Subscribe to Our Blog
Receive updates from our team as we share application notes, customer spotlights, educational tools, spectroscopy how-to’s, and more.
Ocean Insight experts offer their insights on NIR spectroscopy, including its advantages versus other techniques, considerations with sampling, and applications where it’s especially effective.
About Our Experts
Originally captured as part of the Applied Spectral Knowledge Podcast series, this conversation on near infrared (NIR) spectroscopy was led by Yvette Mattley, Ocean Insight’s Lab Services Manager, and features Ty Olmstead, Vice President, Engineering and Product Management; Joe Bonvallet, Senior Application Scientist; and our former colleague Troy McKay, a systems engineer and expert in imaging science.
Note: These comments have been edited for length and clarity. To hear the complete discussion, tune in to our NIR Spectroscopy: Expert Insights podcast.
Yvette Mattley (YM): I've done a lot of work in the UV and Visible wavelength ranges, but it's my time at Ocean Insight that introduced me to the NIR. We talked before the podcast how each of us defines the NIR wavelength range. So, as a life scientist and biochemist, the NIR wavelength range to me is from just above where the Visible wavelength range cuts off – so, 780 nm out to 2500 nm.
Troy McKay (TMK): Well, my background began in remote sensing. So, I view the different spectral ranges based on how they're applied in remote sensing. And that means where they fall within the atmospheric windows: NIR would be from 800-1400 nm and then the SWIR [shortwave infrared] would begin at wavelengths greater than 1400 nm out to 2500 nm.
Joe Bonvallet (JB): To me, it's a little bit different. I like to think of it more in terms of wave numbers. I think of it as a broader range -- 4000 wave numbers out to 14000 wave numbers, where you're looking at aliphatic or aromatic O-H stretches or your amine-type groups.
TMK: I have used the NIR and the SWIR extensively because there's a lot a lot of good information in those ranges about chemical composition, soil composition and things that you can't really see in the Visible. For a lot of chemicals and soils, the Visible spectrum is quite boring and not a lot of features are typically found there. So, if you're looking to do chemical identification, plastics identification, soil classification and vegetation analysis, you really want to be in the wavelengths greater than 800 nm.
JB: Another big advantage that you have with NIR is the depth of penetration that you would have compared with your Visible region. If you're trying to probe into a surface or are looking at the surface chemistry of something and you're trying to look tens of microns into a surface or into the structure of that compound or material, NIR is a great technique versus your Visible. But the biggest disadvantage is you get some strong signatures from O-H groups that may be present in water. And so, if your compound is an aqueous solution of something where you have very low concentrations of other compounds in a water solution, sometimes that O-H can saturate your signal and can wreak havoc on your measurement.
TMK: With NIR, there's very little sample preparation to be done. There's lots of different sampling techniques that can take place. You can use flow cells; you can use cuvette holders. There are lots of options and it's very flexible in the kind of sampling arrangement that you can have, which makes it very useful in a lot of different situations.
Ty Olmstead (TO): The biggest success I've seen with NIR is in food processing and identifying the quality of food. The penetration depth for organic samples can be on the order of a couple of millimeters into the actual products, so we’re able to see blemishes or get information below the surface of the food that we're looking at. We've been very successful at Ocean Insight in predicting tomatoes’ life spans from the harvester to within about a day of when the actual decay will form on the tomato. This is powerful stuff that can help food quality across the world, getting good food to people and being able to sort through produce based on different kinds of conditions.
YM: That's my favorite NIR application as well. NIR spectroscopy is a nice, non-destructive way to potentially determine the sweetness of a product or the ripeness of a product without having to cut it open and destroy it in the process.
JB: There’s also some interesting stuff with biological applications. I was at a customer site where they're looking at different proteins in the drug discovery process and looking in the 4000-6000 wave number range, looking at some aliphatics and aromatic rings out there for the identification of a compound that could potentially be harmful in a drug. So, when you make these drugs, they're stored in different containers and sometimes those containers will shed a layer. And that's where you get these aromatic rings in your solution. And so, this company is looking at quantifying that and seeing if it's harmful to the body.
YM: Wow. So, we have a lot of specificity out there in the NIR as opposed to maybe looking at signals in the Visible, where you're getting color and things that are going to interfere. My favorite way to explain to my family what I do is to talk about oximeter chips. Now they're disposable. Back in the day they were plastic clamps. And in that case, because of the increased penetration depth of the signal, and because you get a nice isosbestic point that occurs out there with the hemoglobin, the NIR has been a great region for oximetry. But also looking for harmful chemicals in therapeutics is another great application in the biological arena.
TMK: NIR can be a useful tool for industrial applications where you're looking at dissolved chemicals in aqueous solutions or even in solvents. The nice thing about NIR is you have some very good, strong signatures out there so you can calculate chemical concentrations fairly easily and break down the absorption or the transmission spectra into a combination of multiple different chemicals based on their concentrations.
TO: One of the neat things that we have at Ocean Insight is the ability to use machine learning to help understand what the spectra are telling us as part of the process. We have the product Ocean Intelligence, which really gives us a powerful tool to help in determining things like Brix [a measure of sugars in a fruit] on tomatoes, to figure out how sweet they actually are.
JB: With NIR spectra you build robust calibration models, and if you are trying to predict outside those calibration models, it's not very viable. Collecting all this data, it becomes a question of what do you do with it and how do you make useful decisions on all this data? And I think that's where Ocean Intelligence has a very nice role in this type of measurement.
YM: Now that we have machine learning capability in-house, it puts the final piece in the puzzle to make these robust solutions involving NIR spectroscopy or whatever wavelength range. When you think about the complexity of a tomato in terms of a reflection spectrum, it's amazing the actual information content there. But it can't be unlocked without some type of advanced data processing – chemometrics, machine learning, and those types of algorithms. I'm glad we have that tool here at Ocean Insight.
TMK: With reflection, I think fixturing is always an issue, no matter which spectral range that you're in. So, holding your sample and holding your reference in a very precise and repeatable fashion is critical to making good measurements, just good practice across the across the board in any sort of reflectance spectroscopy. In the NIR you can still utilize Spectralon® or PTFE standards, which are nearly Lambertian and very high reflectance, which makes them excellent standards to use almost down into the UV all the way out through the 2500 nm range.
YM: In my experience, working in the NIR requires a lot of light to get a good signal to noise ratio because of the inherent noise associated with the detector. I'm wondering if you guys have seen similar challenges, and if you have suggestions for how we can overcome that.
JB: As you mentioned, one of the big issues in the NIR is that noise floor and keeping that under control. So, we always try to get as much light into the detector as possible. And there are lots of tricks you can play to do that. Also, one of our newer products, the NIRQuest+, achieves this. With NIRQuest+ we get more light into the system, allowing us to achieve lower detection limits and higher throughput.
TMK: Also, I've always had to use quartz halogen-type light sources for the NIR, which is a challenge. They are not always the most compact. They make a lot of heat. But recently, I've noticed some applications where you can get some light from an LED in the broadband infrared region. That's great news because LED power levels are relatively low, and NIRQuest+ will allow us to utilize them a little bit more.
YM: What about dealing with sample mixtures? Have you found that deconvoluting those signals, the spectral signatures, which are so complex, is easier when you go out into the NIR?
TMK: Yes, provided the features are stronger and higher frequency. You don't want to see a gradually changing, really smooth signature, otherwise you're going to confuse your computer model much simpler than if you have nice, distinct, sharp absorption features, which tends to be the case out in the NIR.
JB: I think it has to do with the biologics, especially in the time we're living in, in this drug discovery phase, being able to kind of get that fingerprint of what is in your solution, especially as vaccines are developed for COVID-19. I think there are going to be a lot of applications in this area for this type of measurement.
Also, NIR allows us to make measurements in situ in different processes, such as the drug discovery. It allows us to make GMP-type measurements where we're not interfering with that solution, and we're able to interface with it and make sure that things going into the body are safe.
TMK: I think vegetation, crop health and food in general are good applications for the NIR that I'm excited to see where they go. The Vegetation Index was one of the first things they did with the NIR back in the ‘60s, just looking at vegetation health. That's come a long way since then. And with the size of spectrometers shrinking and the size of hyperspectral imagers shrinking, they can be used a lot more in agriculture to ensure that the crops are getting the proper amounts of nutrients and pesticides.
TO: The most exciting space for me with the NIR and SWIR is their continued introduction into commercial processes. As we start to think about how we move food or move different substances into production to get it to the masses, NIR gives us a tool on which we can discriminate and make decisions about quality.
When Joe is talking about using NIR for drug discovery, that's going into a larger manufacturing process where we're going to need to make decisions on very large volumes of material. When Troy is talking about food quality, we think about the food we see at the grocery store before taking it home. But before that food ever gets to the grocery store, there are massive amounts of processing of onions, tomatoes, pomegranates and more. And decisions need to be made of how is that food is to be distributed in the best way to give everybody the best quality they can. And I think that NIR and SWIR really unlock the space to give us a new way into making better decisions about how we take care of our planet.
Measuring soil composition with a near-infrared diffuse reflectance system helps characterize soil quickly, economically and effectively. In this demo, we use the NIRQuest+2.5 NIR spectrometer as part of a diffuse reflectance system for soil analysis.