Insights from the Applications Lab
In these tech tips, members of our Ocean Lab Services team share some simple yet important insights gleaned from their work with customers on sample testing, feasibility studies and consultation.
Tip from Derek Guenther, Senior Application Scientist:
Unit Precision and Global File Writer
Have you ever spent a good deal of time preparing a spectroscopy experiment, taking care to ensure every detail has been covered, only to find afterwards that the data didn’t quite save in the format you had expected or needed? They say that to err is human, but to really screw things up you need a computer.
Thankfully, OceanView software is a very approachable and intuitive software interface in the world of spectroscopy. But there are still some tips to ensure the spectral files you save contain the numbers you want to see.
When you go through a typical OceanView Wizard to create a “processed view” of something such as transmission, reflectance or absorbance, the software will display and save these processed numbers to the last unit precision specified. When you first install and open the software these units may not go out far enough for the resolutions or optical activities you hope to see and analyze. To change this, right-click anywhere on the graph and open Graph Layer Options. In the Unit Precision tab, you can change both the x- and y-axis decimal places to whatever is most applicable to your current experiment, and this will carry over to the files you save from that active graph (Figure 1).
Figure 1. Graph Layer in OceanView software.
But here’s the insider tip: When I’m running an experiment, it’s important to me to also acquire the raw intensity feed seen by the detector. Why, you ask? Two main reasons:
- It allows troubleshooting of bizarre occurrences in the processed view by clearly showing what the detector is seeing versus the reference (assuming you’ve also saved the reference intensity spectrum), and …
- It provides total control of unit precision by allowing manual post-calculation of the processed parameters from the intensity count data. If we’re looking at an absorbance value at a specific wavelength, our processed view may say the absorbance is 0.38. But if we save the intensity data as well and see that our sample intensity was 23629 counts and our reference intensity was 57210 counts, now we can calculate absorbance ourselves as being 0.38403, providing much tighter unit precision.
The way to acquire both intensity and processed views at the exact instance is to use the Global File Writer at the top of the software. In the screenshot in Figure 2, graphs are shown for the raw scope view at the bottom and the processed transmission view on top. Configure file saving for each in its respective menu, and when you click the Global File Writer button on top, you will have both data sets saved at the same time.
Figure 2. Process View (top) and Raw Scope View
Tip from Dr. Anne-Marie Dowgiallo, Application Scientist:
Cleaning Up Raman Peaks
When collecting Raman spectra, a useful tool that can be employed in OceanView software is known as “CleanPeaks.” This is a built-in algorithm that can be applied to the raw Raman spectra to remove the baseline and any fluorescence. As Figure 3 demonstrates, the difference between an unprocessed Raman spectrum and one with the CleanPeaks tool applied is dramatic.
Figure 3. Differences between raw and processed spectra are dramatic.
Here are the steps to implement “CleanPeaks” and help your data look better!
Note: These steps assume you already have the Raman application open and running.
Step 1: Open the “Schematic Window.” Right-click to display the menu seen below:
Step 1. Schematic View in OceanView software
Step 2: Click “Advanced Math,” followed by “Filtering,” and then “Clean Peaks.”
Step 2 guides you via prompts to the CleanPeaks feature.
Step 3: The “CleanPeaks” icon should now appear.
Step 3. The CleanPeaks icon is now visible.
Step 4: Press “Ctrl” and right-click on the “RamanShift” icon to drag an arrow (A) to the “CleanPeaks” icon.
Step 4 begins the process of applying CleanPeaks to your Raman spectrum.
Step 5: Press “Ctrl” and right-click the “CleanPeaks” icon to drag an arrow to the “RamanShiftView” icon.
Step 5 ensures the CleanPeaks spectrum will appear on the same plot as the raw Raman spectrum.
Now, the “CleanPeaks” spectrum will appear on the same plot as the raw Raman spectrum.
If you prefer the “CleanPeaks” spectrum to appear in a new window, then follow the steps below instead of “Step 5” from above.
Step 6: Right-click to reveal the menu shown below. Click on “Views” and “Add Graph View.”
Step 6. Here we activate Graph View.
Step 7: A new graph view will appear (“View_10” in this demo). Press “Ctrl” and right-click the “CleanPeaks” icon to drag an arrow to the “View_10” icon.
Figure 7. The new Graph View is labeled "View_10" in this demo.
Step 8: Rename the icons by right-clicking each one and naming them “CleanPeaks” and “RamanShift” as shown below.
Step 8. Right-click icons to rename them.
Step 9: Once you save a spectrum, the icons will automatically connect as shown below. This will ensure that when you save your raw data, the “CleanPeaks” data is also saved.
Step 9. All the icons are now connected, ensuring that when raw data is saved, CleanPeaks data is also saved.
Tip from Adison Fryman, Application Scientist:
Basic Baseline Adjustments to Save the Day (… I Mean Data)
Researchers are familiar with running experiments, saving piles of data, and then finding during data analysis that some unknown factor has affected their observations negatively.
In the case of spectroscopy measurements, setting up an optically stable measurement that yields consistent, repeatable results can be a challenge. Even with best practices, the spectra for identical samples can vary. Using data from a recent feasibility study, we demonstrate the simplicity and effectiveness of baseline adjustments to combat 1) measurement variability associated with ambient light and temperature changes, 2) setup changes including the dreaded bumped fiber, or 3) variability due to inconsistent, unmatched cuvettes or vials.
In this example, we show absorbance measurements performed in triplicate to assess the repeatability of the measurements. Since the three samples are identical with respect to analyte and concentration, the expectation is that all three spectra will be the same. As shown in Figure 4, this is not the case. The spectra appear to have the same spectral shape but different intensities. This consistent spectral shape with shifted intensity of replicate measurements is the perfect situation for a simple baseline correction.
Figure 4. Raw (unprocessed) spectra of samples at different concentrations.
To clean up this data and reduce variability, we applied a simple baseline correction to each of the spectra. Baseline correction is accomplished by subtracting the absorbance value at a wavelength where no absorbance occurs from every data point across the entire spectrum. In this example, we chose 850 nm as the wavelength for baseline correction because there is no absorbance in this region as evidenced by the relatively flat spectrum. The results of applying baseline correction are shown in Figure 5, where the spectra are now much more consistent.
Figure 5. With baseline correction applied, spectra are much more consistent.
When we expand the experiment to include multiple analyte concentrations to generate a standard curve for the determination of unknown analyte concentrations, the impact of baseline correction is even more evident. As shown in the spectra that have not been corrected for baseline (Figure 6), the spectra vary wildly in the region where no absorbance is occurring. This baseline variability makes it very difficult to see trends in the data that correlate with analyte concentration.
Figure 6. Baseline variability in these raw (unprocessed) spectra makes it difficult to discern trends.
When the spectra are corrected for the shifting baselines as shown in Figure 7, the trend in spectral data is much more correlated with analyte concentration.
Figure 7. Baseline correction reveals that the spectral data is closely correlated with analyte concentration.
The impact of this baseline correction is also observed in the standard curve of absorbance at 600 nm versus analyte concentration shown in Figures 8-9. In these plots, we see a relatively linear trend in absorbance with analyte concentration, but this doesn’t tell you everything you need to know about the quality of your standard curve.
Figure 8. In this standard curve, variability at each data point mitigates one's confidence in the relatively linear absorbance trend.
Figure 9. Baseline correction lessens variability in data points and improves correlation.
Because these samples were run in triplicate, we can see the impact of baseline correction in the size of the error bars around each data point. The uncorrected data show a relatively linear trend but have so much variability (as shown by the error bars) that the concentration of an unknown sample would be questionable at best. Which standard curve would you want to use for determining the concentration of an unknown sample?
Spectroscopy measurements are not over once the spectra are measured. Numerous spectral post-processing techniques are available to improve your results and the outcome of your measurements. In this example, the impact of a very simple baseline correction easily carried out in Microsoft Excel or similar program is evident by the improved correlation and lower variability of the data points in our standard curve.
Additional Technical Resources