Dear Dr. Paneva,

[....]

Simply monitoring the transmission of your samples at a known wavelength and
using the absorption coefficient to evaluate the sample thickness is not a
good method. It is too difficult (unless you have the right equipment) to
account for reflection losses.

For thin samples (<10um) try using your samples as an interferometer in a
region of the optical spectrum where Si absorption is flat and low. Place
the sample at normal incidence to the source, in a IR spectrophotometer. By
recording the optical transmission spectrum over a suitable wavelength
interval, a series of maxima and minima in transmission will be observed.
Given that a max/min in transmission occurs when the condition
(4*pi*n*d)/lambda = m*pi is satisfied, a simple plot of 1/wavelength vs. the
max/min order (m) will yield a slope of 1/(4*n*d). Something to this effect
anyway, it is very easy to evaluate. The refractive index, n, of the silicon
can be located in any optical properties book (assuming you chose a flat
spectral response region to scan). Hence you can evaluate d, the sample
thickness.

Another way to do this for thin samples is to use the sample a  fabry perot
interferometer. With a laser (quasimonochromatic light) of a suitable
wavelength incident on the sample at normal incidence, record the
transmission of the sample as a function of incident angle. This means
rotating the sample, with the detector on the far side of the sample,
aligned with the laser light. If the sample is optically thicker than the
wavelength of the laser, the graph of Transmission vs. Angle will look like
an Airy Function. By fitting this function,  the optical thickness of the
sample may be determined. From a knowledge of the dispersiver properties of
the Si, the refractive index can be obtained at the wavelength of the laser,
and thus the geometrical thickness can be determined.

Yet a third method for thinner samples is to use m-line spectroscopy -
coupling light into the Si film so that it guides. This can be done by
endfire coupling or the preferred method for this technique - prism
coupling. By observing the angular position of the mode on a screen, the
thickness AND refractive index of the sample may be determined (for
multimode guiding) with a high degree of accuracy.

        I have personally used ALL these techniques successfully in polymer
thick
and thin film devices. The method will also work for semiconductors - just
refer to the linear optical constants of Si before proceeding. For the
thicker samples (> 50um, a good micrometer should suffice.

If you require any further details, please do not hesitate to contact me.


        Dr. Karl H. Cazzini

        Conifer Group & Associates,
        116 Cochituate Road,
        Framingham MA 01701
        USA




-----Original Message-----
From: mems-mgr@ISI.EDU [mailto:mems-mgr@ISI.EDU]On Behalf Of Dr.
Roumiana Paneva
Sent: Tuesday, November 10, 1998 3:02 AM
To: MEMS@ISI.EDU
Subject: Si membrane thickness measurement


Dear colleagues,

I am looking for an equipment for precise nondestructive thickness
measurement of Si membranes with thicknesses from 1um up to about 20 um. The
membrane size is larger than 200um. The equipment should allow measurements
on 6" (and 5") wafers, has a x-y-table for measurements over the wafer and a
spot size smaller than 200um. I would prefer to measure the real thickness
of the membrane and not to use a profilometer or change of microscope's
focus to calculate the membrane thickness from the wafer thickness.
Information about IR spectrometer with a spot size smaller than 200um or
another type of measuring system is deeply appreciated.

Dr. Roumiana Paneva

        X-FAB GmbH
        Haarbergstasse 61
        D-99097 Erfurt
        Deutschland
        Tel: (+49) 361 42 053 21
        FAX: (+49) 361 42 053 11