Developments in Dioxin Analysis

Bryan Vining from SGS Environmental Service in Wilmington, North Carolina, reveals some of the cutting-edge research his team has performed involving dioxin analysis, and proposes some future possibilities for the direction of this field.

Your team has been involved in a wide range of projects involving dioxin analysis, including work on the release of dioxins from reservoir sources during beach nourishment programs. How did the work on beach nourishment arise and why are dioxins from reservoir sources a concern?

Vining: That story is a good one, actually. Dr. Yves Tondeur, who wrote one of the first methods for the analysis of all 17 2,3,7,8-chlorinated polychlorinated dibenzo-p-dioxins and polychlorinated dibenzofurans (PCDD/Fs) (1), was strolling along the beach following a beach renourishment program. Beach renourishment is a fancy term for, basically, dredging some of the marine sediment and depositing it on the beach to be spread around. This process is used to reverse the effects of beach erosion due to various natural processes. If this process is not carried out, homes and businesses on the beach may eventually not be sustainable. Dr. Tondeur observed that balls of clay were scattered across the beach as a result of this process.

He immediately thought of "so-called" ball clays, which are mined for various industrial purposes and are notable for being both millions of years old and containing dioxins. Interestingly, the dioxins in these ball clays have a distinctly non-anthropogenic signature and can be present at rather high levels. He grew concerned that beachgoers might pick up and handle one of these balls of clay and potentially expose themselves to relatively high levels of dioxins. So, he sampled some of the clay balls and we analyzed them in the laboratory.

Such renourishment efforts are relatively common, so there was a real concern that dioxin exposure could be taking place with beachgoers. The biggest danger would be from ingestion of some of the clay particulate left on dirty fingers, most likely with food. Of course, some children like to eat dirt, in which case exposure could be substantially higher because of their size and behavior relative to adults. If the balls of clay contained ball clay, the exposure could be a fairly significant concern. Luckily, in this case, the dioxin levels turned out to be quite low, but we found some other interesting things too and we published our findings (2).

What analytical technique did you choose for this project from the wide array of approaches that are available to analyze dioxins?

Vining: The wide array of approaches you mention really all boils down to the same basics. An extraction of some sort takes place, followed by (usually) an extract cleanup and analysis on a high-resolution gas chromatography–high-resolution mass spectrometry (HRGC–HRMS) system using a nonpolar GC column (predominately DB-5 or DB-5MS columns). These assays usually employ isotope dilution, a technique in which the target analyte or analytes are measured relative to an isotopically labeled version of itself. Method 8290 featured isotope dilution when it was published because this technique is very resistant to analytical difficulties and challenges, resulting in excellent analytical performance. Most, if not all, methods written for the analysis of PCDD/Fs have continued that practice.

The key elements of a successful isotope dilution analysis are often unexamined and unaddressed in these approaches, leading to variation in parameters where it is undesirable and allowing (or even promoting) variation where it isn't (such as in the number of labeled standards used).

We used our own implementation of US EPA Method 8290. Tondeur wrote the original version in 1987, and spent his entire professional life perfecting the original version. To date, we've introduced a number of enhancements in our approach over any of the published protocols currently in place. Most notably, we use all 17 ¹³C₁₂-2,3,7,8 chlorinated dioxins and furans as isotope dilution internal standards. That is, each of the dioxins and furans that have a toxicity equivalence factor associated with it is monitored with an isotopically labeled standard in our analysis. Determinations of toxic equivalence to 2,3,7,8-TCDD are therefore achieved with the best possible accuracy and precision (3).

Another very important enhancement we developed, refined, validated, and have used for the past decade and half is our batch control spike system. This system essentially allows us to verify the correct spiking of our samples with the labeled standards mentioned above. The same aspect of isotope dilution technology that makes it so very powerful also makes it very easy to make an irretrievable mistake. Namely, you can get data that look really great from a QC perspective but that are actually really wrong if you believe you have spiked your samples with the labeled standards correctly when, in fact, you haven't. Unfortunately, the QC samples built into the published methods don't consider this aspect of the analysis or the unpleasant potential consequences of it. We took it on ourselves to address this problem in our own way, since the methods did not. Finally, the batch control spike system allows us to correct for some systematic errors that might occur in the analysis.

The enhancements also include ways to rule out false positives, significantly better-than-required GC resolution, and, of special import, a performance-based assessment of the fitness for purpose of the data. This latter aspect allows us and end data users to put the laboratory performance into a situational perspective. Not all data need to be generated with the same analytical uncertainty for all purposes. For example, for exploratory work, a relatively high analytical uncertainty might be tolerable, equating to an "is it there or not?" type of situation. On the other hand, if the end data user wants to answer the fundamental question of whether the levels present are safe, the lowest possible analytical uncertainty may be desirable. Our performance-based data quality assessment allows us to quantify these desires and determine whether the analysis we've performed satisfies those desires.

Were there any analytical problems you encountered on this project?

Vining: The analyses of the samples in this project actually went quite well, and we really had very few issues or problems with the analyses. Really, the single biggest challenge with a sample of this sort is ensuring it is spiked correctly. As we mentioned earlier, a failure to spike correctly can lead to very wrong answers. A review paper on this topic was published by our laboratory in 2009 (4). Since these were clay samples, the challenge was to have the labeled spikes in the same environment as the molecules you want to analyze. If it's not, you may not get a representation of the actual extraction performance.

Within a sample of any decent size, you can have "microenvironments," where the extraction behavior may differ from another part of the sample. For example, you might have a pocket of soil with much different organic carbon content than the sample as a whole. If you make no effort to integrate the spike thoroughly with the sample, you run the risk of getting different extraction behavior for your labeled standards than for your target analytes. With isotope dilution assays, which most dioxin assays are, such a difference means you won't get the right answer. One can use a variety of approaches to address this concern. What matters is that it's addressed in some way.

What were the results of your findings?

Vining: First, we did observe dioxins in the samples, indicating that the dredging used for beach renourishment does reach reservoir source of dioxins. The WHO-2005 TEQs ranged from 0.411–5.78 ng/kg of clay. Sand samples showed lower levels. To put those numbers in perspective, guidelines from the Agency for Toxic Substances and Disease Registry call for dioxin WHO-2005 TEQs to be below 50 ng/kg in soil (residential). The US EPA set its oral reference dose for chronic exposure at 7 × 10^-4 ng per kilogram of body weight WHO-2005 TEQ per day in 2012. Now, you'd have to ask a risk assessor whether these clays could be considered a problem, but we think one of the interesting findings is that the clays in this case did contain dioxins.

Second, there were some really interesting observations in the specific dioxins that were found. Most dioxin assays only look for the 17 2,3,7,8-tetra- to octachlorinated dioxins/furans, which are the ones the scientific community labels as toxic, and total dioxins/furans. However, these analyses were a great case in point about why you might want to look beyond just those. A number of reports have speculated on the origin of dioxins in ball clays, marine sediments, and other sources that are secluded from exposure to dioxins formed by human activities (5–7).

Generally, these sources show a predominance of octachlorinated species with significantly lower levels of dioxins at each lower chlorination level. Also, in general, these sources tend to be found with only dioxins in them and little or no of the polychlorinated dibenzofurans that accompany them in anthropogenic sources. No one knows for sure where these dioxins came from or why octachlorinated dioxin (OCDD) is the dominant dioxin in these soils. Some have speculated that all the dioxins one finds in such samples originated with OCDD that underwent one or more types of dechlorination reactions.

We found similar patterns in these clays. OCDD was the most prominent, and lesser-chlorinated dioxins were found at progressively lower levels. More interesting, though, was the prominence of 1,4,6,9-chlorinated dioxins. We found a lot more of these species than the 2,3,7,8-chlorinated dioxins that are usually looked for.

It's not clear why these species predominate. If one accepts the premise that the dioxins in these types of samples originate with dechlorination of OCDD, this pattern suggests that dechlorination in the marine environment that these clays came from was strongly preferred at the lateral (that is, 2,3,7,8) positions (that is, detoxification), rather than at the peri (that is, 1,4,6,9) positions. This 1,4,6,9-chlorination pattern has been observed in several other worldwide marine environments.

It's also possible that all chlorinated species were formed millions of years ago under conditions that thermodynamically and kinetically allowed each possible reaction to go to completion. It is over time (millions of years) that the (relatively) slow degradation of the various species took place, and depending on their respective redox potential, the composition of the clays as we observe them today evolved to the point where the species that are the most resistant to degradation (anaerobic) such as the (relatively speaking, highly oxygenated) dioxins are the only ones detected today. And within that group of compounds, specific isomers/congener's redox potential contributed to the current (detoxification) profile.

Regardless of the origin of the dioxins found in the clays, the data show that looking beyond the 17 2,3,7,8-chlorinated dioxins/furans leads to interesting observations one would miss otherwise.

Have you been involved in any other projects involving dioxin analysis?

Vining: I've been involved in many interesting dioxin analysis projects. In 2009, our team demonstrated that chemical impurities (in this instance, predioxins) present in solid samples contaminated with pentachloropenol form dioxins during the sample extraction process for analysis (8).

If one performs the extraction with acetone and heat (and most extractions involve heat in some way) present, predioxins present in the sample form dioxins. This formation can be attenuated through the addition of acid by avoiding the use of acetone (and potentially similar solvents) during extraction, or both. This formation reaction rises to a level of some concern because many protocols for dioxin analysis (such as EPA Method 1613) specify the use of acetone in the spiking of samples.

Similarly, I was part of the team involved in the analysis of dioxins formed as microcontaminants in the manufacture of pentachlorophenol. Dioxins in pentachlorophenol make a challenging target because the concentrations can easily extend over a range of six to seven orders of magnitude from the tetrachlorinated dioxins (at the lowest concentrations) to the octachlorinated dioxins (at the highest concentrations). A paper published in 2010 detailed our approach to this challenging analysis (9).

For the past 15 years, our laboratory has been involved in a number of efforts to better define and assess the performance of dioxin analyses. In particular, I'm part of the effort to better define and refine the estimation of measurement uncertainty in dioxin analysis. We noticed that while every accreditation standard for environmental analysis, including dioxins, talks about being able to state an uncertainty of analysis, relatively little thought or effort seems to go into this key part of an analysis.

Many laboratories in the United States don't report an uncertainty unless asked to, which often involves some significant efforts on the part of the laboratory staff because they don't get asked very often for the measurement uncertainty. Other laboratories may report an uncertainty but do so as a static percentage of the result obtained, regardless of the result. Analytical chemists know, though, that the variation (and, hence, uncertainty) in an analysis grows larger in proportion to the result as the result itself gets smaller. So, simply reporting a static percentage of the result is an easy way out that doesn't reflect the reality that we know exists.

The system for uncertainty estimation we've developed over the past 15 years in our laboratory reflects this reality. In fact, it's a great tool for telling when something is wrong with the analysis of a particular sample because its uncertainty won't fit the trend demonstrated by the other samples. A lot goes into this process, and we're continually refining it. Our software is currently undergoing a revision that will include some of the latest refinements in this process. I've even given some thought to how we would extend the ideas underlying the assessment of dioxin measurement uncertainty to other assays, including those that don't rely on isotope dilution, but that's a complex story for another time.

One other particularly interesting study involved helping a facility track down a process change that had increased its dioxin emissions (10).

We realized that something interesting and unexpected was going on precisely by looking at the non-2,3,7,8-chlorinated dioxins/furans in the results. Indeed, it turns out, much like with the clay results discussed earlier, a distinctive pattern could be seen in the particular dioxins that were prominent in the samples. That pattern suggested that something had changed in the facility's process. To make a long story short, the facility did indeed find a very subtle change in its process that one wouldn't have thought would have the impact it did. The facility sorted out the issue (and ended up making a number of other changes), and its emissions are significantly much lower now.

Do you have any other comments about the analysis of dioxins generally?

Vining: First, we would like to really encourage users of dioxin data to think beyond the 17 2,3,7,8-chlorinated dioxins. The examples described earlier show how useful such data can be. However, if you don't look at it, you can't use it. Many laboratories don't even look at the non-2,3,7,8 dioxins/furans, and many of them are reluctant to share raw data with a client who asks for it. We think that trend is an unfortunate one because there may be quite useful information contained therein. We especially urge this for anyone looking to analyze dioxins for forensic purposes.

Second, we strongly recommend that the end users of data think about the uncertainty of the measurements of dioxins and how that might affect the assessment of the data produced by a laboratory. We have a system where we can quantify the desires of end data users in terms of the uncertainty they're willing to tolerate in their results and even tell them what the effects might be on the needed sample size! That an uncertainty is associated with a measurement seems to be often forgotten and, when not forgotten, largely ignored.

Finally, we really need a revamp and update of all the published dioxin analysis protocols. Most of the methods for dioxin analysis are superficial updates of the original EPA Method 8290 published in 1987 with few, if any, substantive improvements. You need only consider that none of them specify the use of all 17 2,3,7,8-chlorinated dioxins or address the importance of proper spiking. To add to those problems, the methods mostly specify the use of sector mass spectrometers, typically called high-resolution mass spectrometers (HRMS), for the analysis of dioxins.

However, I think all the HRMS instruments currently used will become obsolete, and it's likely that by 2025 GC–MS-MS systems or new technologies with a similar performance will be the preferred platform for dioxin analysis - once the dioxin analysis community grows more familiar with them.

The fundamental issue with the current methods is simply that they haven't gotten a real, substantive revision in nearly 30 years, but the technologies for dioxin analysis - most notably isotope dilution - have been moving forward in those three decades. The situation is something like having the traffic and automotive laws from the mid-1980s in effect just as they were then, completely ignoring all the technological advances in automobiles since then.

The unfortunate aspect of this situation is that US EPA's Office of Water publishes an annual method update rule, so the opportunity is already there but is unused.

All of that said, dioxin analysis is here to stay, in whatever form, for quite some time to come. As part of our culture, our laboratory aims to push the frontiers of isotope dilution technology in relation to dioxin analysis forward. It's a challenge that I, for one, greatly look forward to. I am also grateful to Dr. Tondeur and Jerry Hart for the opportunity to be part of this team's efforts.

References

(1) EPA Method 8290 (1987).

(2) Y. Tondeur, B. Vining, K. Mace, W. Mills, and J. Hart, Chemosphere88(3) 358–363 (2012).

(3) G. Becher, L.S. Haug, and C. Thomsen, Talanta63(5), 1115–1122 (2004).

(4) Y. Tondeur and J. Hart, TrAC, Trends Anal. Chem.28(10), 1137–1147 (2009).

(5) Y. Horii, B. van Bavel, K. Kannan, G. Petrick, and K. Nachtigall, Chemosphere70(7), 1280–1289 (2008).

(6) Environ. Sci. Technol.34(2), 4524–4532 (2000).

(7) Environ. Sci. Technol. 36(16), 3542–3549 (2002).

(8) Unforeseen Formation of Polychlorinated Dibenzo-p-Dioxins During the Extraction of Pentachlorophenol-Contaminated Guar Gum, Dioxins 2009, Beijing, China.

(9) Y. Tondeur, T. Nestrick, H.A. Silva, B. Vining, and J. Hart, Chemosphere80, 157–164 (2010).

(10) Y. Tondeur, B. Vining, K. Mace, W. Mills, and J. Hart, Chemosphere88, 358–363 (2012)

(11) Y. Tondeur, B. Vining, J. Serne, and J. Hart, Chemosphere126, 47–52, DOI:10.1016/j.chemosphere.2015.01.046 (2015).

This interview has been edited for length and clarity. Please visit our websites for more interviews like this.