A Study of “Soft Grounding” of Tools for ESD/EOS/EMI Control

Al Wallash

Hitachi Global Storage Technologies, 5600 Cottle Road, San Jose, CA USA tel.: 408-717-8342, e-mail: albert.wallash@hitachigst.com

Abstract – The effect of a discrete resistor placed in the ground wire connected to a metal tool block or hand tool on ESD/EOS/EMI control is studied. Experimental results and SPICE circuit simulations show that in most cases “soft grounding” results in no reduction in CDM current, noise voltage or radiated fields (EMI) because of capacitive coupling. However, using a high-resistance, static dissipative material at the interface between the device input and metal tool was extremely effective in reducing the voltage and current associated with ESD/EOS/EMI and increased the CDM failure level of a Class 0 ESD sensitive device from 5V to over 200V.

I. Introduction Figure 1 shows a schematic representation of an input to a charged, floating device making metal-to-metal contact to a fixture or tool connected to a tester chassis. The device and fixture each have a capacitance to the chassis and the fixture has a “ground” wire to a tester chassis. Vn represents common-mode electrical voltage transients on the chassis, or “ground noise”. Using a ground wire with Rtool < 1Ω DC is generally accepted to “hard ground” tooling to insure that the

voltage on the tooling is the same as the chassis. But a severe CDM event will occur when the input of a charged device is touched to the metal surface of a hard grounded metal fixture or tool. In an effort to reduce the CDM ESD current, it is sometimes proposed to “soft ground” metal fixtures and hand-held tools using a discrete resistor in the tool’s ground wire (Rtool). The expectation is that an increase in the DC resistance between the tool and ground will reduce the severity of the CDM current transient to the tool. A second situation where soft grounding of tools has been proposed is to reduce common-mode noise transients on tooling [1]. Common-mode electrical voltage transients can be present on ground, testers and tools [2-4]. These repetitive, low voltage noise transients can result in ESD-like electrical overstress (EOS) damage to Class 0 ESD sensitive devices [5].

DeviceMetallic

Fixture/Tool

+ + +

RtoolCdeviceCtool

Lwire

Groundwire

Vn

Tester ChassisRchassis

Lchassis

Cchassis

Fig. 1. Lumped element model for charged device and fixture/tool connected to a tester chassis with common-mode noise voltage Vn.

High-frequency current flow depends on the total impedance in the current path. The impedance is determined by the resistance, inductance and capacitance in the current path to ground, not just the DC resistance. For example, at high frequencies capacitive coupling between the tool and chassis (Ctool) could offer a lower impedance path than Rtool. Therefore, it is not immediately clear how effective soft grounding, using only a discrete resistor in the tool’s ground wire, will be in controlling ESD/EOS/EMI in real-world tooling and handling operations. The purpose of this paper is to understand the overall effectiveness of soft grounding metal tool blocks and hand-held tweezers for controlling the CDM ESD

2B.8-1

164

EOS/ESD SYMPOSIUM 07-152

current and common-mode electrical noise pickup for a Class 0 ESD sensitive device. Lumped element SPICE models are used to simulate and provide a better understanding of the current-flow in each experimental setup. Metal Block

Rtool

VoltageSupply

Device1 MΩ

Insulator

Ground Plane

TEKCT6

Fig.2. Schematic representation of a charged device input touching a metal block with a resistance in the ground wire (R ). tool

Rtool

VoltageSupply

Device1 MΩ

Ground Plane

Stainless steel tweezers held in Nitrile gloved hand

TEKCT-6

Fig. 3. Schematic representation of a charged device input touching metal tweezers with resistance in the ground wire (R ). tool

Metal Tool BlockR tool

Device

Insulator

Linear stage with stepper motor

Rtweezers

Ground Plane

TEKCT-6

Stainless steel tweezers held in Nitrile gloved hand

Static dissipative mat

Fig. 4. Schematic representation of the device with one input touching a tool block, and the other input in contact with grounded tweezers.

II. Experimental The Class 0 devices used in this study were tunneling magnetoresistive (TMR) sensors on a magnetic recording head gimbal assembly (HGA). The TMR sensor is a two-terminal device that can be modeled as a resistor in parallel with a capacitor. The capacitor is formed by two magnetic metal layers separated by an ultra-thin <1nm-thick tunneling barrier with a breakdown voltage ~ 1V. The voltage and current from a CDM ESD event can result in electrical breakdown of the tunneling barrier, resulting in a decrease in resistance and degradation of the magnetic properties of the TMR read sensor. The TMR devices used in this study were damaged by a single 500ps-wide, 10mA (5V) peak current (voltage) transient. The failure voltage and current decreases for longer duration transients. For example, the failure current (voltage) was ~2mA (1V) for a 3ns-wide current (voltage) charged device model (CDM) transient. Figures 2 and 3 show the setups for studying the effect of a series resistance to ground for a CDM event to a fixture or metal tweezers. In both cases, a TEKTRONIX CT6 (2 GHz) current probe was used to measure the CDM current transient. A carbon composition resistor (Rtool) was used between the fixture or tool and ground. The device was charged directly with a voltage source and then touched to the grounded metal tool block or tweezers. The wire length (inductance) between device and tool block or tweezers was 0.5cm (4 nH) and 20 cm (160 nH), respectively. Figure 4 shows the setup for studying the effect of a series resistance to a tool block and hand tool for a fixture with common-mode noise from a stepper motor. The metal block was placed on a linear translation stage with a Compumotor stepper motor and SX6 controller. A LeCroy 5005 (5 GHz/20 GS/s) oscilloscope was used with the CT6 to measure the CDM current transients.

III. Results a. CDM Testing

Figure 5 shows an example of the CDM current transient from the device at 1 V to the hard grounded tool block and to metal tweezers held in a Nitrile gloved hand. The peak current is significantly reduced for the CDM event to the tweezers because of the increased inductance in the much longer ground wire (20 cm) to the tweezers, which increased the impedance in the discharge path. [6] The effect of increasing the series DC resistance between the tool block and tweezers on the CDM peak current is shown in Fig. 6. The device voltage was fixed at 1V, which was below the damage threshold of the TMR head. Note that there was no

2B.8-2

165

EOS/ESD SYMPOSIUM 07-153

decrease in peak current as the series resistance was increased to as high as 20 MΩ. Since the peak current did not change with increased series resistance, it would be expected that the failure level was also unchanged. Figure 7 shows the device resistance change vs. device voltage for CDM events between the device and tool, with Rtool = 1Ω and 1MΩ. The device was damaged (resistance decrease) at the same voltage, independent of whether the tool was hard grounded (1Ω) or soft grounded (1MΩ)! The data in Figs. 6 and 7 show conclusively that soft grounding tools does not decrease the CDM current or increase the CDM failure level. SPICE modeling can be very useful in understanding ESD current waveforms. Figure 8 shows a lumped-element SPICE model for a charged device touching a

grounded tool block or tweezers. Note that the tool has a capacitance (Ctool) in parallel with a resistance (Rtool) to ground.

-4

-2

0

2

4

6

8

10

0.0 1.0 2.0 3.0

Time (ns)

Cur

rent

(mA

)Tool Block

Metal Tweezers

Fig.5. CDM current vs. time for a device at 1V touching a tool block and hand-held metal tweezers.

Fig. 8. SPICE model for CDM event between charged device and grounded tool block or tweezers.

0

50

100

150

1 10 100 1000 10000

R tool to ground (ohms)

Peak

Cur

rent

(mA)

Ctool = 20 pFCtool = 10 pFCtool = 1 pFCtool = 0.1 pF

Fig. 9. Peak CDM current vs. resistance of tool to ground for different values of tool capacitance. The device capacitance was fixed at 10 pF.

Figure 9 shows the SPICE simulation result for the peak current vs. tool resistance-to-ground for different values of tool capacitance. The device capacitance was fixed at 10 pF. Note that the peak current was reduced significantly only when the tool capacitance was less than 1/100th of the device capacitance.

0

2

4

6

8

10

12

1.E+00 1.E+02 1.E+04 1.E+06 1.E+08

Rtool (ohms)

Pea

k Cu

rren

t (m

A)

Tool block

Tweezers

Fig. 6. Peak current for device at 1V touched to the tool block and tweezers vs. series resistance to ground.

-500

-400

-300

-200

-100

0

100

0 1 2 3 4 5 6

Device Voltage (V)

Dev

ice

Res

ista

nce

Cha

nge

(ohm

s)

The CDM current transient is determined by the impedance to ground in the discharge path. If the series resistance was the only contribution to the impedance to ground, then the CDM current would be reduced (assuming no other parasitic inductance or capacitance). However, the tool’s capacitance offers an alternate path to ground at high frequency, so the resistance to ground is in effect “shunted”. The impedance due to tool’s capacitance is given by

fCZc π2

1= , (1) 7

Rtool = 1

Rtool = 1 Meg

Fig. 7. TMR sensor resistance change vs. device voltage Rtool= 1Ω and 1 ΜΩ.

where f is the frequency and C is the tool capacitance to ground. For example, the impedance at 1 GHz for a 100 pF tool block is < 2Ω.

2B.8-3

166

EOS/ESD SYMPOSIUM 07-154

EMI event detection was performed using a Credence EM Aware. The EM Aware indicated the same level of EMI for CDM events to the hard and soft grounded tool block. For the case of a very small electronic device that touches a much larger fixture and tool, it would generally not be practical to have a tool capacitance that is smaller than the device. Thus, it would be very difficult to increase the CDM failure level using soft grounding. In summary, soft grounding would reduce the CDM current and increase the device failure level only in the limit of zero tool capacitance.

b. Common-mode electrical noise The common-mode noise on the linear stage in Fig. 4 was measured with an HP 1141A active voltage probe [6]. Figure 10 shows an example of the ~1μs-long voltage transient on the stage with a peak-peak voltage of about 11V. Figure 11 shows the peak-peak voltage on the tool block vs. series resistance to ground (Rtool) for the block on and off the stage. Note that the voltage on the tool block is unchanged by soft grounding when the tool block is on the stage. This is again due to capacitive coupling, this time between the noise source (stage) and the tool block. The voltage on the tool is reduced when Rtool is increased only when the

tool’s capacitive coupling was reduced by removing it from the stage.

-10

0

10

20

30

0 1 2 3Time (ns)

Cur

rent

(mA

)

Fig. 12. Current transient when one input of the floating TMR device was touched to the block on the stage with a common-mode noise voltage.

Figure 12 shows the measured current transient when one input of the floating TMR device was touched to the block on the stage with a common-mode noise voltage. In this test, the tool block was soft grounded with Rtool=22MΩ. The device was damaged after this test, as indicated by a large resistance decrease. This shows that soft grounding a tool is ineffective in preventing ESD damage when the tool is capacitively coupled to a chassis with electrical noise transients. Fig. 13 shows a SPICE model for the setup shown in Fig. 4. It consists of a tool block on an electrically noisy translation stage that is touched by tweezers connected to ground. The tool block is capacitively and resistively coupled to the noisy stage. The voltage transients on the stage are a result of the capacitive coupling between the voltage signals to the motor windings and the stage.

-6

-4

-2

0

2

4

6

0.0 1.0 2.0 3.0Time (microseconds)

Volta

ge (V

)

Fig. 10. Measured voltage transient on the linear stage with stepper motor running.

0

2

4

6

8

10

12

1.E+00 1.E+02 1.E+04 1.E+06 1.E+08

Rtool Resistance (ohms)

Pk-P

k Vo

ltage

(V)

Tool block on stage

Tool block off stage

Fig 11. Pk-pk noise voltage on a tool block that was placed on and off of the linear stage with the stepper motor noise.

The experimental and SPICE simulation results for the pk-pk current vs. the resistance to ground of the tweezers are shown in Fig. 14. In this case, there is a reduction in the steady-state current as the series resistance of the tweezers is increased. The soft grounding was successful in reducing the steady-state current flow from the noisy tool block through the

Fig. 13. SPICE model of the linear stage with the tool block on stage and the tweezers to ground.

2B.8-4

167

EOS/ESD SYMPOSIUM 07-155

Metal Block

Device

Insulator

Linear stage with stepper motor noise

Ground Plane

TEKCT-6

Static dissipative mat

Static dissipative ceramic

VoltageSupply

Hard ground

Fig. 15. CDM event to noisy tool block with static dissipative material at the device-tool interface.

Rtool = 1Ω

0

200

400

600

0 50 100 150 200

Device Voltage (V)D

evic

e R

esis

tanc

e (o

hms)

Device input to metal block

Device input to static dissipativesurface on metal block

Fig. 16. Device resistance vs. device voltage after charged device input is touched to metal tool surface or static dissipative surface.

tweezers to ground because the resistance was in series with the tool capacitance.

0

50

100

150

1.E+00 1.E+02 1.E+04 1.E+06 1.E+08

Tweezer Resistance to Ground (ohms)

Pk-P

k C

urre

nt (m

A) Measured

SPICE simulation

Fig. 14. Measured and SPICE simulated current vs. resistance to ground of the tweezers.

c. Correct approach: Insert distributed resistance between device and tool

It has been shown that inserting a discrete resistance between a tool and ground, “downstream” of the CDM event, is ineffective in reducing the CDM current transient because the parallel capacitance of the tool offers an alternate, lower impedance path to ground for high-frequency current. In contrast, it was shown that a resistance in the ground wire to a hand tool was effective in reducing steady-state noise currents because the resistance was in series with the tool’s capacitance. This offers a clue as to how to make the original intent of soft grounding work. The high resistance approach can reduce the CDM current when it is in the form of a uniform, distributed resistance placed at the correct location, which is at the interface between the device input and metal tool surface. This places the resistance in series with the tool capacitance and reduces both CDM and noise current transients. Also, the distributed resistance of the static dissipative block does not suffer from the parasitic capacitance that would be present if a discrete resistor were used. Figure 15 shows a setup that is a hybrid of Figs. 2 and 4 that tests for both CDM and noise current transients at the same time. The major change over previous setups is the insertion of a static dissipative ceramic (CerastatTM) plate at the interface between the charged device and noisy tool block. The setup includes a CT6 to measure current transients as a function of device voltage. The use of a static dissipative material with a surface resistance of 100MΩ was so effective in reducing current that no current transients or EMI events were measured down to 0.1mA for a device voltage as high as 210V.

Figure 16 shows the resistance of TMR devices as a function of device voltage when the input was touched to the hard grounded metal tool or to the static dissipative ceramic. The use of the high resistance, static dissipative material at the interface increased the device failure voltage from 5V to greater than 200V.

IV. Summary and Conclusions In order to understand high-frequency ESD and noise currents, it is important to include all sources of impedance in the discharge path and all coupling between the device, tools and ground. Table 1 summarizes the experimental findings and compares the effectiveness of a resistance in the tool’s ground wire vs. a static dissipative material at the device input-tool interface in reducing CDM and noise currents. Inserting a resistor in the tool’s ground wire does not protect against CDM, noise voltage or EMI from sparks. Soft grounding tools through a resistance in the tool’s wire to ground was ineffective in reducing transient currents or EMI events for CDM ESD events. For

2B.8-5

168

EOS/ESD SYMPOSIUM 07-156

most practical cases, the impedance of the tool’s capacitance to ground provided a lower impedance path to ground than the resistor. The tool capacitance then acts as a “shunt” for high frequency current. Therefore, soft grounding does not protect against transient CDM ESD events for small electronic devices touching larger tools.

Soft Grounding Static Dissipative Interface

CDM: charged device to tool block

Ineffective Extremely Effective

Floating device to tool with noise

Ineffective Extremely Effective

EMI from ESD spark Ineffective Extremely Effective

Steady-state current from noise

Effective if no capacitive coupling

Extremely Effective

Table 1. Comparison of the effectiveness of soft grounding, using a discrete resistor in the ground wire, vs. a static dissipative material at the device/tool interface in reducing CDM and noise currents and sparks.

References [1] R. Tag-at, R. Galinggana, Jr. and L. Li, “ESD

Grounding Setup for Ground Noise and Leakage Current Protection”, Proc. Association of Semiconductor and Electronic Manufacturing Engineers of the Philippines (ASEMEP) 16th National Technical Symposium, June 2006. The only case where soft grounding was effective was

in reducing steady-state noise transients, as long as the capacitive coupling to the noise source was negligible. [2] V. Kraz, P. Tachamaneekorn and D.

Napombejara, “EOS Exposure of Magnetic Heads and Assemblies in Automated Manufacturing”, Proc. 2004 EOS/ESD Symp., pp. 344-9.

Placing a uniform, distributed static dissipative material at the interface where the device input touches the tool was shown to be extremely effective in reducing CDM current and increased the device failure voltage from 5V to greater than 200V. In this case, the large resistance is placed in series with the tool capacitance so that it increases the impedance to ground.

[3] KP Yan and R. Gaertner, “An Alternative Method to Verify the Quality of Equipment Grounding”, Proc. 2005 EOS/ESD Symp., pp. 220-8.

[4] W. Farwell, K. Hein and D. Ching, “EOS from Soldering Irons Connected to Faulty 120V AC Recepticles”, Proc. 2005 EOS/ESD Symp., pp. 23844.

In summary, placing a large resistance in parallel with the tool capacitance is ineffective in preventing CDM ESD current. However, when the large resistance was placed in series with the tool capacitance, in the form of a static dissipative material at the device/tool interface, it was extremely effective in controlling ESD/EOS/EMI damage to any static sensitive device under all test conditions.

[5] A. Wallash, H. Zhu, R. Torres, V. Kraz and T. Hughbanks, “A New Electrical Overstress Test for Magnetic Recording Heads”, Proc. 2006 EOS/ESD Symp., pp. 131-5.

[6] A. Wallash and V. Kraz, “ A Comparison of High-Frequency Voltage, Current and Field Probes and Implications for ESD/EOS/EMI Auditing”, Paper 2B-8, to be presented at 2007 EOS/ESD Symposium.

2B.8-6

169

EOS/ESD SYMPOSIUM 07-157