| Author | Message | ||
     David Cameron (david_cameron) Junior Member Username: david_cameron Post Number: 6 Registered: 3-2011 |
Take a look further down in the NIR or IR, if you can. The Q factor of quartz is measured, by one method, by measuring the extinction coefficient of the fundamental water band near 3500 cm-1 (2.85 um). I don't know how much water is in there, but not much. Nor do I know if your glass is transparent here, but some are as the major band blocking transmission is water. | ||
| Howard Mark (hlmark) Senior Member Username: hlmark Post Number: 359 Registered: 9-2001 |
Jim - a 12-bit A/D converter can't resolve signal differences smaller than 1 part in 4096. A 1-meter pathlength is 1/25,000 of what the fiber is capable of operating at. Therefore, before you spend too much time or energy trying to improve the measurement, I would suggest the following theoretical calculation: from any estimates you have of the absorbance from a measurement over a long length of fiber, however poor, calculate the amount of transmittance difference you would expect to see over a 1-meter length between zero moisture content and however much is in the fiber. If that's less than 1/4,000, or more realistically a few parts in 4,000, then your A/D converter is the limiting factor in your measurement, and it simply will not be capable to detect such a small difference. Also, if there's any measurable noise then your sensitivity will be evel lower, and you can do a calculation to see if the difference between zero moisture and the amount in the fiber (from your estimate based on the long fiber) is theoretically capapble of rising above the noise level. \o/ /_\ | ||
| David W. Hopkins (dhopkins) Senior Member Username: dhopkins Post Number: 176 Registered: 10-2002 |
Hi Jim, Yes, I think it can be done, but I don't know whether your hyperspectral imaging setup is up to it. Maybe. I wonder whether you could invest in a large enough corner cube to put at the end of your rod, so that you could image 2 traverses of the path? I think the band at 1420 nm should be evident. I'd try a 1 or 2 factor PCR calibration using the 2-Der pretreatment, with 2 to 4 rods as training set and the others as validation. How were you able to get any spectra from such a long rod? Did you have the light source at one end, and the camera at the other? What did the spectra look like? You certainly have my curiousity up. I think the OH absorption would be too high in the IR, judging from the spectrum that David von Boisman posted. Best regards, Dave | ||
| Jim Burger (jburger) New member Username: jburger Post Number: 2 Registered: 11-2010 |
Thanks for the comments everyone. Yes this is glass for optical fibers - and I have yet to find a real spectrum for the stuff. (With or without the OH) The reference method for determination of OH and attenuation is to time a laser pulse sent down a 25km fiber (KILOMETERS!!!) and I am attempting to measure absorbance in a 25mm diam rod 1000mm long. Can it be done? Preliminary results are inconclusive: I have 10 factor PLS models for 18 samples, that are of course useless for predicting anything new. My current instrumentation setup involves scanning the rod end (hyperspectral image) to see rod cross section spatial differences - but with a poor light source. I am using a 12bit InGaAs camera for 960-1660nm measurements. Would I gain much (other than cost!!!) by looking at 1000-2450nm or 8-12 um regions? What would I use as a light source for the IR? And yes Dave, there were some problems with my account, so yesterday I was reset to 'newbie' status! We had snow in Latvia 3 weeks ago, but nothing since... no need to warm my hands on my halogen lamps, yet. Jim | ||
| Howard Mark (hlmark) Senior Member Username: hlmark Post Number: 358 Registered: 9-2001 |
Dave - I wouldn't worry about scattering. I suspect that the glass is intended for use in making optical fibers, which are intended for use in "thicknesses" (actually lengths) of many meters, so scattering would have to be practically zero as well as absorbance being minimized in order for it to fulfill its functions. Also, Jim's spectra might be qualitatively appropriate (but see caveat below), but since Jim's glass (which from the name I suspect is a borosilicate like pyrex) has such high absorbance in what are relatively small thicknesses, it shows quantitatively orders of magnitude higher ansorbance than pure silica would. Also, it's highly likely that B-OH has absorbance at different wavelengths than Si-OH, so that the spectrum of the borosilicate glass would have extra bands, that the pure silica glass wouldn't have. Howard \o/ /_\ | ||
| David W. Hopkins (dhopkins) Senior Member Username: dhopkins Post Number: 175 Registered: 10-2002 |
Hello, David Thanks for the spectrum. I see I was not remembering very well. There is an obvious band at about 1420 nm, and nothing at 1900 nm. I assume that the 1420 nm band is the second overtone of the OH stretch? It should be very useful for OH determinations. Best wishes, Dave | ||
| David von Boisman (david_von_boisman) Advanced Member Username: david_von_boisman Post Number: 21 Registered: 2-2006 |
This datasheet will give you an idea what a spectrum may look like: http://www.schott.com/borofloat/english/download/optical_transmit.pdf | ||
| David W. Hopkins (dhopkins) Senior Member Username: dhopkins Post Number: 174 Registered: 10-2002 |
Hi Jim, Welcome to the discussion group! I thought you had already logged on, but I see that this is post #1. Anyway, I think that the peak at about 1900 nm (as I recall) is very strong, and should enable you to make some determinations, as long as you have 1 or more standards, to set up some kind of primitive calibration. It is very easy to detect the difference between glass and quartz at thicknesses of 1-4 mm, so you should see very large signals at 1000 mm. I would be tempted to use a second derivative, to remove any baseline problems. I suspect that the biggest problem you would have is that at such pathlengths, the glass would no longer be a clear solution, but some scattering would be encountered, that might complicate your calibration. You might have to assume that all samples would have essentially the same scattering properties. With such pathlengths, the measurements would be pretty tricky too. What kind of instrumentation are you using? Is it getting cold in Latvia? We have been getting frost in Battle Creek now for several mornings. Best wishes, Dave | ||
| Jim Burger (jburger) New member Username: jburger Post Number: 1 Registered: 11-2010 |
I am trying to measure trace amounts of OH in high purity silica glass. I have glass samples which are 1 meter long (1000mm pathlength) with 'known' but questionable OH and attenuation reference values obtained from a secondary calibration method. I suspect that my present detector (12 bit InGaAs) and light source do not provide the necessary sensitivity / stability for this project. Before I invest in additional hardware, I thought I would check and see if anyone has any direct experience with measurement of OH in glass. What level of transmission loss should I see in 1 meter of glass that is 'contaminated' with OH? I know that's a wide open question - but in this case, any feedback would be appreciated. |