How to Choose the Correct Wavelength in ICP-OES

When initially developing a method for inductively coupled plasma–optical emission spectrometry (ICP-OES), one common source of confusion is deciding which wavelength to use for each analyte. While there is no single rule that applies to all analytes in every conceivable type of sample, the steps below serve as a guide in selecting analytical wavelengths:

1. Select wavelengths from a standardized method.

2. Pick the top one or two wavelengths suggested by the software.

3. Do your research.

4. Know your sample matrix. What are the major components?

5. Remember that single element standards are your friends.

6. Review peak shapes after analysis.

7. Look at the spectra around the analytical wavelength.

Let's look at each of these steps in more detail.

Select Wavelengths from a Standardized Method

Many standardized methods (american Society for Testing and Materials [ASTM], International Organization for Standardization [ISO], Environmental Protection agency [EPA}) specify recommended wavelengths for each analyte listed in the method. These wavelengths should be used whenever possible.

If multiple wavelengths are listed for a target element, employ at least 2–3 wavelengths, especially during method development. However, given that not all wavelengths are present on all ICP-OES instruments, there is a chance that the top recommended wavelength may not be available. In this case, choose another recommended wavelength that is present on your instrument.

Pick the Top One or Two Wavelengths Suggested by the Software

Instrument software can provide recommended wavelengths for each element. In many cases, multiple wavelengths are listed, in the order of preference, usually based on intensity, background, and common interferences. During initial method development, it's a good idea to pick at least 2–3 wavelengths for each element to evaluate which is better suited for typical samples. Choosing more than 2–3 wavelengths is acceptable, but could result in a significant and overwhelming amount of data to review.

Do Your Research

There is a wealth of information available in application notes, on-line libraries, and journal articles regarding wavelength options. It is possible that the type of sample you wish to analyze has, in fact, already been analyzed. As such, wavelengths used by others can be used as starting points for choosing wavelengths for your analysis.

Know Your Sample Matrix. What Are the Major Components?

In most situations, samples are not completely unknown; there is usually some knowledge of what the sample is. For example, it may be a digested soil, an alloy, or an oil sample. In these cases, it is relatively easy to identify the main sample components; this information is valuable in identifying potential interferences.

When major components are identified, the interferences they produce can be predicted and tested for. This knowledge will help identify which wavelengths to use for routine analysis.

But what if you receive a completely unknown sample? ICP-OES instruments have the ability to perform semiquantitative analysis, which allows you to identify the major components. Sure, since it's only semiquantitative, the concentrations of the major components may not be accurately measured, but they are sufficient to help determine the major components, which is the goal.

Remember That Single Element Standards Are Your Friends

Now that the major components have been identified, determine if they will interfere with the analytes of interest. Some interferences are well known and can be predicted, but others cannot. For example, consider a steel sample that contains high concentrations of iron (Fe) and chromium (Cr). Sure, you can look up the major wavelengths for both of these elements-that is a start-but at high concentrations, even minor wavelengths can cause interferences. In a steel sample, minor iron and chromium wavelengths will be strong, due to the high concentrations of these elements and can cause interferences.

In this case, it's best to run single-element standards for the major components. If it's known that a component will be present at a concentration of ≈1% (10,000 mg/L) in the sample to be analyzed, run a 1% standard using the method that contains the analytes. Do peaks appear where the analytes should be? Or close to where the analytes are? If so, it most likely means that the major component is producing an interference, so it's time to look for another wavelength for the analyte.

Even if the exact concentration of a major component is unknown, running a single-element standard of 1000 or 10,000 mg/L will give an idea of where interferences can appear.

Let's look at an example: the determination of phosphorus in a nickel alloy. Phosphorus (P) has 4 main wavelengths: 177.434, 178.221, 213.617, and 214.914 nm. Which is the best wavelength to use?

First, determine the major components of the alloy and their concentrations after sample preparation. As seen in the tables in Figure 1, there are 14 major components. Therefore, 14 single-element solutions representing the major components and a solution of 0.1 ppm P were measured at the four P wavelengths, and the spectra overlaid.

Figure 1: Spectra at (a) P 214, and (b) P 178 in single-element standards of major components within a nickel alloy.

For simplicity, Figure 1 only shows P 214 and P 178. As can be seen at P 214, the P peak is very small, with larger peaks from molybdenum (Mo) and tungsten (W) sitting directly over P. This means P 214 cannot be used.

However, for P 178, the P peak is much larger than the background, meaning that none of the major components interfere. As a result, P 178 was chosen for this analysis, and yielded good results.

One word of caution: It is possible that single-element standards may be contaminated. To be sure, it is good practice to save the certificate of analysis for each single element standard so that background levels are known. It is also good practice to order single-element standards in relatively small quantities (100 mL or less). This way, they will be replaced more frequently than larger quantities, decreasing the chance that they become contaminated over time.

Review the Peak Shapes After Analysis

The wavelengths have been chosen; now it's time to analyze samples. After the analysis is complete, the sample spectra should be evaluated. Do the peaks look normal? Or do they look a bit strange? Perhaps they appear as if they are tailing? How do you know if a peak's appearance is not normal? Compare the spectra of the sample to a standard that contains just the target analytes. When overlaying the peaks from the standard and the sample, how do the peak shapes compare? If they are similar, that's good-no interference. But, if the peak shape from the sample is significantly different from the standard, then it most likely means there is an interference, which means that it's time to find a new wavelength, or correct for the interference.

As an example, let's look at Figure 2. The spectrum in Figure 2a shows the sample, looking at As (arsenic) 193.696 nm. It looks like the peak is visible, although very close to another peak. To confirm, a 0.3 ppm standard was run, as shown in Figure 2b. The peak shape is noticeably different than in Figure 2a, and at a slightly different position. This suggests that the peak in Figure 2a may not be As. Figure 2c shows spectra of 2 ppm Pt, 220 ppm Cr, 0.3 ppm As, and the sample at As 193. These spectra confirm that the "As" peak in 2a is really Cr, with an interference from Pt, and that both Pt and Cr interfere with As. Once this is known, an interference correction can be made to obtain correct results for As 193.

Figure 2: (a) Sample spectrum as arsenic (As) 193.696 nm; (b) spectrum of 0.3 ppm As standard; (c) spectra of 2 ppm Pt, 220 ppm Cr, and 0.3 ppm As single-element standards overlaid, along with the sample spectrum.

Look at the Spectra Around the Analytical Wavelength

What if two wavelengths for an analyte have been found for a sample and neither have interferences? Which do I choose?

Look for other peaks around the peak of interest. Why? Because one or two background points have to be defined so that a baseline can be drawn for the analyte peak. If there are a lot of peaks around the analyte peak, or if the baseline is noisy or slanted, the background point(s) may not consistently fall in the same position, which will affect the accuracy of the results. In this case, it's better to choose the wavelength that is "cleaner" on one or both sides of the analyte peak.

Once wavelengths have been selected, accuracy can be verified through the analysis of reference materials or known samples, or both.

Kenneth Neubauer is a Senior Application Scientist at Perkin Elmer, Inc., in Shelton, Connecticut. Direct correspondence to: kenneth.neubauer@perkinelmer.com