Grinding and Polishing Infrared Optical Windows

2. Grinding and Polishing Infrared Optical Windows

Patrick Hendra
Dept of Chemistry,
University of Southampton,
SO17 1BJ,

We started IJVS with an article [Int. J. Vib. Spect., [] 1, 1, 2 (1996)] by Geoff Dent telling readers exactly how to make good KBr disks and what to look out for if they had done a job badly. The article has proved very popular because as we all know it’ far easier to make a lousy KBr disk than it is to produce a good one.

Some time before Christmas I encountered an undergraduate trying to record a spectrum of a nujol mull in the organic teaching lab. The spectrum was hideous for two reasons – the student had no idea how to make a decent mull and the two KBr flats she was using looked as if they had been through hell and high water – probably literally. Cross-examination revealed that she had not been taught how to make a mull – article in a future edition – and the “technicians polished the flats”. So – I thought some of you might find it useful to learn how to polish windows properly.

† Grinding solid compounds with liquid parafin and squeezing the paste produced between rock salt or KBr windows was the classic method of examining the mir spectra of solids. KBr disks began to rival the “mull technique” as it was called, only in the 1960’s. Elderly spectroscopists still refer to the method as the Nujol Mull method although these days everyone describes it as the liquid parafin mull procedure.

Why “NUJOL”?. Back in the pre 1960’s, a high quality liquid parafin to British Pharmaceutical Standard was on the market called Nujol – recommended for the ease of constipation. At that time Nujol gave a clearer spectrum than non-branded parafins so the use of this particular brand of laxative was always recommended. Simple, ain’t it?

Can I suggest you find a battered old KBr or NaCl window and then follow what I tell you to do. Do the work near the spectrometer and check progress as I indicate.

Let us assume that the ‘optical flats’ we have available are neither optically clear or flat – a lump of optical material cut into a round or square shape but looking as if some idiot had run it under the tap! Laugh not – I have actually seen a postgrad do just that! I searched around and found myself a 30mm diameter x 5mm thick KBr window which satisfied the description – very badly corroded, light transmission not too bad but the surfaces clearly uneven. I decided to grind and polish this particular window as follows:

  1. Record the mid ir spectrum of the window –

 [High-Resolution Spectra]
As you can see, the transmission is poor particularly at short wavelength (high cm-1). Using two of these to sandwich a liquid parafin mull or to close a cell would produce hopeless results.

  1. Grind the surfaces of the window flat. The best way to do this is to use carborundum (SiC.) powder on glass.
  2. a) Find a scrap piece of glass preferably 5mm thick or more. 150mm square is quite big enough. Carefully wash the glass under water (if dirty, use a domestic cream cleaner), rinse and carefully dry with a cloth. Put the plate on the bench with a new sheet of white paper underneath it. Pass your hand over the glass (an excellent way of removing the last trace of grit).
  3. b) Obtain a supply of SiC of coarse and fine grades. I used grades 160 coarse and 600 fine acquired from the glass shop at the University. Mechanical workshops often have these grades of carborundum.

Put about 10mg of the coarse SiC on the plate and wet it fairly heavily with ethanol. Grip the flat between the tips of your thumb and index finger, now using a circular motion alternating clockwise with anti-clockwise grind one surface of the flat. After 2 clock/2 anti-clockwise movements, turn the flat by 30º under your grip and repeat….




Don’t press hard – use very gently pressure or the surface you produce will not be flat. After about 1 minute wipe the surface with a tissue and see how you are doing. Eventually you will produce a completely uniform white opaque surface.

TIPS: Keep the plate, abrasive and flat swimming in alcohol. Don’t let it dry out

When you have wiped the surface with the tissue, drop the tissue onto the floor. If by mistake you re-used a tissue, it might have a crystal or two of grit on it and this could ruin all your subsequent efforts.

  1. c) Turn over and repeat with the second surface. Using a new piece of tissue soaked in alcohol carefully clean the surfaces and sides of the flat. Do so at least twice each time dropping the tissue at your feet and using a new piece.

I checked the i.r spectrum of the flat at this stage and got the following –

 [High-Resolution Spectra]
The very low transmission at high frequencies arises from light scatter. The i.r beam is scattered most when the irregularities are smaller than the wavelength of the light so obviously at 400cm-1(25µ=) the light is less scattered than it is at 7800cm-1(1.2 µ wavelength).

  1. d) Remove the glass plate and re-wash it in warm water. Wash your hands and wipe over the bench. Replace the dried plate on new paper on the bench and wipe your hand over the plate. Why the ritual? We are going to grind the surfaces with much finer SiC than we used at first. One rogue coarse crystal of SiC will ruin our efforts – thus the precautions.
  2. e) Put on rubber gloves. Put ~10mg of the fine SiC on the plate, add ethanol and do exactly as you did before in b). No need to grind each surface for more than 1 minute.

I checked the ir at this next stage and found that the transmission was a bit better than before particularly at larger wavelengths. Very much as I expected.

You now have two good-looking clean uniform flat surfaces.


You will need to have access to a commercial polishing kit. All the accessory makers offer these – they are all very similar and simple to use and highly effective. A kit was kindly loaned to me by Specac Ltd of St. Mary Cray in Kent, UK.

The Specac kit consists essentially of a “smoothing lap” and a “polishing lap” an optical flat and various spares and supplies.

The smoothing lap is a heavy piece of glass about 80 x110mm in size, covered on one side with an abrasive sheet. The sheet is adhesive backed, replacements are supplied and the abrasive looks very like fine “wet & dry” abrasive paper. Before you start – wipe the bench down, remove the silicon carbide, used tissues etc. Make sure no grit could contaminate the polishing kit or you will be wasting your time. Finally wash you hands, put on rubber gloves and re-clean the flats with tissue wetted with alcohol. Again, drop them onto the floor once used.

  1. a) Put out the polishing lap. Wet the part furthest from you with alcohol. Gently drop the window onto the alcohol and use the same motion as you did before – round and round clock and anti-clockwise moving the window in your fingers after 4 rotations.DO NOT PRESS HARD.After about 1 minute, lift and dry the surface. At a glancing angle you will see that it is slightly shiny i.e. a new much smoother surface is being produced. I recommend repeating with a little more alcohol i.e. spending 2 minutes per side.
  2. b) Repeat exactly as before on the second surface. Using the section of the lap nearest you that you didn’t use before. Wipe clean and run the spectrum. I got –

 [High-Resolution Spectra]
which as you will see is much better than we got before. The smoothing process is certainly reducing the scatter. Now put the smoothing lap aside and change to the polishing lap.

The polishing lap consists of a very fine velvet-like cloth (one of the trade names is Selvyt) stuck onto one surface of another heavy glass plate. The principle is that if a polishing abrasive is put on the cloth and wetted with alcohol, abrasion will remove a thin layer of material from the flat and produce a polished finish. The problem is that the cloth will crush under pressure hence the optical window will polish away more at the edges then at the centre.

So  :-


I exaggerate of course!

To counter this effect, the glass under the smoothing lap is often deliberately made convex so that after smoothing the surface is slightly concave. Polishing then flattens this out – in theory!

  1. c) Put ~5 mg of polishing powder (usually alumina and often called jeweller’s rouge although it is white not red!) on the lap, wet with alcohol and proceed as you did for smoothing. At this stage you must wear rubber gloves.  Use very light pressure, move quickly and wipe the flat on a dry section of the polishing cloth before inverting and inspecting. I found ~40 seconds polishing give quite a good result.
  2. d) Repeat on the second surface, gently polish the surfaces with dry tissue and run the spectrum. You should get something like this –

 [High-Resolution Spectra]

i.e. about 95-96% transmission.

TIPS: Don’t press too hard. Don’t keep on polishing – get acceptable transmission and stop. Too much polishing will make the surfaces convex.

For most purposes these windows are fine. You may however need to make the surfaces really flat. This is a job for the professionals but with care (just the right amount of smoothing and then polishing) you can achieve a surface to ~4-10 ‘fringes’ i.e. with an error from flatness of about 2-5 microns.

To check flatness, you need the optical flat provided in the kit. Clean it thoroughly and then pass the palm of your hand over it. Place the window on it and gently press out the air (remember to put on rubber gloves before you do this). Hold as shown in the diagram below –


You should see a set of blue/green rings at the interface between the glass and the KBr. They are very feint and hard to see. The number of rings you see indicates the flatness. They arise from interference of the blue/green lines from the fluorescent strip-light as the radiation passes through the thin air layer between the surfaces. Don’t move the window across the flat as you will certainly scratch the soft KBr surface.


There are many myths about storing KBr and NaCl. If kept warm, they will survive quite high humidity so keeping them in a warmed cupboard is fine in temporate or dry climates or in air-conditioned rooms. [NOT in the tropics]. For long term storage or in really humid environments, store in a dessicator over silica or in a sealed jar or can again over silica. In humid labs – ALWAYS warm the dessicator or can under an infrared lamp before opening. What kills these materials is condensation on the surface.

TIP: Ever since I was a postgraduate (more than 40 years ago!) I have always kept my KBr and NaCl flats coated lightly in liquid parafin. A quick wipe before use removes all but a minute trace and fogging is very much reduced by the oily layer on the surface.

A flat in reasonable condition can be improved, ready for use in less than a minute by using just the polishing lap. So, I recommend having the polishing kit ready for repeated use.HOW LONG DOES ALL THIS POLISHING TAKE? I hear you plaintively ask!!

Grinding and polishing a roughly cut window or ruined one takes about 15-20 minutes per window. If you look in the catalogues you will see that rough cut windows are MUCH cheaper than polished ones – is the difference in price worth more or less than 15 minutes of your time? I leave you (or your boss) to judge.

Good Luck.


REF: Patrick Hendra Internet J. Vib. Spec.[] 4, 1, 2 (2000)