[labnetwork] LED Lights for Lithography

[martin at algoshift.com]

I am currently working on the development of LED-based lights for 
lithography applications.  I came across this list and was kindly 
allowed to join.

I have a background in high-performance, high accuracy LED-based 
applications going back some twenty years.  In addition to that I worked 
in aerospace engineering, robotics and other work I can't talk about 
(most recently, SpaceX).

At this stage in my new mission I am trying to confirm what I have 
learned in order to start developing a few prototypes for testing.  This 
is what I know and don't know so far:

- Energy above 500 nm should be below 0.001%
- Operator metamerism doesn't seem to be much of a concern in these 
environments (?)
- Outgassing is not desirable (I don't have any kind of a specification 
for this)
- No specification on acceptable flicker
- No specification on required efficiency (Lumen/Watt)
- No specification on the amount of light required, either:
     - Illuminance (intensity of light on a surface, lux) or,
     - Luminance (light energy emitted, lumens)

Frankly, there really isn't very much data out there.  It also seems 
that semiconductor companies keep their lithography illumination 
requirements somewhat close to the vest.  At least this is what I've 
come across.  I wonder if this is because these kinds of specifications 
might reveal process details?  Don't know.

The three main trades I have in front of me at the moment for this 
design are:

- White LEDs with carefully selected film or coating-based filter to cut 
blue + UV
- Green and Red LEDs only, no blue; filtration is still needed
- A combination of carefully selected white LEDs with low spectral power 
above 500 nm along with, perhaps, green and red to enhance; filtration 
still needed

To clarify, the Green+Red LED option still requires filtration because 
green LEDs produce some energy above 500 nm.  If I am to take the 0.001% 
specification to be true, an optical filter would still be required.

White LED's, which, of course, are nothing more than blue LEDs with a 
phosphor coating are the most readily available high efficiency units in 
the market.  Frankly, if high light output at the lowest possible cost 
is a requirement it is hard to beat them with a combination of red and 
green LEDs.  That said, depending on how they are selected, a 
significant portion of the spectral power they emit will have to be 
converted into heat at a filter or bad things will happen in the lab.

The third option involves selecting very warm white LEDs that have 
almost no blue spike.  This means less heating of the filter element 
and, likely, longer life.  This could be an interesting solution.

Plastic film based filters degrade over time, particularly if there's a 
lot of heating due to having too much energy in the undesirable portion 
of the spectrum.  This is where thin film deposition (sputtering?) could 
exhibit far more favorable band-pass characteristics as long a 
longevity.  Cost, of course, could be an issue.

I am very familiar with material out-gassing issues in the context of 
aerospace applications.  Not so for lab usage.  Understanding where 
these limits might lie would be very useful.  The perfectionist in me 
wants to design a T5-class 4 ft LED light fully encased and 
appropriately sealed in a durable glass tube that is both internally and 
externally coated to not pass light above 500 nm.  At the same time, I 
do understand that a real solution has to fit a budget as well as 
technical specifications.  Not sure where that intersection lies but I 
am aware of it.

I introduced a term above that might not be familiar to everyone here; 
observer metamerism.  This is a by-product of the spectral power 
distributions of light, reflection and the human vision system 
interacting in such a way that two colors that are different might 
appear the same (or, in general, you have trouble discerning colors that 
are easy to see under different conditions).  If you've ever tried to 
determine if a steak is well done under a typical white LED light and 
could not, that's observer metamerism.  Mitigation requires "filling in" 
the emitted spectra in areas relevant to the task at hand.

This is why I asked myself this question in the red+green LED case.  
Both of these have narrow emission spectra.  Our brains can function 
with this kind of light and, yes, we will see it as yellow.  However, 
any colors in the portion of the visible spectrum lacking energy will 
become challenging to deal with.  It's like being color blind.  Given 
that lithography labs are already built to work with yellow light, I 
find myself wondering how much of a problem, if any, might be posed by 
observer metamerism in the case of the proposed red+green LED solution.

I think that's the basics at this point.  I would appreciate any and all 
feedback, questions and even a good shove in the right direction.

Thank you,

Martin Euredjian
AlgoShift, LLC
Los Angeles, CA
661-305-9320