Inductively coupled plasma (ICP) systems have an optimal observation height, but measurements made at other observation heights in the ICP should also yield the same accurate analytical results - if there are no interferences.
Inductively coupled plasma (ICP) systems have an optimal observation height, but measurements made at other observation heights in the ICP should also yield the same accurate analytical results - if there are no interferences. This means that a spatial-varying determined concentration of the analyte can reveal that something - such as matrix interference - is wrong with the measurement. George Chan of the Laser Technologies Group at Lawrence Berkeley National Laboratory recently spoke to us about his work exploiting the spatial heterogeneity of the ICP.
In several recent papers (1–3), you explained how spatial emission maps can be used to flag matrix interferences and system drift. Can you provide a brief explanation of how the technique works?
Chan: The underlying principle of this interference-flagging method is very simple - we exploit the spatial heterogeneity of the ICP. Although there is an optimal observation height in the ICP (such as for maximum sensitivity, highest signal-to-background ratio, and best precision depending on the specific optimization criteria), measurements made at other observation heights in the ICP should also yield the same accurate (albeit less sensitive) analytical results - if there is no interference. In contrast, because plasma behavior and excitation conditions vary from location to location in the ICP, the relative magnitude and even the direction of the changes in emission intensity caused by a matrix interference are also spatial-location dependent. Now, because the determined concentration of an analyte in a sample is directly proportional to measured emission intensity in the ICP, a spatially dependent matrix effect will cause a corresponding alteration in the determined concentration - that is, the apparent analyte concentration will also vary, depending on the measurement location in the ICP, under the influence of matrix interference. Clearly, a spatially varying determined concentration of the analyte is a very strong indication that something is wrong with the measurement; because it is the same sample, how can we get different analytical results depending on where in the ICP we are performing our measurements?
What led you to develop the method that could detect these interferences?
Chan: All of the developments on this interference-flagging method were performed at Indiana University, in Bloomington, and started when I was a graduate student in the research group of Professor Gary Hieftje. I was studying fundamental matrix-effect and analyte excitation–ionization mechanisms in the ICP. Because my work showed that matrix effects in the normal analytical zone are the most severe from matrix elements with low second ionization potentials, I wanted to characterize the interference mechanisms from these matrix elements along the whole vertical profile. The reason to study the vertical profile is that the dominant mechanism for plasma-related matrix effects changes with vertical height in the plasma, as demonstrated by Blades and Horlick in their classical paper published in 1981 (4). I thought if the remaining two categories (sample-introduction related and spectral) of matrix interferences also exhibit a spatial dependence, then the vertical profile can be used as an indicator for matrix interference in the ICP. I then characterized the vertical profiles in the presence of different categories of matrix interferences and found that the vertical profile is indeed a powerful and effective indicator to flag matrix interference from all known categories of matrix effects. I initially worked with the conventional lateral-viewing ICP and further extended the study to axial viewing.
A commercial lateral-viewing inductively coupled plasma–atomic emission spectroscopy (ICP-AES) system was modified for axial viewing to perform your study. What was involved in that modification?
Chan: Three modifications were involved to modify radial viewing to axial viewing. First, we turned the plasma torch box 90° so that the central channel of the plasma was aligned with the optical axis of the entrance optics of the spectrometer. Second, we used an elongated torch, with its end extending 15 mm from the top of the load coil. Third, we applied nitrogen as a cutoff gas at roughly 25 L/min to deflect the cooler downstream part of the plasma upward from the optical axis.
Why did you think it was necessary to develop an automated diagnostic approach for flagging matrix interferences for ICP-AES?
Chan: Nowadays all ICP spectrometers are computer-controlled and in many cases, we take for granted that the analytical results reported by the computer are correct. A computer is, no doubt, very powerful in performing data processing, statistical analysis, and reporting. However, we have not yet reached the level of an intelligent system that has the ability to warn an operator when the analytical results are compromised by the presence of matrix interferences. It is very important for the analyst to know whether the results reported by the computer can be trusted or are just a train of numbers! As Jean-Michel Mermet once commented (5), "Knowledge of the origin of matrix effects remains probably one of the last challenges in ICP-AES in order to obtain highly accurate results. . . . This is certainly one of the most important remaining challenges, because the first quality that an analyst expects is accuracy, which cannot be obtained if calibration leads to a bias." Clearly, there is a critical need to have a simple indicator that can guide the accuracy of ICP-AES analyses, preferably one that can be easily automated and programmed into the computer software.
How does the diagnostic technique distinguish between spectral, plasma-related, and sample-introduction-related matrix interferences?
Chan: For a laterally (side-on) viewed ICP, this diagnostic tool is quite powerful to distinguish the three categories of interferences. For purely plasma-related matrix interferences, we found that the matrix-effect crossover location (that is, a location where there is no net matrix effect because the effect from a matrix-induced signal enhancement is cancelled by the matrix-induced signal suppression) does not change with sample dilution. Therefore, if we perform a serial dilution of sample (a minimum of three separate concentrations) and if the vertical profiles of these determined-analyte concentrations (after taking the dilution factor into account) all intersect at one location, we can confidently tell that the interference is plasma-related in nature. In fact, the intersection of these curves will give the proper analytical result and no further correction is necessary. In contrast, if these vertical determined-analyte profiles do not intersect at a single point but rather at multiple points, that is a characteristic of a sample-introduction-related matrix interference. Because of the additive nature of spectral interference, the extent of spectral interference will not be affected by sample dilution. Therefore, the vertical profiles of the determined-analyte concentrations will exhibit identical curvatures and effectively overlap. The spatial behavior in axial-viewing (end-on) ICP emission spectroscopy is more complex because not all emission lines exhibit crossover points for plasma-related matrix interferences.
Which type of interference is the most difficult to detect in ICP-AES?
Chan: In my opinion, direct spectral-line overlap is the most difficult to detect. First, when one inspects the spectrum, the two overlapping emission lines appear as a single perfect line. Second, testing based on signal recovery (such as spiking) does not successfully conquer spectral interference. Let's forget about spatial-resolved emission profile for a moment and consider conventional ICP measurement at a fixed single location. An established technique to recognize the existence of nonspectral (that is, plasma-related or sample-introduction-related) matrix interferences in ICP–AES is to compare the analytical results from the original sample and a diluted one. If the two agree (again, after correction for the dilution factor), nonspectral matrix interferences are likely absent. Alternatively, one can also spike a known amount of the analyte into the unknown sample and measure the recovery. If the recovery is 100% (within experimental uncertainty), nonspectral matrix interferences are likely absent. However, all of these techniques are unable to recognize spectral interference. In the presence of a spectral interference, recovery is still 100%. Fortunately, the spatial emission profile in the ICP is able to recognize interferences that are of additive (spectral) or multiplicative (nonspectral) nature.
What impact has this technique had on ICP-AES research so far?
Chan: This matrix-effect flagging indicator is simple to implement, does not require prior knowledge of sample properties, and can be used on the fly during an analysis with no time penalty. We also developed a robust statistical protocol that can be easily automated and programmed to guide analysis accuracy. I hope that we are now closer to an intelligent system for ICP–AES and that analysts can, as a result, avoid reporting inaccurate analytical results because of the presence of undetected matrix interferences.
What are the next steps in your research?
Chan: Currently, a spectrometer with a Czerny-Turner configuration is being used. A drawback for this configuration is that only a limited spectral window can be measured at once. I would like to test this technique on an imaging spectrometer that is capable of simultaneously measuring spatial profiles at all wavelengths (for example, an astigmatism-corrected Roland-circle spectrometer). Also, I would like to extend this technique to other analytical emission sources. Yan Cheung and Andrew Schwartz, current graduate students at Indiana University, are coupling this method to gradient-dilution analysis.
(1) Y. Cheung, G.C.-Y. Chan, and G.M. Hieftje, J. Anal. At. Spectrom.28, 241-250 (2013).
(2) G.C.-Y. Chan and G.M. Hieftje, Anal. Chem.85, 50-57 (2013).
(3) G.C.-Y. Chan and G.M. Hieftje, Anal. Chem. 85, 58-65 (2013).
(4) M.W. Blades and G. Horlick, Spectrochim. Acta Part B36, 881-900 (1981).
(5) J.M. Mermet, J. Anal. At. Spectrom.20, 11-16 (2005).
This interview has been edited for length and clarity. To read more interviews from Spectroscopy, please visit www.spectroscopyonline.com and look under the "Interview of the Month" section on our homepage.