Comparing SCiO with Professional Devices
We want to know how well the SCiO device preforms when compared with professional devices and if it can give us reliable results. Firstly, we got a little background information about the SCiO technique. Secondly, we wanted to compare the SCiO device with a similar professional device.
The SCiO device makes use of NIR spectroscopic analysis. It is a little different from the widely used mid-range spectroscopic analysis, like FTIR. Researchers thought there was little to be gained by studying NIR spectroscopy, because mid-range spectroscopy was giving better results.
Eventually Karl Norris did research in the NIR spectroscopy range. He found out that it has an important advantage over mid-range spectroscopy. NIR spectroscopy needs little to no sample preparation because it makes use of reflectance.
Since sample preparation isn’t needed, NIR can be used for a wider range of samples. Food and skin are some examples. Although there is no complete mathematical theory of NIR spectroscopy yet, it relies upon a database to give reliable results. 
We wanted to compare the SCiO device with a professional NIR-spectroscopy device. Unfortunatly we couldn’t get our hands on a professional one. Therefore, we tried to compare the SCiO device with a professional FTIR device. When we compared the SCiO device with a professional mid-range spectroscopy device, two things came up:
First of all, scanning with the SCiO device was really easy. There was no need to prepare the samples and every time you scanned something you got a unique result for the material. With the mid-range spectroscopy device we needed to get thicker samples and change parts of the machine so that we could reliably measure our samples. So although the reliability of the SCiO device has yet to be proven, it was way easier to get decent results from the samples.
Secondly, the results from the mid-range spectroscopy device was of a higher resolution than the SCiO device. Peaks where clearer to see, there were more peaks and there were way more differences between the high and low. This meant that different materials can be more easily detected by mid-range Spectroscopy.
How Precise is SCiO?
Even without a direct comparison with a professional device we can tell that the precision of the SCiO device is not that great. Although little sample preparation is needed and the scan conditions don’t need to be prefect (scan angle, environment light, underlay) these conditions will still influence the results.
The hardware of the SCiO device is not of the same quality as professional NIR spectroscopy devices. Without knowing the exact differences, there will be a difference and the SCiO device will not be as precise. The readings differ anywhere between 5-10 percent of the actual samples. 
The SCiO device makes use of a database created by users of the SCiO device. The problem is that the average user of a consumer product like the SCiO device is not a scientist. Therefore it is hard to say if the database can be trusted.
NIR spectroscopy sits within the 700-2500 nm range. Within this range different materials show multiple peaks in the graph. The first overtone peak sits around 900-1100 nm and the second around sits around 1200-1400 nm. The range of the SCiO device is only 700-1200 nm. This means that it will measure only one or maybe two overtone peaks and the results from the SCiO device will give less data to compare, thereby giving less reliable results.
SCiO Scan Results of PP
SCiO Scan Results of Unknown Material
As seen in the graphs above, there are two different results from two different SCiO analysis (each analysis contains three scans). The top result shows the result from the scan of a piece of Polypropylene. When we scanned the second sample we didn’t known what material it was. After looking at the results we saw a lot of similarities between the two materials. Although the graphs aren’t the same, you can recognize some of the peaks. Later on we found out that the second material was indeed PP.
We therefore conclude that you can recognize different materials. But we aren’t sure how well the machine can reliably see the differences between two different samples of PP. If you look closely there is a lot of fluctuation in the graph, especially at the end of the wavelength range.
After looking at lots of scan results, it was noticed that most of the time the third scan differs a lot from the first two. For the most part it has the same shape but it lags a little behind or in the front by a couple of nm. The conditions were the same with the scans, so the differentiation comes from the device itself.
How far does the result of the SCiO scan differ from the FTIR scan results?
- The results can’t be compared 1:1 because they work in a different spectrum and with a different technique.
How far can the result of the SCiO scan results differ from the FTIR scan results s that it is still reliable?
- If both SCiO and FTIR could be compared. The results of a SCiO scan can’t differ far from the results of the FTIR device. We already had problems with the scans of the FTIR Device itself. The FTIR needs good samples to give reliable results. The SCiO result can be obtained with less prepared samples.
How does the differentiation in the results of the two techniques compare?
- The SCiO device has more differentiation in its results. Two scans are never the same. FTIR scans, if done correctly, are almost exactly the same.
Does the amount of scans influence the differentiation in the results of the SCiO scan?
- If you scan a sample more times than the minimum amount of three, you see the results differ quite a bit. Although it differs, a reliable average can still be obtained. The problem is that the SCiO software does not have that option.
Does the surface that the sample lies upon influence the result?
- Yes it does, but not as much as when you scan with FTIR.
Taking all of our findings in consideration, we doubt that the SCiO device with its user database will give reliable results. It will be possible to recognize different materials. However the SCiO device will not be able to reliably recognize the difference between to different samples of the same material.
The question now is: Where can the SCiO be used for if you take the low precision into consideration?