WAVE Report
Photonics West 2002
By John.N.Latta
Wave Issue 0205 1/22/02

January 22
San Jose, CA

The optics industry is bifurcated by era and time: pre fiber optics and
post fiber optics. Unfortunately the post segment hit the wall when the
telecommunications stocks went bust along with the capital spending.
SPIE’s Photonics West 2002 was a mixture of the old and new. Photonics
implies the integration of electronics and optics and there was much of
that on the floor. As we looked over the exhibition is was the optical
equivalent of Toy's R Us. Many optics goodies. As reflective of the SPIE
conferences this was a big optics event. There were 2,600 technical
papers, 100 short courses and 600 exhibitors in two halls.

This was also an international event with papers from many countries and
likewise with the exhibitors. Even the Russian Academy of Sciences had a
booth here, as did the Scottish Optical industry. China and Japan are also
well represented.

We picked up these top-level threads:

Two major markets have created significant opportunity and also injected
research activity into optics: optical fiber communications and optical
playback (CD and DVD).

The technologies of laser emission in semiconductors and diffractive
elements have spawned new optical components of which fiber optics
communications has been largely the beneficiary.

There is a micro optics revolution taking place but it is too early
estimate when and who this will impact industrial or end user products.
Although many of these products are for coherent optics systems, i.e.,
they are diffractive, these are mostly stand-alone components and not
integrated at the wafer level.

In spite of huge investments, the integrated optical circuit is not a
reality but the show illustrates that significant advances have been made.

A major stumbling block to integrated optics remains packaging of the
components into modules.

All of the players, who have had a focus on the telecommunications market,
are looking for other markets and opportunities. A consistent thread from
the floor is that the bottom has fallen out of telecommunications.

Integrated Optics

One of the major challenges in the development of micro optics lies
between the polar ends of on-chip integration vs. module integration. The
latter then becomes a packaging issue and here is where costs climb. In
fact, the statement was made more than once: The package is more costly
than the components. Achieving on-chip integration has been a holy grail
in optics for decades. What optics would like is the same economics of
scale that digital has gained in CMOS with silicon process technologies.
Liking it and achieving it is still the North and South Pole.

Avalon Photonics
We asked about on-chip extensions to VCSELs the answer was YES. The
execution is a little difficult. The fundamental problem is that VCSELs
are based on III V compounds. Thus, the low cost 850 nm VCSELs use a GaAs
substrate. This limits the level of on chip integration in addition to
being a costly process technology. Avalon Photonics is working to extend
its VCSELs with on chip lenses and photo detectors. One of the major
advantages of both is alignment, which is especially important in fiber
systems. The on-chip photo detector makes sense in the telecommunications
applications because the speed performance of the emitter and detector can
be matched. However, for many non-communications applications a photo
detector with GaAs technology is just not required.

Silicon Light Machines
This company, now owned by Cypress Semiconductor, has a Grating Light
Valve (GLV) that is a MEMS structure done on silicon. The GLV is
diffraction grating in MEMS where in the on-state the GLV acts like
mirrors and in the off-state it likes a grating. The advantages claimed
Fast operating speeds
Low insertion loss
Variable attention
Optical angular repeatability based on the silicon process
Scalability and
Fabrication with CMOS process.
GLVs have application in displays. A production part using GLV was shown
in the booth which is used for displays. Laser light for GB hits three
linear GLVs which act as modulators. An HDTV display can be created which
supports 1088 X 1920 with a refresh rate of 250kHz.

Prospects for a Silicon Laser
There are huge advantages to a silicon laser for all the integration
opportunities. On the academic side both University of Trento in Italy and
University of Surrey in England are reported to be close to showing lasers
using silicon. Each operates in region from 1.1 to 1.5 microns. The UK
group has actually shown light emission.

VCSEL - Is there life after 850nm?

VCSEL is a general-purpose laser light emitter? Wrong.
VCSELs have found broad applications in laser/coherent optics systems?

What is the story?

VCSEL (Vertical Cavity Surface Emitting Laser) use has focused on the
telecommunications market. It is an EO converter. Pump in Gb/s and out
comes modulated light. Line up with a fiber and it is the beginning of a
communications system. In discussion after discussion it became clear that
VCSEL technology has not risen above this telecommunications use. Yet,
with the collapse of the telecommunications market virtually all the VCSEL
players are seeking other markets. It is more complex than just shifting
markets with the same products. Honeywell stated that it introduced at
670nm laser to have a product in the visible red. Yet, this did not meet
the needs of the printer market and a 780 nm laser was developed. The
reason cited was the characteristics of paper required a different

The potential of VCSEL is quite broad. Consider the VCSEL as just another
source of light. When put into an array the bandwidth of the
telecommunications goes up by the number of array members. One company
spoke of going to 1Tb/s with only an 8 X 8 array. Yet, an array of VCSELs
can also be used in displays, and there are many issues to be addressed
including tight packaging to accomplish display density. VCSELs in
spectrometers are another application. Yet, an additional application is
measurements and this can include fluid flow, for example. Thus, the use
of VCSELs in high volume applications has not happened yet. As Honeywell
stated in its invited presentation - the era of hypergrowth is over. We
may log 30% growth this year not 200%. To continue to grow the VCSEL
market many changes will take place.

There have been laser edge emitters for some years and these have not
gained the traction expected of VCSELs. VCSELs have the following

Use less power,
Have narrower line width,
Have uniform beam,
Take smaller die area,
More reliable and
Are relatively easy to package into the TO can.

All cited the expectation that VCSELs should be lower cost than
conventional edge emitters. The factors cite above, and especially the
power consumption, favor VCSELs in wireless applications. VCSELs also face
packaging issues due to a lack of standardization and a limited number of
wavelengths, mostly just at 850 nm.

On the Show Floor - Focus on VCSELs

Honeywell gave a paper on its current manufacturing and fab. An appeal was
made in the talk to standardize on a package that is not the TO-46. One of
the reasons given is that this will not scale to the high bandwidths. But
more importantly the industry will gain much by package standards.
Honeywell proposed a new package, which was shown. Honeywell is also
focused on improving the production efficiency. Number pictures were shown
of their fabrication facilities. These were of world class but did not
have the comprehensive look of a foundry fab. One objective they have is
the building of VCSEL products without the use of human hands. An area
where VCSELs have another advantage over edge emitters is the ability to
test them on the die before slicing. Also being developed is the ability
to do burn in again at the wafer level. At the same time these objectives
are accomplished the hand processing of the die or individual components
goes down or is eliminated.

Honeywell also believes that there is the prospect for continuing growth
in telecom as the need for bandwidth extends to the home. However, when
asked if VCSELs have any place in PON, the obvious answer was no. They
responded, but we will serve the metropolitan market. Implying, of course,
that VCSELs will be present in markets that demand active components.

The focus of ULM is to continue to drive the bandwidth with arrays of
VCSELs. They showed a 1/2Tb/s 8 X 8 array. The key to this implementation
was the use of flip chip. That is, all the bonding and leads are out of
the bottom of the chip with the emissions from the bottom side facing up.
I thought this novel in that there is the potential to improve the VCSEL
density, but this was not addressed in the talk. Another advantage would
be die integration and in this design the VCSEL driver was also
integrated. Results were shown that even in a multimode VCSEL the spectral
width was only 2nm. The drive current for a TO-46 can part is 7ma. In
another example, a VCSEL with a 15mw of continuous output had total power
consumption, including driver, of 960mw. The MTBF of an 8 X 8 array was
predicted at 10**7 hours.

United Epitaxy Co.
This is a Taiwanese company moving into VCSELs. They make LEDs and edge
emitting lasers in the range of 650n, to 1550nm. They are now making
VCSELs and sought to show the audience with an invited paper that UEP is a
qualified supplier.

Avalon Photonics
Headquartered in Switzerland this company had made a strategic decision to
focus on just the data communications business. However, 20% of the
business is outside of this and as a result of the turn down they now have
a broader market focus. One market cited is that of measurements but this
is still too early to emerge. From Avalon’s perspective they have seen no
displays surfacing using VCSELs.

Laser Components
This is a distributor of components or small systems applications. They
had a section of the booth with edge emitters and when I asked about
VCSELs they directed me to the other part of the booth. This was the
instrument section and they have just received exclusive NA distribution
rights with a major VCSEL company for the spectroscopy market. However,
even this market has yet to form and they are less certain of other
markets. "All of this is very new."

Fuji Xerox
Tucked away in a small table was Fuji Xerox with table pictures of their
VCSELs. When I asked where are the applications, the response was much to
my surprise. “It is telecommunications.” They originally got into VCSELs
for laser printers, left this application, moved to telecommunications and
are now going back looking at the use in printers. Development remains to
be done but it was not clear how much and when products might surface.

Photonic Products
Photonic Products, located in the UK, is a distributor of conventional
laser diodes. 60% of their business is distribution and 40% is in creating
and manufacturing small assemblies that use laser diodes. When asked about
VCSELs this was a turn off - nothing. They are interested in the red
VCSELs at 625nm but to date no applications have emerged.

Optics Software

Optical analysis and design software has exploded in offerings and
capabilities. Optical analysis programs fit within 5 categories:

(1) Sequential Optical Analysis and Design
Traditional optics with lenses which relies on ray tracing from an object
point to an image point
(2) Non-Sequential Optical Analysis and Design
Illumination systems.
(3) Optics for Communications
A hierarchal stack of tools from semiconductor device design, i.e.,
lasers, to communications link and network assessment
(4) Laser Optics
EM and physical optical analysis for lasers and coherent optics
(5) Optics Visualization
Seeing the impact of optics systems

The best way to address this is on a company-by-company basis.

Applied Optics Research
Applied Optics Research has a laser and coherent optics program that is
widely seen as the best. This software fits within category (4).

Photon Engineering
Incredible. We almost missed this. This fits into the Optical
Visualization category (5). Quite simply this is optical system analysis
by seeing. One can lay out a system and look at the results in the form of
rays or interference patterns. Everything is in 3D and all the structures
can be readily magnified, rotated and manipulated. This is characterized
as software for the small organization that does not do optics design full
time. Optics is put into a visual environment. Its strength is
non-sequential optics (2) but FRED is really not a design program. The
output of FRED would then be used by the more conventional lens and
illuminator design programs. The program is in Beta and for this it costs
$2,500. The final version is due by July and is expected to cost $2,900.
Those that purchase now will receive 1 year of support after the release
and the final version.

Breault Research Organization
ASAP is to non-sequential optics (2) what GLAD is to Laser Optics (4).
That is, when it comes to analyzing illumination systems ASAP is the
software tool. Its median user has a MS or PhD in optics with many years
of experience. However, in another booth a user said - an effective ASAP
user must be one that does this nearly full time. It is a very
sophisticated tool and to use it effectively requires years of experience.
The costs begin at $7k and can reach up to $20k based on the modules

CODE V is the most mature and complete lens design software available. It
is also the most expensive. Only a lease is possible. The cost is
$900/month per seat. CODE V has a comprehensive interface that begins with
a lens table, like an Excel table. Such an interface is common with lens
design programs.

Multiple times in discussions on the show floor ORA came under fire for
their pricing. In fact, ORA was blunt in stating - the lease model
provides us with a continuing revenue stream. It also provides a target
for other lower cost programs that can be purchased.

Light Tools
This is the non-sequential design program (2). The basic program costs
$400/month and for the illumination module it is another $400/month.

For a complete software suite, CODE V and Light Tools, a company could
spend $20,400 per year per seat.

Focus Software
Many times this software was referred to. Based on the demo I received the
user interface seemed as easy to use as CODE V but possibly not as rich.
What is significant is that this program combines both sequential and
non-sequential analysis. The pricing is striking: SE $1,500; XE $2,500;
and EE $3,500. A database of 500 lens designs is available for $300.

Lambda Research
This is the other most talked about lens design program. There are
multiple versions that range in purchase price from $1,000 to $5,000. Fits
into category 1.
The non-sequential design program which has three versions: all reflectors
$2,500; LC version for complex illuminator design $5,000; and the expert
version which will do analysis of LCD panels $15,000. This is a category 2

This program combines non-sequential optics with visualization. After some
booth discussion I can only describe this as a marketing clueless European
company - this is a French firm. They cannot quote a price for the
software unless they know the application in detail. The best I could get
was a price range from $15 - 20k. There is a lens design program also but
they ran out of literature. The price begins at $2k. I have no idea why
they bothered to show up.

Engineering Synthesis Design
IntelliWave is a unique program that is different from our set of 5 above.
It does the analysis of Interferograms. One can image the Interferograms
and it will process from 1 to 12. There are 130 analysis functions.

From the Floor

Corning IntelliSense
Merged with Corning, IntelliSense is a nearly $1b investment in the
company by Corning. IntelliSense is a MEMS only company and has 100
employees. They have fab and tools. The tool is the industry-leading
IntelliSuite software, which they label CAD for MEMS. The software costs
$30k on an annual basis and $72k for a perpetual license. In optics,
IntelliSense describes devices that include: Micromirrors, Silicon Benches
and Waveguided structures. The applications include optical self-assembly,
scanners and beam steering. Another interesting application of MEMS is in
RF components. RF MEMS devices include: capacitors, high-q inductors,
tunable capacitors and switches.

Digital Optics
This is a "reformed" optical telecommunications company. That is, they had
a major play in optical components in telecommunications and the market
collapsed. Now they are selling their abilities in integrated optics to a
broader audience. Digital Optics described their capabilities to create
the Photonic Chip platform. This can include: passive optical components,
such as a silicon bench, diffractive elements, refractive elements,
mirrors and other micro optics, with other active components such as edge
emitters, VCSELs, detectors and electronics all at the wafer level. A
major issue in creating these components is packaging. In the booth they
discussed how the telecommunications market has shifted from performance
as first priority to cost before performance. What cost $5,000 is now
being pushed to $1,500.

Canyon Materials
The product here is HEBS-glass. This is a technology that allows for the
creation of micro optical elements on glass using traditional photo resist
technology. The process begins with an e-beam writing onto unexposed
HEBS-glass. This is then exposed and creates a grey scale whose variations
are on the micron level. The HEBS-glass is used as a mask with photoresist
on the glass substrate below. The resolution will go to .1 micron per grey
level. When a reactive ionic material is used on the photo resist
structure the result is etching on the glass. This leaves a micro-optical
element. Both refractive and diffractive elements can be created. The
examples shown included the following:

Surface profiles to lambda/10;
Variable f number microlenses on the same array of lenses;
100% fill factor, packed, aspheric microlens arrays;
A 60 micron lens with 24 micron surface depth;

Canyon Materials just makes the HERBS-glass. Its core competency is the
materials and the e-beam writing. The cost of a mask is determined by the
amount of e-beam time needed to create it. It typically takes 4 - 6 weeks
to create a mask. There are from 5 - 10 companies who can take the mask
and create the optical elements. The mask once created is suitable for
volume production in that there should be no wear and tear factors in the
production process. Canyon Materials has 10 employees.

MEMS Optical
This company was showcasing its Grey Scale technology that allows for the
creation of many optical shapes including microlenses, beam shapers, beam
splitters and movable mirrors. In the booth was a scanning two axis tilt
micro-mirror. They also had a free space optical path length compensator.
This was all fabricated on a single die, which used a variable height
surface MEMS surface to introduce path length compensation. They called
this Continuous Membrane Deformable Mirrors (CMDM). MEMS Optical has a
9,000 sq ft fab facility that supports both MEMS and traditional silicon
wafer processing. They also do front to back alignment in their
photolithography. They will do complete component fabrication including
silicon components or provide services on wafers that are created in other

Leister Microsystems
They also make microlens arrays and diffractive structures. The array
sizes can go up to 130mm with array elements to 1mm. The f/numbers can
range from .8 to infinity. Materials include Silicon, Fused Silica and
Plastic. They will also create shims for injection molding of the
components. When I spoke with them they were clueless about marketing.
Defensive on talking about anything to do with their capabilities - they
had no "pitch." I watched this scenario unfold with a booth walk-up who
quickly walked away wondering what is the story here?

This is a Finnish start-up that creates micro optics. The array elements
range in size from 20 micros to 100 microns and includes both refractive
and diffractive elements. Elements may be created by e-beam writing or
lithography. They have the ability to create elements on both sides of a
surface. In the booth they were showing an elegant cell phone illumination
application of microlenses. This was a glass surface with many micro
optics lenses, which is used to back light the display. The objective is
to minimize the number of LEDs providing the illumination. This lens array
was lit from the side with the LED and the lens array created a uniform
illumination over the surface of the lens array for back lighting. The
company has a staff of 20.

This is a traditional glass and plastic optics company in Japan. Yet, they
see the potential of micro optics. They have developed technology for
micro optics fabrication and created nano-3D structures in plastic for
communications applications. This was an excellent example of how even a
plastic optics company, with production in China, is moving into these new
optics components.

Founded in 1944 this is a optical component company. They have a facility
in China for fabrication. Extensive capabilities in glass and plastic

This is a Berlin based company that makes, DOE, diffractive optical
elements, which shape and direct beams. The use a polymer to create the
surface profile in glass. The process is inexpensive and stable. The
examples shown in the booth include a circular pattern of dots that come
from a laser that illuminates the center of the element. I would call this
just beam shaping but Holoeye was excited about the market potential

Advanced Microoptic Systems
This is a German company that makes both diffractive and refractive micro
optics. They use chemical etching to achieve deep parts which have
f/numbers to 1.8. The surface characteristics are less than lambda/10.
Production runs vary from 100's to 1,000 units. Pricing of a plastic lens
array can be as low as $10. The company has 21 employees and has been in
business 10 years.

Optical Sciences Center, The University of Arizona
OSC has over taken Rochester as the optics center in the US. I was very
impressed with the materials they had describing the program. Included is
an Industrial Affiliates program that seems to offer some attractive
elements of company participation at the OSC. The program has 100 PhD
students and graduates from 10 - 20 per year.