Grounding personnel in ESD-sensitive environments
Conventional practices may not be sufficient for protection and reliable production
By David E. Swenson and Dr. Robert W. Wilson, 3M Electronic Handling and Protection Division
Grounding has been a principle method of static control since the earliest days of manufacturing gunpowder. However, the creation of semiconductor-based electronic devices dramatically increased the need for reliable grounding to manufacture products. Today, the read-write head elements manufactured in the disk drive industry are susceptible to low levels of electrostatic discharge (ESD). Some manufacturers are concerned that conventional practices in grounding may not be sufficient for protection and reliable production.
Even though grounding of personnel to prevent ESD damage is a reasonably well understood concept in the electronics industry, there seem to be gaps in understanding when periodic testers or monitoring systems are used to verify wrist strap performance. Specifically, there seems to be a desire to control and verify the grounding of personnel at progressively lower resistance levels while at the same time minimizing the number of alarms from the monitors. These goals are mutually exclusive and may necessitate compromises in grounding and monitoring im ple mentation schemes.
Wrist straps are the most common static control device for people handling static susceptible parts. They consist of a cuff or band that has electrically conductive elements on its inner surface for the purpose of making electrical contact with the skin of the wearer. The band is connected to ground through a cord that usually contains a 1-megohm resistor. A person be comes charged through contact and separation with environmental surfaces. If the charge is not removed at the same rate as it is being generated, a potential difference between the person and ground will be developed. The role of the wrist strap is to allow the excess charge to drain from the body to ground rapidly enough to prevent harmful potentials from developing. The rate of this discharge is determined by the resistance of the body to ground. Figure 1 attempts to quantify the maximum potential on a vigorously moving person at different resistance levels1. This work was recently repeated in a study conducted by the International Electrotechnical Commission (IEC) in Technical Committee (TC) 101 (See figure 2)2.
While there are some differences in the reported results, these two tests demonstrate that moderate voltages can be generated when the person`s resistance to ground is more than a few megohms. The significance of these voltages with regard to ESD damage depends on the sensitivity of the items being handled.
In recent years, advances in the disk drive industry have resulted in a major concern regarding electrostatic potential and discharge during the manufacture of the head gimbal assembly. The development of the magnetoresistive head (MR) and the giant MR head (GMR) have increased the performance of disk drives and have also caused the static control industry to evaluate previously accepted electrostatic control practices.
It is well accepted that these structures are among the most ESD-sensitive items ever mass produced. This sensitivity to ESD damage leads to challenges in a production environment. Static-sensitive elements must be soldered or otherwise attached by hand, generally while being watched through a microscope. The assembly of the MR head element to the head gimbal assembly is a critical operation in which the MR head element is sensitive to discharge energies of approximately 10 nanoJoules (1 ¥ 10-8 Joule) or less3. GMR head elements are expected to be 10 times more sensitive (1 nanoJoule or 1 ¥ 10-9 Joule or less). After assembly of the head gimbal assemblies into a head stack assembly, the sensitivity may decrease to manageable levels4. Recent work has shown that changes in parametric performance of MR recording heads can occur with very low machine model types of discharge, down to a level of 4 or 5 nanoJoules5. To illustrate how low these energies are, it takes 1,000 to 10,000 times this amount of energy to cause an ignition of most flammable air-fuel mixtures.
To establish an appropriate threat level for a manufacturing process, it is common to assume a “worst case” scenario. In considering static electricity, this usually involves establishing a threshold value for a direct discharge of the electrostatic charge held on a modeled circuit that will damage the most sensitive element of the most sensitive component being processed. This threshold level usually is used to determine a maximum allowable resistance specification using charts like those of Figure 1 and Figure 2. Assuming that magnetic recording heads can be damaged by direct electrostatic discharges with energies on the order of 10 nanoJoules, one approach would be to calculate the voltage necessary to produce such a discharge from a person.
Using the formula E =1/2CV2, and assuming the commonly accepted capacitance of a person of 100 pF, this corresponds to a voltage of 14 volts. From Figures 1 and 2, it can be seen that it is necessary to maintain the human body at a resistance to ground on the order of a few megohms to prevent discharges with resulting energies above 10 nanoJoules.
We repeated these experiments but confined our study to less vigorous motion. While attached to ground through a known resistance and seated, the test subject simply slid one foot over the floor while keeping the other foot still and in contact with the floor. Flooring was ordinary non-conductive tile and nonconductive footwear was worn. Voltage measurements were made with an HP Model 54111D oscilloscope. The oscilloscope probes used have an impedance of 20 megohms, which limits the maximum resistance-to-ground under the experimental conditions. This impedance was taken into account in the resistance to ground reported in the tables. As might be expected, we saw considerable variation between individual measurements under identical conditions because of our inability to exactly reproduce the sliding motion. The results shown in Figure 3 are the averages of several individual tests.
Column 1 in Figure 3 is the additional fixed resistance added to a wrist strap ground cord. The resistance was added in this fashion to simulate the normal mode of failure in wrist strap systems, that being a change in electrical contact between the wrist band and the wearer`s skin. Column 2 is the resulting actual resistance to ground of the person under test, taking into account the parallel resistance paths established by the wrist strap ground cord and the oscilloscope probe. The discharge energy shown in Column 4 is calculated based on a capacitance of 100pF.
This again supports the notion that, in order to assure that a discharge from a person of more than 10 nJ does not occur, it is necessary to maintain resistances under a few megohms to ground.
Use of constant monitors
Constant monitors that evaluate the performance of wrist straps while in use are a relatively new addition to the products available to the static control practitioner. While there are numerous types and styles of wrist strap monitors sold today, all of them apply some sort of measurement signal that will appear on the body of the wrist strap wearer to evaluate the condition of the grounding system. The nature of this signal and the possibility that it might lead to damage of sensitive parts is of concern to the industry. The monitor style that applies an electrical current to measure the resistance of the entire interconnected electrical system, including the ground cord, wrist strap and the contact between the wrist strap and the person, is our focus here. These systems require the use of a dual-conductor ground cord and a special wrist strap that has two electrically isolated halves. A simplified circuit diagram of this type of wrist strap monitor is shown in Figure 4.
The monitor will alert when the measurement circuitry detects a voltage level that is associated, through Ohm`s law, with a maximum allowable resistance. This resistance is the sum of the two 1-megohm resistors in the cord and the resistance that results as current is passed between the two halves of the wrist band through the skin of the wearer. The person is included as part of the measurement circuit and will thus be raised to some potential simply by the action of the monitor. This voltage level depends on several things, including the current being supplied by the measurement circuitry and any resistances between the person`s wrist and the two halves of the wrist band.
In Figures 6 and 7, we have repeated the information of Figure 3 except that, in this case, the person is attached to one of two different constant monitors. The two monitors differ in the maximum voltage that can be sourced by the monitor to make the resistance measurement. In one case, the monitor can apply up to 9 volts in an open circuit (infinite resistance to ground) situation. The other monitor can source up to 16 volts in the same situation. Referring to the experimental setup shown in Figure 5, the variable resistances are used to emulate skin resistance and are kept equal to simulate the most likely situation of having both halves of the band make equally good contact with the skin of the wearer. The actual skin resistance was kept very low by the use of a moisturizer between the band and the skin of the wearer. The 20 megohm fixed resistance that is in parallel with the variable resistor on the measurement circuitry side of the circuit is there to mirror the effect of the 20 megohm oscilloscope resistance that is present on the ground side of the circuit. Rmc is the equivalent resistance to ground through the measurement circuitry. It is difficult to assign a value to Rmc because it includes the effects of active electronics. For purposes of calculating the resistance to ground in Figures 6 and 7, we assume its value to be very large.
As explained earlier, the 20 megohm resistance on each half of the wrist band limits the ability to investigate the effects at high resistance-to-ground levels but, as shown in Figures 1, 2 and 3, the voltage on people is already of concern to some for ultra-sensitive work environments at a resistance to ground of 4 to 5 megohm.
Comparison of Figures 6 and 7 with Figure 3 shows that the voltages and energies measured with the monitors are less than those measured with only a wrist strap at equal resistances to ground. We believe that this is due to the presence of the additional ground path through the active electronics in the monitor circuitry. This ground path was ignored by setting Rmc very high. Study of Figures 6 and 7 shows that the normal motions of a person when attached to these monitors can generate voltages high enough to produce discharge energies in excess of 10 nJ at resistances to ground as low as approximately 6 megohms. Further study shows that the constant monitors used in this study did not add sufficient voltage to impact this value and actually alerted before this value was reached.
There is an understandable desire to use a constant monitor that applies the lowest possible voltage as to the user to measure the effectiveness of the grounding system. In actual use, the source of additional resistance to ground is not resistors added by the experimenter as above, but rather the resistance at the interface between the wrist band and skin that can arise from a variety of causes including loosely worn straps, dry skin, etc. There are numerous potential sources that cause this resistance to vary from individual to individual, to vary with time, and even with measurement method. Figure 8 shows a compilation of resistances that were taken from a number of wrist strap wearers at different times of day and over different days. In addition, each resistance was determined by measuring the current passed at two measurement voltages (10 volts and 20 volts). For this test, new wrist straps were used and the wearers were instructed to adjust the straps to obtain a tight fit; no lotions or creams were used under the wrist bands.
Notice that, even under the controlled conditions of this test, there is a great deal of dispersion in the resistance values measured and a large number of them exceed 4 megohms, the arbitrary limit we had suggested for the disk drive industry. If we had been using constant monitors, each over-limit value would have resulted in an alerting monitor. Notice also the differences between the values measured at the two different voltages even though they were obtained at the same time. Figure 8 shows, and it has been our experience in a variety of tests, that the resistance measured between a wrist band and the skin of a wearer increases as the voltage used to make the measurement decreases. Therefore, as the voltage used to monitor resistance decreases, the number of alarms will increase.
When we established our threat level by calculating the voltage needed to produce a 10 nJ discharge, we made the “worst case” assumption that all of the energy available from a charged person could be delivered to the sensitive element to cause damage. To understand the risk associated with having an electrical potential on a body, a study was conducted in which the discharge energy was measured from a monitored person holding a pair of tweezers. The experimental arrangement is shown in Figure 9. At the point at which the alarm was sounding due to added resistance on both halves of the wrist strap ground cord, the person touched the tweezers to the lead of a grounded 33 ohm resistor6. This resistance value was chosen because it is in the range used by investigators of recording head susceptibility. A current probe encircled the resistor lead and was attached to the input of a 1-gigahertz oscilloscope with 500-megasamples-per-second sampling rate. The current pulse was captured by the oscilloscope and analyzed. The method used in this study is essentially the same as that reported by Wallash et al and C. Lam in their work7 in the disk drive industry. While several touching methods were tried, including bare hand, gloved hand, moist fingers, dry fingers, dissipative tweezers, and metal tweezers, only the metal tweezers allowed enough current to pass to cause the oscilloscope to trigger routinely and capture the signal. The discharge energy from other contacting forms was below the detection limit of the equipment used. This is in keeping with the results reported by several researchers that investigated the amount of energy transferred from a charged person by a finger touch. In the cases reported to date, obtaining a discharge from a person`s dry finger at less than 500 volts has been reported as difficult. A moistened finger may allow a discharge at 100 volts7.
Comparing the information in Figure 10 with that in Figures 6 and 7, we find that the actual energy delivered to the resistor is only about 2 percent of the total amount calculated to be available. The fraction of the available discharge energy that is dissipated in any target load will depend on the electrical nature of the load and the specifics of the discharge contact. Full exploration of this dependence is beyond the scope of this study, but the energy provided by a discharge from the load should always be less than the total available energy based on calculation. As shown in Figure 10, the discharge energy is much less than anticipated by calculation of the delivered human body model circuit.
All of this data supports the idea that, to prevent possibly damaging voltage levels from being present on personnel, it is necessary to control their resistance to ground. The lower this resistance, the lower the voltage level present on a body. Investigators have shown that some products such as MR and GMR recording heads are sensitive to human body model discharges under 10 nJ and even less if a machine model is considered. The calculated voltage necessary for a human body model discharge of 10 nJ is 14 volts. It has been shown that, at a resistance to ground of about 4 megohms, a typical person can easily develop an electrical potential with respect to ground of 10 to 20 volts, regardless of whether or not the person is being monitored.
Using such models, it is tempting to set resistance-to-ground limits lower than necessary and employ monitors with very low measurement voltages to assure that no product damage can occur under any situation. However, this will cause difficulties in factory implementation because the inherent variability in skin resistance coupled with low voltage measurement may cause the number of monitor alarms to be intolerable. We have already heard of instances where the resistance trigger levels on monitors have been increased to reduce the number of alarms. Soaking wrist bands in water and wearing them wet has also been reported as a method used to reduce the number of alarms, but this cannot be considered a healthy situation for the person being monitored.
Rather than make unrealistic requirements for the grounding system, we support the implementation of a complete control program in which all of the elements are designed to prevent damaging discharges. Most of the investigators of MR and GMR recording head sensitivity have stated that the best protection is the use of the correct type of dissipative tweezers and other tools in the assembly operations. These tools further reduce the energy delivered to the sensitive elements in the event of a discharge. In our experiments, the currents produced in discharges from tools of this type were below our ability to detect them. CR
David E. Swenson is applications development and technical service manager for 3M Electronic Handling and Protection Division in Austin, TX. He is responsible for introducing new product in the static control area, training 3M personnel world-wide, and providing application assistance to users of static control products globally with particular emphasis on Asia/Pacific and Japan.
Dr. Robert W. Wilson is a senior research specialist with 3M. He has a Ph.D. in physical chemistry and has worked for 3M in research and development for more than 18 years. For approximately the past 14 years, he has been involved with the development of products for the static control and electronics industry, including 3M`s wrist strap monitor systems, ionizers and detection systems. He has authored or co-authored numerous papers presented in various symposia around the world. Dr. Wilson has been issued three patents related to products used in the control of static electricity in the work place.
1. D.M. Yenni, “Basic Electrical Considerations in the Design of a Static-Safe Work Environment,” Proceedings, 1979 NEPCON/West, Anaheim, CA.
2. IEC TC 101 61340-5/1 and 2 scheduled for release in 1998.
3. Lam, C., “Characterization of ESD Tweezers for Use with Magnetoresistive Recording Heads,” EOS/ESD Symposium, 1996, Proceedings, pp 14-21.
4. Cheung, T. and Rice, A., “An Investigation of ESD Protection for Magnetoresistive Heads,” 1996 EOS/ESD Symposium, Proceedings, pp 1-7.
5. Lam, C., Salhi, El-Amine, and Chim, S., “Characterization of ESD Damaged Magnetoresistive Recording Heads,” 1997 EOS/ESD Symposium, Proceedings pp. 386-397
6. Cheung, T., “Direct Charging: Charge Device Model Testing of Magnetoresistive Recording Heads,” EOS/ESD Symposium , 1997, Proceedings, pp 398-404.
7. Wallash, A., Hughbanks, T., and Voldman, S., “ESD Failure Mechanisms of Inductive and Magnetoresistive Recording Heads,” EOS/ESD Symposium, 1995, Proceedings, pp 322-330.
Figure 2. Moderate voltages can be generated when the person`s resistance to ground is more than a few megohms.
Figure 4. A dual-conductor ground cord and a wrist strap with two electrically isolated halves is required in interconnected electrical systems.
Figure 5. Variable resistances emulate skin resistance and simulate both halves of the band making equally good contact with the wearer`s skin.
Figure 8. Resistance measured between a wrist band and skin increases as the voltage used to make the measurement decreases.
Figure 9. Experiment to measure discharge energy from a monitored person holding tweezers.
This paper was presented at DataStor Asia `98. For a copy of the DataStor Asia conference proceedings (available for $95 via ground shipping or $130 via international air delivery), please contact Maureen Kane, PennWell Publishing Co., 10 Tara Blvd., 5th Floor, Nashua, NH 03062. Tel: 603/891-9423, Fax: 603/891-9490.