OR WAIT null SECS
Inductively-coupled plasma-optical emission spectroscopy (ICP–OES) and ICP–mass spectrometry (ICP–MS) are often considered mature techniques, but researchers know that some aspects of the techniques are still not fully understood. Probing the mechanisms involved can lead to more accurate results, and in some cases may cast doubt on accepted explanations. John W. Olesik, who directs the Trace Element Research Laboratory (TERL) in the School of Earth Sciences at The Ohio State University, has spent many years studying the processes that control signals in ICP techniques, and yet he is still encountering surprises. He is also in a unique position to get involved in exciting, applied research collaborations, because the TERL also provides elemental analysis and access to ICP–OES and ICP–MS instruments to groups throughout (as well as to clients and collaborators outside) the university, leading to all sorts of collaborations. This interview is the second part of a two-part series. In this part, Olesik discusses environmental research, data analysis techniques, and his collaborative research philosophy.
You are currently involved in a research study to analyze the effects of environmental exposures of metals on newborns and child health (1). What can you share with us about the challenges in analyzing human placenta samples for a variety of metals?
There have been two main challenges. First, some of the elements of interest (including Pb, Pt and Gd) are normally present at extremely low concentrations (low parts per trillion) in many of the digested sample solutions. We also need to measure the pattern of naturally occurring rare earth elements, all present at extremely low concentration, to identify anthropogenic Gd (Dramatically elevated anthropogenic Gd concentrations in placentas come from retention of a small fraction of the approximately 1 g of Gd injected for a Gd chelate contrast enhanced MRI, even if the MRI was done years earlier. In some cases we are detecting much lower, but elevated anthropogenic Gd in the placentas from mothers who never had a Gd-contrast enhanced MRI. We are trying to identify the source of anthrogenic Gd in those cases.). This requires careful control of potential sources of contamination, and a means to avoid potential spectral overlaps. Second, the samples are not all in the same form as we receive them, because they are coming from a variety of sources that use different sample preservation processes. As a result, we have had to develop different sample preparation procedures for the different sample forms, and in many cases for a very limited amount of sample.
You are working on a project to identify and geochemically characterize atmospheric mineral nanoparticles in pre-industrial Antarctic ice during the last climatic cycle (2). How did this project come about? What is your role in analyzing these nanoparticles? What would you expect to find?
Many of our collaborations have begun after another research group designed experiments and came to us about the trace element measurements, either before or after submitting a proposal, or even after they had funding. This project is one where we were involved from the conception of the idea. Lonnie Thompson and Ellen Mosley-Thompson lead a group in the Byrd Polar and Climate Research Center (BPCRC) at Ohio State that has built one of the largest collections of ice cores in the world through many incredible expeditions, not only to the polar regions but to locations far from the poles at high elevations. They are incredible explorers and scientists. Ice core drilling equipment has to be brought to the site (up mountains at high elevations in non-polar regions). After the ice cores are drilled, the cores need to be brought back down the mountain, and then transported, without melting, back to Ohio State laboratories.
The ice cores, the gases trapped in them, pollen and other biological materials, and particles (deposited from the atmosphere and entrapped in the ice) provide a historical record of climate through time. The depth within the ice core can be related to the date it formed, providing a time scale.
Paolo Gabrielli is a research scientist who has worked at Ohio State with the Thompsons since 2007. Paolo is a glaciologist who also has extensive experience using ICP–MS to measure trace elements in the ice or entrapped as particles. Using the “elemental fingerprints,” sources of particles (“dust”) may be identified. However, the vast majority of elemental chemical measurements to date have been done using acid to digest and dissolve the “dust” (consisting of many particles with a range of sizes) and then measure the bulk (average) elemental composition. Both BPCRC and TERL have the same model ICP-sector-field–MS instrument, so we have sometimes borrowed components from each other’s instrument to troubleshoot instrument problems. Paolo was also very interested in the studies of the fundamental processes that occur in the plasma being done in my laboratory, as well as new applications, including single particle ICP–MS.
Following our discussions, Paolo pointed out: there were very little data on the elemental chemical composition of individual particles entrapped in ice cores, and what data there were had been based on small numbers of particles measured using electron microscopy with energy dispersive detection. Furthermore, there was little information on the number concentration of particles less than about 0.5 μm even though the number of particles smaller than 0.5 μm is expected to be orders of magnitude more the than the number of particles larger than 0.5 μm.
Upon further discussion, we also realized that if we could measure the composition of many individual particles that had been trapped in ice cores, we could potentially much more effectively determine the types of particles (for example, particular minerals or anthropogenic materials) present, and therefore their sources, compared to bulk measurements that provided an average composition. Furthermore, small percentages of unusual particles were likely lost in the “noise” of bulk measurements.
We then worked together on a proposal to the National Science Foundation (NSF), not having a feel for the chances of getting funding and expecting it would require more than one iteration and submission before it might be successful. Our plan was based on a combination of complementary measurement techniques, including two that were typically used in previous studies of trace elements in ice cores—Coulter counter based particle size distributions and ICP-sector field–MS measurements of bulk measurements of particles following acid digestion—along with techniques to measure individual particles (single particle ICP–quadrupole MS, single particle-ICP–TOF–MS, and transmission electron microscopy with energy dispersive X-ray spectroscopy). We added two world-class collaborators to support the proposal and to fill in capabilities that were needed but not available in our labs: James Rainville of the Colorado School of Mines, to measure nanoparticles sizes with field flow fractionation-ICP–MS, and Frank von der Kammer, who had an ICP–TOF–MS in his laboratory at the University of Vienna. Both also are among the world’s experts in studying the fate of nanoparticles in the environment. Paolo and I are both involved in nearly aspect of the project other than acquiring and cutting the ice (Paolo is also involved in that, but I am not).
Four of the five reviews we got back on the first submission of the proposal were among the most glowing I have ever seen regarding the importance and novelty of our goals, and the quality, logic, and completeness of our plans to try to accomplish those goals. Occasionally, when I’m having a “difficult” day, I go back and re-read them with a big smile. The fifth reviewer submitted a very short review that was not positive and commented: “The proposal reads more like ‘We want to use every analytical tool we have’ than ‘Here are the questions we are trying to answer and here is how we will attempt to answer those questions.’” Despite that comment, the project was funded.
The pre-industrial Antarctic ice will provide insight into natural nanoparticles and microparticles, transported to Antarctica from the rest of the world, and how the number and elemental composition of those particles have changed over the last 40,000 years. In addition, we want to determine how much the nanoparticles (<200 nm) contribute to the total trace element concentrations, and how the microparticles (0.2 to 5 μm) contribute. Currently, this is a total unknown. Particles in the atmosphere can affect climate because they reflect, scatter or absorb sunlight and act as nucleation sites for formation of clouds. In addition, Fe-containing nanoparticles, transported through the atmosphere, may be an important source of Fe in the oceans. We also plan to measure particles entrapped in ice cores from other locations, such as Mt. Ortles in the Alps, over the last 7000 years to assess how nearby human activity, including the production of anthropogenic particles from industrial processes, has affected the number and composition of particles deposited from the atmosphere over time.
Have you had to develop any specialized chemometric techniques for your various applications of ICP–MS methods? What has been your greatest data analysis challenge in performing your research? In a more general sense, what were some of the major challenges you have encountered during your career of laboratory research? How important is sound experimental design in your work?
A number of chemometric techniques, including discriminant analysis and principal component analysis, are used to determine the best elemental “fingerprints” for comparing water composition to fish ear bone composition. Our collaborators relate trace element fingerprints in the bone samples to the trace element patterns in water in order to compare the element patterns in fish ear bones. We are also developing approaches using MATLAB programs, Excel filtering, and potentially machine learning approaches, to identify different mineral types in large numbers of nanoparticles and microparticles.
However, we also always look manually at raw data, and are constantly asking what factors other than the ones we are trying to assess could affect the data. For many samples, this includes potential sample matrix-induced changes in sensitivity and potential spectral overlaps in optical emission or mass spectra. We have spent a lot of time studying both matrix effects and spectral overlaps in both ICP–OES and ICP–MS, so we apply what we have learned to samples we analyze. For samples we measure for clients, we are constantly looking for multiple ways to measure the same element. This can include three or four different lines for optical emission, or sets of isotopes where possible for ICP–MS. We look for diagnostic signals that tell us about the plasma temperature, such as Ar emission or mass spectrometry intensities, concentrations based on both ion and atom emission intensities, ion-to-atom emission intensity ratios for the same element, or ratios of ion signals for elements with high ionization energy to elements with low ionization energies. These provide hints to when matrix effects or spectral overlaps are occurring for particular samples. The ultimate goal is to develop “intelligent instruments” that can diagnose analysis problems and take the appropriate steps automatically to overcome or avoid the potential sources of error. At a minimum, the intelligent instrument would warn the operator when the results for a particular sample might not be accurate.
Experimental design in terms of appropriate sample preparation, and careful measurement of process blanks to assess and avoid contamination (especially when measuring part per trillion concentrations) is critical. Statistically appropriate design of experiments prior to sample analyses is also essential for questions some of our clients and collaborators are trying to answer.
Occasionally, we are frustrated with clients who just want numbers now, or who are not willing to do the painstaking, time-consuming work to ensure contamination is not a problem, or who have not thought ahead about how they will interpret the analytical results. In one case, the client expected to see increasing concentrations by about a factor of 10 across a set of 10 samples. After refusing to measure blanks first because they needed results now, they measured their samples; all had the same concentration (most likely because their reagents were contaminated so the measurement was of the contamination in the reagents rather than the results of their experiment). My PhD research advisor used to ask us, “If you don’t have time to do it right, how are you going to have time to do it over?” Occasionally, clients also tell us we must have made an error in our measurements because the results are not what they expected. We treat these situations as a challenge for us to teach them.
Too often when we ask what accuracy and precision is needed to address the research question of interest, the answer we get is either “I don’t know” or “The best you can do.” Neither is a good answer. If you don’t know what accuracy and precision you need, how will you know if your results are statistically significant? “The best we can do” could mean using techniques developed by a group at NIST (National Institute of Standards and Technology), and extended by our group that is limited by the precision and accuracy of weighing out standards, not the precision of the measurement instrument. We have shown we could obtain 0.1% relative uncertainty in element concentration measurements rather than the more typical 10% uncertainly. The down side: time and cost of “perfectly” matrix matching standards to the samples and using “perfectly” matched internal standards with sample-standard-bracketing so it might take 20 hours to make the standards and measure 10 samples instead of one hour to make standards that are not perfectly matrix matched and 5 minutes per sample measurement. Another teaching opportunity!
I often remind my students that “weird results are good,” as long as you can prove the weird results are not due to a measurement error or a factor that was not considered. “Weird” results are an opportunity to learn something new from unexpected results. The lack of dependence of analyte mass on ICP–MS matrix effects is one of those “weird” results.
What would you consider to be the most meaningful contributions of your research work? What is most exciting to you about your work?
My first reaction is that is something that is probably best answered by others. I think the biggest impact in general has been to make the processes that control ICP–OES and ICP–MS signals, potential sources of error, and potential means to overcome those errors more clearly understood by others. Others may include researchers, ICP–OES users, ICP–MS users, and students both in our laboratory and in the laboratories we have had close collaborations with who have had an impact in their positions after finishing at Ohio State.
Certainly, there are several investigations done in our laboratory that I am particularly fond of. Our laboratory, and Paul Farnsworth’s group at Brigham Young University, independently, but almost simultaneously, unexpectedly discovered that some sample aerosol droplets and desolvated particles survive in the >6000 K ICP and can cool a surprisingly large volume (about 2 mm wide) of the plasma surrounding each. Several students in our laboratory did a superb job using those effects to gain a better understanding of fundamental processes in the plasma.
Later, a series of experiments injecting monodisperse droplets into an ICP to separately investigate droplet desolvation, particle vaporization, atomization, ionization, diffusion, and ion transport from the atmospheric plasma through the mass spectrometer was challenging, exciting, and provided clarity that was previously unattainable. Those experiments led us, about 10 years later, to join other groups developing and using single particle ICP–MS to measure large numbers of engineered nanoparticles and most recently mineral and other nanoparticles and microparticles entrapped in ice cores.
Some of the students in my laboratory also made great efforts to investigate ion-molecule reactions to overcome spectral overlaps in ICP–MS, and others study the fundamental basis, requirements, capabilities, and limitations of single particle ICP–MS for micro- and nanoparticle analysis.
A small fraction of the work we have done on the fundamental processes involved in ICP–OES and ICP–MS was funded by NSF. NSF then decided that all the related questions had been solved. Certainly, our recent work on matrix effects in ICP–MS showed that was not correct. Fortunately, we have had an almost 30-year long collaboration with PerkinElmer, who have supported our continuing research in that area.
While I’ve been criticized for not publishing more, and we certainly have made mistakes in some of our publications, I have been demanding that only data that are proven to be reproducible, that we are convinced are not biased by processes other than those we used in explaining the results, and that have something new and potentially important to say, are published. We have learned lots from experiments that “failed” or were not complete enough for us to decide to publish them in scientific journals; many of those were important to our learning process, and were included in theses and openly discussed at conferences.
What words of wisdom do you have for any young people interested in a scientific research career?
Find something you enjoy doing and are passionate about, and do it the best you can. It will not be easy (most easy things have already been done), but with determination, the opportunities are incredible. Before joining a research group, find out how the group is operated (there are many successful ways to run a group; you need to find a style that fits with yours). That may be as important or more important than the research topic. It is more important to learn how to do research in graduate school than what specific research you work on. That is much easier to do in the right environment for you.
Don’t limit yourself. Grades are not necessarily an indicator of future success. Learning and being able to use what you learn is more important. In my view, just “learning” to get an A and then forgetting the information so you are not able to use it in the future is a waste of time and someone’s tuition money, and does not lead to future success. The most successful graduate student I have ever had in my research group ended up in my group because no faculty in the Chemistry Department would take him; his grades and test scores weren’t great. It turned out he could learn independently and from others, he could apply what he knew to entirely different problems, he had great determination even after “failed” experiments, and he openly shared what he knew and learned.
1. Ohio State subaward: T.H. Darrah (PI), J.W. Olesik (coPI), “Human Placental Morphology, Function, and Pathology: Relationship to environmental Exposures and Newborn and Child Health” (National Institutes of Health Study, Bethesda, Maryland), NIH R011ES029281-01.
2. P. Gabrielli and J.W. Olesik, “Atmospheric Mineral Nanoparticles in Antarctic Ice during the last Climatic Cycle” (National Science Foundation Study, Alexandria, Virginia), NSF #1744961.
Jerome Workman, Jr. is the Senior Technical Editor for Spectroscopy. Direct correspondence to: firstname.lastname@example.org●