discharge (ESD) properties and realistic tests for them. Non-ESD properties are also reviewed. Then promising packaging designs that are now available or on the horizon are discussed, with pros and cons of the two major catego• ries, plastic and paper.

ESD Properties Desired

ESD properties have been discussed' in tenns of damage mechanisms: Human Body Model (HBM), Machine Model (MM), Charged Device Model (CDM), and Field-Induced Model (FlM). See Refs. 2 and 3 for explanations of the models. The key ESD properties desired are high-voltage discharge resistance, CDM safety, low current-carrying capability to protect personnel as well as devices, a static-dissipative surface even at low relative humidity (RH),permanence of the static-dissipative property (including

305 306 APPENDIX

lack of undesirable antis tat transfer to other surfaces), abrasion resistance (no loss of static• dissipativeness and no sloughing of conductive particles), minimal triboelectric charging propensity for nonconductive surfaces touched by the packaging material, and volume con• ductivity to prevent stored energy. These properties will be further explained in the follow• ing discussion of tests.

Realistic Tests for Key ESD Properties

Nine tests have been developed as practical criteria for the above key properties. In the following discussion, the tests are referenced, briefly described, and illustrated with data on commercial materials.

High-Voltage Discharge Resistance. External discharges can damage ESD-sensitive items, through the wall of a package, by the HBM or MM. A realistic tesr uses a 3M Static Event Detector (SED) (Fig. 1) inside a bag or box (Fig. 2). A charged steel probe touched to the package represents either a sharp toolS or a blunt tool held by a person (Table 1); the blunt tool throws a longer, more damaging arc. A ceramic capacitor and a person serving as the capacitor give similar results. Discharge protection ratings are assigned according to Table 2. This test is realistic because a spark is thrown, as often happens in reality. The Electronic Industries Association standard EIA 541 test, with large, flat electrodes, low volt• age (1 kV), and no spark, allows metal-out, partially transparent bags to pass, but the metalli• zation is destroyed by a spark from a fmger (Fig. 3). A potential of 8 kVon a person, as shown in Fig. 3, commonly results from walking on a nylon carpet at 30% RH.

CDM Safety. In the CDM damage mechanism, a charged ESD-sensitive device discharges to a conductive surface. Tests have been devised to simulate this process,6-9 and Fig. 4 shows a method using an SED.· An indirect method for judging CDM safety is to measure the resistance to ground of materials with a buried conductive layer (Fig. 5) or resistance between electrodes for other materials (Fig. 6). The resistance reading can be related to tests with metal oxide semiconductor field-effect transistors (MOSFETs) in simulated CDM events.6 Table 3 gives examples.

Current-Carrying Capability. The resistance is measured between two 5-lb, 2.5-inch• diameter National Fire Protection Association (NFPA) 99 electrodes on the surface of the

Fig. 1. Protected SED. PAPER NO.2O 307

Fig. 2. High-voltage discharge test.

Table 1. Test Conditions.

SHARP TOOL BLUNT TOOL

Voltage,tV 10 6 10 Capacitance, pF 200 350 350 Probe diameter, inch 0.025 0.25 0.25

Table 2. Discharge Protection Ratings.

BLUNT-TOOL 'I1!ST SHARP-TOOL 'I1!ST 6kV 10kV RA11NO F Poor p F F Fair P p F Good P P P Excellent

Code: p =Pass, F =Fail

Fig. 3. Discharge tests on metal-out bags. 308 APPENDIX

Package Surface '--

Discharge

Probe

Tel

Fig. 4. CDM test schematic. material (Fig. 6). Resistance for a paperboard packaging material versus voltage and RH is shown in Fig. 7. Table 4 gives current calculated from resistance for several materials. Obviously, low current-carrying capability is desired for safety for people as well as de• vices. An alternating current of 4-21 rnA causes reflex action (which could result in injury), 21-40 rnA causes muscular inhibition, and 40-100 rnA causes respiratory block.1O there• fore, only the fIrst material in Table 4 is a current-carrying hazard for personnel at Ito V.

Surface Resistivity versus RH. Surface resistivity values are only approximate because the measurement depends on factors such as the pressure of the electrodes against the sample. Nevertheless, this property is useful as a guideline. At approximately 1012 ohms/square,

Megohmmeter

~ LL2

Sample Buried Conductive Layer

Fig. 5. Resistance to ground. PAPER NO. 20 309

ohmmeter

~ LD

Fig. 6. Current-carrying capability. materials become nonconductive and hold static charges for several seconds or more. Re• sistivity rises with falling RH, and the RH at which 1012 ohms/square is reached is the cutoff for retention of the antistatic property. Surface resistivity at various RH levels is measured by sealing samples above humidity-regulating saturated chemical solutions (or anhydrous calcium sulfate for 0% RH) as diagrammed in Fig. 8 and explained in detail in Ref. 11. Examples of curves of surface resistivity versus RH are shown in Fig. 9. A low-RH cutoff is desired for packaging materials because dry air may be encountered in shipping and han• dling. During unpacking, for example, triboelectric charges not drained to ground (through the work surface or operator's fingers) could cause damage by the FIM.

Permanence of the Anmtatic Property by Solvent Extraction. Materials depending on fu• gitive antistatic additives (antistats) can become nonconductive and hold static charges when the antistat is lost by volatilization, absorption by contact with materials such as paper products, or solvent extraction. Table 5 gives examples of surface resistivity measurements

Table 3. ResIstance to Ground and COM Danger.

RESISTANCE PREDICTED MOSFET MATERIAL RH(%) (OHMS AT 10 V) FAILURE RAl1l (%)

Conductively coated paperboard 50 3xl(}l 87 with thin antistatic varnish, Brand A Conductively coated paperboard 40 8 x 1()3 80 with thin antistatic varnish, Brand B Paperboard impregnated with carbon 50 2 x 1()6 37 Paperboard with buried 43 2xlOS 0 conductive layers Conductively coated paperboard 36 6 x 109 0 with heavy antistatic varnish 310 APPENDIX

10 8 en E .c 0 cC 107

10V 6 • 10 l00V •0 ___ 500V

o 10 20 30 40 50 60 70 80 90 100 R!i%

Fig. 7. Resistance to ground for cardboard with buried conductive layer. before and after exposure of samples to two common solvents. The antistatic coating failed because the antistat in it was easily leached.

Antistat Transfer. Transfer of antistat to the surface of packaged items can cause unde• sirable effects such as lowering of surface impedance of resistors, stresscracking of poly• carbonate plastic, and discoloration of epoxy paint.2 In a simple test (Fig. 10), samples are pressed against nonconductive plastic (cellulose acetate) sheets, which become antistatic if antistat is transferred. In Table 6, antis tat is seen to have been transferred from an antistatic

Table 4. Current-Carrying Capability at 110 V.

RESISTANCE CURRENT MATERIAL RH(%) (OHMS) (MA)

Conductively coated paperboard 43 3 x l()1 32 with thin antistatic varnish (measured at I v) Paperboard impregnated with carbon 50 I x 1()6 0.11 Paperboard with buried conductive layers 43 9x 10' I X 10-3 Antistatic viny I 54 Ix 10" I x IQ-4 Conductively coated paperboard 36 3xl0" 3 x 10-5 with heavy antistatic varnish PAPER NO. 20 311 coating. Superior antistatic materials are inherently antistatic, without depending on an ad• ditive, or they contain a polymeric or chemically bound additive that cannot migrate.

Abrasion ResIstance. This property is measured with a Thber Abraser in which an abra• sive wheel presses the surface as shown in principle in Fig. 11. The difference in cycles required to wear holes through the liner (skin) of the four conugated paperboard samples in Thble 7 is caused by varying liner thickness. Superficial conductive coatings were quickly removed, whereas a buried conductive layer was not reached for hundreds of cycles. Abra• sion is of concern because it might cause loss of BSD shielding and, more importantly, result in sloughing of conductive particles that could bridge circuit lines in open devices.

Triboeledrle Charging. This property is notoriously erratic and unreproducible. Rather than attempting to bracket all materials with the extremes of the triboelectric series, quartz and TeflonTN, testing with the materials of concern in practice is recommended. A method has been described' in which coupons of circuit-board laminates (epoxy or polyimide) are stroked against a packaging material and the voltage on the laminate is read with a static meter.1b prevent buildup of contamination, the least amount of which can give false charges, the coupons are cleaned with isopropyl alcohol between determinations. Because of the scatter of the data, as seen in Thble 8, the best that can be done is to sort samples into the rough categories of low, high, and medium chargers. Very low chargers, e.g., antistatic

~ ~

Fig. 8. Surface resistance vs. relative humidity. 312 APPENDIX

.. ...~ ....> ...~ cI I

1=PIp8I'boenI 2 • ConducIIvely coated papeltloerd with heavy antIatatlc vamlah 3. hpeltMMlrd with bUried conductive layers 4. Paperboercllrnpragnated with carbon

105 ....~~ __.. __ .. ____ .. ~ .. o 10 20 30 40 50 60 70 80 90 100 RH,%

Fig. 9. Surface resistivity vs. relative humidity. polyethylene, may owe their success to the undesirable transfer of antistat, causing liquid to separate from liquid on a microscopic level. For rigid containers, which do not rub the packaged item on loading or removal as bags do, a medium-charging rating should suffice.

Crypotocharges. These hidden charges are an example of stored energy. They are diffi• cult to detect with a static meter and occur on packaging material with a buried conductive layer that suppresses the voltage on the charged (electron-rich or electron-depleted) non• conductive surface. 12 Cryptocharges are brought out of hiding by contacting the surface with a conductor, which becomes polarized, and then momentarily grounding the conduc• tor, e.g., by touching it as shown in Fig. 12, so that electrons flow on or off the conductor to PAPER NO. 20 313

Table 5. Solvent Extraction Tests.

SURFACE RESISTIVITY (OHMS/SQUARE, 100 V. 31-41% RH)

AF"lER ONE WEEK

ISOPROPYL WATER RINSE WATER SOAK ALCOHOL SOAK MATERIAL ORIGINAL (30 SECONDS) (30 MINUTES) (30 MINUTES)

MIL-B-SI705C. Type I or 10'0 >1013 >1013 >1013 TypeID Plastic with antistatic coating 9 x lOS >1013 >1013 >10'3 Paperboard with buried 3 x 109 2 X 109 4 x lOS 3 x lOS conductive layers Conductively coated 1 x 10'0 1 X 1010 9 x 1Q9 2 X 1010 paperboard with heavy antistatic varnish

give it a net charge. This charge then manifests itself as an E field (electrical field) when the conductor is lifted from the surface. This special case of compound induction. called con• tact compound induction (CCO. can impart several kV to the conductor, which could dam• age ESD-sensitive items by a direct discharge according to the MM or, if the conductor is part of an ESD-sensitive item, by the CDM. Scenarios for damage to devices by cryptocharges are unlikely but must be considered, especially for critical hardware such as defense or Iife• support systems.

Non-ESD Properties Desired

Transparency. This property is easily obtained with plastics but not with paper products. Cellophane, like paper. is cellulosic, and the correct grade, e.g .• plasticized and nitrocellu• lose-coated. is permanently antistatic, noncorrosive, and heat-sealable but has low tear strength and is best used as a window, liner, or intimate wrap. '3

Samples

Fig. 10. Antistat transfer. 314 APPENDIX

Table 6. Antlstat Transfer Teste.

SURFACE RESlSnvITY (OHMSISQUAItE, 100 V. 36-59'1 RH) OF CEU.ULOSE ACETA11! (ORlOlNAUY 1014)

MA11!RIAL AFI1!R ONE DAY AFI1!R ONS WEEK

Plastic with antistatic coating 3 X 1010 1 X 1010 M1L-B-8170SC. TYpes I or III; antistatic vinyl; paperboard >10" >10" with buried conductive layers; conductively coared paperboard with heavy antistatic varnish

Low Corrosivity. A stringent test13 is to expose copper or Sn63 solder in contact with a packaging material at 100% RH for three months at 7S-80"F or one month at 1lO°F. There are many standard tests. e.g .• FED-STD-IOI. Method 3005. Paper products should have a low reducible sulfur content by the Technical Association of the Pulp and Industry (TAPPI) test T406 or FED-STD-IOl, Method 3005.

Low Toxicity. By Committee of Northeastern Governors (CONEG) regulations, heavy metals, e.g., lead, mercury, cadmium, and hexavalent chromium, cannot exceed low levels in packaging. These metals can be minimized in both plastic and paper.

Strength and DurabiUty. The American Society for Testing and Materials (ASTM) has many standard tests for properties such as tensile strength (ASTM D 882), tear resistance (ASTM D 1004, D 1922). puncture resistance (ASTM D 2582), impact resistance (ASTM D 3420, D 4272), and resistance to being dropped (ASTM D 959, D 4(69). Less general properties, e.g., score resistance, are important in specific applications.

tr,\ Rotating 8 -Sample

Fig.ll. Abrasion resistance. PAPER NO. 20 315

Table 7. Taber AbraserTest.

CYCLES 1'0 ABRADE CYCLES 1'0 CREATE RESISTANCE (OHMS)· (CAUSE DULL HOLES IN 3% OF SURFACE) 50% OF UNABRADED AREA UNDER MATERIAL AREA UNDER WHEEL AREA ABRADED AREA WHEEL

Conductively coated 10 3 x l()l Ix10"-3x1OS llOO paperboard with thin antistatic varnish Paperboard impregnated 140 4 x l()l 9 x l()l 540 with carbon Paperboard with buried 470 >2 x 109 >2 x 10" 700 conductive layers (3 x 10" inside holes) Conductively coated 2 I x 10'0 5 X 109 1050 paperboard with heavy antistatic varnish

• AI 1V. measured between 0.063-inch-diameler conical probes 0.2 in. apart.

Recyclability. Plastics are recyclable but require sorting for best results; a single type of plastic can be added as regrind to virgin plastic for , but mixed plastics can be used only for noncritical items such as low-pressure sewer pipe. Another approach is to depoly• merize mixed plastic waste to basic oil feedstock. 14 Germany is especially strict about recy• cling plastics, with a goal of 80% of plastic packaging to be recycled by July 1995. Of course, reuse of plastic items is preferred to ; some companies reuse plastic trays, for example, to lower costs and protect the environment. Paper packaging has an advantage in recycling because it need not be sorted (unless, for example, it has a high carbon content) and can be mixed with general paper waste, such as newspapers, for repulping. The solid• waste debate is discussed in a recent book" that challenges many conventional positions. For example, conversion by U.S. fast-food chains from disposable to would require consumption of huge arnounts of water, detergent, and energy for dishwashing.

Biodegradability. This controversial property requires moist conditions. Dry landfills contain intact paper products after decades, but this is no surprise (witness Egyptian pa• pyri), and it might be argued that landfill material beneath permanent building sites need never degrade. In the United Kingdom, abandoned coal mines provide low-cost, permanent disposal sites for all ordinary waste. IS The biodegradability of paperboard packaging is superior to that of wood or leaves and is at least useful in preventing long-term littering of the landscape.

Flame Resistance. There are many tests for plastics, e.g., Underwriters Laboratories test UL94 for heavy sections or ASTM D 568 or D 1433 for sheet and film, and for paper products, e.g., ASTM D 4433 for fire-resistant, treated paperboard. A new fire-retardant, antistatic, recyclable, and biodegradable treatment for liner board for packaging was an• nounced late in 1993. Some plastic materials are inherently fire-resistant, for example, one-mil 316 APPENDIX

Table 8. Trlboelectrlc Charging Test.

RATING OVERAU (PROPENSITY ro VOLTAGE ON COUPON (tV). 11fREE AVERAGE (kV). CHARGE CIRCUIT READINGS AND AVERAGE NEGLECTING BOARD MATERIAL EPOXY I EPOXY 2 POLYIMIDE SIGNS LAMINATES) Plain paperboard +1 +1 -7 (corrugated cardboard) +1 +2 -1 3 Medium ±l ±ll ::2 +1 +S -3 Carbon-loaded polyethylene +4 +18 +3 +1 +11 +1 8 High ill ...n ±2 +S +16 +2 Antistatic polyethylene 0 0 0 (Mll..-B-8170SC, Type II) 0 0 0 0 Low Q Q Q 0 0 0 Paperboard with buried +2 +2 -4 conductive layers +1 +4 -2 3 Medium 1:3. +7 =l +2 +4 -2 Paperboard impregnated +1 +10 -1 with carbon +O.S +12 -2 4 Medium -.Jl ill ::l. +1 +11 -1 Paperboard with heavy +14 +14 +1 antistatic coating +11 +11 +1 8 High +11 ±ll .J! +12 +12 +1 polyvinyl chloride (PVC) packaging film that is made permanently antistatic by a propri• etary polymeric additive. Aluminum trihydrate and magnesium hydroxide are environmen• tally friendly flame retardants because they function by endothermically decomposing to release harmless water rather than to toxic gases. 16 Other inorganic, nonhalogenated flame retardants include intumescents like ammonium polyphosphate.1 7 All these additives can be used in (or on) both plastic and paper packaging.

Other Properties. The many other properties of importance in packaging include perme• ability, markability, heat-sealability, and gluability with both resin and hot-melt adhesives. It is important to realize that just one "Achilles heel" in some special property can dis• qualify an otherwise ideal package. In fact, it is questionable how many applications a single "ideal" package design could satisfy. The "ideal package" may in fact be a family of designs with intrinsic advantages, e.g., permanent antistaticity and volume conductivity, in common. PAPER NO. 20 317

Fig. 12. Contact compound induction (CCI).

Promising New ESD-Protective Package Designs

Plastic. Conventional plastic packaging is typified by MIL-B-81705C (Table 9), which depends on antistatic additives and is volume-nonconductive. However, a commercial poly• meric antistatic additive confers permanence (with no antistat transfer to other surfaces) and volume conductivity (to avoid stored energy such as cryptocharges) to a variety of thermoplastics including acrylonitrile-butadiene-styrene (ABS), acrylic, polycarbonate, PVC, nylon, polyester, thermoplastic polyurethane, and polypropylene. PVC and nylon versions were found to be antistatic at 0% RH when measured by the method of Fig. 8. New products of this type have appeared as both sheets and thin films. Another approach to permanent antistaticity involves dispersed powdered metal, e.g., copper, or metallic or carbon fibers. Some plastic packages use protective spacing (air gaps or antistatic foam between the wall and the packaged item) to mitigate the effect of external discharges.s In one design, a trans• parent, antistatic bag is simply inflated with air. Spacing in conjunction with a fair shielding layer can raise the discharge rating to good (Table 2), and spacing in conjunction with a good shielding layer can raise the rating to excellent.

Paper. Paperboard has the advantages of being inherently volume-conductive and biode• gradable, as well as reusable and recyclable. Its weaknesses are permeability, opacity, flam• mability, and limited durability. Several commercial corrugated paperboard designs, which are stiff, strong, and lightweight, were tested with the following results. Paperboard with carbon impregnated throughout the skin was too conductive (Table 3, Fig. 1), but conductively coated paperboard with a heavy antistatic varnish was CDM-safe (Table 3), safe to person• nel (Table 4), and permanent (Tables 5 and 6). Though the varnished paperboard lacked high abrasion resistance (Table 7), a highly conductive surface was never exposed. Conductively coated paperboard with a thin varnish was undesirably conductive (Tables 3 and 4). A particularly effective design has a conductive carbon layer on the inside of each 318 APPENDIX

Table 9. Plastic vs. Paper Packaging.

PAPERBOARD WITH BURIED CONDUCTIVE LAYERS MIL-B-81705C METALLIZED (PLASTIC) CARBON PAPER RJTURE PROPERTY TYPE I TYPE III LAYER LAYER PLASTIC PAPER High-voltage Excellent Good Fair* Excellent Excellent Excellent discharge resistance (good with (Table 2) O.5-in. anti- static foam between wall and packaged item) CDMSafety Excellent Excellent Excellent Excellent Excellent Excellent Volume conductivity No* No* Yes Yes Yes Yes (to prevent stored energy) Antistat transfer No No None None None None Permanence No* No* Yes Yes Yes Yes (solvent extraction) Triboelectric Low Low Medium Medium Low- Low- charging Medium Medium Biodegradability No* No* Yes Yes Yes Yes RecycJability Yes Yes Yes Yes Yes Yes Transparency No* Yes No* No* Yes Yes (cellophane window) Vapor barrier Excellent Fair Poor* Poor* Excellent Good (permeability) Flame resistance Poor* Poor* Poor* Poor* Good Good Toxicity Low Low Low Low Low Low Durability High High Fair* Fair* High Medium Costlow Low Low Medium Medium- Medium- High High Overall comments Definite deficiencies, Definite deficiencies, In general, approaches e.g., permanence, e.g., opacity, permeability the ideal, but cost stored energy to vapors will not be low (reuse will help)

• Deficiency. In some cases, e.g., biodegradability, criticality is arguable. liner (skin) in contact with the medium (fluted paper). Thus, the conductive layer is safely buried and cannot be easily abraded or be a CDM or personnel hazard. This design has generally excellent ESD properties (Tables 3-7 and 9, Figs. 7 and 8) but gives only fair shielding against external discharges. Excellent shielding is achieved in this design if metallized paper is substituted for the carbon layer, while all the other desirable proper• ties are retained (Table 9); aluminum is, in this context, nontoxic, and the amount used (approximately 1000 A) is too small to interfere with repulping or . Two layers of metallized paper, one inside each liner, are needed for the excellent shielding PAPER NO. 20 319 rating; a spark discharge creates a hole in the metallization on the first layer but leaves the second layer intact and is intercepted by it. Metallized cellophane is a barrier to gases but is more expensive than metallized paper and, not being fibrous, can cause fish-eyes in re• cycled paper. Aluminum foil, incidentally, gives excellent high-voltage discharge protec• tion but presumably interferes with repulping.

Future ESD-Protective Package Designs

Plastic. The antistat permanence and volume conductivity achieved for plastics in recent years has been a major breakthrough. This corrects the deficiencies in these properties found in MlL-B-81705C, Types I and III, which are used as a reference point in Table 9, and leaves only lack of biodegradability and lack of flame resistance. These may not be major deficiencies, but efforts can be expected to correct them, at least for special packaging applications. In fact, plastic packaging already available meets most of the "future" ratings in Table 9--{)r can meet them if the right materials are combined. For example, aluminum metallization buried between layers of the PVC film mentioned above might give a perma• nently antistatic, flame-resistant version of MIL-B-81705C, Type I. Biodegradability is a more "blue sky" property because, for one thing, the plastic might degrade prematurely on the shelf or in use, but new products are appearing on the market. IS Also, transparency and excellent (rather than good) high-voltage discharge resistance are incompatible properties with present methods of metallization, and an extremely good vapor barrier requires metal foil, an opaque layer of vapor-deposited metal or, possibly, a layer of expensive polychlorotrifluoroethylene (PCTFE) transparent film.

Paper. Paper has the inherent limitations of being opaque and highly permeable. Coated cellophane is transparent and a good vapor barrier and is volume-conductive and biode• gradable, 13 but it is not recyclable for repulping (only for producing more cellophane). Flame retardants can be added to paperboard, but corrosivity must be avoided. A flame-retardant treatment for liner board recently announced claims to be colorless, odorless, nontoxic, biodegradable, and recyclable. Durability of paperboard packages can be improved by us• ing heavy sections, but there will be penalties in weight and cost. The paperboard with buried metallized paper proposed in Table 9 remains to be commercialized and would be a good future product on which to improve with flame retardants, etc.

Plastic versus Paper. In general, plastic is stronger and more durable than paper, but paper is more "natural"; it is made from a renewable resource, is biodegradable, and is easily recycled. Also, in regard to ESD, paper has the advantage of being permanently antistatic and volume-conductive, but state-of-the-art plastic compounds also have these qualities. Therefore, paper and plastic are moving toward the ideal package from differ• ent positions. In the foreseeable future, neither material will preempt the other, and paper and plastic constructions will coexist along with mixed constructions, e.g., paperboard boxes lined with antistatic plastic foam. Of course, combinations like the latter could be recycled only by separating the materials. A foam-lined paperboard box would be reused as many times as possible and then peeled apart for separate disposal of the foam and paperboard. 320 APPENDIX

Conclusion Much progress toward an ideal ESD-protective package has been made in recent years. Advanced materials are available, realistic tests have been devised. and ingenious, cost• effective packages can be created by metallization, lamination, incorporation of cushion• ing such as foams, rigid mounting of items to prevent triboelectric charging, protective spacing with air gaps, and other design features. Biogradability for plastic packages will be difficult to achieve but may be unimportant, while flame retardance for either plastic or paper can be attained to a considerable degree if demanded. Certainly, overly conduc• tive surfaces, less than good high-voltage discharge resistance, fugitive antistats, and stored energy will be - or should be - problems of the past. Without major breakthroughs, consolidation and extension of present techniques should give products approaching the "ideal" packages in Table 9. These may be more expensive than current packages, but greater durability as well as emphasis on reuse will lower the effective cost. Meanwhile, less toxicity and more recycling and reuse will protect the environment.

References

I. J. M. Kolyer, "Fundamentals of ESD Control," EMC Technology Magazine 1992 Expo Techni• cal Record, May 1992, pages 154-161. 2. 1. M. Kolyer and D. E. Watson, ESD from A to Z: Electrostatic Discharge Controlfor Electron• ics, Van Nostrand Reinhold, 1990. 3. 0.1. McAteer, Electrostatic Dischange Control, McGraw-Hili, 1989. 4. 1. M. Kolyer, "Realistic Testing of ESD Materials," EMC Test and Design, September 1993, pages 28-31. 5. 1. M. Kolyer and D. E. Watson, "Packaging for High-Voltage Discharge Protection," Evalua• tion Engineering, March 1992, pages 96-100. 6. 1. M. Kolyer and D. E. Watson, "COM and Work Surface Selection," Evaluation Engineering, October 1991, pages II 0-117. 7. D. C. Anderson, "A Simple Approach to ESD Damage Prevention," EMC Technology, Marchi April 1991, page 38. 8. "New Test Proposed for Checking ESD Safety of Materials," Compliance Engineering, Fall 1990, page 77. 9. J. M. Kolyer, "Is Your Work Surface COM-Safe?," EOSIESD Technology, February/March 1992, pages 27-28. 10. DoD-HDBK-263, May 2,1980, page 46. II. 1. M. Kolyer and R. Rushworth, "Humidity and Temperature Effects on Surface Resistivity," Evaluation Engineering, October 1990, pages 106-110. 12. J. M. Kolyer and D. E. Watson, "Hidden Charges on ESD-Protective Packaging," Evaluation Engineering, September 1992, pages 94-100. 13. News notes in EOSIESD Technology, OctoberlNovember 1989, page 9, and Evaluation Engi• neering, March 1990, page 96. 14. P. L. Layman, "Advances in Feedstock Recycling Offer Help with Plastic Waste," Chemical and Engineering News, October 4, 1993, pages 11-14; News note: "Group Formed for Recy• cling Plastics into Feedstocks," ibid, November 29,1993, page 41. 15. 1. H. Alexander,ln Defense ofGarbage, Praeger Publishers, 1993. 16. S. Ainsworth, "Magnesium Oxide Finds New Applications," Chemical and Engineering News, October 25, 1993, pages 15 and 16. 17. "Flame Retardants: Processors Learn to Work With Halogen-Free Systems," Modern Plastics, September 1993, pages 55-60. 18. "New Players Emerging in Biodegradable Polymers," Modern Plastics, October 1994, pages 33-37. Index

Abrasion resistance, 311; illus., 314; grit-blasting, 68 table, 315 paper forms and , 58 Aclar shipping containers, 71 charged by heat gun, 65 vinyl chair, 58 in cryptocharge test, table, 300 categorization of, table, 23 in operator charging test, 245 definition of, 23 in special test on chain link fence, discussion of, 26 69 Antistat, topical. See 1bpical in special test on walls, 58 antistat usually negatively charged, 43 Antistatic Air approach to ESD control, 28, 196 space for ESD protection, 296; bags and tore boxes, permanence table, 295; table, 297 of, 175-180; table, 179; table test for triboelectric charging by, 313 12,66 box, transparent, 101 Air guns, nuclear and electrical, bubble-wrap, ·63-64 criteria for selection, 100 Aluminum foil decay of charge on, by room for conductive masking tape, 193, ionization, illus., 193; table, 198 192 for shielding, 39 for cushioning, 48, 173 for SSP, 48; table, 262; table, 275; categorization of materials, table, table, 295 23 tape for holding shields on module, clean nylon, criteria for selection, 68 99 Amber, charging of, 3 cubical container, 101 Analysis definition of, 23, 83, 110 applied to discussion of, 27 automated process, 72 foam, brushes, 60 criteria for selection, 100 charged windows, 59 shunt, 104 cleanroom gloves, 62 testing of, table, 275; table, 295

321 322 INDEX

Antistatic (continued) measured in disposition of intimate wrap in SSp, 48 mishandled hardware, 145 materials, on circuit-board laminates, by disadvantages of, 28, 159, sliding on stainless steel, 66 238-243 on walls, 91 in future, 159 ARTG pennanence and shelf life of, categorization of, table, 23 175-180; table, 179 definition of, 24 resistance measurements on, 170 discussion of, 27, 207, 210, 250; plastic liner for trash cans, 92 table, 157 polyethylene, in model handling and assembly cleaning of, 89-90 specification, 81 criteria for selection, 99 lower limit of one megohm, 81 discoloration of paint by, 242 lower limit violated with wet shoes effect on adhesive bonds, table, on conductive floor, 91 241 upper limit of 10 megohms, 81, fogging of instrument mirrors by, 210,250 239,241 Assemblies, definition of, 89 precautions with, 89-90 Assembly and handling, model testing of, 263, 286; illus., 287; specification for, 77-106 illus.295 Audits and reviews, 87-88 smocks, 100; table, 141 Automated process surfaces, 253 control of hazards in, 72, 88 vinyl, 75-76, 315-316 coupons for, 72, 233-237 workbench top, 185 monitoring of, 237 Antistatic lotion recertification of, 88 categorization of, table, 23 criteria for selection, 105 Badges, identification definition of, 23 among necessary nonconductors in discussion of, 27, 245, 250 SSW, 84 Antistats hazard of, 57-58 corrosion and contamination by, treated with topical antistat, 58, 238-243 84 effect on adhesive bonds, table, 241 Bag materials transfer of, 310-311; table, 314 commercial, 201; table, 202 Apparel of operators. See Clothing, conclusions from tests on, 170-174, operator, and Smocks, 205-206,294-296 ESD-control criteria for selection, 99-100 Apparent charge in future, 159-160, 174, 319 categorization of, table, 23 properties of, table, 202; table, 205; definition of, 24 table, 295; table, 300; table, discussion of, 27; table, 157 318 in special test of air guns, 64 screen layer for excellent shielding, in special test on chain link fence, 206; table 202; table, 205 69 tests on, 274; illus., 204; illus., INDEX 323

275; illus., 295; illus., 307; for boxes for SSP, 48, 311; table, tables, 195; table, 203; table, 275; table, 295; table, 300 205; table, 227; table, 262; in packaging specifications, 111 table, 275; table, 295; table, tote box, cost of, table, 142 307 Carpet Bags, Faraday-cage. See Faraday antistat-treated, 18, 83, 91 cage charge from walking on, 14, 18, 73; Basic Rule illus., 9; mus., 17; table, 36 categorization of, table, 23 charge generation on plastic by, 299 central in operator training, 105, with conductive fibers, criteria for 106 selection, 73, 102 definition of, 24, 80, 109 Carriers, requirements for, 111 discussion of, 28 Cathode ray rubes (CRrs) importance of, 1-2 charges on, 59 in model handling and assembly screens for, 59, 105, 141 specification, 80 Caution tag, 87 in model packaging specification, CCI (Contact Compound Induction), 109 299,313; mus., 301; illus., Benchtops. See Workbenches 317 Bins, criteria for, 101 CDM Biodegradability, 315 categorization of, table, 23 Bleed-off time (decay time), 41, 264, definition of, 24 277 discussion of, 7-8, 16, 29,44-46, Blow-off nozzles, 37-38, 73; table, 115, 162, 253, 254, 260, 190 263-271, 281, 285, 305, 306; Boxes, criteria for selection, 101 illus., 7; mus., 15; illus., 277; Boxes, tote. See Tote boxes allus., 308; table, 309; table, Brushes 318 Criteria for selection, 105 involving operator, 45, 162, 186, evaluation of four types, 60-61 285 Bubble-wrap, antistatic. See Anti• test on tote boxes, 231; table, 229 static bubble-wrap CD Rule Buyers' guides for packaging mate• applied to rials, 49 brushes, 61 clothing, 57 Capacitance gloves, 85-86 categorization of, table, 23 screwdriver handle, 61 definition of, 24 walls, 58, 91 discussion of, 29 windows, 59 of capacitor (150 pF) used in as primary defense, table, 158 shielding/discharge test, 183, categorization of, table, 23 294; table, 307 definition of, 24, 198 probe used in shielding/discharge discussion of, 29,45, 116 test, 48, 262; illus., 295 experimental derivation of, Cardboard 188-189; illus., 189 324 INDEX

CD Rule (continued) static charges on, 10, 57, 162 for 12-volt-sensitive devices, equa• Complementary equipment/materials tion and table, 161-162 and techniques, table, 197 in model handling and assembly Components, definition of, 89 specification, table, 84 Conductive in operator training, 105 approach to ESD control, 28, 196 Cellophane, 74, 286, 292; mus., 288; boxes, 101; table, 309; table, 310 mus., 293 categorization of materials, table, Certification 197 of ionization equipment, 88 chairs, 102, 139, 141, 156,254; of operators, 55, 86, 88, 105-106 table, 197 of packaging, 113 definition of, 24, 83, 109 of SSWs, 51, 80, 88, 93-94 discussion of, 30 Chain link fence fibers in carpet, 73, 102 for shielding, 69-70 floor mat, 82, 102, 139, 141 near SSZ, 86, 129 floor tile, 75, 102; mus., 74 Chair, vinyl, static charge on, 58; floors, table, 187 hazard of, 91, 132, 143 Chairs, conductive. See Conductive in model handling and assembly chairs specification, 91 Charged "finger" test, mus., 171 foam for shunt, 40,104 Charged Plate Monitor, 67, 151 footwear, 72-74; table, 139; table, Charges, hidden. See Cryptocharges 141 Check lists, 148, 150, 151 hazard by carrying lethal current, Chloride ion 143,306,308; mus., 309; corrosion of silicon wafers by, 42, table, 227 52 heel strap, 100 in workbench laminates, 63 masking tape, 193, 198 Circuit-board laminate, charging of materials, by grit-blasting, 68 disadvantages of, 30, 143, 178, by rubbing with antistatic packag• 180,226 ing, 176; mus., 177 in future, 160, 319 by sliding on stainless steel, 66 when to use, 53, 143, 196-197 Classification of ESDS Items, 79 packaging materials, high tri• Cleaning of antistatic polyethylene, boelectric charges generated 89-90 by, 178, 226, 279; table, 228; Cleanliness table, 279; table, 316 in model packaging specification, plastic, table, 262 lll, 114 polyethylene bags, 263; table, 275; of SSW, 85, 90 table, 282; table, 295 Cleanroom gloves. See Gloves, polyolefin, criteria for selection, 99 cleanroom seat covers, 102 Clothing, operator shunt for conectors, plastic, 40, 62, as hazard, table, 197 104 rules for, 85, 135 surfaces, 253 INDEX 325

tape for holding shields on module, required in SSW, 55, 94, 157,250 68 safety of, 95, 143 threads in smocks, 61 springs for safety, 143 trays, 101 two-conductor (resistive) type, 214, Conductive tote boxes. See Tote 250; illus., 215; illus., 247 boxes, conductive versus periodic monitoring, 213, Conductor 250 categorization of, table, 23 Contracting, 136 checking of grounding with Corona discharge, 9, 11 ZapOasb,247 Corrosion definition of, 24 by workbench laminates, 63 discussion of, 30 of silicon wafers, 42, 52, 63 isolated, charging by ionizers, of solder by antistat containing 66-67 n-octanoic acid, 238-239, when to allow in SSW, 53 242; table, 240 Connector dust cover, special test and special test for, in SSp, 49 standard test for, 59-60 Corrosivity in packaging, 110, 114, Connectors 314 dust cover for, 59-60 Cost-effectiveness of ESD-control electronic box, packaging of, 70 program, 134-135, 153 power turned off before inserting Cots. See Finger cots ESDS item, 86 Coupon shunt for, plastic, 39-40, 62, 104 categorization of, table, 23 Contact charging, 10 definition of, 24 Containers for storage or in-plant discussion of, 30, 234 transfer, 112 for automated process, 72, Contamination 233-237; illus., 236 by antistatic polyethylene, for testing SSP, 47 238-243 CRrs. See Cathode ray rubes of packaging, 111, 114 Cryptocharges, 299-304, 312-313 Continuous wrist-strap monitor Current-carrying capability, 143, 306, alert signals, 95 308; illus., 309; table, 227 as primary defense, table, 158 Curve-tracer, 146, 151, 182 band,95 Cut-off relative humidity, 292, 309 calibration of, 94 capacitance type, 213; illus., 214 Damage mechanisms categorization of, table, 23 check list of, 148 cord,95 in automated process, 236-237 cost-effectiveness of, 216 relation to defenses and hazards, criteria for selection, 94-95, 100 table, 157 definition of, 24 Decals, on certified operators' badges, discussion of, 30,157,207-217, 106 250 Defenses procurement requirements for, primary and secondary, table, 158 94-95 relation to hazards and damage 326 INDEX

Defenses (continued) Ius., 204; illus., 275; illus., mechanisms, table, 157 295; illus., 307; table, 203; cabinets table, 307 criteria for selection, 104 pulse attenuation measured by, static charges in, 69 194-195; table, 195 Detergent (water scrub) test, illus., results of, table, 195; table, 203; 221 table, 205; table, 227; table, DI 262; table, 275; table, 295; allowable voltage on conductors for table, 318 12-volt-sensitive devices, Disposition of mishandled hardware, 161-162 88, 145-147 categorization of, table, 23 Documents, reference, check list of, definition of, 24 149-150 discussion of, 9-10, 16, 30,209, Double-bagging 281; illus., 14 cost of, 174 from people as principal ESD haz• method, 110, 169 ard,244 Drag test, 40; illus., 222 grouding not always needed for Drain time, 52, 184-185, 231; table, damage by, 68-69 230 in automated process, 236 Dual-in-line package (DIP), 8, 44, Discharge 263,283,304 air,8-9 Dust cover for connector, special test categorization of, table, 23 and standard test for, 59-60 defmition of, 24 discussion of, 31 E Field (see also Field) guarding against, 155 categorization of, table, 23 high-voltage, protection from, definition of, 24 294-298 discussion of, 16, 31; illus., 15; rate increased by conductive sur• table, 157 face, 29, 285 Electron, 3-10,45, 105-106; 160; rate slowed by antistatic surface, illus., 5 31,264,277 Electronic box, packaging of, 70 relation to damage mechanisms and EMIIRFI shielding, 115, 201, 206 defenses, table, 157 Enforcement problem spark, 12, 31, 196, 253, 262, 265, check list, 154 268,296 in ESD-control program, 138-139 Discharge test Engineering function, 137 categorization of, table, 23 Equipment, ESD-control definition of, 24 and materials, discussion of, 31, 110; illus., 204; approved, 95-106 table, 203 complementary, table, 197 in model packaging specification, criteria for selection, 96-106, procedures for, 113-114 196-198 on tote boxes, 229-230; table, 227 what to buy, check list, 150 procedures for, 171; illus., 184; il- approved, as primary defense, table, INDEX 327

158 FFB elimination of unnecessary, 249 categorization of, table, 23 in future, 160 definition of, 25 in SSW, 54-55 discussion of, 33, 45, 256; ilIus., ESDS 15; iIlus., 45 categorization of, table, 23 in automated process, 236 definition of, 24 Field discussion of, 32 categorization of, table, 23 ESDS item controlled within SSW, 50 categorization of, table, 23 definition of, 24 definition of, 24 discussion of, 4-9, 11, 31, 116, 162, discussion of, 32 299, 313; ilIus., 6; ilIus., 7; in future, 161-162 ilIus., 17 selection of packaging materials emission, 11 for, 169-174, 294-320 from lights and light fixtures, 92 Evaporation, charge formation by, 12 from transformers or electric motors, 91 Facilities and Industrial Engineering, from various objects, MOSFE1S 136-137 damaged by, 187 Failure analysis, iIlus., 20 guarding against, 155 Faraday cage hazard at the work station, categorization of, table, 23 188-194,247 definition of, 24-25, 109 penetration of walls by, 91 discussion of, 32-33, 48-49, 116, read by rIeld meter to determine 170, 249, 251 charge on operator, 208 foil laminate relation to damage mechanisms and bags, table, 170; tables, 195; defenses, table, 157 table, 202; table, 205; table, strategy of keeping weak, 155 262; table, 275; table, 295; Field meter table, 318 as noncontact voltmeter, 34, 191; criteria for selection, 99-100 table, 191 in model packaging specifica• categorization of, table, 23 tion, 109, 129 checks with, to enforce CD Rule, MIlrB-81705, 1YPe I as, 170, 56; table, 197 201; table, 318 cost, 140 part of conservative approach, criteria for, 102 156,249 definition of, 25 for SSP, 48-49 discussion of, 9-10, 12, 33-35, tote boxes, 49, 228, 331 140,141 Faraday cup for monitoring of necessary non• categorization of, table, 23 conductors, 84 definition of, 25 in future, 160 discussion of, 33, 114 measuring apparent charge with, in triboelectric charging tests, 27,181-182 230; iIlus., 221 use of, check list, 152 328 INDEX

FlM Freezing, charge generation by, 10, 38 categorization of, table, 23 Future definition of, 25 equipment, 160 discussion of, 4-7, 10, 256, 260, ESDS items, 161-162 278, 279, 281, 299; illus., 6; materials, 159-160 illus. 15; table, 283 properties of bags, 174, 319-320 in automated process, 236 standards, 118, 122, 161 in special test of connector dust cover, 60 Gloss test, illus., 220 Finger cots, 45, 85-86, 100, 162, Gloves 285; table, 282 antistatic, 85-86, 100 Flame resistance of packaging, cotton, 85, 188 315-316 latex, 62-63, 85-86 Floor finish. See Static-limiting Door nonconductive for cleanrooms, finish 62-63,85-86 Floor mats, conductive, 82, 102; vinyl,62 table, 139; table, 141 Grid, ionizing, 103 Floors Grit~blasting as necessary nonconductors in discussion of, 11 SSW,84,255 of heat sink on module, 68 conductive, of module to remove conformal hazard of, 91, 132, 143 coating from components, 68, in model handling and assembly 193-194 specification, 91 Ground fault circuit interruptors in SSW, 51 (GFCls), 82, 143-144 materials, criteria for, 102 Grounding mats for. See Floor mats, as primary defense, table, 158 conductive categorization of, table, 23 tile or concrete, treatment of, 91 check, in SSW, 93-94 PM (Floating Model), illus., 15; table, cord, criteria for, 100 261 definition of, 25 Foam, antistatic. See Antistatic foam discussion of, 35, 162; illus., 17 Foam, conductive. See Conductive in SSW, 80-81 foam lug, for workbench tops, 35, 52, Foil laminate bags. See Faraday cage, 81, 185; illus., 82 foil laminate of personnel, 81, 245, 250; illus. 82 Footwear resistor needed for, 52, 81; illus., hazard of, on conductive floors, 91, 82 132,143 versus DI damage, 68-69 special, needed for conductive floors, 74-75,82,91 H fields. See Magnetic fields with leather soles, low charging, Hair 224 arm, as ESD hazard, 56; table, 197 with vinyl soles, high charging, charge induced on, 8 224 head, control of charges on, 50, 57, INDEX 329

85; table, 197 compound, 10,299 Handling and assembly, model speci• contact compound. See CCI fication for, 77-106 discussion of, 4-8 Hazards, ESO Installation site, procedures, 88 check list of, 148-149 Insulator. See Nonconductor of static charges and fields at the Ionization work station, 181-200 as primary defense, table, 158 relation to defenses and damage categorization of, table, 23 mechanisms, 157 definition of, 25 HBM discussion of, 19, 36-38, 92, 144, as principal ESO hazard, 244 146-147,224,250,251,302; categorization of, table, 23 ilIus., 17; table, 197 definition of, 25 equipment, criteria for, 103 discussion of, 4-10, 15, 161, 256, for controlling charge on silicone 272-273; illus., 5; illus., 15 rubber probe, 72 Heat guns, 64-65 grid,103 Heel mark test, illus., 220 hazard of nuclear, 37, 144, 190 Heelstra~conductive, 100 in future, 160 Hidden charges. See Cryptocharges in grit-blasting, 68 High voltage, nature of, 8-10 in model handling and assembly History of ESO control, 31 specification, 83 Housekeeping in SSW, 85, 90 pros and cons of, 38 Humidification room, as backup, 155 applicability of, 97 as secondary defense, table, 158 caution recommended, 249 categorization of, table, 23 charge decay rates, illus., 193; definition of, 25 tables, 192 discomfort from "mugginess," 224 criteria for equipment, 103 discussion of, 35-36, 147, 249, effect of humidity on, 191-192 250 testing of systems, 183, 191-193, in model handling and assembly 198 specification, 83 when to use, 38, 53 when to use, 52-53, 92 Ionizers Humidity AC, control of, 83, 234 for controlling charge on silicone effect on cellulosics, 255, 292; rubber probe, 72 illus., 288; ilIus., 293 old,67 effect on room ionization system, safe distance from, 66-67; table 191-192 190 relative, 116,277,286-293,302, air guns, electrical and nuclear, 308-309, iIlus., 278; ilIus., 63-64, 73 310; illus., 311; illus., 312 blowers, criteria for, 103 voltages at high and low, table, 36 certification of, 83, 88 charging of isolated conductors by, Induction 66-67 charging, 10 330 INDEX

Ionizers (continued) 92 cost, 139, 141 Manufacturing function, commitment electrical vs. nuclear, 37, 189-190; to ESD-control program table, 190 required, 137 evaluated with MOSFET board, Marking, caution 189 in handling, 87 fanless, 67 of packaging, 111, 115 nozzles, criteria for, 103 Masking tape, conductive, 193, 198 ozone generation by, 38, 89 Materials, ESD-control "piggyback" for grit-blasting, 68, and equipment, 193-194 approved, 95-105 precautions, 89 complementary, table, 197 pulsed DC, criteria for selection, 96-105, criteria for, 103 196-198 safe distance from, 66-67; table, what to buy, check list, 150 190 approved, as primary defense, table, shadowing effect, 89 158 space-charging by, 75, 275-276; in future, 159-160 illus., 276 in model packaging specification, Items, definition of, 89 109-110 in SSW, 53-55 Labeling. See Marking, caution Materials, packaging. See Packaging Labels, paper, allowed in SSW, 58 Mats, floor. See Floor mats Laminate, foil. See Faraday cage, foil Mats, table. See Table mats laminate Megohmmeter, 93,104,151; illus., Laminates, workbench. See Work• 311 benches, laminates Metallized (see-through) bags, limita• Latent fallure tions of, 31, 48-49; illus., categorization of, table, 23 297; table, 205; table, 262; definition of, 25 table 275; table, 295 discussed in video training tape, Metals, criteria for selection, 104 106 Microscopes discussion of, 38-39, 49, 138, 147, in SSW, 54, 247 234,244,256; illus., 15 ungrounded, with rubber feet, 37 possible, of mishandled hardware, MIlrSTD-1686, replacement of, 146 122-129 Lights and light fixtures, ftelds from, Mishandled hardware, disposition of, 92 88, 145-147 Logistics, 136 MM definition of, 25 Magnetic fields (H fields) discussion of, 8, 9, 16, 39, controlled by soldering standard, 273-276; illus., 15 54 MOSFET shielding from, by iron foil, 115 categorization of, table, 23 Maintenance of ESD-protected areas, damage tests, procedure, 182-183 INDEX 331

damaged by, in special test with UV light, 69 charged operator touching lead to role of grounding in test with, bench tops, table, 185; table, 68-69 186 test procedure with, 182-183; fields, illus., 15; illus., 189 illus., 182 faelds from ionizers, 190 Mottoes, check list of, 152 fields from various objects, 187 Myths grounded operator above ARI'G, check list of, 153-154 246 discussion of, 134-135,251-257 grounded operator touching lead to charged tote boxes, table, Neon bulb, 8, 270, 281, 283; illus., 186; table, 187 270; illus., 306 unbalanced room ionization Nitrogen, ionization of, 73 system, table, 191 Nonconductive damaged in CDM test on tote categorization of materials, table, boxes, 231; table, 229 23 definition of, 25 definition of, 25, 110 discussion of, 39, 265 discussion of, 39 in coupons for automated process, plastics, as hazard in work station, 234; illus., 236 137,187,197-198 in demonstrations for training tote box, cost of, 142 personnel, 106 Nonconductor in discharge test, 113 categorization of, table, 23 in packaging tests, illus., 171 definition of, 25 in shielding! discharge test for bag discussion of, 4, 8, 278 materials, 196 excluded from SSW, 85, 247 in shielding test for SSP, 48-49 exclusion from SSW as secondary in special test defense, table, 158 for ARI'G, 246 fields from, damage to MOSFETs for mishandled hardware, 146 by, 187; table 187 on air guns, 64 necessary, in model handling and on connector dust cover, 59-60 assembly specification, 84, 91 on heat gun, 64-65 not charged by nitrogen in desicca• on packaging of electronic box tor cabinet, 69 with connector, 70 not charged in test with UV light, on pulsed DC ionizer, 66-67 69 on screwdriver handle, 61 Nylon, antistatic, 75, 87, 99, 277, in various tests, 4-8, 12, 14-15, 317; illus., 278; illus., 287 44,252,258,260,264,266, 267,272,273,274,281-283, One-meter rule in DoD-STD-1686, 51 294,302 Operator disciplines punch-through of gate oxide of, as primary defense, table, 158 173, 260; illus., 15 categorization of, table, 23 MOSFET board certain, as secondary defense, table, in shipping test, 70-71 158 332 INDEX

Operator disciplines (continued) Permanence of the antistatic property check list of, 151-152 of bags and tote boxes, compatible with work-place para• 175-180,309-310; table 231; phernalia, 135 table 318 cornerstone of cost-effective pro• Personnel Voltage 'lester gram, 134, 156 categorization of, table, 23 definition of, 25 data in ARTG test, illus., 246 discussion of, 39, 138-139, 188, definition of, 25 194,196-197,199,245,250 discussion of, 39, 151, 161 in certification of personnel, 105 for walk test, 83, 248 in model handling and assembly in disposition of mishandled hard- specification, ware,l46 inspection of, 88 Photoelectric effect, 11 list of, 84-86 Piezoelectric effect, 11-12, 15, 38 in SSW, 55 Plant Services, 137 regarding chairs, 58 Plated wire , retraining needed, 86,245 analysis of, 71 Ozone, generation by ionizers, 38, 89 Polarization, 4 Polonium-210, 37, 64, 83, 144, 190 Packaging Polyester fabric, for trlboelectric antistatic bags for, shelf life of, charging in special tests, 56, 176, 178; table, 179 61, 67, 69, 100, 101 biodegradable, 292, 315 Polyethylene, antistatic. See Anti· configurations of, 71 static polyethylene flame-resistant, 315-316 Polystyrene foam, 8-9 for delivery, 88 Probe tools, antistatic nylon, 75, 277 history of, 169-170 Program management, 136 materials in model specification for, Program organization and 109-110 implementation materials, relative costs, table, 174 administration, 249 model specification for, 107-115 advice to small companies, new developments in, 305-319 139-142 paper, 317-319 check list for, 152-155 plastic, 317, 319 cost-effectiveness, 134-135, 250 recyclability of, 315 discussion of, 130-144,249,250 selection of materials for, 169-174 enforcement problem in, 138-139 summary of test results on mate- objectives, 131-132 rials, table, 174; table, 227; plan of action, 136-138 table, 262; table, 275; table, role of program coordinator in, 296; table, 297; table, 318 142-143 Paper safety, 143-144 in packaging, 317-319 variables in, 156 labels and forms, allowed in SSW, weakened by myths, 134-145 58 what, when, and how, 132-133 Particle beams, 11 Proposals, 136 INDEX 333

Puncture resistance of bags, 314; by bag materials, table, 174 table, 174 categorization of, table, 23 Purchasing, 136 definition of, 25 of CKI's by screen, 59, 105, 141 Quality assurance provisions required for cable, 60 in ESD-control program, 137-138, test for. See Discharge test, pro• 143 cedures for in model handling and assembly test on tote boxes, 229-230; table, specification, 88-89 227 in model packaging specification, Shielding/discharge test 113-115 apparatus, mus., 184; mus., 204 QA representative notified when flaws in, 31 hardware mishandled, 145 test conditions, table, 203 test results, table, 205; table, 227 Radar Shipping container, test on, 71 shielding against, 206 (see uncon• Shipping test, 70-71 densed paper for data) Shoes. See Footwear test with packaged MOSFE1S, Shunt, conductive mus., 172 foam, 40, 104 Radiation effects, 121 plastic bar, 40,62, 104 Radioactive decay, II Shunting Rails, requirements for, 111 antistatic plastic for, 104 Relative humidity. See Humidity categorization of, table, 23 Resistor conductive foam for, 40,104 for grounding, 52, 81; mus., 82 conductive plastic for, 40, 62, 104 simulated, mus., 236 definition of, 25 thin-film, 44, 267; table, 283 discussion of, 39-40 Roller test, 40; mus., 221 foam for, antistatic or conductive, Room ionization. See Ionization, 39-40,104 room in model packaging specification, 111, 114 Safety lead-shorting devices for, 104 check list, 155 material, criteria for selection, 104 in ESD-control program, 143-144 to protect ESDS items in SSp, 47 of personnel by grounding, 81-82 Silicon wafers, corrosion of, 42, 52, Screen for CKI's, 59, 105, 141 63 Screen, metal, 205-206; table, 202; Silicone rubber probe, triboelectric table, 295 charging of, 72 Screwdriver handle as rleld hazard, Skin 61; table, 187 causing charging of nonconduc• Scuff test, 40 tors, 34-35 Seat covers, conductive, 102 control of voltage on, 245-249, Shelf life of antistatic packaging ma• 250; mus., 246 terials, 175-180; table, 179 voltage suppression by, 56, 57 Shielding Slip test, 151, 161; mus., 220 334 INDEX

Small companies, advice to, 139-142, for elecronic box as package, 70 154 for fanless ionizer, 67 Smocks, ESD-control for footwear, 74-75 antistatic, 100 for grit-blasting, 68 as basic requirements in future, 162 for heat guns, 64-65 conductive threads in, 61 for ionizing nitrogen, 73 danger of stainless-steel fibers in, for materials and equipment, 97 62,194 for old AC ionizers, 67 evaluation of three designs, 61-62 for plain polyethlene bag as pack- in SSW, 50-51, 135, 207 age, 70-71 pros and cons of, 57 for screwdriver handle, 61 Soldering irons for selecting materials, 53 magnetic fields from, 54 for shunt bar, 62 voltage on tips, 54 for smocks, 61-62 Solvent extraction test, 309-310; for space-charging, 75 table, 313 for static charging, illus, 177; illus., Space-charging. See Ionizers, space• 221; illus., 222 charging by for tribolelectric cbarging by air, Special test 66 accelerated aging of bags, illus., for UV light, 69 179 for workbench laminates, 63 categorization of, table, 23 for wrist-strap bands, 63 corrosion and contamination test in disposition of mishandled hard• for SSp, 49, 238-239 ware, 146 discbarge test, illus., 171; illus., Specifications 204 model, handling and assembly, discussion of, 40, 118 77-106, 124, 126-127 for air guns, 63-64 model, packaging, 107-115, 125, for automated process, 72, 128-129 234-237 referenced in model handling and for brushes, 60-61 assembly specification, 97 for chain link fence, 69-70 referenced in model packaging for charges on specification, 109 arm hair, 56 Spraying of liquids, charge generation clothing, 57 by, 11 CKfs, 59, 105 SSP head hair, 57 categorization of, table, 23 identification badges, 57-58 corrosion, special test for, 49 walls, 58-59 definition of, 25 for c1eanroom gloves, 62-63 discussion of, 40, 47-49, 116, 117 for connector dust cover, 59-60 in model packaging specification, for cordless wrist strap, 72 108-109 for DC ionizers, 66-67 requirements, check list of, 149 for desiccator cabinet, 69 SSW for DI and grounding, 68-69 categorization of, table, 23 INDEX 335

certification of, 51, 80, 88, 93-94 specification, 78, 97-98 cleanliness requirement, 85, 90 in model packaging specification, decal for certification of, 93, 94 108 defmition of, 26, 80, 93 Static charge designation of, 90 apparatus for producing, illus., 117; discussion of, 40, 50-55, 117 ilIus., 221; illus., 222 for highly sensitive devices in fu- categorization of, 23 ture, 162 definition of, 26 housekeeping in, 85, 90 discussion of, 40-41 identification of, 93, 94 hazard at the work station, 184-188 in model handling and assembly Static-dissipative specification, 80-87 categorization of materials, table, inspection of, 88 23 minimal, check list for, 149 definition of, 26, 83, 110 operator skills vital in, 106, 245 desoldering handpiece, 96 work surface in, 90, 184-185, 198 discussion of, 41 SSZ table mat, 102 categorization of, table, 23 tote box, 101 definition of, 26 workbench tops, advantage of, discussion of, 40 184-185,268,285; illus., 282 in model handling and assembly Static Event Detector (SED), 12, 44, specification, 80 272, 281, 283; illus., 273; in model packaging specification, illus., 284; illus., 306; illus., 109 307 maintained in SSW by operator, Static rleld (see also rreld) 50-51,55,84-85 categorization of, table, 23 maintained inside SSP, 47 definition of, 26 Stainless-steel fibers, hazard of discussion of, 41 in smocks, 61-62, 194 Static-OmitinR noor finish in wrist-strap bands, 63, 194 as "safety net" or backup, 52-53, Standard test 155,225 categorization of, table, 23 as secondary defense, table, 158 definition of, 26 categorization of, table, 23 discussion of, 40 cost, 140; table, 223 for brushes, 60 cost-effectiveness of, 219, 221, 224 for connector dust cover, 59 criteria for, 102 for floor finish, ilIus., 200 definition of, 26 for packaging materials, table, 174 discussion of, 18, 21, 41, 73, for paper forms and labels, 58 218-225,248,273; illus., 17; for various packaging configura- table, 158 tions, 71 effectiveness after wear, 250 Standards in maintenance of ESD-protected industry, 118-121, 123 areas, 92 in future, 161 in model handling and assembly in model handling and assembly specification, 84 338 INDEX

Static-limiting (continued) in SSW, Tesla coil test, 228; illus., 171; illus., 52-53 204; table, 203; table, 205 test methods for, mus., 220; mus., Test equipment 221; mus., 222; illus., 223 check list of, 150-151 Statistical process control, 123 criteria for selection, 104 Steel, painted, 84, 263; mus., 270 Testing of ESDS items Steel work surfaces, 91, 93 precautions during, 86-87 Stresscracking of polycarbonate by procedures, inspection of, 88 antistats, 238, 239, 242 Test, multiple-choice, for certifying Summary of book, 158 personnel, 115-117 Surface resistivity Thermionic emission, 11 categorization of, table, 23 1bpical antistat definition of, 26 applied to plastic parts of equip• discussion of, 41-42 ment, 54, 96 effect of relative humidity and tem• applied to screwdriver handle, 54, perature on, 277, 286-293, 61 310, 317-319; illus., 311; categorization of, table, 23 illus., 312 chloride-free, 42, 104 measurement of, for packaging, 113 cost, 110 meter, for static-limiting Door fIn• criteria for selection, 104 ish, 83,150 definition of, 26 meters in future, 160 discussion of, 42; table, 197 method of measuring, 182; illus., for treating 178 antistatic tote boxes, 179 of brush handles, 60 identification badges, 58, 84 of smocks, 61 necessary nonconductors in requirements for packaging, no SSW, 84, 91, 198 versus time for aged antistatic walls or windows, 58-59 bags, illus., 179 possible contamination effects of, Symbols, 112 238-243 Tote boxes Taber abraser, 311; illus., 314 antistatic, need for antistat treat• Table mats, 101, 117, 263; illus., 270; ment of, 179 table, 282 cardboard, 142,226-229, 285; Tape table, 231 antistatic, 74, 255-256 conductive, charging by stretching, 12 cost of, 142; table, 231 conductive, for holding shields on disadvantages of, 180, 185-186, module, 68 226 conductive, masking, 193, 198 Corshield, 227-228; table, 227; Techniques vs. materials and equip• table, 231 ment, table, 197 cost of, 142; table, 231 Teflon criteria for, 101, 185-187 charged by heat gun, 65 current carried by, 227 in triboelectric series, 43, 279 discussion of, 4, 226-232 INDEX 337

drain time of, 231; table, 230 resistivity for static-limiting Faraday-cage construction, 49, Door finisb, 41 228-229 limiting of, in packaging, 110 hazard of static cbarges on, not caused by pure gases, 12,59, 179-180; table, 186; table, 228 66 lid used in special test, 66 of operator at work station, 248, materials, comparison of, table, 231 250; illus., 246 new materials for, 160,226-229 polyester fabric for, 56, 61, 67, 69, sbielding/discbarge test on, table, 100,101 227 reduced by increased relative static-dissipative, 101 humidity, 36, 83 techniques for using, table, 197 test, in model packaging triboelectric cbarging of, 228 specification, 114 vulcanized fiber, 232; illus., 227 tests for liner of SSp, 47 Toxicity of packaging, 314 tests for packaging, 278, 311-312; TQM (Total Quality Management), table, 279; table, 316 19,123; illus., 20 Triboelectric series Training of operators, 55 discussion of, 3 Transient personnel in SSWs, 86 example of, table, 42-43 Transparency of packaging materials, limitations of, 43 313 Troubleshooting Trays, criteria for, 101 illustrative examples of, 56-76 liiboelectric cbarging major tools for, check list, 151 Aclar for, 27, 43, 58, 65, 69, 245 1\veezers, 73-74 avoidance of by preventing packaged item from sliding, Ultraviolet (UV) light for inspecting 71, 110 ESDS items, 69 bag shaker test for, 43 Unglazed cardboard by air, 12, 59, 66 for packaging, 111 by body movement, 85, 248 for tote boxes, 83 by conductive packaging materials, Units, defmition of, 89 178, 226, 279; table, 228; Unrealistic testing, example with table, 279; table, 316 example with plastic, 31 by grit-blasting, 68, 193-194 example with wire, 53-54 by spraying with conformal coating, 188 Vehicular bounce test for packaging, by stainless steel, 66, 69 illus., 172 by tote box materials, table, 228 Vinyl, antistatic, 75-76, 315-316 categorization of, table, 23 Voltage suppression definition of, 26 by conductive masking tape, 198 discussion of, 42-43, 278; ilIus., by skin, 56-57 17 categorization of, table, 23 in blow-off operations, 38 definition of, 26 in operator training, 105-106 discussion of, 43 lack of correlation with surface hiding field on tote box, 52 338 INDEX

Voltage suppression (continued) criteria for, 101 in brushes, 60-61 testing of, 184-185, 263, in cleanroom gloves, 62-63 269-271, 281-285 in packaging, 110, 299 surface, of charged nonconductive plastics corrosivityof, 52 on workbench, 187 drain time of, 52, 184-185, 231, of necessary nonconductors 269, 277, 285; table, 230 (painted or plastic-coated metal (stainless steel or aluminum), grounded metals) in SSW, 84 91,93 Vulcanized fiber for tote boxes, 233; with buried conductive layer, 93 table, 227 Wrist straps V-Za~35, 132, 137, 153 bands, metal fibers in, 63, 194 bead-chain, 194, 198; table, 212 Wafers, silicon carbon-loaded plastic as question- corrosion by chloride ion, 42, 52, able material, 194 63 cloth band, looseness and soiling ESD testing of, 258-262; mus., of,212 259; illus., 264; mus., 303; cordless, 72, 272 table, 275 discussion of, 251 Walk test expansion band, table, 212 apparatus for, mus., 222 factor in determining resistance to apparent vs. actual voltage, 219; ground of operator, 212 illus., 223 increased AKfG by loose bands, applied to connector dust cover, testing and monitoring of, 216, 250; 59-60 table, 212 categorization of, table, 23 various designs, 194 definition of, 26 discussion of, 43, 73, 74 X-rays, charging of particles by, 11 Walls charged, field hazard, 58-59 Zapflasb in SSW, 51 as basic equipment for operators, penetrated by fields, 91 245 precautions, 91 brightness of light vs. resistance, Water analogy, 4-8 211 Windows, charged, 59, 91, 197 categorization of, table, 23 Woolen fabric, 34, 50 cost, 140, 141 Workbenches criteria for selection, 104 conductive surfaces, 91,93 definition of, 26 damage to MOSFE'lS by touching discussion of, 43,54,55,93 surface of, table, 185 for checking grounding of conduc• design and grounding of, 51-52, tors, 247 93-94,184-185 for locating AC power leakage, 52 laminates, various uses of, 43 chloride ion in, 63