>>> 'ellips' problem from hebert -- Tue Jan 24 17:39:17 1989 <<<

Nf for steam ox (thermal oxide) is 1.538 to 1.560 by ellips.
It should be 1.46. It appears that there may be a problem with 
the ellipsometer.

>>> 'ellips' problem from hebert -- Tue Jan 24 17:50:10 1989 <<<

Misaligned polarizer is causing Nf of thermal oxide on silicon to
be 1.54 to 1.56. It should be 1.46.

>>> 'ellips' problem from hoagland -- Thu Jan 26 10:57:54 1989 <<<

problem:  User reports measured Nf of SIO2 at 1.53 rather than
          standard 1.46.

cause:  Unknown.  Reported misalignment of polarizer was in error.
        Polarizer optics are sealed and polarizer drum is locked
        into position by set screw.  Optical orientation test revealed
        no problems.  Possibly stage height was mis-set prior to
        measurement.

solution:  None.  Measurement on lab standard showed Nf of 1.458 .

>>> 'ellips' problem from chenwei -- Thu Feb 23 17:41:17 1989 <<<

?

>>> 'ellips' problem from jchen -- Wed Mar  8 17:49:55 1989 <<<

sorry, hit the wrong key.

>>> 'ellips' problem from carlos -- Wed Mar 15 05:04:39 1989 <<<

It cannot read Nf properly. Please someone look at it

>>> 'ellips' problem from bill -- Thu Mar 16 10:01:06 1989 <<<

I tried the ellipsometer and measured films that I
had measured last week - I obtained almost the exact same
results.
I looked for the ellipsometer "standards" so I could check these
also - but could not find them in the ellipsometer drawers.
Perhaps Steve Hoagland knows where the "standards" are.
I suggest the ellipsometer is working fine.

>>> 'ellips' problem from hoagland -- Thu Mar 16 16:47:42 1989 <<<

Measurements on lab standards of oxide show consistent readings
5-10% above actual thickness.  Further investigation of a potential
problem here will continue Friday morning and results will be posted.

>>> 'ellips' problem from hoagland -- Fri Mar 17 12:44:39 1989 <<<

Testing thus far on the ellipsometer is inconclusive.  Measured
Nf values appear low for oxide and nitride films, but standard
values seem to yield reasonably accurate film thickness readings.
More troubleshooting will continue Monday- users may use this
equipment at their own risk in the meantime.

>>> 'ellips' problem from hoagland -- Tue Mar 21 11:38:41 1989 <<<

problem:  Film thickness readings measure consistently high by 5
          to 10 percent.  Film Nf values measure consistently low.

cause:  Unknown. 

solution:  After rebooting system program off hard disk measured
           thickness of oxide standards was correct.  The keyboard
           cursor "wheel" appears inoperative but use of ellipsometer
           and CV station are unaffected.  Subsequent rebooting of
           system off floopy disk also produced correct measure of
           film thickness standards.

>>> 'ellips' problem from dcarl -- Tue Mar 28 09:20:59 1989 <<<

Well, not REALLY a problem yet - the paper is running out on the HP
print-out device.  It needs a new reem.  So, I thouhgt that I would report
it. Otherwise, the ellips seems to be running fine.

>>> 'ellips' problem from dcarl -- Thu Mar 30 17:59:07 1989 <<<

PAPER IS OUT !!!!

>>> 'ellips' problem from bob -- Wed Apr  5 13:05:31 1989 <<<

The paper has been installed in the HP printer of the ellips. A good 
stock is on hand and it is now in our inventory database. We should 
not run out of stock again.

>>> 'ellips' problem from hoagland -- Thu Apr  6 16:24:28 1989 <<<

I would like to remind all users of the ellipsometer that it
is CRITICAL you adjust the stage height at the same time you
are rotating the analyzer drum to achieve a maximum reading
of beam intensity.  This "maximizing" procedure is outlined in
section 2.F of the lab manual.  In essence, it is necessary 
to raise or lower the stage after rotating the analyzer to 
a peak, to achieve the greatest beam incidence on the analyzer
objective.  Failure to do this results in film thickness readings
which are absolutely incorrect.

>>> 'ellips' problem from hebert -- Mon May  1 12:36:58 1989 <<<

Floppy disk (712-AK) for SC6A, SC7A, and GC5A + SubCA appears to be
damaged. When I try booting, I get "ERROR 88 Read data error."

As a result, I can't measure the thickness of my film.

>>> 'ellips' problem from bill -- Tue May  2 13:22:32 1989 <<<


Re : Dave Hebert's problem of 5/1/89

I sent Dave a message describing loading from the hard disk
- so floppy damage would not prevent operation

However, today I tried to load the two film program
( which is not on the hard disk ) from a different
floppy and received the same error message
"ERROR 88 - Read data error ".

I received the same message from a number of floppies.

All the above load attempts were made from the right hand 
disk drive = ( Internal,4,0 )

But all programs loaded successfully from all floppies
when loaded from the lefthand disk drive = ( Internal,4,1 ).

-------------

The problem seems isolated to the right hand disk drive
and is intermittent -
when I tried to demonstrate the problem to Steve Hoagland
both drives worked fine !

>>> 'ellips' problem from hoagland -- Wed May  3 10:49:30 1989 <<<

For the time being please use drive 1 for disk reads; internal,4,1.
I will service drive 0 drive ASAP.

>>> 'ellips' problem from sylee -- Thu Jun  1 14:41:20 1989 <<<

Ellips did not give a reliable reading. We tested the SiO2 thickness
using a standard. The thickness was about 1000 A off from the expected
value which is 3000 A. The standard was recheccked by nanospec and found to be 
O.K. Ellips need to be cheched for accurate measurement.

>>> 'ellips' problem from bob -- Thu Jun  1 15:02:53 1989 <<<

The problem reported for the ellips was a result of the condition
of the sample. The ellips is up and running.
bob

>>> 'ellips' problem from sylee -- Thu Jun  1 15:06:49 1989 <<<

The previous problem, i.e. reliable measurement, was investigated by
Tom using another standard wafer and using fixed refracteive index.
The measurement was pretty reliable and repeatable. However, variation
in measurement readings were found using non-fixed Nf and thickness mode.

>>> 'ellips' problem from jimmoon -- Fri Jun  2 10:47:22 1989 <<<

Repeatability is not as good as usual.  Usually, repeated
measurements at a fixed point on the wafer are within an 
Angstrom or two.  Now it is varying by 10-20A.  Time for some
sewing machine oil, Steve?

>>> 'ellips' problem from dcarl -- Fri Jun  2 15:46:31 1989 <<<

I didn't see the non-reproducibility (sp?) on thermal oxide films or
on plasma CVD nitride films.  I re-measured an oxide from quite some
time ago and got the same answers.  So, I don't know.  Use at your own
risk.

>>> 'ellips' problem from hoagland -- Fri Jun  2 17:10:39 1989 <<<

After users complained of inconsistent readings the laser polarizer
shutter was oiled as a precaution against sticking.  This did not
prevent subsequent inconsistent readings(~5%).  The  ellipsometer
is up, but should not be used for critical measurements.  further
efforts will be undertaken Monday to isolate the cause of these poor
measurements.

>>> 'ellips' problem from hsin -- Fri Jun  2 22:55:40 1989 <<<

Measure SiNx film on GaAs, results not consistant each time on the 
same point. Nf vary from 1.95 to 2.04 from run to run on the same
point.( Thickness vary from 4800 to 5150 A AT THE SAME POINT ).

Wei Hsin

>>> 'ellips' problem from dcarl -- Mon Jun  5 09:44:44 1989 <<<

Sporatic readings - sometimes the readings for Nf and thickness are
very tight for the same point on a SiNitride film, sometimes they are
off by up to 20 Angstroms and 10% Nf.

>>> 'ellips' problem from hoagland -- Mon Jun  5 16:46:38 1989 <<<

Some tests were performed on the ellipsometer today to isolate
possible causes of recent fluctuating measurements.  A power meter
placed directly in the path of the laser beam indicated relatively
stable power levels, at least over the short duration of several
measurements.  The analyzer drum was manually rotated to verify no
pulse counts are missing during drum rotation- a nominal 72/rotation
were observed as expected.  A lead was attached to the pulse count test
jack and the signal monitored with a scope as the drum rotated at normal
speed; considerable jitter was observed, which could be an indication
of trouble. Tomorrow the drum jitter will be looked into, as well as
the condition of the photodetector.  Hopefully the ellisometer will
be restored to it's normal repeatability by Wednesday.

>>> 'ellips' problem from hoagland -- Tue Jun  6 14:38:33 1989 <<<

According to Gaertner, pulse jitter is normal in the analyzer.
Checks on the photo-detector power supplies likewise turned up no
problems.  Gaertner felt the most likely cause of our fluctuations
was a bad laser power supply, which they claim cannot always be
detected by a power meter.  After leaving the analyzer drum running
over the lunch period and taking additional readings however, repeated
measurements at the same point inexplicably stabilized to within 3
or 4 angstroms.  Gaertner claims no warmup of the analyzer assembler
should be necessary and that our unit probably has a weak power supply.
I was told schematics for the ellipsometer were not available, but Bob
will nonetheless attempt to persuade the company to provide them.    

In the meantime, my advice to users is to switch on the analyzer drum
15 minutes before taking any critical measurements.  This should
insure relalively repeatable numbers for most ranges of thickness.
If schematics are ever available, some effort will be made to further
troubleshoot the chassis power supplies.  The ellisometer is up and
available for use.

>>> 'ellips' problem from moazzami -- Thu Jun  8 14:38:54 1989 <<<

Similar to previous problems:
Using the general program to measure the thickness of a thin SiO2 film
and the refractive index:  thickness readings varied from 90A to 170A,
refractive index varied from 1.3 to 2.4.
With fixed refractive index, thickness measurements were more repeatable
(+/- 2A) but still not as good as it used to be ( >6 months ago).
Will this problem ever be fixed or will we just have to go on faith
from now on?

>>> 'ellips' problem from bob -- Fri Jun  9 07:44:01 1989 <<<

There are problems with repeatability of measurements on the ellips. 
These problems can be greatly reduced by allowing the machine to run for
15 minutes prior to use. Steve Hoagland has schematics coming to trouble-
shoot the photodetector/power supplies. It required a lot of negotiating
to get Gaertner to send the schems as they coonsider them propritory and
we have to sign a non-disclosure agreement. The ellips will be worked
on when the schems get here which should be early next week.

>>> 'ellips' problem from rwilliam -- Fri Jun  9 16:51:15 1989 <<<

It's having real problems.  Even though
-I had it running for ~15 minutes
-films were a thousand A away from period
it STILL gave be ludicrous results...

Running a silicon nitride expected to be ~200 A thick, it said
Nf=1.373
thick= 1973

(yes I had the laser shutter open)

Another user used a different program (not autost) and
got similarly afwul results- Nf 1.38, not 1.9 as expected,
and thickness 8130, not 600 as expected.

Ellips is NOT HAPPY...

>>> 'ellips' problem from moazzami -- Sat Jun 10 17:59:28 1989 <<<

See previous problem reported by rwilliam

Ellipsometer is not measuring anything.  The thickness and refractive
index are the same regardless of film material/thickness, etc.
measured psi and del are ALWAYS the same which explains why the thickness
and refractive index are always the same.

>>> 'ellips' problem from hoagland -- Mon Jun 12 16:50:28 1989 <<<

The problem reported this weekend with repeated identical
measurements was corrected after connectors to both the HP computer
and ellipsometer were removed and reseated.  Repeated testing 
with a standard however revealed that the problem with bad
repeatability still persists.  Even with a fixed Nf thickness 
measurements varied by as much as 15 angstroms on a 1000A 
standard.  The back panel was taken off the instrument and both
+/- 15 volt and 24 volt power supplies tested good.  Gaertner
suggested next scoping the output of the photo detector, but
noise on lab's earth ground (20MV avg) prevented any useful
measurement.  A DVM however showed no drift whatsoever(<1 MV)
with the laser blocked, so that is probably not the problem.
I will consult further with Gaertner on this tomorrow, but
their recommendation at this point is that we put in a call
for field service.  Information will be posted as available.
The ellipsometer is up for use, but may not be suitable for
very accurate measurements.

>>> 'ellips' problem from hebert -- Tue Jun 13 17:03:28 1989 <<<

The ellips doesn't give accurate thickness and Nf data. It also
does not give repeatable data. I measured Nf and thickness for
thermal oxide on silicon ten times at the same place on a wafer
(i.e. the wafer was held down with vacuum and not moved at all
between measurement). Below is a summary of the thickness and
Nf data. The thickness is 3212A by the Nanometrics, and the Nf
is expected to be 1.45 for thermal SiO2.

SUMMARY OF ELLIPS DATA FOR SiO2:

Thickness (A)        Nf
3029             1.517
2808             1.593
3006             1.524
3031             1.516
2936             1.547
2946             1.543
3133             1.486
3002             1.526
2947             1.542
2964             1.539

The ellips clearly isn't working properly. It is a valuable lab
instrument that should be repaired as quickly as possible. Those
people who are doing CVD oxide and nitride should beware! 

>>> 'ellips' problem from hoagland -- Wed Jun 14 16:40:15 1989 <<<

Work on the ellipsometer has been hampered by a lack of spare
parts to swap out.  Dave Hebert has graciously agreed to allow us
use of parts from his ellipsometer in an effort to further isolate
a cause for our machines lost repeatability.  Work will continue
on this tomorrow and any further information posted as available.

>>> 'ellips' problem from hoagland -- Thu Jun 15 16:40:49 1989 <<<

Swapping another known good power supply for the laser failed to
correct it's measurement problems.  An alignment prism also failed
to reveal any problem with alignment of either the laser or analyzer.
Gaertner will be contacted tomorrow to arrange for field service at the
earliest opportunity.  In the meantime the ellipsometer is still up
with somewhat reduced measurement accuracy.

>>> 'ellips' problem from hoagland -- Fri Jun 16 09:23:50 1989 <<<

Gaertner has been contacted for field service of our ellipsometer.
The local technician was not available this morning but we have
been assured he will be available early next week.  The ellisometer
should be up to full specs by Wednesday next week.

>>> 'ellips' problem from hoagland -- Mon Jun 19 13:13:47 1989 <<<

Peter Brown of Gaertner worked on our machine for 3 hours this
morning and determined the likely cause of our measurement errors
was a faulty chassis or "motherboard".  He will return tomorrow
to isolate the failure to a component level using his swap board,
and the unit will hopefully be returned to service in the afternoon.
At some point the scanner motor brushes will also need replacement
(current ones are badly worn) which will require another recalibration
as well.

>>> 'ellips' problem from hoagland -- Fri Jun 23 17:02:25 1989 <<<

Somehow the last report on the ellipsometer was directed to
an erroneous destination- sorry.
                        Repairs on the ellipsometer were
completed earlier this week with replacement of the entire 
electronic chassis.  A complete calibration of the instrument
was also accomplished resulting in better measurement accuracy
now than has been available for several years.  At some point
later this year brushes on the scan motor will need to be replaced-
The current ones are badly worn.  Notice will be given far in advance
of this maintenance procedure.  Measurement accuracy and repeatability
are excellent; the ellipsometer is up and available for use.

>>> 'ellips' problem from dcarl -- Wed Jul  5 09:52:29 1989 <<<

Out of paper - that's all.

>>> 'ellips' problem from hoagland -- Wed Jul  5 11:15:36 1989 <<<

Paper has been restored to the ellipsometer printer.

>>> 'ellips' problem from han -- Wed Jul 19 16:04:26 1989 <<<

The vacuum holder is malfuntioning.  It's hard to lift the wafers even after turning off the toggle valve.

>>> 'ellips' problem from bill -- Tue Jul 25 15:43:32 1989 <<<

The stage height adjustment wheel no longer moves the stage.
The wheel turns freely on it's threaded stem -
rather than move the stem up and down.
The wheel does not seem stripped at all,
but perhaps has lost a lock nut of some kind.

I did not have this problem the last time
I used the ellips, and no one has enabled it since,
but a new ellips program has been loaded into the hp...
I'm just complaining - I wish users would report problems
rather than discover them and leave them unannounced.

>>> 'ellips' problem from hoagland -- Wed Jul 26 10:44:24 1989 <<<

problem:  User reports height adjustment screw on ellipsometer
          stage not working- screw turns but stage height remains
          unchanged.

cause:  Stage was moved completely forward to facilitate removal
        of wafer following measurement.  In this position stage 
        cannot be lowered because stage height adjustment post
        does not clear interior surface of X-Y stages.  During
        normal stage movement stage post moves within inside
        cavity of stages.

solution:  Move stage inward to normal range of measurements-
           about +/- 1 in. of stage center.

Another user reported different measurements across wafer comparing
manual displacement of wafer over stage verses vernier adjustment.
A check of stage level verses position on both axises should no
discernible change, but that individual user will be contacted for
more information.  In the meantime an effort will be made to better 
level the entire ellipsometer chassis.

>>> 'ellips' problem from hoagland -- Wed Jul 26 14:42:42 1989 <<<

During efforts to pinpoint a cause for the recent stage
problem some screws were inadvertently lossened on the adjustment
stem which might have effected the stage orientation.  Proper level
of the stage, to very exact tolerances, is required for accurate
repeatable measurements.  Gaertner will be contacted tomorrow(no
techs in now) to determine if a releveling of the stage is necessary.
If it is, the work can be done in-house and the machine should be
back up by Friday.  In the meantime the ellipsometer is working fine
for all but the thinnest films and still remains usable.

>>> 'ellips' problem from hoagland -- Wed Jul 26 16:09:09 1989 <<<

I spoke with Peter Brown of Gaertner this afternoon and he indicated
that in the worst case our stage level is likely only altered slightly-
with almost no impact on measurements.  Nonetheless, we will still take
the machine downstairs to the machine shop ASAP to verify positioning of
the stages just to be certain.  He also said that for thicknesses of
nitride over oxide in the hundred angstrom range, a measured refractive
index span of .1 to .3 is acceptable and in fact quite good for most
ellipsometers.

>>> 'ellips' problem from voros -- Thu Jul 27 07:56:38 1989 <<<

The ellipsometer has been repaired and it is up for use.
Katalin

>>> 'ellips' problem from moazzami -- Thu Jul 27 11:50:55 1989 <<<

I tried to use the ellipsometer to measure the thickness and refractive
index of thin (<200A) oxide films (both thermal and LTO). I was
using the general recipe.  The measured values were way off of the
expected values: unexpectedly high thicknesses and low refractive indices
(~1.26 rather than 1.46, note that this includes THERMAL oxide on bare silicon).However, for films thicker than 200A, I was able to get reasonable values.
We have been able to get accurate results for films thinner than 200A before
(see Bill Flounders' comments.)
The ellipsometer used to be a valuable instrument.
Please check into this problem.

>>> 'ellips' problem from hoagland -- Fri Jul 28 16:53:14 1989 <<<

Efforts to correct apparent tilt of the sample stage proved ineffective.
Since the tilt is not in fact linear across the entire stage, it appears
more likely the linear bearing raceway for the Y axis was bent recently
when a user cranked the micrometer all the way out to remove a wafer(using
both palms) and pulled against a stop.  Please use 1 hand to turn the
micrometers!! We cannot repair the stage, but a series of measurements will
be taken to find the best X and Y micrometer settings so accurate
readings can be achieved.  This only effects films under 200 angstroms
thick.  

>>> 'ellips' problem from bill -- Mon Aug  7 10:24:00 1989 <<<

HP computer is unable to access the hard disk.

I entered a basic system and the ellips programs to the hp
from the backup floppies -
so the Ellips is UP.

but the hard disk access needs to be restored.

>>> 'ellips' problem from hoagland -- Mon Aug  7 17:15:03 1989 <<<

problem:  User reports PC cannot access hard disk.  
          All efforts to load programs on HD fail.

cause:  Unknown.

solution:  Power off hard disk for 20 seconds and restart.
           PC accesses HD successfully- the ellipsometer is
           up and available for use.

>>> 'ellips' problem from hoagland -- Wed Aug  9 09:13:28 1989 <<<

Bill Flounders has been extremely helpful in characterizing the
very thin film performance of this machine both before and after
recent damage to the sample table.  Based on the refractive index of
a <200 angstrom nitride film, the micrometer settings which yield
the best measurement consistent with previous numbers is;
X-axis : 8 on the left micrometer scale
Y axis : 4 on the outer micrometer scale(towards user)
In all cases the only films affected are those under 200 angstroms.
To take measurements across a sample leave the table stationary and
manipulate your wafer with tweezers.

>>> 'ellips' problem from dcarl -- Thu May 10 18:03:53 1990 <<<

out of paper - all else seems to be fine.

Dan Carl
