JEP155

JEDEC
PUBLICATION
Recommended ESD Target Levels for

HBM/MM Qualification
JEP155
AUGUST 2008
JEDEC SOLID STATE TECHNOLOGY ASSOCIATION

NOTICE
JEDEC standards and publications contain material that has been prepared, reviewed, and
approved through the JEDEC Board of Directors level and subsequently reviewed and approved
by the JEDEC legal counsel.
JEDEC standards and publications are designed to serve the public interest through eliminating
misunderstandings between manufacturers and purchasers, facilitating interchangeability and
improvement of products, and assisting the purchaser in selecting and obtaining with minimum
delay the proper product for use by those other than JEDEC members, whether the standard is to
be used either domestically or internationally.
JEDEC standards and publications are adopted without regard to whether or not their adoption
may involve patents or articles, materials, or processes. By such action JEDEC does not assume
any liability to any patent owner, nor does it assume any obligation whatever to parties adopting
the JEDEC standards or publications.
The information included in JEDEC standards and publications represents a sound approach to
product specification and application, principally from the solid state device manufacturer
viewpoint. Within the JEDEC organization there are procedures whereby a JEDEC standard or
publication may be further processed and ultimately become an ANSI standard.
No claims to be in conformance with this standard may be made unless all requirements stated in
the standard are met.
Inquiries, comments, and suggestions relative to the content of this JEDEC standard or
publication should be addressed to JEDEC at the address below, or call (703) 907-7559 or
www.jedec.org
Published by
©JEDEC Solid State Technology Association 2008
3103 North 10th Street
Suite 240 South
Arlington, VA 22201-2107
This document may be downloaded free of charge; however JEDEC retains the
copyright on this material. By downloading this file the individual agrees not to
charge for or resell the resulting material.
PRICE: Please refer to the current

Catalog of JEDEC Engineering Standards and Publications online at
http://www.jedec.org/Catalog/catalog.cfm
Printed in the U.S.A.

All rights reserved
PLEASE!
DON’T VIOLATE
THE
LAW!
This document is copyrighted by JEDEC and may not be

reproduced without permission.
Organizations may obtain permission to reproduce a limited number of copies

through entering into a license agreement. For information, contact:
JEDEC Solid State Technology Association

3103 North 10th Street
Suite 240 South
Arlington, VA 22201-2107
or call (703) 907-7559
JEDEC Publication No. 155
RECOMMENDED ESD TARGET LEVELS FOR HBM/MM QUALIFICATION

Contents
Foreward ................................................................................................................................................................................ ii
Introduction ............................................................................................................................................................................ ii
1 Scope........................................................................................................................................................................1
2 References ................................................................................................................................................................1
3 Terms and Definitions ..............................................................................................................................................4
4 Historical Perspective on HBM/MM ESD Requirements ........................................................................................5
4.1 Motivation for the HBM Target Level .....................................................................................................................5
4.2 Motivation for Introducing Machine Model (MM)? ................................................................................................5
5 Changes and Improvements in ESD and Control Environment ...............................................................................6
5.1 Historic ESD Handling Procedures ..........................................................................................................................6
5.2 Global Implementation of ESD ................................................................................................................................7
5.2.1 Ground and Bond all Conductors:............................................................................................................................8
5.2.2 Control Charges on Insulators:.................................................................................................................................8
5.2.3 Use Protective Packaging for Transit and Storage ...................................................................................................8
5.2.4 ESD Control Programs and Resulting Data: ............................................................................................................8
5.2.5 Advantage of Process Analysis ..............................................................................................................................10
5.3 Change of HBM Hazard Scenario by Increasing the Automation Level................................................................12
6 Machine Model – Correlation between HBM and MM ESD .................................................................................12
6.1 HBM vs. MM .........................................................................................................................................................12
6.1.1 Consequence of 1kV HBM Target .........................................................................................................................14
6.2 Exceptions to HBM/MM Ratio ..............................................................................................................................15
6.2.1 Bipolar vs. Unipolar Stress.....................................................................................................................................15
6.2.2 Advanced Technologies .........................................................................................................................................16
6.3 Conclusions ............................................................................................................................................................18
7 Consolidated Industry Data on HBM Levels vs. Field Returns..............................................................................18
7.1 Field Return Rates versus HBM Level...................................................................................................................18
7.2 Case Studies ...........................................................................................................................................................21
7.2.1 Devices with Failure Levels below 500V HBM.....................................................................................................21
7.2.2 Devices that Fail between 500-1000V HBM .........................................................................................................21
7.2.3 Devices that Fail between 1000-2000V HBM........................................................................................................22
7.3 Conclusion .............................................................................................................................................................22
8 Impact of ESD Requirements from Customers and Suppliers................................................................................23
8.1 ESD Requirements and Specification Failures.......................................................................................................23
8.2 Impact of “ESD Failures”.......................................................................................................................................23
8.3 Impact of Revised ESD Target Levels ...................................................................................................................25
9 IC Technology Scaling Effects on Component Level ESD ....................................................................................26
9.1 Scaling Effects on ESD Robustness .......................................................................................................................26
9.2 Protection Design Window ....................................................................................................................................29
9.3 ESD Capacitive Loading Requirements .................................................................................................................31
9.4 Package Effects ......................................................................................................................................................34
9.5 ESD Technology Roadmap ....................................................................................................................................35
10 Differences between Component ESD and System Level ESD .............................................................................36
10.1 The History of System Level ESD .........................................................................................................................36
10.2 Differences in Component and System Level ESD Stress Models ........................................................................36
10.3 Case Studies ...........................................................................................................................................................38
10.4 Conclusion .............................................................................................................................................................38
11 Recommendations for New ESD Target Levels .....................................................................................................39
11.1 8.1 New Realistic Target Levels for HBM and MM ..............................................................................................39
11.2 Treatment of Special Pins.......................................................................................................................................40
11.3 Timeframe for Applying New Recommendations .................................................................................................40
11.4 Future Cost of ESD Design ....................................................................................................................................40
11.5 Product ESD Evaluation Criteria............................................................................................................................41
11.6 Looking Forward....................................................................................................................................................42
Annex A: Frequently Asked Questions .................................................................................................................................43
-i-
Foreword
For more than 20 years, IC component level ESD target levels for both HBM (2kV) and MM (200V) have
essentially stayed constant, with no focus on data to change these levels. Today's enhanced static control
methods required by OEMs do not justify these higher HBM/MM levels as data will show in this
document. ESD over-design to these levels in today’s latest silicon technologies is increasingly
constraining silicon area as well as performance, and is leading to more frequent delays in the product
innovation cycle. Based on improved static control technology, field failure rate, case study and ESD
design data, collected from IC suppliers and contract manufacturers, we propose more realistic and safe
HBM/MM ESD target levels. These new levels (1kV HBM / 30V MM) are easily achievable with static
control methods mandated by customers and with today’s modern ESD design methods.
Introduction
This document was written with the intent to provide information for quality organizations in both
semiconductor companies and their customers to assess and make decisions on safe ESD level
requirements. It will be shown through this document why realistic modifying of the ESD target levels for
component level ESD is not only essential but is also urgent. The document is organized in different
sections to give as many technical details as possible to support the purpose given in the abstract.
Additionally, Frequently Asked Questions (FAQ) in the annex are intended to avoid any misconceptions
that commonly occur while interpreting the data and the conclusions herein. All component level ESD
testing specified within this document adheres to the methods defined in the appropriate JEDEC and
ESDA/ANSI specifications.
In June 2009, the formulating committee unanimously approved the addition of the ESDA logo on the
covers of this document.
-ii-
Page 1
(From JEDEC Board Ballot JCB-08-41, formulated under the cognizance of the JC-14 Committee on
Quality and reliability of Solid State Products.)
1 Scope
The intent of this report is to document and provide critical information to assess and make decisions on
safe ESD level requirements. The scope of this document is to provide this information to quality
organizations in both semiconductor companies and their IC customers.
Special Notes on the System Level ESD:
1. This work and the recommendations therein are intended for Component Level safe ESD
requirements and will have little or no effect on system level ESD results.
2. Systems and System boards should continue to be designed to meet appropriate ESD threats
regardless of the components in the systems that are meeting the new recommendations from this
work, and that all proper system reliability must be assessed through the IEC test method.
Special Notes on the Machine Model:
1. The Machine Model (MM) method as specified by some customers and suppliers is not a preferred
methodology by JEDEC for use in place of or in addition to HBM and CDM test protocols.
2. In contrast to HBM testers, MM testers are known to have wide variations in output results and thus
can give relatively less accurate information from user to user.
2 References
1. W.M. King, “Dynamic waveform Characteristics of Personnel Electrostatic Discharge”, EOS/ESD

Symposium Proceedings, EOS-1, 78 (1979).
2. L. Avery, private communication.
3. D.L. Lin, M.S. Strauss, and T.L. Welsher, “On the Validity of ESD Threshold Data Obtained Using
Commercial Human-Body Model Simulators”, Proceedings of the 25th International Reliability
Physics Symposium, 77 (1987).
4. M.S. Strauss, D.L. Lin, and T.L. Welsher, “Variations in Failure Modes and Cumulative Effects
Produced aby Commercial Human-Body Model Simulators”, EOS/ESD Symposium Proceedings,
EOS-9, 59-63 (1987).
5. ANSI/ESD S20.20-1999; Protection of Electrical and Electronic Parts, Assemblies and Equipment
(Excluding Electrically Initiated Explosive Devices)
6. IEC 61340-5-1; Electrostatics – Part 5: Specification for the protection of electronic devices from
electrostatic phenomena – Section 1: General requirements; 12.1998
Page 2
2 References (cont’d)
7. R. Gaertner; Do we expect ESD-failures in an EPA designed according to international standards?

The need for a process related risk analysis; ESD Symposium 2007; Anaheim, CA, USA
8. ANSI/ESD STM5.1-2001 or JESD22-A114C.01
9. ANSI/ESD STM5.2-1999 or JESD22-A115-A
10. M. Kelly et al., ‘A Comparison of Electrostatic Discharge Models and Failure Signatures for CMOS
Integrated Circuit Devices’, EOS/ESD Symposium, pp. 175-185, 1995
11. A. Amerasekera and C. Duvvury, ‘ESD in Silicon Integrated Circuits’, 2002
12. G. Notermans et al., ‘Pitfalls When Correlating TLP, HBM and MM Testing’, EOS/ESD Symposium,
pp. 170-176, 1998
13. M.D. Ker et al., ‘Electrostatic discharge implantation to improve machine-model ESD robustness of
stacked NMOS in mixed I/O interface circuits’, ISQED, pp. 363-368, 2003
14. D. Pierce, ‘ESD Failure Mechanisms’, ESD Symposium Tutorial, 1995-2005
15. J. Whitfield et al., ‘ESD MM Failures resulting from transient reverse currents’, 44th IEEE IRPS, pp.
136-139, 2006
16. C. Duvvury and G. Boselli, “ESD and latch-up reliability for nanometer CMOS technologies,” IEDM,
pp. 933-936, 2004.
17. S. Voldman, “ESD robustness and scaling implications for Aluminum and Copper interconnects in
advanced semiconductor technology,” ESD Symp. 1997.
18. K.H. Oh, K. Banerjee, C. Dvvury and R. Dutton, “Non-uniform conduction induced reverse channel
length dependence of ESD reliability for silicided NMOS transistors,” IEDM, pp. 341-344, 2002.
19. G. Boselli, J. Rodriguez, C. Duvvury, V. Reddy, P.R. Chidambaram and B. Hornung, “Technology
scaling effects on the ESD design parameters in sub-100nm CMOS transistors,” IEDM, pp. 507-510,
2003.
20. C. Russ, H. Gossner, T. Schultz, N. Chaudhary, K. Schruefer, W. Xiong, A. Marhsall, C. Duvvury and
C. Cleavelin, “ESD Evaluation of the emerging MuGFET technology, EOS/ESD Symp. 2005.
21. H. Gossner, C. Russ, F. Siegelin, J. Schneider, K. Schruefer, T. Schlutz, C. Duvvury, R. Cleavelin and
W. Xiong, “Unique ESD failure mechanisms in a MuGFET technology,” IEDM 2006.
22. A. Salman, R Gauthier, E. Wu, P. Riess, C. Putnam, M. Muhammad, J. Woo, D. Ioannou,
“Electrostatic Discharge induced oxide breakdown characterization in a 0.1 um CMOS technology,”
IRPS 2002.
23. A. Ille, W. Stadler, A. Kerber, T. Pompi, T. Brodbeck, K. Esmark and A. Bravix, “Ultra-thin gate
oxide reliability in the ESD time domain,” EOS/ESD Symp. pp. 285-294, 2006.
24. G. Boselli, J. Rodriguez, C. Duvvury and J. Smith, “Analysis of ESD protection components in 65nm
CMOS technology: scaling perspective and impact on ESD design window,” EOS/ESD Symp. 2005.
25. A. Jahanzeb, Y. Lin, S. Marum, J. Scichl, and C. Duvvury, “Investigation of device body size on
CDM tester peak currents and variability,” International ESD Workshop, 2007.
26. M. Mergens et al., “Diode- Triggered SCR (DTSCR) for RF-ESD Protection of BiCMOS SiGe and
CMOS Ultra-Thin Gate Oxides”, IEDM Digest, 2003.
27. ESD Association Road Map: http://www.esda.org/
Page 3
2 References (cont’d)
28. ESD Association White Paper II: Trends in Silicon Technology and ESD Testing, 2006.
29. IEC standard 61000-4-2 edition 1.2, 2001 “EMC – Part 4.2: Testing and Measurement Techniques –
ESD Immunity Test”
30. W. Stadler “State-of-the-Art in ESD Standards”, Proc. of 1st International ESD Workshop, Lake Tahoe,
2007, pp. 127-147, 2007
31. W. Stadler, S. Bargstaedt-Franke, T. Brodbeck, R. Gärtner, M. Goroll, H. Gossner, N. Jensen and C.
Müller “From the ESD Robustness of Products to the System ESD Robustness”, EOSESD Symp.,
2004, pp. 67, 2004
32. S. Marum, R. Watson, and C. Duvvury, “Effects of Low Level IEC 61000-4-2 Stress on Integrated
Circuits” Proc. of 1st International ESD Workshop, Lake Tahoe, 2007, pp. 262-273, 2007
33. W. Stadler, T. Brodbeck, R. Gärtner and H. Gossner “Cable Discharges into Communication
Interfaces”, EOSESD Symp., 2006, pp. 144-151, 2006.
34. B. Reynolds, M. Muhammad and R. Gauthier , “A Test Method to Determine Cable Discharge Event
Sensitivity at the Module Level”, Proc. of 1st International ESD Workshop, Lake Tahoe, 2007, pp.
252-261, 2007
35. T. Smedes, J. Van Zwol, G. De Raad, T. Brodbeck and H. Wolf, “Relations Between System Level
ESD and (vf-) TLP”, EOSESD Symp., 2006, pp. 136-143, 2006
36. N. Lacrampe, F. Caignet, M. Bafleur and N. Nolhier, “VF-TLP based methodology for the prediction
of ESD immunity of a PCB”, Proc. of 1st International ESD Workshop, Lake Tahoe, 2007, pp. 240-
251, 2007.
37. S. Marum, D. Wang and A. Chadwick “Monitoring Clamp Voltage during IEC 61000-4-2 Stress”,
Proc. of 1st International ESD Workshop, Lake Tahoe, 2007, pp. 450-457, 2007
38. J. Barth, J. Richner and L. G. Henry “System Level ESD Radiation Test Far Exceeds Real Human
Metal Discharge”, Proc. of 1st International ESD Workshop, Lake Tahoe, 2007, pp. 467-478, 2007
39. J_H Ko, S_J Kim, K_S Im, K-K Jeon, S-J Song, C-S Kim, C-H Jeon, K-T Lee, H-G Kim, I-H Son
“Abnormal HBM and MM Stress at Mobile LCD Module by Set-ESD gun”, Proc. of 1st International
ESD Workshop, Lake Tahoe, 2007, pp. 450-457, 2007
40. White Paper II: Trends in Semiconductor Technology and ESD Testing, ESD Association,
http://www.esda.org
41. EIAJ ED4701; Environmental and endurance test methods for semiconductor devices
42. JESD22-A114; Electrostatic Discharge (ESD) Sensitivity Testing Human Body Model (HBM)
43. JESD22-A115; Electrostatic Discharge (ESD) Sensitivity Testing Machine Model (MM)
Page 4
3 Terms, definitions, and letter symbols
BGA ball grid array

CDE cable discharge event
CDM charged-device model
CM contract manufacturer
DIP dual-in-line package
DSP digital signal processor
DTSCR diode-triggered SCR
EMS electronic manufacturing supplier
EOS electrical overstress
ESD electrostatic discharge
ESDA Electrostatic Discharge Association; ESD Association
FinFET fin field-effect transistor
FOD field oxide device
HBM human body model
HF high frequency
HSS high-speed serial link
IEC International Electrotechnical Commission
LNA low-noise amplifier
MCM multichip module
MM machine model
MugFET multigate field-effect transistor
PCB printed circuit board
RF radio frequency
SCR silicon controlled rectifier
SERDES serializer/deserializer transceiver that converts parallel data to serial data
SiP system-in-package
SoC system-on-chip
SOI silicon-on-insulator
TDDB time-dependent dielectric breakdown
TLU transient latchup
TVS transient voltage suppressor
ULSI ultra-large-scale integration
VDD positive voltage supply
VSS negative voltage supply
ESD Design Window: The ESD protection design space for meeting a specific ESD target level while
maintaining the required I/O performance parameters (such as leakage, capacitance, noise, etc.) at each
subsequent advanced technology node.
ESD robustness: The capability of a device to withstand the required ESD-specification tests and still be
fully functional.
It2: Failure current under ESD time domain.
Node: Within a circuit, a point of interconnection between two or more components.
Protection impedance: The turn-on impedance of any ESD clamp during the ESD current flow
Page 5
4 Historical Perspective on HBM/MM ESD Requirements
4.1 Motivation for the HBM Target Level
The early 1970s saw the first systematic measurements of HBM on people (e.g., H-P study) including
people in moving chairs. A good summary of some of this early work on the human body model was
published at the first EOS/ESD Symposium [1]. Workers in these early HBM studies found that even
while wearing wrist straps, one could easily generate 1-2 kV HBM. Later in the 1970s and early 1980s,
the automotive industry began instituting ESD pass levels, with Ford adopting the MM and 200V, GM
and Chrysler more focused on the HBM, and Chrysler specifying 2 kV HBM minimum after lengthy
consideration of even higher voltages. Meanwhile, RCA TV division settled on 2kV HBM and a specially
devised "Kinescope" test model, but no one accepted the latter [2].
Following these kinds of requirements from customers, by the mid-1980s, semiconductor companies
began to set internal HBM standards for components, and the 2kV HBM specification became most
common among them. Even at that time, CDM was recognized as a major cause of device failure, not
necessarily predicted with HBM testing, and so to some extent CDM testing was also desired.
Most of this early HBM testing was with various testers that were aligned with HBM as described by Mil
Spec 883C, Method 3015.X. Workers who were active in the greatest improvements in that spec in the
1980s will remember 3015.4 through 3015.7 (the last being in 1989) in particular, where major changes in
the waveform standard followed studies showing that the testers aligned to the earlier version of 3015
produced widely differing failure voltages in semiconductor devices [3,4]. Only after a short-circuit
current waveform spec was introduced were the internal tester parasitics brought under control to an
extent that allowed some consistency among testers aligned to 3015. All the while, 2kV remained as a
convenient target for a “passing” voltage. When the HBM test reached Method 3015.7 (1989), the tester
waveforms were much improved, but at that point the US Military stopped revising the spec and further
HBM spec development passed on to standards committees at the ESD Association and JEDEC.
4.2 Motivation for Introducing Machine Model (MM)
The reasons were:

a) MM simulated the failures caused by an ungrounded soldering iron contacting a semiconductor pin
lead.
b) MM simulated field damage failures such as CDM at that time better than HBM did.
c) The zero-ohm discharging resistance in the machine model results in a higher peak current than HBM
test for the thermal damage, a lower voltage MM test can be done.
Reason A does not exist any more due to the dramatic improvement of ESD controls in the advanced
automated manufacturing. Reason B also does not exist anymore because the newer CDM test method
reproduces these failures better than MM testing. As for reason C, the HBM test itself is not meaningful
now because it does not correlate to overstress failures in the field (see clause 7).
MM has been used for many years to verify the ESD performance of semiconductor devices in Japan.
Historically, the first MM ESD test was reported before 1977. Discharging inductance was not defined as
it is now, and no discharging waveform was defined either. It was adopted as the Japanese standard, EIAJ
IC-121-1982. Because of no discharging resistance and inductance, it was closer to real “Machine Model”,
metal to metal contact ESD than the existing MM standard.
Page 6
4.2 Motivation for Introducing Machine Model (MM) (cont’d)
Several years later, MM standard returned from the US with a discharging inductance and oscillating
waveform. Since then the MM standard does not simulate metal to metal contact which is now simulated
by CDM. Because of these reasons JEITA dropped the MM standard and added it in the HBM standard as
a reference in 1994 (EIAJ ED4701).
However, since Japanese users have accumulated much more MM data compared to HBM and CDM data,
these users are still using the MM test results to decide the ESD control level of their production lines.
This is the reason why many users request to continue the MM test, though they are amenable to accept a
new ESD test method.
5 Changes and Improvements in ESD and Control Environment
5.1 Historic ESD Handling Procedures
ESD control programs have been in place for many years. One of the earliest programs involved the
production of gunpowder. This simple program simply kept the powder wet during manufacturing and
handling. This kept the static charge low enough that the gunpowder would not ignite.
In the 1950’s and 1960’s, electronics were not that sensitive to ESD events. The devices of the time could
withstand most events without a problem. Even if they did fail for ESD events, the failures were a very
small proportion of the overall failure rates.
In the late 1970’s, with the introduction of Large Scale Integration (LSI), ESD was noted as a problem. A
group of industry experts realized that this was a problem and organized the first ESD Symposium in
1978 in the US. At the time, technical papers were exchanged and there were workshops on problems and
solutions. Companies at this time also started to implement ESD control programs. Each company had
their unique program and did not share the information. The need for standardized programs was not
recognized at that time.
The US Military was one of the first organizations to recognize the problems with static electricity and
ESD. The first standard to address ESD process control was Mil-STD-1686 released in May of 1980. This
standard along with its companion handbook Mil-HBK-263 represented the first ESD control standard in
the industry. All of the suppliers of electronics to the military were required to comply with this standard.
However, most of the private sector still followed company developed procedures.
These early standards were focused on people and packaging. Controls in place for insulators were left
mostly to the end user without much consideration except for the removal of non required insulators.
Tools, machines and automated equipment were not addressed or really considered as most of the
processes were manual. The basic instructions were to keep everything and everyone at the same potential.
An additional issue with these first ESD control programs was that the materials that were used to control
static electricity did not have standards to qualify the materials. This lead to many different types of
testing, different methods and different instrumentation that caused different results. In some cases,
materials measured by these methods did not perform well in controlling static. In the early 1980’s, a
technical association, the ESD Association (ESDA) was formed to try to resolve some of the issues
surrounding material testing. The first standards from the ESDA were simple material tests for items such
as wrist straps, work surfaces and flooring.
Page 7
5.1 Historic ESD Handling Procedures (cont’d)
The standards did create a way to compare one product with another product. Suppliers of these materials
were able to use the standards to improve the products to make them better. For example, the simple wrist
strap has gone through many changes in the industry. What started out as a simple metal bead band has
evolved into a system that makes better contact with a person and in some cases allows for continuous
monitoring. They provide a much more reliable connection than before and last longer. The standards also
provide a way to test the wrist straps in a consistent manner so that one that becomes defective can be
removed and replaced. Before this, materials were used until they were physically damaged without
regard to the electrical properties.
In the 1980’s and 1990’s the electronics manufacturing industry changed from each company having all
the manufacturing reside within the company to a model that included many Contract Manufactures (CM)
or Electronic Manufacturing Suppliers (EMS). The military standard and the European standard, CECC
00 015:1991, became out of date. They were either too restrictive or did not address all aspects of a
control program.
The ESDA in 1995 was given the task of replacing Mil-Std-1686 with an industry standard. The standard
ANSI/ESD S20.20-1999 [1] was the replacement for ESD process control. Following this standard, a
third party certification program was established to demonstrate compliance to the standard. Today, this
standard has been updated and replaced by ANSI/ESD S20.20-2007. In parallel, the IEC is updating IEC
61340-5-1 [2] to become technically equivalent to ANSI/ESD S20.20-2007. The current status for this
standard is Final Draft International Standard (FDIS) and is being voted to become an international
standard later this year. These standards when followed are written to safely handle 100V Human Body
Model devices. All of these standards have improved control materials, understanding and ESD control
processes.
5.2 Global Implementation of ESD Control
Manufacturing of ESD sensitive products is currently performed in all parts of the globe. However, since
the late 1990’s there has been an ever increasing trend to move electronics production from high cost to
low cost geographies.
Globally, there is a large difference in the types and levels of ESD programs that are in existence today.
ESD control programs range from:
a) Little or no ESD control

b) Basic ESD controls
c) Advanced ESD control programs
The level of ESD controls is not strictly related to geography but in many cases is driven by customer
requirement. There are many ways to establish an effective ESD control program. This leads to
considerable differences in effective ESD program design and the controls that are ultimately used.
However, every well established and maintained ESD control program is based on the following three
fundamental principles:
• Ground and bond all conductors

• Control charges on insulators
• Use protective packaging for transit and storage
Page 8
5.2 Global Implementation of ESD Control (cont’d)
In the following each of these principles is described in more detail.
5.2.1 Ground and Bond all Conductors:
Maintaining ESD sensitive devices and every item that they come into contact with at the same electrical
potential will ensure that ESD related events do not happen. This equi-potential situation can be achieved
by attaching all of the conductors in the environment to ground (earth) or by bonding them together to
maintain an equal potential. Conductors in this situation refers to people, working surfaces, ESD sensitive
devices and any process related conductors and dissipative materials that come into contact with ESD
sensitive devices. The grounding and bonding of conductors will minimize the chance of HBM and MM
discharges from occurring.
NOTE If all conductors that contact ESD sensitive devices are grounded, companies should be able to handle
devices with a MM robustness of less then 10V.
5.2.2 Control Charges on Insulators
Every good ESD control program will do the following:
1) Remove unnecessary process related insulators from the operations where ESD sensitive devices are
handled.
2) Determine what constitutes an unacceptable electrostatic field on insulators that are required in the
manufacturing / handling process (for details see ANSI/ESD S20.20).
Controlling charges on insulators will help to minimize the chance of CDM related ESD events from
occurring.
5.2.3 Use Protective Packaging for Transit and Storage
Finally, in order to ensure that ESD events do not occur between manufacturing process steps or during
the shipment of ESD sensitive devices to other locations (customer or next processing facility) the devices
should be packaged in ESD protective packaging. The adequate level of protection provided by the
packaging can be achieved by different packaging systems and has to be defined by the responsible
companies.
5.2.4 ESD Control Programs and Resulting Data
A. Little or no ESD control
For the few companies that have not even implemented a basic ESD control program (this means little or
no controls used and not verified on a consistent basis) it is very likely that these companies would not be
able to handle ESD sensitive devices that have an ESD sensitivity of even 2 kV Human Body Model.
Figure 1 shows the voltage on a person’s body as they walk in a manufacturing environment that has no
ESD floor or footwear. In this situation, the person could damage a device with an ESD HBM sensitivity
of 2 kV.
Page 9
5.2.4 ESD Control Programs and Resulting Data (cont’d)
Figure 1 — Voltage on a person’s body when walking in a manufacturing environment without

ESD floor or footwear
B. Basic ESD Controls
Some companies have implemented a basic ESD control program. A basic ESD control has all the
required control elements but no redundancies. The simplest ESD control program consists of personnel
grounded with wrist straps, a grounded surface where ESD sensitive devices are handled and all static
generating materials are removed and protective packaging for movement of ESD sensitive devices
through the process. This type of program is often used by companies where:
a) The manufacturing operation is confined to a small area.

b) The number of employees handling ESD sensitive devices is small.
c) The value of the product is low.
d) The reliability of the products being produced is low.
However, simple does not mean that the program cannot be effective. A well grounded wrist strap system
will keep the voltage on personnel to less than 10V. As long as the program is audited on a frequent basis
this program can be every bit as effective as one where multiple ESD controls are utilized.
Page 10
5.2.4 ESD Control Programs and Resulting Data (cont’d)
C. Advanced ESD Control Program
Finally, many companies utilize advanced ESD control programs to ensure that the devices that they
handle will not be damaged. The use of constant monitors, ionization systems and ESD flooring and
footwear can add a degree of redundancy and convenience to ESD programs where:
a) The devices are very ESD sensitive

b) The value of the finished product is high.
c) There is a large employee population with a high turnover rate.
d) The product has high reliability requirements.
All of these factors and more can drive the need for a more complex ESD control program to be
implemented.
5.2.5 Advantage of Process Analysis
This example shows the difference between just implementing ESD control measures and doing a deeper
ESD control process analysis. Figure 2 shows an example of a manufacturing location that had a
conductive ESD flooring system installed. Unfortunately, the company did not make an effort to evaluate
the ESD footwear system that was used for its employees. As you can see the voltage on personnel was
well above the 100V HBM threshold that the company had established for itself. The footwear used was
chosen based solely on price. However even these shoes would provide adequate protection for a 500V
device.
Figure 2 — Voltage of a person in a manufacturing environment when a conductive ESD flooring

system was installed (in comparison to Figure 1)
Page 11
5.2.5 Advantage of Process Analysis (cont’d)
However, once the company understood the implications of their decision, properly selected footwear was
implemented and the company was now able to safely meet their goal of handling 100V HBM sensitive
devices as shown in Figure 3.
Figure 3 — Voltage of a person in the same manufacturing facility / ESD flooring as Figure 2 but
with properly selected ESD control footwear
The other factor that can easily jeopardize an ESD program is infrequent verification that the ESD control
elements are working. Most successful companies audit their ESD programs frequently to ensure that all
of the control elements are functioning as intended.
Two good resources for establishing an ESD control program are through the implementation of
ANSI/ESD S20.20 which is published by the ESD Association or IEC 61340-5-1 which will be published
as a draft international standard later in 2007. Both of these documents will provide the structure and
guidance necessary to establish an ESD control program that can safely handle 100V HBM sensitive
devices and higher.
Conclusion:
By establishing an ESD control program and frequently verifying that the ESD controls are working as
installed most companies can easily handle ESD sensitive devices with a sensitivity of 500V HBM or
higher. However, with slightly more attention to the selection of ESD control items a 100V HBM
program is easily attainable.
Page 12
5.3 Change of HBM Hazard Scenario by Increasing the Automation Level
The complexity and automation level of printed circuit board (PCB) manufacturing has increased
significantly in recent years. Years ago most devices used packaging with easily contactable pins, with
pin counts being lower and pin-to-pin pitch relatively high. These devices were generally assembled
manually by operators. In this environment there was the probability of human discharge to a single pin.
Modern packages today can contain up to thousands of I/O, and these I/O can be either pins on the
package periphery, balls (as in BGA) or chip-scale packages. The I/O to I/O pitch has decreased
dramatically to allow high pin count die to be packaged in a reasonable size. As a result the packaged
parts cannot be assembled manually, and the process is automated by non-human handling (grounded
machines / tools / pick and place). Modern ESD control programs have evolved to become very effective
in control in these types of handling environments, as discussed in 5.2. Therefore for automated assembly
lines with modern, frequently audited ESD control programs, the risk of HBM or MM events is very low.
Nevertheless there are still manufacturing areas, such as rework / optical inspection areas, usually smaller
areas, where human contact with devices does happen. ESD control programs if effectively implemented
and audited minimize the HBM / MM discharge risk in these areas.
It is also confirmed by literature, that most of the ESD related field fails are due to CDM like stress and
not due to HBM like stress [5].
6 Machine Model – Correlation between HBM and MM ESD
This clause addresses the relationship between HBM and MM ESD robustness of products. It will be
shown that typically the MM value is 1/30 or more of the HBM value, with a few exceptions. This
implies that if a product passes 1 kV HBM the expected MM performance is > 30V. This is sufficient as
is explained in clause 5.
6.1 HBM vs. MM
The previous clause explained the development of the MM standard. Historically an MM:HBM ratio of
1:10 was commonly used, as is e.g. clear from the classification in the respective standards [8, 9]. For
example, a class 2 HBM product passes at least 2 kV, where as a class 2 MM product should pass at least
200V. This commonly accepted ratio most likely arose from observations on products at the time of
development of the standards. These observations are given in [10], yielding a 1:11.7 ratio.
In [11] a ratio between 1:10 and 1:20 is quoted. In [12] a paper was published that demonstrated ratios of
1:10 and 1:17 measured on test structures, using 2 different types of MM. In [13] a ratio varying from
1:13.5 to 1:18 was achieved for different variations of stacked NMOSTs in an advanced CMOS
technology.
All results are given, while stressing that very similar failures were observed for HBM and MM.
Simply equating the available charge in the HBM and MM models it is easy to see that a ratio of 1:2 can
be expected:
CMM*VMM=CHBM*VHBM -> VHBM = 2 * VMM.

Page 13
6.1 HBM vs. MM (cont’d)
Most HBM and MM failures are related to thermal damage: due to overcurrent creating thermal melting /
reflowing. Pierce [14] has shown that by equating the energy deposited in the IC during the stress and
assuming that all ESD energy is used to create damage the following relation can be found:
VMM= (Rprot*CHBM/(CMM*(RHBM+Rprot)))*VHBM.
Using typical values this leads to an MM:HBM ratio of 1:25. This neglects the facts that power to failure
depends on pulse width and that MM pulses are shorter than HBM pulses. From the above equation it is
also clear that an increase of the Rprot leads to a lower ratio. By reducing the HBM target the protection
elements may have more impedance and thus the MM level will (relatively) be reduced less.
On the other hand, failures can also be induced if the protection does not clamp the voltage sufficiently. In
this case, a small device or fragile oxide may become damaged with just a small fraction of the energy,
because most energy is safely dissipated by the protection. For this reason it is good to compare peak
currents for HBM and MM. According to the standards the peak current into a short is 1.3 A for a 2 kV
HBM discharge and 3.8 A for a 200V MM discharge. Equating peak currents thus leads to a 1:30 ratio of
MM:HBM.
So on theoretical grounds a ratio of MM/HBM will be a minimum of 1/30. This is confirmed on products
and test structures by the publications quoted before. The council collected data both on test structures
and products of several of the members. Figure 4 shows the results on the test structure. Clearly on
average a ratio of 1:20 is found and factors larger than 30 are very unlikely.
HBM-MM correlation
10
9
8
7
average 19.1
Frequency
6
5
4 median 20.0
3
2
３σ 5.0
1 median+3σ 25.0
0
14 15 16 17 18 19 20 21 22 23 24 max 21.9
HBM/MM ratio min 15.7
Figure 4 — HBM/MM correlation of test structures in an advanced CMOS process
Page 14
6.1 HBM vs. MM (cont’d)
Figure 5 shows data collected from several products from different suppliers and technologies. The figure
shows the MM failure voltages vs. the HBM failure voltages. A best fit regression shows a ratio of 1:16.
The whole population is bounded by 1:3 and 1:30 lines.
Figure 6 presents the HBM:MM ratio as a function of the HBM performance. It is clear that the ratio
increases for increasing HBM level. This is mainly due to the fact that to achieve high HBM levels large,
low impedance protections are needed. Since the protection impedance affects the MM peak current, this
leads to a higher ratio for higher HBM performance. Thus assuming a factor of 30 is actually worst case.
6.1.1 Consequence of 1kV HBM Target
The above reasoning and data supports the expectation that a 1 kV HBM performance will imply an MM
performance between 30V and 200V, with a typical expectation value of 60V.
The above reasoning assumes that the MM bipolar stress case can be approximated by two unipolar HBM
stresses. This means that the current and voltage rise times must be similar. It also means that the tester
dynamics should be the same for both cases. Although these assumptions are correct for a large number of
cases, MM and HBM can address different failure modes, in which case correlation is not possible. The
next clauses will describe some of the cases in which MM and HBM do not correlate, detailed with
measurements from different technologies and products.
700
y = 0.3333x
600
y = 0.0646x
MM failure Voltage (V)
500
400
300
y = 0.0333x
200
100
0
0 2000 4000 6000 8000 10000
HBM failure Voltage (V)
Figure 5 — MM failure level vs. HBM failure level for several products of several companies
Page 15
6.1.1 Consequence of 1kV HBM Target (cont’d)
30
25
20
HBM/MM Ratio
15
10
0
0 2000 4000 6000 8000 10000
HBM failure voltage (V)
Figure 6 — HBM/MM ratio vs. HBM level for several products of several companies
6.2 Exceptions to HBM/MM Ratio
6.2.1 Bipolar vs. Unipolar Stress
The most obvious difference between MM and HBM is the bipolar nature of MM. Correlating the two
models as shown in Figure 7 assumes that the device can be approximated in a quasi-static regime when
the voltage crosses 0 during the MM pulse; only then can the MM level be extracted from a positive and a
negative HBM level. In many cases this quasi-static approximation is reasonable, reducing the physics to
the unipolar case.
Figure 7 — MM and HBM pulse, the main difference being the bipolar characteristic of the MM
pulse
Page 16
6.2.1 Bipolar vs. Unipolar Stress (cont’d)
Figure 8 — Diode up was destroyed by a positive MM stress between IO and VDD2
This is however, not a general statement. One of the most notable physical effects for which this
simplification is not valid is dynamic avalanching. In Figure 8 a case is shown in which MM stress
damaged the diode from pin to supply VDD2, which can not be explained by merely looking at the peak
current or energy dissipated by the MM pulse. The well of the diode gets injected with charges during the
first (positive) swing; when the stress reverses, these charges worsen the effect of avalanching at the
Nwell/P+ junction. Dynamic Avalanching is one the most important effects where the pulse reversal has a
dramatic influence on the ESD behavior of the device [15].
6.2.2 Advanced Technologies
Figure 9 shows the results of HBM and MM testing on a wide variety of test structures in a 65nm high
performance technology. The majority (67%) of the test structures correlates well with the expected
correlation (HBM/MM 10-30). A small portion (12%) has a lower correlation; some devices (21%) have
higher correlation. All structures are either self-protective drivers or are measured with a sensitive node in
parallel, such that the given numbers indicate their effectiveness as well as robustness.
Given the formula by Pierce:
VMM= √(Rprot*CHBM/(CMM*(RHBM+Rprot)))*VHBM
higher correlation for lower resistive protection devices can be predicted. Care must be taken, however as
this formula does not take into account the parasitic inductance of both testers (which is about 10 times
worse for the HBM as for the MM tester), the test board capacitances and the parasitic resistance of the
test board capacitances. The formula is plotted in Figure 10. Without the corrections for the tester
parasitics, the correlation reaches infinity for zero Ohm protection devices, meaning the approximation is
not correct for small Rprot. Note also that this correlation factor is closely related to the testers, and does
not correlate with real life. As technology scales down, lower resistive protection devices are needed,
meaning VHBM/VMM is expected to increase.
Also important to note is that the difference in parasitic inductance in both testers gives a very different
rise time behavior. With poorly designed ESD protection, this might lead to large variations in the
correlation factor, and especially the MM value might vary from process (and lot to lot) variations.
Page 17
6.2.2 Advanced Technologies (cont’d)
Figure 9 — HBM and MM Results from a test chip in a 65nm high performance technology
HBM/MM as function of Rprot
200
180
160
HBM/MM Ratio
140
120
100
80
60
40
20
0
0 2 4 6 8 10 12 14
Rprot (Ω)
Figure 10 — Pierce Formula for Different Rprot

Page 18
6.2.2 Advanced Technologies (cont’d)
An example of such apparent miscorrelation between HBM and MM qualification levels was reported for
a product in a sub-micron Smart Power SOI technology. The product passed 8 kV HBM. Higher levels
could not be tested due to tester limitations. The same product passed 1000V CDM in a TSSOP32
package. During MM qualification a problem was observed. Whereas most pins were qualified without
problems one specific pin combination failed at 75 V. This same pin combination passed 50V MM. The
same pin types in other combinations also did not pose problems. The physical failure signature was a
broken gate oxide on a transistor with its gate connected to the discharge path. Most likely the reverse
recovery effect created a voltage overshoot due to the bipolar nature of MM. A rough calculation indeed
shows that the voltage did rise high enough to damage gate oxide for that particular transistor. After a
design fix the product passed 250V MM (not stressed up to failure). No field returns related to this MM
issue have been reported for both the initial and improved version.
6.3 Conclusions
It has been demonstrated that most of the observed HBM-MM relationships on products fall between the
boundaries that are expected on theoretical grounds. Some rare exceptions to this general relation have
been described. In most cases these are related to the use of relatively slow protection elements.
7 Consolidated Industry Data on HBM Levels vs. Field Returns
This clause discusses the relationship between the HBM qualification level of a product and the potential
risk of failure related to that level when using the device. To this end a large set of data was collected.
The next clauses discuss the global findings from that database and present several case studies.
7.1 Field Return Rates versus HBM Level
A statistical comparison between number of shipped ICs achieving certain ESD qualification levels and
their field return rate is given based on the consolidated data of the companies contributing to the Council
(Figure 11). A total quantity of 21 billion parts was included in the statistics. 24% of the parts belonged to
the 500V pass/1000V HBM fail category. 28% were passing levels between 1000V and 1500V. 4% had a
robustness of less than 2000V but higher than 1500V. The remaining 44% met the 2 kV HBM level. The
weakest pin combination determined the level of ESD robustness. Overall more than 600 qualified /
released designs were considered, which were shipped in the years 2000 to 2006. Both designs with a
lower ESD qualification level on a few pin combinations and designs with a reduced ESD robustness on
many pins are included in the data. The IC designs considered belong to various application fields
including communications, consumer, storage, automotive and discrete ICs. They were processed in
several technologies ranging from a 1 µm node down to a 65 nm node. The assembly was done at a large
number of sites located in America, Asia and Europe. All of them are running at least a basic ESD static
control program as defined in clause 5.
Page 19
7.1 Field Return Rates versus HBM Level (cont’d)
1 dpm line all devices

1
based on 21 billion devices
"EOS/ESD" fails per million devices 0,1
with 1.5kV HBM

0,01
with 500V HBM
0.7 billion sold

4.8 billion sold
9.3 billion sold

with 2kV HBM
5.7 billion sold
with 1kV HBM
1E-3
1E-4
500 1000 1500 2000
HBM robustness
Figure 11 — EOS/ESD fails returned to IC supplier versus the achieved HBM qualification level. A
total number of nearly 21 Billion parts is considered.
The above EOS/ESD failure rate comprises all failed devices returned from in-house handling, board
manufacturers’ and OEM’s analysed by failure analysis departments of many IC suppliers. It was
included in the statistics if the root cause in the physical failure analysis report was given as ‘ESD fail’ or
a fail due to ‘ESD or EOS’ (electrical overstress). This means the chart covers all kinds of possible
electrically damaging mechanisms like discharge in the electrostatically protected environment, the
discharge outside electrostatically protected areas and electrical overstress due to malfunction of the
controlling board circuit.
Due to similar failure pictures and the missing information about stress conditions in the field a more
detailed distinction between pure ESD events and EOS related fails cannot be made in this statistics.
However, even including EOS related fails the total return failure rate summed up over all ESD classes is
below 0.1 defects per million (dpm). Clearly not all failing devices are returned to IC suppliers.
Especially in cases of consumer ICs and other high volume, low cost products the effort for analysis is not
often performed by the board manufacturer or the OEM. However, extracting just the data for automotive
parts (where the awareness of defects is very high) provides the same distribution of EOS/ESD fails
versus HBM qualification level as non-automotive parts (Figures 12 & 13). Both graphs provide evidence
that the fail rate due to electrical stress is independent of the achieved HBM level above a threshold of
500V. It can be concluded that a qualification level of 500V HBM is sufficient to safeguard against
increased failure rate due to electrical damage.
Page 20
7.1 Field Return Rates versus HBM Level (cont’d)
1 dpm line automotive products

1
"EOS/ESD" fails per million devices 0,1
with 1000V HBM
with 1500V HBM
with 2000V HBM

0,01
with 500V HBM
2.0 billion sold
0.4 billion sold
1.6 billion sold

1.4 billion sold
1E-3
1E-4
500 1000 1500 2000
HBM robustness
Figure 12 — EOS/ESD fails of automotive ICs returned to IC supplier versus the achieved HBM
qualification level. A total number of 5.5 Billion parts is considered
1 dpm line non-automotive products

1
"EOS/ESD" fails per million devices
0,1
with 1000V HBM
with 1500V HBM
with 2000V HBM

with 500V HBM
3.7 billion sold
0.3 billion sold

3.4 billion sold
7.7 billion sold
0,01
1E-3
500 1000 1500 2000
HBM robustness
Figure 13 — EOS/ESD fails of non-automotive ICs returned to IC supplier versus the achieved
HBM qualification level. A total number of nearly 15 Billion parts is considered
Page 21
7.2 Case Studies
In 7.1 the findings on the overall data were presented. It is not feasible to discuss each data point in detail.
Therefore 7.2 will highlight several interesting cases in more detail. Since these are case studies, each
case is necessarily from a particular company, but examples of several companies of the Council are used.
Also none of the cases by itself is evidence for the general conclusion. However, together they increase
the confidence in the final conclusion.
7.2.1 Devices with Failure Levels below 500V HBM
One company reports that they have 11 designs that fail between 100V HBM and 200V HBM. The die
are produced in external foundries and in their own fabs, whereas contractors and their own assembly
lines do the package assembly. In total over 4 million parts have been shipped without a single field
return.
Another company has delivered several hundred thousand parts representing different ASIC designs, each
failing between 100V and 200V HBM. In general, only a few pins limited the chip robustness level to
these levels. No returns have been reported, despite the fact that several assembly companies were used.
Another example is given by a less than 500V HBM part for a consumer electronics application. The
device is produced in a 90nm SOI technology, assembled by multiple low-cost Far East CMs. The low
HBM levels were observed on approximately 10 out of 900 pins. 11 million parts were shipped with no
customer returns for ESD. For a new version, the root cause of the lower failure levels was identified and
improved. The redesign was qualified as a 1500V part. Of this version, 3.8 million parts have been
shipped to date, again with no customer returns for ESD.
An EPROM product in a relatively old and mature (early 90’s) technology showed handling problems.
PPM levels are not known, but they were high enough to start an investigation. The failure mode could be
reproduced by a HBM test, showing that the device was very weak (HBM robustness < 500V) and would
have needed advanced ESD control during assembly (see classification table). This control was not
available at the time. After redesign the device reached an HBM level of 1500V and did not show any
further fails in the field.
Finally one company reported on a product that passed 400V HBM and failed 500V HBM on a limited
number of pins. 16 million samples have been sold. In the past 3 years 1 incident caused above-average
returns. This incident was traced to a problem in the assembly flow, unrelated to ESD, where it was
subsequently eliminated. An improved design, meeting 2 kV, never showed problems, while 4 million
samples have been sold of this version.
These cases illustrate that it is possible to handle even these extremely sensitive parts, if the necessary
precautions are taken.
7.2.2 Devices that Fail between 500-1000V HBM
A first example is given for a product that passes 500V HBM, but fails 1000V. The only returns that were
received were traced back to system level stresses. The customer demanded a market conform upgrade to
1000V HBM. This was accomplished by a redesign. With the new design the same system level return
rate was observed.
Page 22
7.2.2 Devices that Fail between 500-1000V HBM (cont’d)
A second council company reports on 3 product types failing 1kV HBM stresses. For one product the
1kV level failed on all pins. For the other 2 designs only a few pins did not pass 1 kV. In total only 2 ESD
complaints were reported. The first was due to system level qualification by the customer. The other
showed damage on a pin that directly interfaces to the application. As far as known, this part was not
deliberately stressed, but it is likely that this is also a case of system level damage. Neither of the ESD
failures was on the lower HBM level pins. In addition, some clear EOS fails were reported. These
occurred on random pins, not only on the pins with lower ESD qualification level.
7.2.3 Devices that Fail between 1000-2000V HBM
A product failing 1 kV HBM / passing 1500V HBM had a dpm level well below target. A deep analysis
of the fails showed the following root causes: 58% of the fails had been due to EOS, 29% showed no
problem, and only 1% of the fails were due to ESD. All EOS fails were traced to inappropriate use of the
IC.
Several ASIC designs failing between 1000V and 2000V were reported. All products were accepted by
the customers. Due to the relatively small sales numbers no ppm data is available. No ESD returns exist.
For one product, failing 1200V HBM, several EOS returns were received. All returns came from the same
customer. This same customer also reported similar fails with a 2kV HBM passing product.
Another company reported on two similar designs. One product passes 1500V and fails 2000V HBM. The
other product passes 3000V HBM. The products do not show significant reject rates, ppm levels are well
below target. Also there is no significant difference between the reject rates of both products.
A microprocessor ICs was processed in 130nm CMOS and an ESD robustness level of 1 kV HBM and
300V CDM was achieved. For a shipped volume of 200,000 no field returns are known.
7.3 Conclusion
The conclusion is that ICs with 500V HBM and above can safely be manufactured in existing IC and
board manufacturing environments. This is an on average statement and can be invalid for single
manufacturing sites where fundamental rules of ESD static control, as described in clause 5, are not
obeyed.
Page 23
8 Impact of ESD Requirements from Customers and Suppliers
8.1 ESD Requirements and Specification Failures
With technology scaling and an ever increasing need for I/O performance, it is no surprise that during
ESD qualification of processes, many ESD issues can come up. The failures most often may be related to
only a few pins. The ESD failure debugging can take many weeks or even months of work involving
ESD experts, product engineers, ESD test engineers, I/O design engineers and failure analysis engineers.
Furthermore, participation from quality managers and customer interface engineers might be warranted in
more urgent cases.
During this extensive analysis one usually finds that the original ESD failures may lead to more work if
they are related to non-repeatable random events. Or, as in more recent cases, the tester-induced failures
are not consistent with bench analysis, leading some to wonder if they are generated by the ESD tester
itself.
During these product evaluation efforts a tremendous amount of time and cost is expended, but most
important is the delay in time to market for the product. In most cases the analysis and the eventual
improvement to meet the customer ESD requirement results in a product that is not any more reliable to
the customer than it already was originally. This has been the experience of many IC suppliers.
8.2 Impact of “ESD Failures”
Both suppliers and customers are impacted by a result of ESD failures seen during the qualification
process. Both supplier and customer automatically assume that an ESD failure generated by the ESD
tester means failure certainty in the field. The data of this document does not support this assumption.
Once ESD failure is seen a supplier may ask the following questions:
- Could these failures be replicated and are they consistent?

- Was the root cause confidently identified?
- Will any changes impact the product performance by impacting the pin capacitance? (See Figure 25)
Answering these questions involves costs to both the supplier and the customer. By “cost” this is not only
the expense of the additional mask / silicon wafer production, but also an additional slow down in the
product delivery to the market, and degradation in the expected circuit performance due to the added ESD
protection. Additionally, both supplier and customer must engage in a series of discussions to resolve
these issues and agree on a path of resolution. Questions that must now be answered include:
- What data was used to evaluate and resolve target ESD specifications?
- Will there be a delay in product delivery?
- How many phone/face-to-face meetings will be necessary for satisfactory resolution?
- What additional delay may be experienced by the customers for qualification of re-designs?
- Will the reliability improvement gained justify the efforts; is there real risk in doing nothing?
- What will be the total impact on time to market for both supplier and customer?
Table 1 summarizes real life examples collected from companies for a few products that originally met a
1kV ESD target level to illustrate these issues. Although in some cases there were no disruptions to the
product sales, the effort involved meant cost to both suppliers and customers. In one extreme case, the
product release was delayed by two years.
Page 24
8.2 Impact of “ESD Failures” (cont’d)
The added cost to the customers comes from the meetings and negotiations that have to take place before
the issue is settled as well as impact to the product launch. Also, even if a re-design is completed, there is
no guarantee that the failure rate will be improved and that no random false failures, or even a new failure
mode might occur. For these reasons, the cost of meeting the current specs at 2kV HBM is continuously
going up, accelerated by the technology scaling effects and increased pin count.
Table 1 — Selected Product Example Cycles for Meeting 2kV

Product Disruptions Impact Intro Delay Number of joint
ESD Meetings
P1 No Redesign None 1
P2 Yes Redesign 2 Years 10
P3 No De-rate ESD None 4
P4 Somewhat Had to do minor redesigns 3 Months >5
P5 No No None >5
P6 Yes Some delay 6 Months 19
P7 No No None 2
P8 Yes Redesign 1 Year >5
These examples in Table 1 represent a snapshot of what routinely and typically occurs for products at
each supplier company. Note the number of customer/supplier meetings that had to take place during the
efforts to improve the product ESD from 1kV to 2kV.
Both customer and supplier now must take a look back and try and address the following:
1) Were these efforts meaningful or justified?

2) What other design focus was affected while concentrating on component level ESD and how have
new product innovations been affected?
The data included in this white paper supports the conclusion of the Industry Council in regards to
Question 1 in that the efforts to reach 2KV HBM level are not justified. In regards to Question 2, this is
difficult to quantitatively answer but clearly pulling resources off to focus on a non-risk item deters efforts
to improve product performance in other areas.
Another example, in Figure 14, shows the work-months involved for eight products manufactured at two
different technology nodes. All eight products were passing 1kV and required by the customer to be
redesigned to meet 2kV. It can be seen that the analysis effort can range from 2 to 15 work-months.
Consolidated Industry Council data shows that the average ESD respin causes >7 work-month effort.
Since qualification times add additional delay, the total product delay can be as much as 12 months time
to market.
Page 25
8.2 Impact of “ESD Failures” (cont’d)
16
rew o rk effo rt / red esig n [Wo rk Mo n th s]

14
12
10
8
6
4
2
0
180 nm CMOS 130 nm CMOS
Figure 14 — Actual example of ESD redesign efforts needed to meet 2kV on devices meeting 1kV
8.3 Impact of Revised ESD Target Levels
Revision of the component HBM target levels to a safe level of 1kV would provide several benefits for
both the customer and the supplier:
• Elimination of design respins for HBM performance between 1kV and 2kV and shorter time to
market
o For example, as much as 12 months can be saved as learned from case studies
• IO area savings to accommodate circuit requirements
o For example, some calculations have shown as much as 43% reduction for advanced circuits with
low leakage and high performance demands
o Similar reductions would also apply to analog circuits
• Capacitance savings to help achieve faster circuits
o For example, at 45nm and 32nm technologies 16-18 Gbit/sec cannot be met with 2kV designs but
can be accommodated with 1kV or less requirement
• Short term gains would obviously be faster release of products for production and more focus on next
generation technology ESD development and I/O performance
• Long term gains would be better customer relations and more opportunity for innovation of protection
methods for the more relevant system level ESD
Page 26
9 IC Technology Scaling Effects on Component Level ESD
9.1 Scaling Effects on ESD Robustness
Advances in Integrated Circuit (IC) technologies which were achieved for circuit performance and overall
reliability requirements have had a major impact on the intrinsic ESD design [16]. This is not surprising
since the silicon scaling effects to increase circuit speed with shorter channel lengths or thinner gate
oxides will lead to transistors that are much more sensitive to ESD and will result in lower failure current
(It2). At each new technology node, new adverse effects are noted. Table 2 lists these technology trends
and ESD impact starting in the early 80s to the present time period.
During the early technology applications, the transistor scaling involved increased current density (J)
which led to higher dissipation in power per unit volume,⎯J •⎯E (where E is the electric field) and thus
lower ESD levels. The next major change involved lowering E to improve channel hot carrier reliability
which subsequently increased the power density (and decreased the bipolar efficiency) for NMOS NPN
bipolar operation and thus again reduced ESD. This was rapidly followed by introduction of silicided
source/drain diffusions that led to current crowding effects and even poorer bipolar efficiency. In addition
to the silicide effects, the implementation of lower substrate resistance with epi to reduce latchup effects
caused another problem for the ESD design, especially using SCR type of clamps. However, when the epi
was replaced by bulk substrates for cost effectiveness, this led to yet another unexpected problem –
parasitic bipolar interactions at the IO areas and in the internal circuits.
Table 2 — Technology Scaling Impact on ESD

Feature Process Impact on Factor(s) degrading intrinsic ESD
Size Advance ESD performance
3um Junction NPN Current Density
Scaling Robustness
2 um Graded NPN Power Dissipation
Junction Robustness
1 um Silicides & NPN Ballasting Effects and Avalanche Process
Epi Robustness
Substrates
0.5 um STI SCR Trigger Decreased Parasitic Bipolar Efficiency
0.35 um Bulk Parasitic Increased Bipolar Effects
Substrate Interactions
0.18 um Shorter Lower It2 Localized Heating
Channel
Lengths
0.090 um Ultra Thin Lower CDM Ineffective Clamps
Gate Oxides
0.065 um Thinner Lower HBM, Metal Heating
Metal MM and
Layers CDM
0.045 um Insulating Low Overall Increased Power Dissipation
Substrates ESD
Page 27
9.1 Scaling Effects on ESD Robustness (cont’d)
As the technologies entered the nanometer range, the ESD sensitivity began to get much worse as much
thinner gate oxides and thinner metal interconnects [16, 17] were both introduced to improve circuit speed.
The thinner oxides result in lower CDM performance and the thinner metals cause heating effects with
increased resistance in the ESD connections, making it difficult to keep the potentials at the IO pad low
enough to meet both HBM and CDM protection requirements. Figure 15 shows that for technologies
starting around 130nm the failure current density of the ESD metal interconnect reduces with the effect
becoming significantly worse as technologies shrink to 65nm and below. The metal bus resistance per
square is increasing which is also decreasing the electromigration reliability margin. This means that
supply / ground bus routing will play an even more critical role in ESD design.
450
ESD Current Density (mA/um)
400
350
300
250
200
150
100
50
0
0 50 100 150 200
Technology Node (nm)
Figure 15 — Metal ESD failure current density as a function of technology node.
20
maximum core supply voltage
oxide breakdown @ 100 ns
15
voltage [V]
10
0
L [µm] 0.5 0.35 0.25 0.18 0.13 0.1
GOX[nm] 15 8 5 3.5 2.2 1.7
Figure 16 — ESD regime oxide breakdown voltage and core supply voltage as a function of scaling
In Figure 16 the core Vdd supply voltage is shown as function of the technology node scaling for feature
size transistor length and gate oxide thickness. Also shown in the figure is the simultaneous reduction of
the gate oxide breakdown voltage under ESD like conditions.
Page 28
At the 100nm node the gate oxide breakdown approaches 5V for HBM stress. This means that any
protection clamp at the IO has to keep below 5V for 1.3A or 2kV HBM. With the metal resistance and
current density limitations as discussed in Figure 15, the design to meet 2kV becomes challenging and
will even become impossible with further scaling.
Another new trend is the “Reverse Poly Effect” where the It2 values unexpectedly decrease with
decreasing poly lengths [16]. Two different explanations are offered: 1) a decrease in the volume
available for heating [18] and bipolar effect coming from merging of the pocket implants [19]. Combined
with the local heating for reduced channel lengths, the introduction of SOI can lead to heating at the
channel surface so much so that even gated diodes can have relatively lower failure current performance.
In addition to SOI now the emerging technologies with multi gate (MuGFET) transistors, also called
FinFET devices, have already indicated extremely low It2 and much more complexity to process effects
[20]. A cross-section of the FinFET is shown in Figure 17.
Figure 17 — Cross-section of an advanced FinFET
The It2 data from [20] for the NMOS FinFET is shown in Figure 18. Note that although the device
triggers as a parasitic NPN the failure current for the device (with an effective width of 50 um) is less than
even 1 mA/um. This suggests that an understanding of the device heating under ESD conditions is
required and that the methods to improve the It2 with layout and device structural changes need to be
understood [21].
Figure 18 — Ids-Vds curves for varying gate bias with trigger voltages (Vt1) and failure currents
(It2) shown for a FinFET device structure [20]
Page 29
The next major scaling effect on ESD is the use of ultra thin oxides in the range of 20 Angstroms which
were first introduced at the 130nm/90nm technology nodes. At such low oxide thickness the commonly
used silicide blocking is not an option, since the increased on-resistance with contact to gate spacing itself
will increase the drain potential during bipolar turn on to breakdown the oxide [22]. Even more severe is
the oxide breakdown voltage which is in the 4-5V range for thin oxides in 90nm and below technology
nodes making it very difficult to design for HBM let alone CDM protection. The statistical nature of the
oxide breakdown mechanism is well known to be a very complex topic for time dependent dielectric
reliability (TDDB), but the TDDB extension to the ESD regime is taking ESD design to the next level of
challenges. As described recently in [23] the ESD regime oxide breakdown varies with process variations
thus making it somewhat unpredictable to design for a given CDM spec. There are other issues with
technology scaling that have not yet been investigated in detail for their impact on ESD. These include the
upcoming strained silicon and the elevated source drain junctions and introduction of metal gates and
high-K dielectrics. It is clear that the newer advances in transistor scaling will continue to have an impact
on the ESD sensitivity up to a point that a completely new direction to the ESD protection strategy may
have to be explored.
9.2 Protection Design Window
The ESD protection design has undergone several changes in strategy according to the technology scaling
effects described in 9.1. Whereas the Field Oxide Devices (FOD) in the early 1980’s, and the NMOS and
breakdown SCR devices in the 1990’s have been extensively used, the current technologies make them
difficult for practical implementation leaving only diode clamps and diode-triggered SCRs as mostly the
available options.
A typical diode and rail clamp based protection concept is shown in Figure 19. Note that the entire
protection performance critically depends on the on-resistance of the diodes, the VDD and VSS bus
resistance values, and the efficiency of the Rail Clamp. This is indeed where the ESD Design Window is
facing its constrictions [24]. The diodes sizes cannot be too large to minimize capacitance at the pad,
while metal interconnects (which are becoming large component of the capacitive loading) at the diodes
has to have minimum resistance to keep the pad voltage to a minimum for the ESD current flow in the
range of 1-2A HBM and 12-15A CDM for very large package devices.
Figure 19 — Common Diode with rail clamp based IO protection strategy

Page 30
9.2 Protection Design Window (cont’d)
Even for standard digital IO designs the voltage buildup at the pads during an ESD discharge can damage
the input gate oxide or the output drain junction, especially if the output transistor has low It2 This can
lead to disappearance of the “ESD Design Window” as reported in [24]. This window is defined by the
available ESD protection design space for meeting a specific ESD target level while maintaining the
required I/O performance parameters (such as leakage, capacitance, noise, etc.) at each subsequent
advanced technology node. It should be noted, however, that the relative margin for window is dependent
on the choice of the protection clamp design. To illustrate the point, Figure 20 shows the design
restriction for the common dual-diode protection device represented in Figure 19. As the technology is
scaled the design for 2 A might vanish at the 65nm node. Or, for 1.3 A (2kV HBM), the window is
reduced by 45%. No matter which clamp technique is employed, further technology scaling will further
decrease the ESD protection window.
4000
I_ESD=1.3A
ESD Window (um )
2
3000 I_ESD=2.0A
I_ESD=2.6A
2000
1000
0
130nm 90nm 65nm
Technology
Figure 20 — ESD Design Window Scaling [24]
These effects are even more severe for designs involving High Speed SERDES (HSS) macros or RF
Linear Amplifier (LNA) circuit applications. The design window severity comes from the constraints
shown in Figure 21. Obviously the protection design must not only not interfere with the operating
voltage it must also have low enough on-resistance to protect the input gate oxide. The thermal failures
from the top come mostly due to the metal heating in the advanced technology devices. As the window
closes the challenge to meet 2 kV HBM or 500V CDM becomes increasingly difficult.
Page 31
9.2 Protection Design Window (cont’d)
ESD Protection Window

Thermal Failure Effects
IESD
IC IC
Operating Safe ESD Reliability
Area Protection Constraints
Window
Operating Clamp Trigger Oxide Break-

Voltage Voltage down Voltage
V
Figure 21. — Technology Scaling Effect on ESD design
To summarize, the severe restrictions on ESD design are coming from:
• Heating of the interconnect

• Low breakdown of the gate oxides
• Reduced failure currents of the advanced transistors
• Lower operating voltages for the circuits
• Very low tolerance to ESD capacitance
9.3 ESD Capacitive Loading Requirements
As Radio Frequency (RF) and high frequency (HF) application data rates and frequencies continue to
increase in each new technology generation there is increased pressure to reduce the capacitive loading
and improve the quality factor (Q-factor) of ESD devices. Quality factor is defined as the ratio of the
reactance in Ohms divided by the resistance in Ohms. In a series RLC Circuit, Q = 1/R * (L/C)0.5, where
R, L and C are the resistance, inductance and capacitance of the tuned circuit, respectively. Note for
example, in a parallel RLC circuit, Q is equal to the reciprocal of the above expression. Figure 22 shows a
generic example of the general trend of ESD HBM levels vs. I/O operating frequencies. The operating
frequency increase is due to the technology scaling and performance increasing as technologies continue
to scale. In the labeled region 1 in Figure 22 the combination of chip level bussing resistance and power
clamp resistance dominates the I/O signal pad clamping voltage thus the signal pad ESD protection as it is
scaled larger (more capacitive loading) has a minimal effect on the overall signal pad voltage during an
ESD event. In region 2 in Figure 22 the signal pad ESD protection network is scaled to reduce capacitive
loading and in this region the ESD protection device itself dominates the signal pad voltage during an
ESD event.
Page 32
9.3 ESD Capacitive Loading Requirements (cont’d)
Figure 22 — Generic plot of ESD HBM Level vs. ESD Capacitive Loading
Figure 23 shows a double diode based ESD protection strategy with the first and last stage of a bi-
direction digital I/O receiver and driver respectively as a sample schematic for further discussion purposes.
Included in the simplified schematic are the Vdd and Gnd bussing resistances, the power supply ESD
clamp and the power supply effective decoupling capacitance. For a double diode based ESD protection
strategy one of the typically used power supply clamps is the RC-triggered MOSFET clamp. For
calculating the I/O signal pad voltage during an ESD event in either positive or negative polarity
referencing any power supply or another I/O signal the power supply resistances on the Vdd and Gnd
sides along with the voltage drop across the power supply clamp itself can be equally as important as the
voltage drop across the I/O signal pad ESD protection devices. In Figure 23 secondary CDM clamps
using double diodes are shown but various types of CDM devices (SCRs, non-silicided NFETs for
example) could be used rather than diodes. Figure 24 shows a similar simplified schematic as in Figure 23
except the HBM double diodes have been removed and a diode-string triggered SCR (DTSCR) is inserted
[26].
Figure 23 — Double Diode based with RC-triggered Rail Clamp ESD Protection Strategy
Page 33
Figure 24 — Diode String Triggered SCR (DTSR) with RC-triggered Rail Clamp ESD Protection
Strategy
Figure 25 shows a specific example of the allocated capacitive loading budget for a High Speed Serial
Link (HSS, also referred to as a SERDES Core) where it is assumed in parallel to the ESD capacitive
loading there are cancellation types of networks (such as t-coil) that cancel out approximately half the
ESD capacitive loading. Figure 25 is just one sample showing the capacitive loading of the ESD device
scaling; the specific results for a given technology may vary. However, the general trend is reduced
capacitance loading budget for the high speed I/O.
Also in Figure 25, the calculated ESD results for diode and SCR based ESD protection (see Figure 24)
concepts are added to the previous curve. As can be seen from the left hand y-axis in Figure 25, as data
rates go from 6 gbits/sec to 12 gbits/sec the ESD capacitive loading budget decreases from approximately
300fF down to 150fF. In the right hand y-axis in Figure 25 the calculated HBM ESD results are shown.
The calculations use high current TLP (Transmission Line Pulse, energy equivalent to HBM) data from a
65nm technology and compare both diode based ESD protection with RC-triggered power supply rail
clamps and diode string triggered SCRs (DTSCR) with RC-triggered power supply rail clamps. In the
comparison the worst case HBM robustness is shown vs. capacitive loading requirements for both types
of ESD protection. The diode-based and DTSCR-based ESD protection networks are two of the most
commonly ESD protection networks used for RF and HF applications. Machine Model (MM) results
exhibit similar trends to HBM.
Page 34

100 2000
90 1800
HBM Protection Level [V]

80 1600
Data Rate [gbits/sec]

70 1400
60 1200
50 1000
40 800
30 600
20 400
10 200
0 0
0 100 200 300 400
Capacitive Loading [fF]
High Speed Serial Link Cap Loading Budget

Double Diodes
Diode String Triggered SCR
Figure 25 — High Speed Serial Link Data Rates and HBM Protection levels vs. Capacitive Loading
requirements
The ESD boundary conditions used to create Figure 25 are listed in Table 3 for reference. The capacitance
values for the ESD devices include both FEOL and BEOL capacitances extracted using an extraction tool
from actual ESD devices designed in a 65nm technology node.
Table 3 — Assumptions used in calculations

Signal Signal Pad RC RC Clamp Decap Rvdd Rgnd Max
Pad ESD ESD device Clamp on- on [ohms] [ohms] Pad
Protection on- turn-on resistance Vdd Voltage
Device resistance voltage [ohms] [fF] allowed
[V] [V]
Double Varies with 0.5 0.5 0 1.0 0.5 4.0
Diode capacitive
loading
DTSCR “ “ 0.5 0.5 0 1.0 0.5 4.0
9.4 Package Effects
Package advances are based on requirements of the different market segments. Generally the packaging
of a particular die in a package, given the same pins are bonded out for different packages, has little effect
on HBM ESD performance. However, the variation of package size, bond wire inductance, etc, does
cause a variation in CDM performance for the same die packaged in different packages. This
development in package types from Dual-in-Line (DIP) to Multi-chip Modules (MCM) and from there to
Flip-Chips and Stacked Die would surely determine the achievable ESD performance for CDM since
during this stress mode the package capacitance plays a very dominant role. The original DIP packages
have pins that are readily exposed to handling making them sensitive to HBM, but modern packages such
as the Ball Grid Array (BGA) have pins that are embedded as well as are closely spaced, making them
much less vulnerable, and in fact may be impossible, to HBM stress. Thus, CDM plays the critical role
for the overall ESD reliability. The critical issues for CDM and the relevant level for safe manufacturing
will be addressed in a subsequent white paper.
Page 35
9.5 ESD Technology Roadmap
The roadmap for ESD [27] projects severe restrictions on the achievable HBM ESD levels as shown in
Figure 26. What should be noted from Figure 26 is that the major impact is expected beyond 2010.
Constraints from the circuit designs such as RF could eventually reduce the practical ESD HBM design
levels into the 100V range. Similarly, the CDM level may get reduced to the 50V range. As a final point,
the Machine Model (MM) scaling is not presented here since many in the industry now consider that with
the exception of very few limited applications, this MM requirement is no longer valid and there is a
strong industry drive to eliminate it for mainstream IC ESD reliability tests.
CMOS HBM (Min and Max Levels) CMOS CDM (Min and Max Levels)
10000 2000
CDM Volts
HBM Volts
8000 1500
6000
1000
4000
2000 500
0 0
1980 1985 1990 1995 2000 2005 2010 1980 1985 1990 1995 2000 2005 2010
Year Year
Figure 26 — ESD Roadmap for HBM and CDM [27]
These trends above will only mean that ESD control in the production areas will become absolutely
necessary. Fortunately, the trends towards packages with very close pin spacings and the much reduced
incidences of human handling would certainly alleviate this threat for HBM. However, CDM ESD control,
which is dependent on the package structure, needs to be carefully considered as the protection levels
would drop with technology scaling. CDM control at the factory should be critically improved.
This clause addressed the scaling of IC technologies and the corresponding increased sensitivity of the IC
chips to ESD. What has become clear is that the view and strategy of ESD has to be changed.
Throughout the electronics industry there have been questions on the validity of the 2kV HBM
requirement given the much better factory controls combined with other factors. There are also questions
such as if the MM achieves any additionally meaningful information, and how the real world CDM
discharge events can be more accurately represented with improved CDM tests, especially for large
packages. ESD will continue to be a major reliability issue and some reasonable protection can be
achieved as long as the nature of the threat is more realistically represented. An important objective
should be to consider appropriate modification of the ESD target requirements enabling designs to meet
the design performance objectives, while maintaining safe ESD target levels that represent today’s
realistic component ESD environments. Meanwhile, more attention should be paid to the more obvious
threats from System IEC, CDE, and TLU as outlined in the ESD Association White Paper [28].
Page 36
10 Differences between Component ESD and System Level ESD
10.1 The History of System Level ESD
System level ESD tests have their origins back in the 1960s. It was noticed that when people sitting in
chairs with castors bumped into a mainframe computer system, the system would have to be reset or
unexpected results would occur. As a precaution, mainframe designers developed ESD tools which would
simulate this waveform and subjected operating machines to these events. As long as the machine was
able to continue to operate, the test was considered successful.
This test was designed to generate noise on the signal line and power planes to ensure that the machine
could continue to operate. It was not designed as a destructive test. How the machine handled the noise
was resolved in many ways, from system packaging (from circuit board layout to cover design), system
architecture (can a system recover from an unexpected signal) to signal to noise ratio tolerance.
As this standard has evolved into IEC 61000-4-2, it has also been extended to more than just mainframe
systems. More equipment is now being subjected to this standard including notebook computers, gaming
systems and mobile phones. While the standard does require discharge to connector pins, there have been
many system level designs that still allowed sensitive devices to exist without problems.
10.2 Differences in Component and System Level ESD Stress Models
Standardized component level ESD stress test routines are defined for the qualification of packaged ICs.
The tests are designed to reproduce failure signatures observed in the IC component manufacturing
environment. All pins of an IC are stressed in a large number of stress combinations during these tests.
The IC is not powered when the ESD stress is applied. Passing these levels in the qualification is intended
to ensure safe manufacturing of the IC in an ESD protected area of an IC manufacturing sites.
System level ESD strongly differs from component level ESD both in testing and resulting failure
mechanisms. Both the stress model and the circuit ESD path environment are different between
component and system level models. There is no strict correlation between IC level ESD robustness and
system level ESD robustness. The procedure for system level ESD testing is described by the IEC 61000-
4-2 standard [29].
The system level ESD stress event, which the system level protection has to be designed for, is an ESD
discharge to an interface pin of a complete electronic system which is touched or operated by the end-user
or during the component replacement outside an ESD protected area. Electro-statically, the environment
is uncontrolled and charging levels beyond 10 kV are possible. The waveform of the discharges varies
over a wide range of pulse duration and currents. For testing complete systems a very flexible system-
dependent set-up is required. This is realized, for example, as a desk top placement of the system under
test and application of the stress pulse by an ESD gun. This method simply cannot be extended to a stand-
alone IC test. In general it is not clear which part of the discharge current applied to the system is
reaching the IC. It is also known that EMI (radiated emissions) can result from the system level ESD tests
which cause unique failures not captured by component level ESD tests.
Page 37
10.2 Differences in Component and System Level ESD Stress Models (cont’d)
Apart from this, the testing conditions of a system require the discharge both while the system is powered
and without a connection to a supply. Due to this, both functional and destructive system ESD test
failures have to be considered differently in comparison to IC component level ESD test failures. For
example, in the case of triggering an IC latch-up event, functional fails might not lead to a physical failure
signature and vanish as soon as the system is reset. Destructive fails often show a larger failure picture
compared to component level ESD failures during testing due to higher pulse energy and possible
subsequent dc stress in a powered system.
To determine the system level ESD robustness, an application board is required, where only a few
system relevant pins (e.g. pins attached to connectors) are connected to the discharge points and are
stressed. Currently a standardized set-up has been discussed which allows an evaluation of the system
level related ESD robustness of an IC using appropriate application boards [30]. Typical target levels for
system level ESD tests are 8 kV or 15 kV. The peak currents at these levels may range from 24 Amps to
45 Amps. A comparison of the test parameters is summarized in Table 4.
Table 4 — Comparison of Component level ESD testing according to JEDEC JESD22-A114-B

(HBM) and system level ESD testing according to IEC 61000-4-2
Component level ESD test System level ESD Test
Stressed pin group Multitude of pin combinations Few special pins
Supply Non-powered Powered & non-powered
Test methodology for Standardized Application specific using
‘HBM’ various discharge models
Test set-up Commercial tester & sockets Application specific board
Typical qualification goal 1 ...2 kV HBM 8 …15 kV
Corresponding peak 0.65 … 1.3 A > 20 A
current
Failure signature Destructive Functional or destructive
Obviously, pins exposed to system level ESD stress require quite a different protection concept than pins
addressing component level ESD to sustain pulse energies orders of magnitude higher than component
level ESD. The protection path can be provided at the printed circuit board level (PCB) (e.g., by transient
voltage suppression (TVS) diodes) or at the IC level. Both economical and technical reasons influence the
choice of the approach and it usually differs from application to application. If it is implemented on a
PCB, it is required that the high current characteristic of the on-chip protection is compliant with the
clamping behaviour of the PCB protection. The achievable system level stress fail level is related to the
effective resistance of the on-chip current path. If resistance is too low in the voltage regime below the
clamping voltage of the protection element on the PCB, the IC will inevitably be destroyed due to the
extreme currents provided during the system level discharge.
Page 38
10.3 Case Studies
To highlight the fact that there is no direct correlation between system level ESD (IEC) and HBM
component level ESD, various case studies are presented. Both cases of products with low HBM ESD
qualification levels, perfectly passing IEC tests and cases of products with 2kV HBM and high CDM and
MM qualification levels failing the IEC test are known. The examples show that the assumption that high
component level HBM leads to high system level ESD is too simplistic. This does not mean that HBM
robustness tests are useless for assessment of the system robustness. However, it has to be applied as pin-
specific testing method and has to be accomplished by a high current IV characterization [31]. The
standard HBM qualification testing addresses completely different failure mechanisms and can only lead
to misinterpretation if compared to system level performance.
A DSP IC processed in a 90nm technology had several pins passing only 500V or 1kV HBM. This IC
also passed the IEC system level ESD tests conducted by the customer. The satisfactory IEC system ESD
test has improved customer confidence that this DSP is production worthy even if the HBM level is lower.
One product designed in 130nm technology had 35% of the pins passing below <500V HBM. It had no
handling issues and additionally it passed 8kV IEC (contact method) test by the customer.
Two different IC designs, both with 2000V component level HBM pass level, showed customer returns.
Neither HBM, nor MM nor CDM could reproduce the failure signature. System level ESD testing showed
the same failure signature as found in the customer returns. The device degradation already occurred at
50V IEC pulse stress [4].
One IC processed in 0.35 µm technology passed 1.5kV HBM at first silicon. The required IEC 61000-4-2
level was achieved at first silicon on the pins exhibiting 1.5kV HBM. After a redesign, the same pins
showed an improved 2kV HBM level in the component level testing but failed the previously passed IEC
level. As explained above, even when the ESD robustness of the modified on-chip ESD protection was
improved, the I-V characteristics of the on chip ESD protection were no better at shunting additional
current to the external, PCB protection element.
10.4 Conclusion
To achieve high system level ESD protection it is not sufficient to design for high component level HBM
ESD values. In a number of cases, this assumption has been proven to be misleading and caused problems.
Sufficient system level ESD protection requires a dedicated, pin specific protection development scheme.
In many cases, joint design efforts between the IC and printed circuit board (PCB) protection circuitry or
at least an adequate IV behaviour of the on-chip protection structure and modelling is required to enable
optimization of the protection at the PCB level.
In general, system level ESD as well as similar pin specific system discharges like cable discharge (CDE)
is considered as the more critical threat for electronic systems [33, 34]. Recently, significant effort has
been made to address this on the level of IC design and testing [35 - 39]. However, these are only the first
steps done. It is recommended by the Council that IC suppliers and their customers focus on this topic in
the future. System level ESD protection is the technical challenge in the field of ESD protection of
electronic systems with proven relevance for application at the end customer.
Page 39
11 Recommendations for New ESD Target Levels
11.1 New Realistic Target Levels for HBM and MM
The preceding clauses have discussed the impact of component level HBM and MM withstand voltages
on the capability to manufacture ICs without yield loss and reliability concerns. It has been shown that,
for over 20 billion parts shipped by various Council companies, the field return rate was independent of
the HBM qualification pass level of the ICs ranging from 500V to 2kV. This database includes more than
600 designs covering communication ICs, consumer ICs, discretes, memory products and automotive ICs
that have been manufactured and placed onto PCBs at a large number of different sites worldwide over
the last 5 years. Therefore the conclusion has been drawn that the overwhelming majority of today’s
manufacturing sites have ESD control measures in place that ensure safe handling of 500V HBM parts.
This includes the full manufacturing flow from wafer technology to testing, mounting and final placement
on printed circuit boards when performed in an ESD protected area.
This experience of robust ESD control measures is seen by all companies contributing data and confirmed
by ESD control companies involved in ESD manufacturing control for these components. As a wide
variety of ESD protection circuits are used in these devices, this result can be considered to be
independent of the detailed protection circuit concept.
The component level HBM and MM qualification standards are made to ensure safe handling of ICs in
ESD protected areas until assembled on a printed circuit board. During this phase of the manufacturing
flow, a discharge between any combination of pins of an IC can occur. To guarantee a sufficient level of
ESD robustness in later manufacturing steps outside an ESD protected area or even when the final system
is handled by the end-user, system level ESD standards have to be applied to the endangered pins which
are outside the scope of this white paper.
Based on the data collected and analyzed, there is considerable evidence that a component level HBM
ESD protection level of 1000V is adequate in modern board assembly conditions and for most end
product applications. This target level includes an appropriate margin to the proven safe level of 500V
HBM.
1000V HBM ensures a minimum MM pass level of 30V, which is fully compliant with existing static
control grounding measures addressing MM in the manufacturing environment. Both HBM and MM
target values are referring to the currently applicable test standards of ESDA, JEDEC and equivalents of
several other standardization bodies. The HBM target is the recommended requirement for products and
generally ensures MM robustness for ESD protected areas. Additional testing according to MM, if desired,
can always be done as a verification of the expected correlation. All levels above 30V have margin above
the minimum requirement.
Comparing a qualification level of 1000V to a 2000V HBM requirement, the industry will benefit from
the following factors:
• No change in quality of ICs seen by the customer

• No change in process yield at the manufacturer
• Faster design cycles of ICs
• Elimination of many unnecessary design respins
• Gain in average time to market of electronic systems
Page 40
11.2 Treatment of Special Pins
Pins which are prone to ESD discharge at the printed circuit board level or in the electronic system
handled outside ESD protected areas have to obey system level ESD standards. These are system specific
and are not covered by device level tests. Component level ESD tests do not correlate to system level tests.
Weak system level ESD has been seen for pins with high HBM levels. The number of IC pins exposed to
system level ESD is system specific. For a high pin count IC there are typically only a few pins that need
system level requirements.
For a long time the ESD requirement for a subset of pins, e.g., RF pins, has been traded off with operating
performance. To account for extreme performance requirements it is foreseen that this will continue in the
future.
11.3 Timeframe for Applying New Recommendations
The data shown in this paper reflects experience with ICs over the last 5 years. The analysis of the ESD
handling capability of the manufacturing site is not limited to a special process technology or generation.
As such, the recommendation is general and valid for any design developed now or in the future.
However, the benefit of applying the new recommendation will be highest for ULSI technologies at 65
nm CMOS and below and complex SoC/SiP designs. In these cases the over-design is in average most
expensive (in terms of design resources and die manufacturing / mask cost) and causes the largest delay.
11.4 Future Cost of ESD Design
While the data support a component HBM qualification shift from 2kV to 1kV, one should further
consider the continued “Cost of ESD”. As described in Chapter 5, this cost is going up exponentially,
burdening the suppliers in delivering the products on time and inconveniencing the customers in receiving
products that have the expected performance. These projected cost curves are shown in Figure 27.
Supplier/Customer Cost of ESD Protection
Cost of
ESD design
l
ve
Le
Dependent on
M
HB
- chip area
- respins
V
2k
- resources l
- circuit performance ve
Le
- time-to-market M
HB
V
1k
dling L evel
Safe Han
Technology node @ 90 nm 65 nm 45 nm
product qualification 2003 2005 2008
Figure 27 — The projected cost of ESD requirements as a function of calendar year and the
technology node, comparing current customer requirements versus lower recommended target and
safe level requirements for handling in an ESD protected area with basic ESD control measures.
Page 41
11.4 Future Cost of ESD Design (cont’d)
It is obvious that if safe handling only requires, for example 500V HBM levels, there has been
unnecessary cost for the last few years and this trend is expected to continually get much worse if we stay
at the existing 2kV HBM levels. What is also evident is that even if the ESD target levels were to be
further reduced to 500V in the future, the cost of ESD design will still go up, albeit more slowly, because
the technology impact will continue to play a significant role. This is one strong argument as to why the
Industry Council recommends immediate lowering of the ESD levels independent of the technologies of
the products that are in production now. Furthermore, this realistic shift will enable more focused R&D
to develop effective ESD protection concepts that can be compatible with the very advanced 32 nm and
22 nm technologies in the near future. Equally important is the much-needed focus in understanding and
developing protection for system level ESD tests such as Transient Latchup (TLU) test and the Cable
Discharge Event (CDE) test. These very important tests for the IC products in their electronic
applications are described in reference [40].
11.5 Product ESD Evaluation Criteria
Following the information given in this document, IC suppliers, IC customers, and OEMs should consider
ESD HBM requirements as summarized in Table 5 below.
11.5.1 Supplier to Customer
With this recommendation in place the ESD negotiation between the customer and the supplier becomes
much more realistic and flexible. As specified in the table, when a product passes 2kV it would mean that
with all basic ESD controls in place safe manufacturing is guaranteed with very good margin, likewise if
it meets 1kV instead the product is still very safe since it would have margin as stated in clause 11.1.
According to this study even if it only passes 500V HBM the product still meets the requirements
adequately and is safe for manufacturing. This minimum HBM level also ensures that 30V MM is always
maintained as illustrated in clause 7. Therefore, this should smooth out current misunderstanding between
the supplier and the customer, and eliminate a lot of the unnecessary waivers. It should also pave the way
for ESD requirements for the advanced technologies that are under development.
11.5.2 Catalogue Products
Catalogue products are often dealt with multiple customers and therefore a classification with a
quantitative number may or may not have much meaning. As explained products passing 1kV or 4kV are
just as reliable. Therefore future classification is suggested below to be adopted so that dubious marketing
competition for catalogue parts is eliminated. With this approach, Product X with 4kV HBM exceeds
requirements, Product Y with 2kV HBM exceeds requirements, and Product Z with 1kV meets
requirements with available margin. Thus the marketability of all three products is equally appealing and
mutually beneficial to both the customer and the supplier while the true circuit performance specifications
remain as the only critical factors for consideration.
Page 42
11.5 Product ESD Evaluation Criteria (cont’d)
11.5.3 Realistic Rating of ESD Qualification Levels
Table 5 — Proposed component level ESD targets after testing according to JEDEC JESD22-A114-
B (HBM)
HBM Level of IC Impact on Manufacturing Environment
2 KV Basic ESD control methods allow safe manufacturing with wide margin.
1 kV Basic ESD control methods allow safe manufacturing with sufficient

margin.
500 V Basic ESD control methods typically allow safe manufacturing.
100 V to <500 V Advanced ESD control methods (ANSI/ESD S20.20 or IEC 61340-5-1)
are required to safely handle sensitive parts.
Important note These proposed HBM levels fully ensure that more than sufficient MM robustness
(>30V) is also maintained with basic ESD control methods.
11.6 Looking Forward
As noted in Table 5, products with HBM ESD levels <500V can also be safely handled but do require
advanced ESD control methods such as the ANSI/ESD S20.20 or the IEC 61340-5-1. The cost of such
advanced control implementation would only be incremental with basic ESD protection already in place.
Factories and production areas must seriously consider moving towards these methods if they are not
already doing so. As we further scale down technologies and develop even faster circuit applications
100V or 200V HBM requirement would not be unrealistic in the future.
Page 43
Annex A (informative) Frequently Asked Questions
Q1- Will the lowering of component ESD levels have an impact on the overall system reliability?
This is an often misunderstood concept. There has been no proof that components with lower HBM
performance have seen more system level failures. As described in clause 11, since the system level test
applies only to external interface pins this protection design strategy is quite different. In fact, as stated in
clause 11, the Industry Council is in agreement that system levels ESD protection, cable discharge
protection and transient latchup are critical areas where future focus is needed.
Q2 - If system level ESD testing does not guarantee system level (including component) ESD
performance, isn’t higher target component level HBM ESD better than nothing?
This would only give a false sense of security while again going through extensive cost of analysis and
customer delays and circuit performance impact. Our data and analyses in clause 7 and clause 11
clearly show that the system level ESD and component level ESD are not related to each other while the
system ESD protection depends on the pin application and requires a different strategy. This document
further argues that system level ESD is clearly important and targeting focus on excess component level
requirements could pull resources away from addressing and designing better system level ESD.
Q3 - Is the root cause of EOS failures related to prior ESD damage?

Do devices with lower ESD levels result in more EOS failures?
Do devices with lower ESD levels result in more system failures?
This is not only an incorrect assumption, but it has persisted for many years. Whereas ESD (1 nS to 1 uS)
is a subset of EOS, EOS events are much longer in time domain (microseconds to several tens of milli-
seconds) and represent orders of magnitude higher thermal energy. EOS failures occur for different
reasons. Several major studies in the past and within corporations have found no linkage between the two
after tracking millions of products. The data presented in clause 7 clearly supports this by showing that
the tracked field returns are independent of their respective HBM ESD level when the devices were
shipped. As described in clause 11, the system level protection requires a different test method and a
different protection strategy. We have seen case studies that showed products passing a system level ESD
of 8 kV based on the IEC System Level test method that were shipped with a corresponding HBM level of
500V or less on a few pins. These lower level HBM pins as well as the 2kV pins all equally performed
well for the IEC test in the total system. Thus, lower component level HBM ESD results do not translate
into poor system level ESD performance as the failure mechanisms & protection schemes are not the
same.
Q4 - While we agree that 1kV or 500V HBM is adequate and safe, how would one deal with competition
that uses ESD as a marketing advantage?
Most customers who are informed, especially through this document, we hope would see that the only
things that matter are consistent circuit performance to specifications and on-time product delivery. As
long as the minimum required component ESD levels (as recommended here) are met, and basic ESD
static control methods are in place, having a product ESD level at 1kV, 2kV or 4kV does not enhance its
system level ESD performance nor its susceptibility to EOS failure causes. For details on these please
see clause 7 and clause 11.
Page 44
Annex A (informative) Frequently Asked Questions (cont’d)
Q5 - Will the modified ESD levels as recommended here shift the burden to manufacturers?
No, manufacturers have an obligation to provide basic ESD controls. Verification of these controls in
their manufacturing and handling processes are necessary no matter what ESD levels are accepted. The
current ESD target levels have been over specified. The modified levels reflect what is realistic and
represent no shift of burden to the customers. Basic ESD controls from clause 5 as well as the substantial
data of clause 7, collected on products shipped to different manufacturing sites with no special rigorous
ESD control programs, strongly support these assertions.
Q6 - Would basic ESD control methods be sufficient to tolerate 1kV to 500V HBM, or would one need
special precautions?
Basic ESD controls ensure that devices with a HBM robustness of at least 500V can be handled safely.
The details are given in clause 5. With advanced ESD controls, such as the ANSI ESD S20.20 and IEC
61340-5-1, even devices with a HBM robustness of 100V can be safely handled with only a minimal
incremental cost.
Q7 - Will all CMs be able to guarantee that there is good control to safely meet these levels?
CMs handling electronic components typically have the expertise in basic ESD control programs. They
are already generally required to provide and verify ESD control programs as a condition for doing
business with their customers. As clause 5 describes, just the basic control methods easily ensure that
devices with a HBM robustness of 500V can be safely handled.
Q8 - If these proposals address only HBM and MM, does it automatically mean that the CDM level
requirements are reduced as well?
It was decided to first focus on HBM/MM and after collecting all the relevant data the proposals for
realistic and safe levels for these models are presented. Likewise, CDM is also critical as a real world
failure mode and in fact poses a more serious threat as a technology and design constraint. There has
not been as much research to understand the real necessary CDM requirements level and how the CDM
tester stress is correlated to the real world events. Also, there is much confusion in requiring a specific
CDM level when it is known that the package dimensions have a significant impact on the stress level as
mentioned in clause 9. The Council intends to study the CDM effects and collect data in order to
recommend a safe and realistic level, and these studies are expected to be completed during early 2008.
Q9 - Why is the Industry Council pushing reduction of the component ESD levels now?
It has been widely observed for almost 10 years that the current requirements, while nearly universally
accepted by customers, have been over specified and that lower levels are very safe. This realization
about the component level ESD is also stated in clause 4. As technology scaling continues and demands
for even higher circuit performance prevail, it has become necessary that these specifications be
reexamined and modified to realistically meet the current practices. Clause 9 outlines the IC technology
changes that necessitated the reexamination of the specifications, while clause 5 describes the state-of-
the-art in basic ESD control methods that support a reduction of component ESD target levels. The
recommendations for safe target ESD levels are given in clause 10.
Page 45
Q10 - Will the automotive lines handling lower 1kV or 500V HBM parts require any additional care?
Since basic control methods should apply equally effectively to all manufacturing lines the automotive
products are expected to be just as safe with no additional special care. Clause 7 illustrates this point by
showing automotive versus non-automotive field return rates versus HBM robustness. However, this is
not to be confused with requirements for IC pins with external interfaces that may be required to pass
high ESD stress coming from the IEC System Level test. This differentiation of the component level HBM
test and the IEC System Level test is discussed in clause 11.
Q11 - Some customers always insist on 200V or 250V MM; how is this relevant to your
recommendations?
As far as the collective Industry Council is concerned, there are no comprehensible situations where 200-
250V MM events would or should occur during product handling in an ESD controlled environment.
From the data in this white paper the Council recommends a 30V MM requirement as more than enough.
We believe that 200V is a legacy issue and has no relevance to component or system level ESD reliability,
whether it is for automotive or any other type of product. This can be reviewed in more detail in clause 6.
Q12 - How do we know there is enough data to convince us that the conclusions are correct?
The Industry Council has gathered enough consolidated data to show that the current component ESD
levels are over specified. The data in clause 7 gives ample evidence for this conclusion, and the basic
static control methods that are universally in practice as described in clause 5 gives a high level of
confidence for these recommendations.
Q13 - There has been an assertion in a recent publication that the Industry Council is rushing in for
modified ESD levels. Is this a valid assertion?
No. The Council has carefully researched the topic, collecting substantially relevant failure rate data,
and coordinating this data gathering with product and quality engineers in our respective companies.
These engineers, focused on continuous improvement in product quality, are equally convinced that there
is enough evidence from their own experience after shipping billions of parts representing a wide range
of product types that the modified ESD levels (as proposed here) indeed do make sense. A sample of this
data is presented in clause 7.
Q14 - If these new recommended levels are safe, why was this not done before?
Both suppliers and customers have been comfortable in the past in meeting the required levels as long it
was not a significant cost / schedule constraint to them. But over the last couple of years, it has become
increasingly difficult to meet these current levels. An inordinate amount of time and resources as
discussed in clause 8 is being spent to go through the complicated ESD testing for large pin count devices
in an attempt to meet the target levels even if the failures had no real world consequences. However, the
bottom line is that by bringing the ESD levels to more realistic levels much of the unnecessary cost of
ESD design can be reduced allowing faster time to market as well as higher performance products to
meet the customer needs. All of this can be attained with no impact to product reliability.
Page 46
Q15 - If we as customers are happy with the status quo why would we want modified component ESD
levels? What is in it for us?
We are quite certain that the IC product customers will gain confidence that these modified component
ESD levels are safe and reasonable. By allowing this freedom to the suppliers the customer can expect
higher performance products delivered with shorter design cycles. In other words, this should be a win-
win proposal. This potential impact is presented in clause 8.
Q16 - Suppliers look at products that are shipped at lower HBM levels and say that no field return failures
are seen. If not all the pins are weak, how can we be sure if this type of data has any relevance?
This is somewhat of a misunderstood concept. The data presented in clause 7 clearly show that the
tracked field returns are independent of their respective HBM ESD level when the devices were shipped;
i.e. most the field returns are due to EOS and not due to ESD. What is interesting from the clause 10 data
is that these failures are not often seen on the pins with the lowest HBM levels. The returns seen with
EOS seem to depend more on the function of the pin on the application board.
Q17 - In the early 70’s the 2kV HBM level was discussed as a possible standard and by the 80’s it
became the widely accepted level, so why would this not be relevant anymore?
The 2 kV HBM / 200V MM target levels were set when not much was understood about the ESD control
methods. These have improved dramatically, as discussed in clause 5. At the same time, the assembly
methods for the IC chips have undergone major changes as well. When basic ESD control elements are
installed, HBM and MM are no longer a risk for the devices in modern assembly lines.
Q18 - If, according to some IP suppliers, they can deliver a cheaper more efficient ESD protection
methodology, would it not make sense to keep the same levels as before?
It does not matter what type of protection device is implemented. In the end, all are limited by the same
physics as outlined in clause 9. The “target level” is not independent of the physics. Independent of the
protection device implemented, if the ESD target level is reduced, then the ESD layout area and impact
on normal electrical performance may be similarly reduced. The problem is then to choose the
appropriate ESD target level. ESD over-design is inefficient and wasteful.
Q19 - Do the modified target levels apply only to technologies of 45nm and beyond?
There is enough data to show that products from 180 nm and above were just as safe and that the design
cycles were considerably affected even more than 5 years ago. Some of this information is given in
clause 7 and in clause 8. At 45 nm and 32 nm and below the problem of ESD cost will keep getting worse.
Q20 - Are these modified target levels only limited to CMOS, or are they independent of the silicon
technology?
IC products in all different technologies may suffer from ESD over-design. Hence there is no reason to
consider only CMOS products. The data in clause 7 gives comprehensive information to illustrate that the
same arguments apply independent of the silicon technology.
Page 47
Q21 - Is the Industry Council making these recommendations to save money for the supplier?
The concept of “ESD Cost” applies to both suppliers and customers. The suppliers go through repeated
testing, debugging, and design respins to meet the existing specified ESD requirements. Both customers
and suppliers go through joint meetings to understand and negotiate solutions for improvements. These
efforts delay the product delivery and can also have an impact on product circuit performance. The cost
curves are discussed in clause 11.
Q22 - There is always a concern to some customers that the Machine Model (MM) is relatively more
important and so how do these HBM modifications affect MM?
First, with some few exceptions HBM and MM are intrinsically related. The details of this are given in
clause 6. Second, all machines are expected to be properly grounded as part of a basic ESD control
program as outlined in clause 5. Our experience with shipped units as reviewed in clause 7 has shown
that lower HBM level products, and hence lower MM levels, are just as safe as 2 kV HBM level products.
Thus, the necessary MM level is provided by a realistic HBM protection level.
Q23 - What about multiple cumulative HBM / MM ESD events which are just below the controlled ESD
levels? Is there a risk at these lower levels of higher field failure rate?"
No. There has never been any data that indicates repeated ESD events occur on the same device at any
level.
Q24 - Given that the recommended new HBM ESD target level is 1 kV, which is half the current 2 kV
HBM target level, why isn’t the MM target level recommended to be initially reduced to 100V which is
half the current 200V MM target level? Isn’t the reduction to 30V MM a little drastic given that MM data
is not included in the same manner as for HBM?
This question first assumed that 2kV HBM and 200V MM are uniquely related. The data in clause 6
clearly contradicts this. Besides, as shown in clause 9, 1kV HBM mostly would correspond to 100V MM
with the worst possible rare case being 30V. Given that 30V is more than safe for MM, there should be
no further concern.
Q25 – If the recommendations from this white paper are based on volume of products failing the 2kV
HBM target levels, why were these shipped in the first place? What drove the customers to accept them?
The answer to this question is varied. In some cases the customers agreed to accept the devices based on
the intention of the supplier to improve these levels at the next design cycle. But as large product volumes
were shipped with the waived levels and no field returns were seen, both the customer and the supplier
gained increased confidence that the 2 kV target levels are more artificial than real. Clause 11 documents
several cases where, for products that were passing 1 kV a tremendous amount of resources had to be
expended to meet the 2 kV level while these same products as reported in clause 7 do not show any ESD
returns but only insignificant unrelated EOS returns. The product and quality groups across the industry
have had numerous similar experiences, and this became a driving reason for revising to more realistic
ESD requirements.
Page 48
Q26 - Does the Council propose to make further reductions over the next five years?
The first objective of the Council is to get an industry wide acceptance for the proposed modified ESD
target levels. As technologies progress even further, it is not unreasonable to expect that eventually the
ESD protection responsibility will shift further away from the IC designers to much better static control
throughout the manufacture and application of IC devices. This also would be consistent with industry
focus shifting to System Level ESD performance over the coming years.

JEP155

Uploaded by

Copyright:

Available Formats

JEP155

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

JEP155

Uploaded by

Copyright:

Available Formats

JEDEC

Recommended ESD Target Levels for

JEDEC SOLID STATE TECHNOLOGY ASSOCIATION

PRICE: Please refer to the current

Printed in the U.S.A.

This document is copyrighted by JEDEC and may not be

Organizations may obtain permission to reproduce a limited number of copies

JEDEC Solid State Technology Association

RECOMMENDED ESD TARGET LEVELS FOR HBM/MM QUALIFICATION

RECOMMENDED ESD TARGET LEVELS FOR HBM/MM QUALIFICATION

RECOMMENDED ESD TARGET LEVELS FOR HBM/MM QUALIFICATION

Special Notes on the System Level ESD:

Special Notes on the Machine Model:

1. W.M. King, “Dynamic waveform Characteristics of Personnel Electrostatic Discharge”, EOS/ESD

7. R. Gaertner; Do we expect ESD-failures in an EPA designed according to international standards?

3 Terms, definitions, and letter symbols

BGA ball grid array

It2: Failure current under ESD time domain.

Node: Within a circuit, a point of interconnection between two or more components.

4 Historical Perspective on HBM/MM ESD Requirements

4.1 Motivation for the HBM Target Level

4.2 Motivation for Introducing Machine Model (MM)

The reasons were:

4.2 Motivation for Introducing Machine Model (MM) (cont’d)

5 Changes and Improvements in ESD and Control Environment

5.1 Historic ESD Handling Procedures

5.1 Historic ESD Handling Procedures (cont’d)

5.2 Global Implementation of ESD Control

a) Little or no ESD control

• Ground and bond all conductors

5.2 Global Implementation of ESD Control (cont’d)

In the following each of these principles is described in more detail.

5.2.1 Ground and Bond all Conductors:

5.2.2 Control Charges on Insulators

Every good ESD control program will do the following:

5.2.3 Use Protective Packaging for Transit and Storage

5.2.4 ESD Control Programs and Resulting Data

A. Little or no ESD control

5.2.4 ESD Control Programs and Resulting Data (cont’d)

Figure 1 — Voltage on a person’s body when walking in a manufacturing environment without

B. Basic ESD Controls

a) The manufacturing operation is confined to a small area.

5.2.4 ESD Control Programs and Resulting Data (cont’d)

C. Advanced ESD Control Program

a) The devices are very ESD sensitive

5.2.5 Advantage of Process Analysis

Figure 2 — Voltage of a person in a manufacturing environment when a conductive ESD flooring

5.2.5 Advantage of Process Analysis (cont’d)

5.3 Change of HBM Hazard Scenario by Increasing the Automation Level

6 Machine Model – Correlation between HBM and MM ESD

6.1 HBM vs. MM

CMM*VMM=CHBM*VHBM -> VHBM = 2 * VMM.

6.1 HBM vs. MM (cont’d)

6.1 HBM vs. MM (cont’d)

6.1.1 Consequence of 1kV HBM Target

HBM failure Voltage (V)

6.1.1 Consequence of 1kV HBM Target (cont’d)

HBM failure voltage (V)

6.2 Exceptions to HBM/MM Ratio

6.2.1 Bipolar vs. Unipolar Stress

CMMVMM=CHBMVHBM -> VHBM = 2 * VMM.