FTIR or NIR for dairy analysis?

FTIR spectroscopy is, in principle, very similar to Near Infrared (NIR) spectroscopy, but works at longer wavelengths where the chemical information from the samples is more specific.

While the sensitivity and range offered by the longer wavelengths offers many advantages, it runs into a natural barrier when testing more solid samples. This is because the light can only penetrate a very thin sample (30 -50 um liquid). NIR can depending on wavelength penetrate up to 20 mm of a sample which makes NIR more effective for solid samples. 

Per Waaben a chemometrician at FOSS, explains the difference as well as the benefits of each technology. The whole interview can be seen in the video below. 

Section 1: Per explains the difference between NIR and FTIR

NIR, that's what's just next to the visible spectrum, that starts around 780, goes to 2,500 nanometers and from there to 10,000 nanometers. That's what we normally call mid-infrared. FTIR is a little funny because FTIR is actually the technology used to acquire the spectra. The region is called mid-infrared. That means that when we're looking at the range from 2,500 to 10,000 nanometers, that's mid-infrared. We acquire the spectra with an FTIR instrument, Fourier transform infrared.

Section 2: What are the strengths and weaknesses of NIR and FTIR?

The technology you choose really depends on the application. I would say mid-infrared, that is FTIR, is the best method because you get the best signal-to-noise ratio. You get the clearest signals and so on. That's what you choose. If you could choose just whatever you like, that's what you'd choose. However, signals are very strong in the infrared and that means that you can only shine through a very, very narrow, very, very small sample. That means that sampling errors become a problem.

When you're measuring milk, you have to measure through a very thin film. If it's inhomogeneous, you cannot measure it with the mid-infrared. That's where NIR is good because NIR can penetrate up to centimeters of sample. That means that you can measure cheese almost as is, you can measure milk powders in reflection setup and so on. That's really useful. The drawback of NIR is that the signal is not as strong, and therefore, your accuracy or your repeatability will not be as good as if you're using FTIR or mid-infrared.

Section 3: In which application would you use NIR or FTIR?

The general rule of thumb is that, if you have something with less than 20% total solids, then I recommend mid-infrared, FTIR. If it's above, it would be NIR because that's more of a semi-solid or solid sample. The exception from that would be cream because cream is liquid at very high solid levels. That's why we still use mid-infrared for cream.

Section 4: Can you control your production process with FTIR?

There are basically three ways that you can analyse your milk in a process. You can do at-line, that's when you take a sample out and analyse it in an instrument. The frequency you can do that at is very low, so that means that you don't have fantastic control of the process. Then, you have on-line, where you take a sample out through a little tube into an instrument that analyses so it does it automatically. The problem there is that you don't know what's happening between measurements because you take a sample out.

If there are rapid fluctuations in the process, you won't get that. In-line, you are looking at things happening live and that means that when you have dramatic changes, you can see them immediately. Of course, it requires that you have a technology that you can put into the line. FTIR is difficult because it requires this very thin layer of milk, but it can be solved. The advantage that you get out of it is that, in particular; your protein measurement becomes much more accurate compared to if you're using NIR, because NIR would probably be the method or the choice if FTIR is not possible.

Section 5: Is it at all possible to use FTIR for process control in-line?

Until now, FTIR has not been used in the process. That's because it's very difficult, because you need this very thin layer of milk, and how that is handled has been quite a challenge. You can actually do it - if you have some very fine motors you can control the cuvette so that it can move and it can move into position, so when the system is CIP'd, then you can have it open so that it's cleaned, and then you can close it again for measurements.

The other thing is that stability requires that you need to do some reference measurements from time to time. You need to do some mathematical tricks in order to not have that reference measurement once every hour or so, because you only have the opportunity every time you have a CIP. That might happen every 24 hours, for example. Right now, we're at the stage that we've just released the MilkoStream FT, which solves the problem of doing references all the time. It can do references every time there's a CIP. It doesn't need any more than that.

It can open during a CIP so that it can be cleaned, probably between batches, and then, it can close through some very sophisticated motor control, into a distance of 35 microns, where it's optimal to measure milk in a cuvette. Moving fully in-line makes a huge difference because you can control your process to a very large detail. If something is changing, you can see it immediately.

Also, the maintenance has been reduced quite significantly because we are just using the CIP system in-line in order to clean the instrument. There's no specific cleaning or no reference measurement or so on, everything is happening in relation to the CIP. Finally, there's no flow system. No moving parts to be replaced in the flow system as there is with the current equipment available. The main advantage is that we can measure protein at a very high accuracy compared to competing methods. That's the reason why its FTIR in the line and not any other method.