James, David and I have been discussing what lab tools would be appropriate to have in the new RFC soil + food testing lab (see notes for full details). We came to consensus that a 200 - 2000nm range (or at least 350 - 2000nm) is important because of the decreasing cost of the mid IR (1000 - 2000nm) and known useful applications in that range. If we collect that data now, we may be able to use the collected data in a library in a future miniaturized application.
So, I went looking for some. Here’s the items that fit our specification:
So for that used Lambda 900 looks pretty good. It’s an older model, we’re not super interested in resolution and while noise should be low, we don’t need the newest + greatest.
The other option is to throw a minispec from hamamatsu (350 - 850 (7nm resolution) ) with their mid IR packages (C14272 and C13272-02, 20nm resolution) to cover 1350 - 1850 and word on the street is they’ve got another one which will go to 2150 coming out soon. That would leave a gap between 850 - 1350, and leave out the UV. And we’d need to design the optics, and it would be point based (not like the integrating sphere designs that are done professionally). But then we would have something that we can continue to develop and improve on over time, and someone that we could miniaturize directly in the future. This option would be cheaper, but probably only marginally cheaper after engineering costs.
What is the scan time for the various instruments? While this is also a function of resolution, averaging, etc., if most scans start to get into many tens of minutes or hours we need to rethink things.
Do we need to swap detectors to cover the wavelength ranges of interest? Or if multiple detectors are required does the instrument output data from both simultaneously?
Sample prep and repeatability in the sample prep and measurement - will we need to make custom sample holders? I suspect there may need to be a bit of trial and error.
Above 1450nm, water will basically overwhelm any other signal. So to see anything past that, we would need to dry the sample. That makes the point of correlating it to a handheld device somewhat moot.
Scan time is from 1000nm per minute to 30nm per minute, depending on resolution. I would expect, since we’re not super interested in highly defined individual peaks, we’d run it at full speed.
Yeah - trial and error on sample setup. They do come with integrating spheres, so we’ll be catching the light bouncing in all directions, not just ‘first surface’ reflection.
Ok, based on talking to folks and feedback - here’s my new thought:
We want 200 - 1450nm (cutoff where water takes over).
In this case, two stellarnet specs might be a good fit. We’re have to design our own enclosure, but honestly because of our need for flexibility with different samples and sample times, we may want that anyway.
Also, using the deuterium / halogen combination, we can run fluorescence measurements by turning on only the deuterium lamp (avoiding visible light contamination in our fluorescence response). This may be useful for some protein determination.
In total, this would cost ~$5000 for the equipment, then the time it would take to design + make the enclosure. It hits the feasible ranges, allows us flexibility that we’ll need, and is relatively inexpensive. I’m also going to investigate to see if we can pull the data directly into Our Sci to manage the data with the rest of our data.