Electrostatic
Discharge (ESD): Where are We Heading ?
Steven
H. Voldman
EOS/ESD Symposium 2002 General Chairman
ESD Association Board
of Director
Part
II
In
the last ESD Journal, in the article "ESD: Where are We Heading?" the
question was first measured some of the technical vitality in the ESD industry.
Are we as an industry shrinking or growing? Is there a future in ESD work and
products? What does the Semiconductor International Association (SIA) Technology
Roadmap teach us? How does our ESD roadmap align with the SIA Technology Roadmap?
What should our ESD roadmap look like? Will our ESD technology machine hit the
ESD Technology Wall? Let's keep looking over the dashboard, so we have a sense
where we are going. What is coming down the road? These questions are being asked
by the ESD Association, whose ESD Association Board of Directors have taken steps
to build an ESD Technology Roadmap Sub-committee to pay attention to our industry.
Part
II: Technology and Innovation
ESD
has entered the public domain with the first article published in this year's
Scientific American on the ESD industry. The Scientific American October 2002
article, "Lightning Rods for Nanostructures" rose interest in the public
eye of the kind of issues that the ESD industry will be grappling with - - how
to provide ESD protection in complementary metal oxide semiconductor (CMOS) technology,
silicon on insulator (SOI), radio frequency (RF) technologies, the magnetic recording
(MR) industry, and masks. How can we see the trends? What do we measure ?
A
quick ESD industry metric to show the technical vitality is the activity of patents
and inventions in the field of ESD. For those interested, take a quick look at
what we can see from the US Patent Trademark Office (USPTO) Web site. Go to http://www.uspto.gov,
and proceed with a Boolean Search on ESD. From this, the number of patents per
year are grouped in years. Some interesting trends are evident. Here is what you
will see:
From
the USPTO database, patents are bucketed into the following groups based on the
date of issue: 1976-1980, 1981-1985, 1986-1990, 1991-1995, and lastly 1996-2002.
The data shown in the plot is the number of issues in the groupings based on a
search of ESD, and the logical AND-ing of ESD AND CMOS, ESD AND SOI, and lastly
ESD AND RF. What can be seen is that the trend shows a logarithmic increase of
issued patents in each grouping. In 1976 to 1980 there were 24 ESD patents sited.
From 1981 to 1985, 50 were US patents were issued. In 1986-1900 there was a 5X
increase to 256, and from 1991 to 1995 the number issued was 673 and finally,
from 1996 to 2002, there were 2991 issued ESD related patents. Out of the 2991
ESD patents, 1063 were related to ESD in CMOS technology. In this time frame,
the number of RF-ESD and SOI-ESD related patents were 272 and 195 patents respectively.
Whereas the CMOS ESD patents began at an earlier date, the rate of growth of patents
in other areas are just as rapid, and a key point is that out of the 2991 ESD
issued patents, there is much activity which is not CMOS technology. Interesting
? Yes, very interesting trend.
CMOS
Technology
In
CMOS technology, the nano-structure age has begun where the silicon MOSFET transistor
channel lengths are decreasing below 100 nanometers. MOSFET transistors dimensions
are scaled according to MOSFET constant electric field scaling theory. MOSFET
constant electric field scaling theory scales the dimensions of the transistor
to maintain the same electric field across the oxide film. MOSFET scaling has
reduced the oxide thickness from 1000 A thickness to less than 20 A gate in recent
generations. With oxide scaling, the gate dielectric breakdown voltage decreases.
For reasons of dimensional similitude, the MOSFET channel length and other dimensions
decrease. As a result, the scaling of MOSFETs play a profound role in the ESD
robustness of the MOSFET transistor.
Scaling
and the desire for improved performance has influenced both the silicon devices
and wiring interconnect used in silicon technology. To improve the speed of high
performance semiconductor chips, and to maintain dimensional similitude with the
MOSFET transistor, interconnects are also scaled with the silicon devices and
material changes continue to change. To achieve faster devices, interconnects
have moved from aluminum-based to copper-based interconnect systems to reduce
the resistance. To reduce the line-to-line and coupling capacitance, new inter-level
dielectric (ILD) materials with lower dielectric constants have continued interest.
ESD robustness of the wire interconnect and ILD dielectric are a strong function
of the material melting temperature, stress characteristics, and dimensions. The
material change, wiring hierarchy and architectures of the wire interconnects/dielectric
system have significant influence on the ESD robustness of leading edge high-pin
count advanced technologies.
Yes,
but isn't ESD protection getting easier in CMOS and due to the learning isn't
it getting easier? Yes, CMOS is maturing and the issues to provide ESD protection
to these technologies are decreasing. But - - the new fields are emerging rapidly.
SOI, RF CMOS.. and other logical extensions.
To
maintain performance objectives in CMOS, it is also possible to develop CMOS technology
on insulators to lower capacitance. This can be achieved by using standard CMOS
technology on a starting silicon-on-insulator (SOI) or silicon on sapphire (SOS)
technologies. Many process technology techniques exist to form SOI wafers (e.g.
SIMOX, SIBOND) and SOS wafers. With the development of a mainstream SOI technology,
SOI technology will be integrated with both copper interconnects, low k materials
and features used in sub-100 nanometer CMOS technologies. As SOI MOSFET is scaled,
the buried oxide (BOX) layer will also scale to reduce cost and improve thermal
transport to the substrate material. With the thinning of the silicon film on
the BOX layer, both partially depleted SOI (PD-SOI) and fully depleted SOI (FD-SOI)
will become important. Although it has been demonstrated that good ESD results
are achievable in PD-SOI, continued research and development is needed in the
future for both PD- and FD-SOI devices. As SOI leaves the niche marketplace, it
will be a bigger player with plenty of interesting work needed in ESD.
Radio Frequency (RF) and GHz Technologies
Focusing
on the RF technologies, what do we see? Looking within the RF ESD group, we can
observe that Gallium Arsenide (GaAs) related ESD patents began at an 1985 time
frame due to the long term presence of the GaAs industry - but much later than
the CMOS ESD patents. RF designers did not concern themselves with ESD issues
since RF GaAs products was a low volume niche market. With the growth in the wireless
market, this has driven more innovation and concerns about ESD in GaAs. In more
recent years between 1996 and 2002, the growth of ESD patent activities in the
Indium Phosphide (InP) and Silicon Germanium (SiGe) technology has also increased.
Silicon Germanium and Indium Phosphide technologies are recent arrivals in the
technology arena.
With
the growth of mobility and portability of today's society, radio frequency (RF)
technology and Giga-Hertz (GHz) applications continue to grow at rapid pace. With
the laptop, palm readers, cellular telephones, and the Internet, the wired and
wireless marketplace will have a rapid growth in the next decade. High speed wired
communications for 10 to 40 GHz Ethernet and SONET will fuel the need for RF technology.
With the increased speed of these wired systems, the loading and RF quality of
ESD protection strategies and components will have to be re-addressed. These wired
systems will contain combinations of mixed signal chips, RF CMOS, Silicon-Germanium
technology, Gallium Arsenide (GaAs), RF SOI and optical components. ESD will be
a threat and a concern for high speed circuitry in these technologies.
RF
CMOS
ESD
protection will be a concern in RF CMOS for applications as CMOS moves into the
GHz regime. To achieve high unity current gain cutoff frequencies in transistors,
advanced CMOS technologies will needed to be used. This will require low capacitance
and small MOSFET channel lengths. As a result, ESD devices will be smaller than
standard CMOS technologies. So, the effort and time put into CMOS technology will
be needed as the ESD structures are reduced in size to achieve RF frequency performance
objectives, leading to more ESD challenges. So just as it looked like CMOS ESD
work was finished, RF CMOS will be highly challenged as it chases applications
being fulfilled by its BiCMOS counterparts. Guess what ? There are less options
for ESD in RF CMOS than RF BiCMOS due to less structures and levels….So still
more challenges to come.
Silicon-Germanium
Silicon
Germanium (SiGe) technology will be a strong player in the GHz revolution. SiGe
transistors continue to achieve rapid increases in the transistor speeds pushing
unity current gain cutoff frequencies to 300 GHz. The significant reason which
is fueling the SiGe revolution is that it can be integrated with CMOS technology
in a standard foundry. SiGe applications consist of high speed oscilloscopes,
cellular phones, GPS devices, and high speed wired communication systems. The
understanding of SiGe ESD sensitivity will be of importance as these products
have entered the mobile market. Even today, new devices such as SiGe:SiGeC hetero-junction
bipolar transistors (HBT) are being explored to further extend the application
space and transistor speeds. ESD issues? Yes, work is ongoing and will need to
continue. ESD publications in SiGe have just begun in the last few years with
more work to go !
Gallium
Arsenide
Gallium
Arsenide (GaAs) has advantageous electrical and thermal characteristics (e.g.
higher Johnson Limit) compared to SiGe devices allowing dominance in the power
amplifier and other markets. GaAs will be a significant player in optical communication
systems. GaAs lasers will also play a key role in these systems. As in SiGe technology,
GaAs technology will be extended to GaInP and other Gallium compounds to produce
more efficient and faster devices for the future. Today, the ability to achieve
ESD robust GaAs devices has still been a difficult road. GaAs still has a role
and ESD is very important in places where GaAs chips end up - military, space
applications, satellites, etc…where ESD is important and failure is costly.
What about the magnetic recording industry ?
Magnetic
Recording
Magneto-resistive
(MR) head devices preceded Giant Magneto-resistive head (GMR). From the plot,
it can be seen that ESD related patents on MR heads in the recording
industry
began at the recent discovery of the MR head sensitivity to electrostatic discharge.
The first publication on ESD concerns in magnetic recording heads, written by
Wallash, Hughbanks and Voldman, was released at the 1995 EOS/ESD Symposium. Since
that time frame, a flurry of innovation, patents and publications has continued
in the time frame from 1996 to 2002 as evident from the USPTO patent issue growth.
In the meantime, GMR patent innovation has continued to occur with continued growth
in this field.
Magneto-resistive
(MR) Heads
Magneto-resistive
(MR) read head is used in the hard drive disk industry for storage of information.
The MR head senses a variation in a magnetic field from the disk as it sweeps
over the disk. This signal is translated into a voltage due to the magneto-resistivity
of the thin film magneto-resistive stripe on the TiC wafer. The areal density
of information on the disk continues to increase forcing the size of the MR stripe
to scale with each disk drive improvement. The ESD (HBM) sensitivity of a typical
MR head is 150 V. These thin film stripes do not have ESD protection solutions
because of the physical size of the MR head, cost, and they can not be built on
the wafer due to the lack of a silicon wafer. ESD protection of MR heads is a
continued area of research and growth in MR stripe itself, the armature, head
gimbal assembly and other components of the disk drive system. Areas of research
consists of understanding of magnetic phenomenon, thermal physics, magneto-thermal
research, electrostatics, design characterstics, failure analysis, and failure
mechanisms. A lot of work ahead !
Giant
Magneto-resistive (GMR) Heads
Giant
magneto-resistors (GMR) heads are devices that follow the MR head designs to improve
signal improvement with the decreased areal density of the disk. GMR devices utilize
a spin valve (SV) for initialization of the GMR head. GMR heads are scaled in
comparison of MR heads leading to an increased ESD sensitivity (e.g. 30 V) . GMR
heads experience "spin valve reversal" as a result of the ESD event,
causing de-initialization of the magnetic dipole alignment. This is followed by
MR melt damage at higher currents. Research, design and understanding of the ESD
sensitivities of GMR devices from EOS, ESD and EMI is ongoing today.
What
about the future MR devices ?
Tunneling
Magneto-resistive (TMR) Heads
Tunneling
MR (TMR) head are the next generation which utilizes multiple control layers allowing
the quantum tunneling of current through a thin film in a transverse fashion.
ESD sensitivities of TMR heads have been shown to be below 10 V HBM protection
levels. Significant research will be needed to evaluate the feasibility of manufacturing
and ESD protection of these elements. Additional to this device, there are a significant
number of alternative MR heads for the future. It will get harder to manufacture
these structures in the future, leading the need for more manufacturing solutions
in the equipment, garment, packaging and tooling sectors - - from tweezers to
tools.
So, what
does this tell us ? It shows that innovation and growth in this field is continuing
at a rapid pace. There is a lot of highway on the ESD Technology Roadmap to still
do business. Many of these new fields are just beginning!
