Page 0 van 38
Detection of Elemental and
Synthetic Toxic compounds
in Marine Mammals
A
bioaccumulation
study
Page 1 van 38
MSc Chemistry
Analytical Track
Master Thesis
Detection of Elemental and Synthetic Toxic
Compounds in Marine Mammals
A bioaccumulation study
By
Kai Dollevoet
11820543
42 EC
February 2020 – October 2020
Supervisor/Examiner (UvA):
Examiner (UvA):
Assist. Prof. Bob Pirok
Prof. Dr. Arian van Asten
Supervisor (MU):
Daily supervisor (MU):
Assoc. Prof. Dr. John Harrison
Dr. Marie-Anne Thelen
Daily Supervisor (UvA):
Supervisor (UvA):
Assist. Prof. Andrea Gargano
Assist. Prof. Saer Samanipour
Massey University
Institute of Natural and Mathematical Science
(Albany Campus)
University of Amsterdam
Van ‘t Hoff Institute for Molecular Sciences Faculty of Science
Page 2 van 38
Foreword
Before reading this thesis I would like to thank you as reader for taking the interest. Months of investigation has been included in this thesis, which will hopefully contribute to a better understanding of the effects of pollution of our seas and the marine life in it. I hope it shows that we as humans have a responsibility as the superior form of life on this planet which should not be neglected. Also we should be aware that marine mammals have a similar digestive system to humans and that, though be it in different quantities, accumulation occurs in our bodies too. Unfortunately, parts of this study have not been completed because of precaution measures taken by the Dutch and New-Zealand governments for the COVID-19 outbreak. It has been completed to the best and future steps which would have to be made are taken into account as recommendations for further research.
Page 3 of 38
Abstract
Marine mammals serve as sentinel species for marine ecology due to the absence of gills. This renders them to easily accumulate persistent organic pollutants (POP) in their bodies. This study focusses on the in vitro analysis of two groups of POPs introduced in to environment by human activities. Heavy metal with mercury in particular and poly- and perfluoroalkyl substances (PFAS) were analyzed using microwave induced plasma atomic emission spectroscopy (MP-AES) and untargeted UPLC-QTOF respectively within samples of marine mammals. The mercury studied in dolphin liver required microwave destruction as a sample preparation step. Also the mercury had to be reduced into its elemental state in a multiple sample introduction system (MSIS) using NaBH4 which increased the sensitivity of the MP-AES. This resulted in an 5.114 mg/kg (w/w) mercury concentration in the dolphins liver tissue. The untargeted PFAS analysis in seal tissue was conducted using SPE and activated carbon cleanup. It was then followed by UPLC-QTOF analysis and digital data analysis in the form of a Python script. Unfortunately no results about the tissues could be reported due to technical difficulties, however, standard analysis proved a potential for this method.
Page 4 of 38
Table of Contents
Foreword ... 2 Abstract ... 3 Table of Contents ... 4 1. Introduction ... 6 2. Theoretical aspects ... 82.1 Elemental analysis using MP-AES ... 8
2.1.1 Nitrogen plasma ... 8
2.1.2 Multimode Sample Introduction System ... 8
2.1.3 Auto and FLIC Background subtractions ... 9
2.2 PFAS analysis using UPLC-QTOF MS ... 11
2.2.1 PFAS compounds ... 11
2.2.2 Python programming ... 11
2.2.3 Kendrick Mass Defect ... 11
3. Experimental ... 12
3.1 Mercury analysis using MP-AES... 12
3.1.1 Instrumentation ... 12
3.1.2 Reagents and sample collection... 12
3.1.3 Standard preparation ... 13
3.1.4 Microwave digestion ... 13
3.1.5 Elemental analysis ... 14
3.2 PFAS analysis using LC-QTOF MS ... 14
3.2.1 Instrumentation ... 14
3.2.2 Reagents and sample collection... 15
3.2.3 Sample preparation ... 15
3.2.4 UPLC-QTOF MS analysis ... 16
3.2.5 Data analysis ... 17
4. Results & Discussion ... 18
4.1 Mercury analysis ... 18
4.1.1 Interference study and MSIS mode comparison ... 18
4.1.2 Linearity ... 20
4.1.3 Digested sample analysis ... 21
4.2 PFAS analysis ... 21
Page 5 of 38
4.2.2 Seal tissue analysis ... 24
4.2.3 Digital data analysis ... 25
5. Conclusion ... 27
5.1 Elemental analysis of mercury in dolphin liver ... 27
5.1.1 Future research ... 27
5.2 PFAS analysis in seal tissues ... 27
5.2.1 Future research ... 28
6. Acknowledgements ... 29
7. References ... 30
Appendix A: Python script ... 35
Page 6 of 38
1. Introduction
Unlike most marine species, marine mammals do not have gills. This renders them incapable of breathing underwater. In addition, they also lack the possibility to excrete undesired substances back into the seawater using this organ [1]. This results in larger accumulation of persistent organic pollutants (POPs) and (organo-) metals [2, 3]. This renders marine mammals more vulnerable to poisoning as a result of fish consumption. As these mammals experience similar digestive processes as humans observing these species can contribute to monitor potential food safety for caught fish. Apart from being the top predators, the mammals serve as sentinel species which used for analyzing the ecosystem it previously lived in. Examining this on a regular basis could help monitoring these systems and human food safety.
Therefore, methods have been developed to monitor toxic substances such as heavy metals [3-5]. In particular, the concentration of mercury is of interest as it is toxic in many of its orientations and is as one of the few metals which can become volatile. Many studies have been conducted in investigating and determining several orientations as for example elemental mercury and methylmercury [6-9]. One of these methods was described by Mohammed et. al (2017) who conducted a cold vapor atomic absorption spectrometry (CV-AAS) method after ambient acid digestion on comparable shark samples looking for the total mercury content [9]. Another method using GC-MS and ICP-MS by Duarte et. al. (2013) investigated the ionic speciation of mercury as Hg2+ and CH3Hg+ which granted insight in the extraction using a salt-acid matrix [6]. Tokutaka et. al. (2004) studied hepacytosol enzymes of porpoises and seal. This study observed the mercury-binding properties of the enzyme with the use of HPLC-ICPMS. This resulted not only in information on this aspect but also on metal complexes with selenium [7].
A related study was conducted by Stockin et. al. (2007) which focused on New Zealand Common Dolphins (Delphinus Delphis) and investigated mercury accumulation as well as other trace elements using ICP-OES and ICP-MS [8]. However, to the knowledge of the author, no other research has been reported studying the possibility of using the cheaper MP-AES for trace metal determination in marine mammals. The most-related study in this context was reported by Savoie et. al. (2018), who studied mercury concentrations in Atlantic salmon using cold vapor – microwave induced plasma – atomic emission spectroscopy (CV-MP-AES) [10].
Most methods developed thus far require the use of an ICP instrument which is more costly than MP while capable of reaching similar detection limits and higher compared to Flame-AAS [11-16]. In particular the possibility of using nitrogen produced by a generator coupled to the instrument can significantly reduce the operational costs [17-19].
In this study the use of MP-AES for trace metal elemental analysis, with mercury in particular, of marine mammal samples will be investigated. It will focus on measuring a significant and reliable signal within a linear range. Also it will investigate a microwave assisted microwave digestion of a Common Dolphin sample to create a measurable sample. When accomplished this study produces a cheaper and reliable method for the analysis of heavy metals in marine mammals.
Page 7 of 38
Another common group of POPs which has gained interest over the last years due to their inert nature are per- and polyfluoroalkyl substances (PFAS) [20, 21]. This group of synthetic compounds are concerned to negatively impact human and animal health [22] and are widely distributed throughout the environment [20, 22, 23].
Similar to heavy metals PFAS also accumulate in marine mammals and are usually detected using (UP)LC-MS methods [24-26]. These studies, however, often examine selected compounds within the group instead of the broad range of PFAS as described by the Environmental Protection Agency (EPA) [27]. This study aims to develop a method capable of detecting multiple to all registered PFAS from the EPA database’s master list in a marine mammal sample. Achieving this goal would result in a new untargeted screening method for PFAS analysis in marine mammals.
Page 8 of 38
2. Theoretical aspects
2.1 Elemental analysis using MP-AES
2.1.1 Nitrogen plasma
Nitrogen plasma is commonly used in MP-AES systems [10, 28-32]. This variant is cheaper compared to argon plasma used in the more common ICP systems [17-19], mainly since it can be generated by a nitrogen generator. This instrument can be attached directly onto an MP-AES system as for example an Agilent 4200 MP-AES® [29]. However, apart from this cheap aspect nitrogen plasma does hold negative aspect. Its lower electron activity compared to argon results in a larger redox potential [28]. Though this could be described as a negative aspect it leaves a steady noise because of the lower energy from the lower plasma temperature (5000 K approx.) [30]. This is hardened by a thermodynamic observation as the nitrogen plasma appears to be close to a local thermal equilibrium. This means that the electron density, determined by Thomson scattering and the excitation temperature agree well, demonstrating its stable operation [30]. Another notable negative aspect is that the analyte excitation by nitrogen plasma is not yet completely visual. However, the most dominant molecules have been identified either in dry and wet environment, which covers most of the signal. Because of this steady signal it would be easier to computationally remove background noise rendering a clear signal.
Although ICP is usually assumed a more established technique, MP using nitrogen plasma is also applicable for mass spectrometry. MP-TOFMS (Microwave plasma-time of flight mass spectrometry) as demonstrated by Schild et.al. (2018) [31] proves that this technique has promising futuristic aspects in this related area as well. Therefore it could be questioned whether the use of nitrogen plasma in the future would become more desirable over argon as a whole.
2.1.2 Multimode Sample Introduction System
The multimode sample introduction system or MSIS is a glass instrument which allows the user to use three different modes of sample introduction for MP and ICP systems [33]. The three options use two different modes of operation and allows the user to use them simultaneously. Nebulization mode
In the nebulization mode the most traditional mode of sample introduction is applied. A dissolved sample is nebulized using an inert gas generating very fine aerosol droplets. In this orientation the droplets are able of migrating towards the plasma inlet of the MP-AES instrument where the solvent is evaporated and the atoms become excited. A schematic set-up of the instrument used in this mode can be observed in figure 1.
Figure 1 MSIS operation set-up for Nebulization mode. This figure was derived from Agilent [33]
Page 9 of 38 Cold vapor mode
Cold vapor mode or CV-mode allows the user to introduce more volatile and hydride forming elements into the MP-AES [10, 34-36]. The set-up as pictured in figure 2 involves the use of a reductant which promotes the reduction of specific ionic elements like mercury (reducing the atoms into the neutral elemental state) and the formation of metal hydrides [33-35].
A commonly used reductant in similar
experiments is sodium borohydride (NaBH4) [37-39]. Following the equation below, it reduces ionic mercury into elemental (gaseous) mercury as studied by Bramanti et.al. (1999) [40].
𝐵𝐻4−+ 8𝑂𝐻−+ 4𝐻𝑔2+→ 4𝐻𝑔0(𝑔) + 𝐵𝑂2−+ 6𝐻2𝑂
During operation the reductant which is often stabilized in acidic or basic buffered solutions flows from the top inlet towards the sample entering from the bottom inlet. Between both inlets there is a small space where the two solvents react and flow down as a thin film over the reaction cone in the MSIS (outside of the bottom inlet). An inert gas flows through the nebulizer unit to promote the volatilization and move the elements towards the MP or ICP inlet.
Dual mode
In the Dual mode both modes are operating simultaneously. A schematic set-up can be observed in figure 3. Though this set-up grants the used the possibilities to measure volatile, non-volatile and hydride forming elements it also has two minor disadvantages. One being that sample consumption during a measurement is doubled which can be disadvantageous when the volume is low. Another problem is that possible interferents can also become detectable which worsens
the signal in MP-AES influencing the methods sensitivity. 2.1.3 Auto and FLIC Background subtractions
The MP expert® software developed by Agilent® for operation and data analysis of the 4200 MP-AES provides the user with multiple possibilities for background subtraction. Most important are the auto function and fast linear interference correction (FLIC) function. The auto subtraction uses the blank and provides the user with a common basic background subtraction, whereas FLIC can be used to remove the interference of specific element.
FLIC requires two extra samples in the sequence which provides the software with necessary information to process the subtraction. The first is a standard which holds a nearly equal
Figure 2 MSIS operation set-up for Cold Vapor mode. This figure was derived from Agilent [33]
Figure 3 MSIS operation set-up for Dual Mode. This figure was derived from Agilent [33]
Page 10 of 38
concentration of the interfering element (if more elements interfere than a separate standard must be provided for every element) as present in the sample. The second includes a relatively high concentrated sample, for example the most concentrated standard used for the determination.
An example of the auto and FLIC subtractions are presented in figure 4which shows the total signal and modelled analyte signal for each sample. The figure shows the interfering element signal, in this case iron, which causes a decrease in the baseline signal, can be corrected when applying FLIC. Key points demonstrating the application are the unreduced right shoulder of the FLIC 0.75Hg10Fe sample in the analyte spectrum compared to the same sample when FLIC is not applied. The formation of the positive baseline of the FLIC sample signals is another as well as the complete reduction of the Fe peak at 253.57 nm in the analyte model (although this is also compensated in the auto background reduction).
Figure 4 Overlay of the samples Fe standard (10mg/L of Fe in 2% HNO3) & 0.75Hg10Fe sample (0.75 mg/L of Hg + 10 mg/L Fe
in 2% HNO3; measured using auto and FLIC background correction).
The major drawback of FLIC is that a comparable concentration of the interfering element is required for a correct subtraction to make it useful for accurate determinations. This however can be compensated by preparing and measuring multiple interferent standards in one sequence as the software has the option of choosing the interferent and analyte solutions post-analysis without loss of information.
Although FLIC and auto background subtractions can be useful they still cause a loss in information. Therefore the raw data should always be observed as this could also be used as a reference to whether the correct concentration of the interfering element is applied.
253,45 253,5 253,55 253,6 253,65 253,7 253,75 253,8 253,85 In te n si ty Wavelength (nm) Fe standard sample 10 mg/L Analyte model Fe standard sample 10 mg/L Raw data 0.75Hg10Fe sample raw data
0.75Hg10Fe sample analyte model
FLIC 0.75Hg10Fe sample raw data
FLIC 0.75Hg10Fe sample analyte model
Page 11 of 38
2.2 PFAS analysis using UPLC-QTOF MS
2.2.1 PFAS compounds
PFAS compounds are a class of fluorinated alkyl chains (R-(CFn)n) of variating length, orientations and functional groups [41, 42]. According to the American Environmental Protection Agency (EPA) over 3000 species exist [27]. The substances are often manufactured for different applications like protective coatings, surfactants, lubricants, fire retardants and pesticides [41]. PFAS are due to their fluorinated structure highly persistent and can have both hydrophobic and hydrophilic properties [43]. These properties render the compounds to be accumulative and as a result of that hard to analyze as they are wide spread through all sorts of substances as for example drinking water [22, 23] and laboratory equipment [44].
2.2.2 Python programming
Python is a multilanguage open source programming/data modelling tool. By importing common programming packages as for example pandas it is capable of processing datasets in multiple programming languages in the same script [45]. Python can be accessed through multiple writing programs like Jupyter Notebook & Spyder [46]. The application has the capabilities to read and write .CSV files and use its content to perform calculations and replicate info related to that data.
2.2.3 Kendrick Mass Defect
The Kendrick Mass Defect (KMD) is a tool used in mass spectrometry Imaging to define groups of compounds with the same reoccurring group in its backbone and its functional groups [47]. It is often used to analyze alkyl-compounds in organic and fossil fuel samples using the reoccurring CH2 unit to determine the KMD [48]. However, the KMD can also be used for other repeating units as for example the CF2 unit in PFAS compounds [49, 50]. The KMD in mass spectrometry is calculated as described in the equations below [48].
𝐾𝑒𝑛𝑑𝑟𝑖𝑐𝑘 𝑀𝑎𝑠𝑠 (𝐾𝑀) = 𝑚𝑧 ∗ 𝑅𝑜𝑢𝑛𝑑𝑒𝑑 𝑢𝑛𝑖𝑡 𝑚𝑎𝑠𝑠
𝐸𝑥𝑎𝑐𝑡 𝑢𝑛𝑖𝑡 𝑚𝑎𝑠𝑠 = 𝑚𝑧 ∗
12+2∗19 (=𝑟𝑜𝑢𝑛𝑑𝑒𝑑 𝐶𝐹2)
12.011+2∗18.998403 (=𝑒𝑥𝑎𝑐𝑡 𝐶𝐹2)
𝐾𝑒𝑛𝑑𝑟𝑖𝑐𝑘 𝑀𝑎𝑠𝑠 𝐷𝑒𝑓𝑒𝑐𝑡 (𝐾𝑀𝐷) = 𝑅𝑜𝑢𝑛𝑑𝑒𝑑 𝐾𝑀 − 𝐸𝑥𝑎𝑐𝑡 𝐾𝑀
By plotting the KMD versus the detected m/z values in a scatter plot a relation between similar compounds becomes clear. For structures with the same repeating unit a KMD will be found which has only got a negligible deviation from other related compounds. By plotting these values versus the m/z will result in a slightly tilted line.
Page 12 of 38
3. Experimental
3.1 Mercury analysis using MP-AES
3.1.1 Instrumentation
The tissue samples were weighted using an analytical balance (ABJ-NM/ABS-N, Kern, Balingen, Germany) and then digested in a Multiwave Pro Microwave Reaction System SOLV by Anton Paar (Graz, Austria). This machine was equipped with an ROTOR 16HF100 carrying 16 high pressure reaction vessels of which one contained a controlling pT-probe for pressure and temperature monitoring. The commercial 4200 MP-AES system by Agilent Technologies (Santa-Clara, USA) was used to perform the elemental analysis of the digested samples. The system was equipped with a multimode sample introduction system (MSIS) spray chamber and SPS 3 Autosampler both manufactured by Agilent Technologies (Santa-Clara, USA). The sample (bottom inlet) and reductant (top inlet) where pumped individually using the peristaltic pump system of the MP-AES system through complementary PVC Solvaflex (orange/green tabs) pump tubing with an 0.38 mm id. The waste was drained using PVC tubing with 3.18 mm id (black and white tabs) equipped to the same peristaltic pump. To prevent the nebulizer gas from escaping the sample outlet the inlets not in operation were capped off. The nitrogen and argon gasses where both supplied by BOC (Linde Group, Auckland, New Zealand) in the absence of a nitrogen generator producing sufficient purity. The instrument was operated and data acquired using the MP Expert software (version 1.6) provided by Agilent Technologies.
3.1.2 Reagents and sample collection
Trace elemental grade metal solutions of mercury and iron (Sigma-Aldrich, St. Louis, USA) were purchased for the preparation of standards. As these standard should be prepared in nitric acid a 68% trace analytical grade (HNO3 68%, Thermo Fisher, Waltham, USA) was used in the microwave digestion and in dilution as a solvent for the standards. For the dilution and also to add up to volume for the digestion, fresh 18.2 Mcm filtered water (Milli-Q Reference, Merck, Kenilworth, USA) was used.
The reductant solution containing 1% sodium borohydride (NaBH4, 99.99% trace metal grade, Sigma-Aldrich, St. Louis, USA) in a prepared 0.1 M sodium hydroxide solution (NaOH, 99.99% trace metal grade, Sigma-Aldrich, St. Louis, USA) prepared using 18.2 Mcm filtered water (Milli-Q Reference, Merck, Kenilworth, USA). The NaOH was prepared in a larger volume for while the reductant solution was prepared freshly before analysis. The NaOH would stabilize the NaBH4 in aqueous environment during the analysis.
Samples were provided by the Marine Biology section of Massey University (Auckland, New Zealand). All samples consisted of either common or bottle-nose dolphin liver derived from dolphins washed ashore or in incidental bycaught in or near the Huraki Gulf and Tasman Sea. After necropsy the livers were cleansed and stored frozen at -80°C wrapped in aluminum foil. Before analysis a part of the liver was cut of using sterilized surgery equipment and left in a closed plastic bag to thaw at room temperature. After thawing the sample was diced finely before further preparation.
Page 13 of 38
3.1.3 Standard preparation
During this study multiple standards were prepared to determine which MSIS operation mode would be most effective, which elements caused interference and to observe linearity within two different ranges all for the element mercury as described in table 1. All standards were prepared using the trace element standards manufactured by Sigma-Aldrich and added up to volume using a prepared 10% HNO3 trace metal grade solution.
Table 1 Standard range concentrations used for mode effectiveness, interference study & linearity
Standard number Concentration (mg/L) range low Concentration (mg/L) range high
1 0.10 1 2 0.25 2 3 0.50 4 4 0.75 6 5 1.00 8 6 - 10
Apart from the standards used as calibration standards also different sample standards were prepared containing iron only or a combination of iron and mercury to study the effects of iron interference while doing elemental analysis. The samples were prepared in 10% nitric acid following the compositions described in table 2.
Table 2 Iron containing samples used for interference effects on mercury using MP-AES.
Label Fe concentration (mg/L) Hg concentration (mg/L)
10Fe 10 0 0.75Hg5Fe 5 0.75 0.75Hg10Fe 10 0.75 0.75Hg15Fe 15 0.75 4Hg10Fe 10 4 4Hg15Fe 15 4 3.1.4 Microwave digestion
Prior to the microwave digestion the vessel liners and caps were rinsed using filtered MilliQ water and left upside-down to dry. Then the dry liners were tared on an analytical balance before approximately 500 milligrams of thawed sample was added. Next 3.5 mL of 68% nitric acid (trace elemental grade) was added along with 3 mL of MilliQ filtered water to reach the required volume in the liners. A pT-probe was placed in the first vessel to monitor the pressure and temperature within the vessels during the microwave digestion. All vessels were then tightly capped and inserted in the instrumentally decided places of the rotor.
During digestion the maximum power limit was set to 900W and the maximum pressure to 35.0 bar. A thermal step program was programmed commencing with a temperature ramp towards 190°C over 20 minutes after which the temperature was maintained for 35 minutes before dropping to 30°C over 27 minutes. Afterwards the microwave door was opened carefully and ventilated to remove the escaped nitrogen oxides (NOx) as a safety measurement and extra cooling.
Page 14 of 38
The digests were then transferred into 25 mL beakers, the liners were flushed twice using 2% nitric acid to remove the remaining digest and transferred into the beakers. Eventually, the digest was filled op to volume in 25 mL volumetric flasks before being transferred into falcon tubes. These tubes were filmed and stored in the dark at room temperature before analysis containing an approximate 10% nitric acid concentration (determined by raw titration).
3.1.5 Elemental analysis
The elemental analysis was conducted using the MP-AES 4200 system by Agilent Technologies. The samples in 15 mL falcon tubes were placed in racks of the SPS 3 autosampler. The samples were introduced utilizing the peristaltic pump and pump tubing with a speed of 15 rpm of normal speed and 70 rpm of fast speed. All measurement conditions are presented in table 3.
Table 3 MP-AES operating conditions used during elemental analyses (using 2% nitric acid samples)
Setting Value Unit
Nebulizer Concentric -
Spray chamber MSIS -
Nebulizer Ar flow rate 0.5 approx. L/min
Nitrogen consumption 20 L/min
Read-time 2 Sec/measurement
Number of replicates 5 -
Rinse time* 230 Sec
Sample uptake time* 100 Sec
Stabilization time 30 Sec
Number of pixels 3 -
Wavelength(s) 253.625 Nm
* Using fast speed pumping
During measurements the sequence commenced with a blank sample followed by the appropriate standard range. After the standard range one or more blanks were measured to observe possible carry-over or memory effects. This was repeated between sample groups of different content (e.g. iron containing samples, digested samples, etc.) and at the end of the sequence both for the same purpose.
3.2 PFAS analysis using LC-QTOF MS
3.2.1 Instrumentation
The tissue samples were homogenized using a Braun Vario 350W commercial kitchen stick-blender (Frankfurt, Germany) and purified using Waters Oasis WAX 3cc 60mg SPE cartridges (Milford, MA). The SPE manifold was kept at 5mmHg pressure using a KNF Labport N86KT.18 vacuum pump (Freiburg, Germany). All centrifuging steps were executed operating a 5804R Eppendorf centrifuge (Hamburg, Germany) equipped with an appropriate tube probe (Epp tube and falcon 50mL tube sizes).
Page 15 of 38
The sample separation and mass detection was done using a Waters Synapt G2 UPLC-QTOF system (Milford, MA) equipped with an Aquity Binairy Solvent manager UPLC pump, an Aquity Sample manager autosampler, HTCH column oven and Synapt G2 quadrupole time of flight (QTOF) mass spectrometer. The samples were separated using an Aquity UPLC HSS T31 C18 (1.8 um; 50x2.1 mm 100Å) column manufactured by Waters (Milford MA).
3.2.2 Reagents and sample collection
All seal tissues were supplied by Diergaarde Blijdorp in Rotterdam. The tissues were obtained from two imprisoned seals in the zoo which had died because of unknown reasons in 2019 and 2020. The animals were treated with care during their lifetime and with the greatest respect. Directly after death, kidney and liver tissue samples were extracted and partly stored either frozen at -20°C or in formalin (37% formaldehyde solution) for the sample acquired in 2020 and frozen only for the sample acquired in 2019.
Extraction solvents and mobile phases were all prepared using LC-MS grade solvents. Formic acid 99% in water (Biosolve Chimie, Dieuze, France) was diluted to a 0.2% solution in acetonitrile (Biosolve Chimie, Dieuze, France) to be used for the PFAS extraction from the seal tissues. The 2% formic acid solution in water was prepared using the same bottle of formic acid and high purity LC-MS grade water (Biosolve Chimie, Dieuze, France). The methanol Lichrosolv used was purchased from Supelco/Sigma Aldrich (Bellefonte, Pennsylvania, USA). Using this methanol a 1% ammonium hydroxide (Sigma Aldrich, Bellefonte, Pennsylvania, USA) was prepared as well as the mobile phases containing 2mM ammonium acetate (Sigma Aldrich, Bellefonte, Pennsylvania, USA).
To develop the LC-MS method a PFC-24x reference standard was purchased from Accustandard (New Haven, CT, USA). This standard contained 24 common PFAS species at a concentration of 20 mg/L and was diluted to 25, 50, 75, 100, 125 and 150 ng/L concentrations using high purity LC-MS grade water and methanol (40:60). All standards and samples were spiked with 50 ng/L C13-labeled caffeine (Sigma Aldrich, Bellefonte, Pennsylvania, USA) to correct for ion suppression effects.
3.2.3 Sample preparation
The tissues stored in formaldehyde solution were first carefully washed using MilliQ filtered water before further treatment while the frozen tissue samples were thawed out before being washed with MilliQ filtered water as well. Then approximately 500 milligrams of tissue were cut off using a surgical scalpel blade, finely diced and then transferred into a 250 mL beaker glass. 30 mL of 0.2% formic acid in acetonitrile was then added to the sample and was then homogenized using the commercial kitchen blender for approximately 1 minute until a suspension formed. The suspension was then transferred into 50 mL tubes and then reduced to a volume of approximately 5 mL under a stream of nitrogen gas.
The sample was then applied in two aliquots to the SPE cartridges which prior to the applications were conditioned with 3 mL methanol and high purity water (LC-MS grade) respectively. The cartridge was then washed using 1 mL 2% formic acid in high purity water followed by 2 aliquots of 1 mL high purity water. After washing the first fraction was eluted using 1 mL of methanol. Then the cartridge was washed again using 1 mL of methanol. Then the second fraction was
Page 16 of 38
eluted using 2 aliquots of 1% ammonium hydroxide in methanol. Next the fractions were evaporated to dryness under a stream of nitrogen and reconstituted in 1 mL of 40:60 water:methanol. 500 µL of each of the fractions were then added to an Epp tube containing 20 mg of active carbon (Supelco, Envi-Carb). The tubes were first vortexed and then centrifuged for 30 minutes at 10.000 rpm. 300 µL the supernatant of each tube was then transferred into UPLC vial inserts to which 1 µL of labeled caffeine (15µg/L) was added as an internal standard. Afterwards the samples were capped and stored at 5°C for analysis.
3.2.4 UPLC-QTOF MS analysis
The separation and mass detection were conducted using the Waters G2 Synapt UPLC-QTOF system operated using the Masslynx operation software. The UPLC settings are presented in table 4.
Table 4 UPLC settings for the analysis of PFAS
Type Setting
Flow 0.2 mL/min
Injection volume 50 uL
Column temperature 30 °C
Mobile phase A 2mM ammonium acetate in 40:60 MeOH:water
Mobile phase B 2mM ammonium acetate in 95:5 MeOH:Water
Gradient (start with 8 minutes stabilizing at 40% B, integrated in gradient for the next sample)
0.0 min 40% B 10.0 min 100% B 20.0 min 100% B 22.0 min 40% B 30.0 min 40% B
The QTOF system used in negative electrospray mode was operated using the Masslynx software as well. The applied settings used for the PFAS analysis are presented in table 5. This method was derived from Xiao et.al. [51].
Table 5 Synapt G2 settings for the analysis of PFAS using electrospray in negative mode
Type Setting
Ionization mode ESI negative mode
Desolvation temperature 350 °C Source temperature 110 °C Cone voltage 20V Capillary voltage 1.8 kV Mass range 50 to 1000 Da Acquisition rate 1 Hz Cone/desolvation gas N2
Cone gas flow 10 L/h
Desolvation gas flow 1000 L/h
Collision energy 6 eV
LockMass aquistition 20s (acquired but not applied)
Page 17 of 38
3.2.5 Data analysis
The data was extracted from the Masslynx software and later edited in MZmine, an open source software tool for mass spectrometry analysis. Using this application the istopes and background peaks were removed. The result was converted into a csv file.
The script, found in Appendix A, in principle subtracts all monoisotopic masses of the PFAS substances in the EPA database from every m/z value found at each retention time. To correct for the negative ionization the mass of a proton is added to the obtained m/z value. For every value bellow the set threshold, the value for ‘mass_tol’, the row ID, the retention time, IUPAC name, common name, mass difference, and database’s monoisotopic mass are printed. The output is a new csv file containing all m/z values matching with a PFAS compound.
Apart from this difference it also generates the Kendrick Mass Defect for every m/z value detected. Using all KMDs for PFAS compounds in the EPA database and a random selection of approximately 5000 non-perfluorinated compounds from the NORMAN Substance Database [52] it generated two probability distributions which were converted into a false positive rate (FPR) for every KMD. The FPR and KMD were then appended to the feature list for every detected m/z value to indicate the possible error of a false positive identification. The feature list was then written into an output file.
Page 18 of 38
4. Results & Discussion
4.1 Mercury analysis
4.1.1 Interference study and MSIS mode comparison
According to the MP expert software mercury has multiple interfering elements with iron being the most probable in higher concentrations within mammals (as part of hemoglobin complexes used for oxygen transfer in blood). Therefore its effect was studied using prepared standard samples containing either mercury, iron or both of them. The used samples in this comparison are pictured in the overlay of figure 5 which shows results in nebulizer mode. The figure clearly shows a negative effect of iron decreasing the signal of the mercury.
Figure 5 Overlay of an iron sample, mercury sample, combined sample and blank with in the marked area the peaks which should show mercury presence. These results were obtained operating in nebulizer mode of the MSIS.
Using FLIC it was attempted to virtually remove the negative effect of the iron. A 10 mg/L iron sample was used as the reference standard as could have been observed before in figure 4. The effect of FLIC on the measurements is depicted in figure 6 which shows that it has a significant effect, though it clearly doesn’t completely remove it.
-2000 -1500 -1000 -500 0 500 1000 1500 2000 2500 3000 253,5 253,55 253,6 253,65 253,7 253,75 253,8 In te n si ty Wavelength (nm) Fe sample 10mg Hg sample 0.75mg HNO3 blank end Sample 0.75Hg10Fe
Page 19 of 38
Figure 6 The effect of FLIC on the intensity change due to increasing iron concentration in 0.75 mg/L mercury samples measured using nebulizer mode.
On the other hand when the CV-mode is applied a total different image becomes visible as presented in figure 7. A sample containing exactly the same amount of mercury shows hardly any difference between the one without and the one with iron. A sample containing iron only shows a low peak, however due to the high concentrations and a bad positioned nebulizer this has previously been observed as a form of carryover. This was removed in later experiments. Due to the abrupt cancelation of this study it was not possible to present a later measurement without the carryover while measuring the iron only sample. It was however clearly removed as was observed in later measured blanks.
Figure 7 Overlay of an iron sample, mercury sample and combined sample with in the marked area the peaks which should show mercury presence. These results were obtained operating in nebulizer mode of the MSIS.
Comparing the two different modes of operation it can be observed that the interference of iron is reasonably lower. If intensities of the same concentrations are compared as shown in the overlay of Figure 8 a slight shift of the peaks wavelength can be observed. There are three possible scenarios which can lead to this shift.
-1000 -500 0 500 1000 1500 2000 2500 0 5 10 15 In te n si ty Fe conc (mg/L) FLIC No FLIC 0 20000 40000 60000 80000 100000 253,45 253,5 253,55 253,6 253,65 253,7 253,75 253,8 253,85 In te n si ty Wavelength (nm) 10 ug Fe sample Analyte signal 4ug Hg sample Analyte signal 4Hg10Fe sample Analyte signal
Page 20 of 38
Figure 8 Overlay of a 0.75 mg/L mercury sample measured in cold vapour and nebulizer mode
The first explanation is that there was no possibility available to calibrate the wavelength detector of the instrument in CV-mode due to the absence of a fitting calibration standard. As a result of that a minimal shift is observed because the settings had to rely on older (yet not expired) data from a previous nebulizing mode wavelength calibration. Second it could be possible that the detector would be saturated, however that seems questionable since there is no plateau formation and the peak is Gaussian distributed. And last a third explanation would be that the actual peak lies between the two measurement wavelengths and that this is a result of a lack in the detectors resolution. A possible interferent was ruled out as the measured sample was prepared from a high purity mercury standard along with trace elemental grade solvents. Because the wavelength shift seemed likely to be explainable, the major increase in intensity and removal of iron interference the CV-mode was chosen as the most ideal mode of operation and used for the linearity determination and digested sample analysis.
4.1.2 Linearity
Two concentration ranges were examined to determine which would be most appropriate. The low range varying from 0.1 to 1.0 mg/L and the high range from 1.0 to 10.0 mg/L as previously noted in Table 1. Both calibration curves are presented in figure 9 with calculated correlation coefficients (R2) and regression lines by Microsoft Excel. The low range shows the highest linear relation as can be observed in either the straightness of the measured points and the R2 value, however this might be explained by the more tight range. In the high ranges the R2 seemed less sufficient, however, if parts of this range would be observed it could be that it would be higher.
0 4000 8000 12000 16000 253,45 253,5 253,55 253,6 253,65 253,7 253,75 253,8 253,85 In te n si ty Wavelength (nm) CV-mode Nebulizer mode
Page 21 of 38
Figure 9 Concentration curves of the observed ranges for linearity. The concentrations for both ranges have been noted in Table 1. The intensities used for these lines are averages over 5 replicate measurements.
4.1.3 Digested sample analysis
Due to a shortage in time only a single try-out sample had been analyzed. The sample was successfully digested following the method described in section 3.4 considering that no solids were left in the liners. Due to dissolved NO2 the digest was colored yellow, however after addition of the MilliQ filtered water this was reversed to a non-colored solution.
Following the developed elemental analysis method as previously described in section 3.5 a common dolphin liver tissue (marked as sample KS19-14tt by Massey’s Marine Biology section) was examined. This resulted in a measured intensity for mercury of 24235.47 which when using the low range regression line of figure 9 translates to a concentration of 1.006 mg/L. Reversed to a wet weight (w/w) mass fraction this would mean an amount of 5.114 mg of mercury in 1 kg of undried liver tissue.
4.2 PFAS analysis
4.2.1 Standard analysis and method validation
To develop the LC-MS method a combined standard of 4 PFAS compounds (perfluoropentanoic acid, perfluorooctanoic acid, perfluorononanoic acid and Heptadeca-perfluorooctane sulfonate) supplied by the Institute of Biodiversity and Ecosystem Dynamics of the University of Amsterdam, was injected in concentrations of 200 ng/L and 2 µg/L. These standards were used to test the MS settings and observe the retention times for these compound. The results as shown in figures 10-13 indicate clear peaks in the extracted ion currents (XIC) with counts around and above 1000. These peaks were also identified using the digital Python script with each a reported FPR lower than 0.18 in either of the concentrations.
Page 22 of 38
Figure 10 XIC of m/z 262.9785 (perfluoropentanoic acid) at RT 4.66 min in 4 compound standard (200 ng/L). This standard was measured using the correct settings
Figure 11 XIC of m/z 412.9724 (perfluorooctanoic acid) at RT 8.25 min in 4 compound standard (200 ng/L). This standard was measured using the correct settings
Page 23 of 38
Figure 12 XIC of m/z 463.9717 (perfluorononanoic acid) at RT 9.40 min in 4 compound standard (200 ng/L). This standard was measured using the correct settings
Figure 13 XIC of m/z 498.9305 (heptadeca-perfluorooctane sulfonate) at RT 8.72 min in 4 compound standard (200 ng/L). This standard was measured using the correct settings
Page 24 of 38
Unfortunately, during the set-up for the measurements of the PFC-24x standard and seal samples in the final phase of the research there were some complications on the instrument of which one was wrongfully entered settings. The other complications included leakages, large fluctuations in the source voltages and damage on the switching valve and ESI capillary. This caused major problems for the detection of the PFC-24x standard and tissue sample solutions. Therefore this research can furthermore only include the results used under the wrongfully entered settings. The wrong settings applied are presented in table 6.
Table 6 Wrongfully entered settings for MS analysis of PFAS using electro spray ionization in negative mode
Type Setting
Ionization mode ESI negative mode
Desolvation temperature 150 °C Source temperature 80 °C Cone voltage 20V Capillary voltage 2.0 kV Mass range 50 to 1000 Da Acquisition rate 1 Hz Cone/desolvation gas N2
Cone gas flow 0 L/h
Desolvation gas flow 500 L/h
Collision energy 6 eV
LockMass aquistition 20s (acquired but not applied)
Collision gas Ar
When applying the digital data analysis, none of the PFAS compounds assumed to be present in the PFC24-x standard were identified and neither was the C13-caffeine internal standard. However, when extracting an XIC the compounds: Sodium 1H,1H,2H,2H-perfluoro-hexanesulfonate, Sodium 1H,1H,2H,2H-perfluoro-octanesulfonate and Perfluorooctane sulfonamide could be identified for the respective m/z values of 448.9454 (RT 5.68 min), 548.9467 (RT 7.78 min) and 497.9486 (RT 9.08 min).
4.2.2 Seal tissue analysis
Similar to the PFC24-x standards the measurements for the seal tissue samples did work as planned resulting from multiple causes. As a possible result of this none of the compounds detected in either of the standards was identified by the Python script. However, there were still results identified as a PFAS compound as can be read in the supplementary data of Appendix B. The data outputs regularly show most datapoints in the first two minutes of the chromatogram and often with a high FPR. For the later eluting peaks the observed FPR is often lower in relation. Another feature which is reported in some cases is that for one m/z multiple isomers could be related. Also in others the same database entry is related to two compounds with a different mass error. These duplicate results make it more difficult to correctly assign a database entry to the results based on this data only.
Page 25 of 38
4.2.3 Digital data analysis
The digital data analysis using the developed Python script aimed to enable the user to perform untargeted PFAS screening using LC-MS(1) results. To start the data would first need to be cleaned up and converted into a comma separated values file (.csv). Using the MZmine tool the data was cleansed by removing background noise lower than 150 counts, peak deconvolution (max 30 sec.) and isotope peak removal. The resulting .csv files were then read and used in the Python script.
In the python script the differences between a m/z value and every EPA database entry were compared. Then for every m/z value if a match within 5mDa tolerance was found it was appended to a features list along with relevant data as described in the script in Appendix A. Apart from that, KMD values were calculated for all m/z values within the data, EPA database and 5000 randomly selected non-fluorine containing substances from the NORMAN substance database. The KMD values of the EPA database which only contained PFAS compounds were plotted in a scatter plot as shown in figure 14. In this figure it can be observed that nearly all KMD values varied between -0.4 and 0.4. Given this information a matrix was created which was named bindata. This raster of dimensions equal to all number of m/z values reported (3997 counted datapoints) and the length of -0.4 and 0.4 (0.01 height for every cell) named bind.
Figure 14 Scatterplot of the Kendrick Mass Defect versus the related m/z values of all EPA database entries.
Then to each cell the corresponding KMD was appended and every empty cell was defined as -99. Afterwards for every row in the bindata the frequency of a value larger than -99 was reported and plotted. The same procedure was then repeated 5 times for the 5000 non-fluorine containing compounds within the same bind range. The result of these six plots are presented in figure 15. A clear difference between the two types of compounds was observed, while the 5 plots of non-fluorine compounds did not show much variance.
Page 26 of 38
Figure 15 KMD frequencies of data entries in the EPA PFAS database (brown) and 5 of approx. 5000 randomly selected compounds (multiple colours) from the NORMAN substance database. The difference between the right dotted line representing the relative number of KMDs for which the substance would be a true positive perfluorinated compound and the left dotted line representing the false negative for which the compound would be assumed a perfluorinated compound. Because of this slight variance the random selection of the approximate 5000 compounds could not influence the FPR too much. Therefore the datapoints on these lines could be observed as virtual false positives. On the other hand, the values in the EPA database did not include any non-PFAS substances and could therefore be observed as only true positives. Using these two variables the FPR was calculated for every bind and as a result of that for every KMD value. This resulted in the rate as presented in figure 16. The corresponding values were then appended to the feature list and printed in the output file.
Figure 16 Graph of the false positive rate (FPR) compared to the bins in the bindata frame (i.e. the KMDs) representing the possibility of the occurrence of a compound identified as a false positive based on the KMD value.
Page 27 of 38
5. Conclusion
5.1 Elemental analysis of mercury in dolphin liver
This study has proven to be a potential method for mercury determination in marine mammal tissue. However, as there is no information about accuracy and reproducibility it still needs more validation. The method did show promising potential for mercury determination as there is a reasonable signal intensity and very few interference of other elements when applying the Cold Vapor mode. As a first digested sample was measured and a linear range with a reasonable correlation coefficient was examined it should be easy to adapt the dilutions within the method to increase the accuracy. But it is worth noting that as the recovery of the digestion procedure has not been determined that more adaptations would be necessary. In conclusion this method shows high potential but will need more research to be validated. Some suggestions for further research are described in the next section 5.1.1
5.1.1 Future research
To establish a recovery for this method, a reference sample (e.g. DOLT-5 [53] or DORM-3 [10]) should be measured. As there is a known concentration of many elements in this sample it would be possible to examine the amount lost during the digestion procedure and the measurement itself. Also it could help optimizing the digestion procedure which would contribute significantly to the effectivity of this method.
Another important feature to look into would be the possibility of using a dried sample. Because a wet sample loses moist while being out of a freezing environment it is very hard to determine the real stable weight [54]. As a dried sample doesn’t contain moist the weight should be more or less stable. However, as mercury is a slightly volatile element this might decrease the recovery as it is lost while drying. Therefore it would be interesting to see what the effects would be. For further exploitation of the MSISs functionalities the developed elemental analysis should be conducted for other hydride forming elements to observe whether it could possibly measure them as well. This would open a doorway to process multiple element determinations in a single run, decreasing costs and analysis time. Also it would be of great interest to observe whether the Dual mode operation of the MSIS could be used for multiple element analysis as well [55, 56]. As more elements will be introduced this could possibly mean a bigger chance of encountering an interfering element, but as has been observed in this study the fast linear interference correction could help counter this.
5.2 PFAS analysis in seal tissues
The method developed in this study for untargeted PFAS analysis has shown that it has potential to identify PFAS compounds using the Python script. However, based on the current results it cannot be concluded that the sample preparation for the seal tissues extracts PFAS. What should be noted is that the Python script was able to identify PFAS compounds in the tissue samples regardless of the functionality of the sample preparation. Although, the output data of the Python script could still be more accurate, there still are some duplicates and the results with a high FPR could be excluded. Considering these deficiencies, the Python script does certainly provide an important first step in the untargeted analysis of PFAS using the comparison of
Page 28 of 38
masses and false positive rate determined from the Kendrick Mass Defect. These two indicating features really enable the user to get a first look on what PFAS could be present and to what certitude it could be a right identification.
What can also be stated is that the LC-MS method did show good resolution of the 4 compounds present in the prepared 4 compound standard as well as for the 3 detectable compounds in the PFC-24x standard when observing the XIC, which shows that the method can be used for clear separation and that the instrument under correct settings is able to ionize PFAS substances. In conclusion this method has potential, however needs validation and fine-tuning for which suggestions will be done in section 5.2.1.
5.2.1 Future research
For further research the final measurements of this study should be repeated using the correct settings and a fully functional system. From that point the method could be optimized for the PFC-24x to verify that al 24 compounds can be measured and identified by the Python script. Apart from the PFAS compounds also the identification of the internal C13-caffeine standard should be studied so that it can be used for semi-quantitative analysis. Also it should be
suggested that the PFC-24x standard could be used as a spike to determine the recovery of the sample preparation and observe how well that performs.
Another interesting feature would be the investigation of MSe as part of the Python Script and using that information to identify in what subclass of PFAS the target should be classified as these fragments can provide very useful information [57]. Also this information could help indicating what isomer of a certain mass the target could be as the fragmentation pattern could help solve this. Using this form of MS could provide the required information to achieve a very accurate untargeted analysis.
When both of these suggestions would be implied and the experiments work then it would become interesting to look at different types of tissue possibly of other marine mammal species to examine whether this method could be used as a universal method for untargeted PFAS analysis.
Page 29 of 38
6. Acknowledgements
I would like to thank my supervisors assoc. Prof. Dr. John Harrison and Dr. Marie-Anne Thelen for their confidence and support even during the tough time they and the other employers of Massey University experienced in the last months. Also I would like to thank Erin Moffet for her assistance during the experiments related to the microwave digestions and runs of the MP-AES, assoc. Prof. Dr. Karen Stockin and her team of researchers for providing the dolphin samples and related information used in this study as well as Alex Burton for supplying the shark tissue used for practicing and the pleasant collaboration during research. And finally the rest of the chemistry department of the Albany campus for the wonderful time I had during this study in beautiful Aotearoa (New-Zealand).
Also I would like to thank Assist. Prof. Bob Pirok and Prof. Peter Schoenmakers for providing me with the possibility to finalize my MSc assessment at the University of Amsterdam, help and advises during and in preparation of that part of this research. And finally Assist. Prof. Saer Samanipour and Assist. Prof. Andrea Gargano for their enormous help and advises regarding PFAS analysis and help in creating the Python script for the digital data analysis.
Page 30 of 38
7. References
References
[1] M. Palzenberger, H. Pohla, Gill surface area of water-breathing freshwater fish, Rev Fish Biol Fisheries 2 (1992) 187-216.
[2] Y. Jeong, Y. Lee, K. Park, Y. An, H. Moon, Accumulation and time trends (2003-2015) of persistent organic pollutants (POPs) in blubber of finless porpoises (Neophocaena
asiaeorientalis) from Korean coastal waters, J Hazar Mat 385 (2020) 121598.
[3] S.S. Monteiro, M. Bozzetti, J. Torres, A.S. Tavares, M. Ferreira, A.T. Pereira, S. Sá, H. Araújo, J. Bastos-Santos, I. Oliveira, J.V. Vingada, C. Eira, Striped dolphins as trace element
biomonitoring tools in oceanic waters: Accounting for health-related variables, Science of The Total Environment 699 (2020) 134410.
[4] V. Liem-Nguyen, H. Nguyen-Ngoc, G.A. Adediran, E. Bjorn, Determination of picomolar levels of methylmercury complexes with low molecular mass thiols by liquid chromatography tandem mass spectrometry and online preconcentration, Anal. Bioanal. Chem.(2020) .
[5] J. Alava, P. Calle, A. Tirapé, G. Biedenbach, O. Cadena, K. Maruya, W. Lao, W. Aguirre, P. Jiménez, G. Domínguez, G. Bossart, P. Fair, Persistent Organic Pollutants and Mercury in Genetically Identified Inner Estuary Bottlenose Dolphin (Tursiops truncatus) Residents of the Guayaquil Gulf, Ecuador: Ecotoxicological Science in Support of Pollutant Management and Cetacean Conservation, Fron. Mar. Sci. 7 (2020) 122.
[6] F.A. Duarte, B.M. Soares, A.A. Vieira, E.R. Pereira, J.V. Maciel, S.S. Caldas, E.G. Primel, Assessment of Modified Matrix Solid-Phase Dispersion as Sample Preparation for the Determination of CH3Hg+ and Hg2+ in Fish, Anal. Chem. 85 (2013) 5015-5022.
[7] T. Ikemoto, T. Kunito, Y. Anan, H. Tanaka, N. Baba, N. Miyazaki, S. Tanabe, Association of heavy metals with metallothionein and other proteins in hepatic cytosol of marine mammals and seabirds, Environmental Toxicology and Chemistry 23 (2004) 2008-2016.
[8] K.A. Stockin, R.J. Law, P.J. Duignan, G.W. Jones, L. Porter, L. Mirimin, L. Meynier, M.B. Orams, Trace elements, PCBs and organochlorine pesticides in New Zealand common dolphins (Delphinus sp.), Science of The Total Environment 387 (2007) 333-345.
[9] E. Mohammed, T. Mohammed, A. Mohammed, Optimization of an acid digestion procedure for the determination of Hg, As, Sb, Pb and Cd in fish muscle tissue, Methodsx 4 (2017) 513-523.
[10] J. Savoie, R. St-Louis, M. Clement, Facilitating local analysis in northern regions: microwave plasma-atomic emission spectrometry for mercury determination in wild Atlantic salmon, Int. J. Environ. Anal. Chem. 98 (2018) 582-591.
Page 31 of 38
[11] V. Balaram, D. Vummiti, P. Roy, C. Taylor, P. Kar, A.K. Raju, K. Abburi, Determination of precious metals in rocks and ores by microwave plasma-atomic emission spectrometry for geochemical prospecting studies, Curr. Sci. 104 (2013) 1207-1215.
[12] M.R. Hammer, A magnetically excited microwave plasma source for atomic emission spectroscopy with performance approaching that of the inductively coupled plasma, Spectroc. Acta Pt. B-Atom. Spectr. 63 (2008) 456-464.
[13] P. Niedzielski, L. Kozak, M. Wachelka, K. Jakubowski, J. Wybieralska, The microwave
induced plasma with optical emission spectrometry (MIP-OES) in 23 elements determination in geological samples, Talanta 132 (2015) 591-599.
[14] A.F. Lima, M.C. Da Costa, D.C. Ferreira, E.M. Richter, R.A.A. Munoz, Fast ultrasound-assisted treatment of inorganic fertilizers for mercury determination by atomic absorption spectrometry and microwave-induced plasma spectrometry with the aid of the cold-vapor technique, Microchem. J. 118 (2015) 40-44.
[15] W. Li, P. Simmons, D. Shrader, T.J. Herrman, S.Y. Dai, Microwave plasma-atomic emission spectroscopy as a tool for the determination of copper, iron, manganese and zinc in animal feed and fertilizer, Talanta 112 (2013) 43-48.
[16] Y. Zhao, Z. Li, A. Ross, Z. Huang, W. Chang, K. Ou-Yang, Y. Chen, C. Wu, Determination of heavy metals in leather and fur by microwave plasma-atomic emission spectrometry, Spectroc. Acta Pt. B-Atom. Spectr. 112 (2015) 6-9.
[17] N. Ozbek, S. Akman, Method development for the determination of calcium, copper, magnesium, manganese, iron, potassium, phosphorus and zinc in different types of breads by microwave induced plasma-atomic emission spectrometry, Food Chem. 200 (2016) 245-248. [18] N. Ozbek, S. Akman, Microwave plasma atomic emission spectrometric determination of Ca, K and Mg in various cheese varieties, Food Chem. 192 (2016) 295-298.
[19] Y. Wang, L.A. Baker, E. Helmeczi, I.D. Brindle, Rapid high-performance sample digestion of base metal ores using high-intensity infrared radiation with determination by nitrogen-based microwave plasma optical spectrometry, Analytical Chemistry Research 7 (2016) 17-22. [20] C.A. McDonough, S. Choyke, P.L. Ferguson, J.C. DeWitt, C.P. Higgins, Bioaccumulation of Novel Per- and Polyfluoroalkyl Substances in Mice Dosed with an Aqueous Film-Forming Foam, Environ. Sci. Technol. 54 (2020) 5700-5709.
[21] J.N. Dodds, Z.R. Hopkins, D.R.U. Knappe, E.S. Baker, Rapid Characterization of Per- and Polyfluoroalkyl Substances (PFAS) by Ion Mobility Spectrometry-Mass Spectrometry (IMS-MS), Anal. Chem. 92 (2020) 4427-4435.
[22] D.J. Piekarski, K.R. Diaz, M.W. McNerney, Perfluoroalkyl chemicals in neurological health and disease: Human concerns and animal models, Neurotoxicology 77 (2020) 155-168.
Page 32 of 38
[23] G. Codling, H. Yuan, P.D. Jones, J.P. Giesy, M. Hecker, Metals and PFAS in stormwater and surface runoff in a semi-arid Canadian city subject to large variations in temperature among seasons, Environ. Sci. Pollut. Res. 27 (2020) 18232-18241.
[24] G. Boisvert, C. Sonne, F.F. Riget, R. Dietz, R.J. Letcher, Bioaccumulation and
biomagnification of perfluoroalkyl acids and precursors in East Greenland polar bears and their ringed seal prey, Environ. Pollut. 252 (2019) 1335-1343.
[25] S. Valsecchi, M. Rusconi, S. Polesello, Determination of perfluorinated compounds in aquatic organisms: a review, Anal. Bioanal. Chem. 405 (2013) 143-157.
[26] G.D. Villanger, K.M. Kovacs, C. Lydersen, L.S. Haug, A. Sabaredzovic, B.M. Jenssen, H. Routti, Perfluoroalkyl substances (PFASs) in white whales (Delphinapterus leucas) from Svalbard - A comparison of concentrations in plasma sampled 15 years apart, Environ. Pollut. 263 (2020) 114497.
[27] Environmental Protection Agency, PFAS Master List of PFAS Substances, 2020 (2020) . [28] I. Berg, Validation of MP-AES at the quantification of trace metals in heavy matrices with comparison of performance to ICP-MS, (2015) .
[29] Agilent, Agilent 4200 MP-AES; specifications, 5991-3445EN (2014) .
[30] N. Chalyavi, P.S. Doidge, R.J.S. Morrison, G.B. Partridge, - Fundamental studies of an atmospheric-pressure microwave plasma sustained in nitrogen for atomic emission spectrometry, - J. Anal. At. Spectrom. 32 (2017) 1988.
[31] M. Schild, A. Gundlach-Graham, A. Menon, J. Jevtic, V. Pikelja, M. Tanner, B. Hattendorf, D. Gunther, Replacing the Argon ICP: Nitrogen Microwave Inductively Coupled Atmospheric-Pressure Plasma (MICAP) for Mass Spectrometry, Anal. Chem. 90 (2018) 13443-13450.
[32] S. Karlsson, V. Sjoberg, A. Ogar, Comparison of MP AES and ICP-MS for analysis of principal and selected trace elements in nitric acid digests of sunflower (Helianthus annuus), Talanta 135 (2015) 124-132.
[33] Agilent, Flexible sample introduction with the Multimode Sample Introduction System, 5991-6509EN (2016) .
[34] A. Asfaw, G. Wibetoe, A new demountable hydrofluoric acid resistant triple mode sample introduction system for ICP-AES and ICP-MS, J. Anal. At. Spectrom. 22 (2007) 158-163.
[35] H. Matusiewicz, M. Slachcinski, Method development for simultaneous multi-element determination of hydride forming elements (As, Bi, Ge, Sb, Se, Sn) and Hg by microwave induced plasma-optical emission spectrometry using integrated continuous-microflow ultrasonic nebulizer-hydride generator sample introduction system, Microchem. J. 95 (2010) 213-221.
Page 33 of 38
[36] M. Slachcinski, Modern chemical and photochemical vapor generators for use in optical emission and mass spectrometry, J. Anal. At. Spectrom. 34 (2019) 257-273.
[37] R. McLaughlin, P. Cheese, M. Ding, A. Ptolemy, A. Conn, I. Brindle, A convenient sample introduction system for ICP, ICP-MS, and AA, Am. Lab. 38 (2006) 13-+.
[38] A. Howard, (Boro)hydride techniques in trace element speciation - Invited lecture, J. Anal. At. Spectrom. 12 (1997) 267-272.
[39] J. TOFFALETTI, J. SAVORY, Use of Sodium-Borohydride for Determination of Total Mercury in Urine by Atomic-Absorption Spectrometry, Anal. Chem. 47 (1975) 2091-2095.
[40] E. Bramanti, A. D'Ulivo, L. Lampugnani, G. Raspi, R. Zamboni, Cold vapour atomic fluorescence studies on the behaviour of mercury(II) and mercury(II)-thiol complexes. An alternative route for characterization of -SH binding groups, J. Anal. At. Spectrom. 14 (1999) 179-185.
[41] A. Seyoum, A. Pradhan, J. Jass, P. Olsson, Perfluorinated alkyl substances impede growth, reproduction, lipid metabolism and lifespan in Daphnia magna, Sci. Total Environ. 737 (2020) 139682.
[42] N.A. Aly, Y. Luo, Y. Liu, G. Casillas, T.J. McDonald, J.M. Kaihatu, M. Jun, N. Ellis, S. Gossett, J.N. Dodds, E.S. Baker, S. Bhandari, W.A. Chiu, I. Rusyn, Temporal and spatial analysis of per and polyfluoroalkyl substances in surface waters of Houston ship channel following a large-scale industrial fire incident, Environ. Pollut. 265 (2020) 115009.
[43] C.M.A. Eichler, J.C.C. Little, A framework to model exposure to per- and polyfluoroalkyl substances in indoor environments, Environ. Sci. -Process Impacts 22 (2020) 500-511. [44] T. Anumol, S. Mohsin, M. Woodsman, P. Abrahamsson, Recommended Plumbing Configurations for Reduction in Per/Polyfluoroalkyl Substance Background with Agilent 1260/1290 Infinity (II) LC Systems, 5991-7863EN (2017) .
[45] Python Software Foundation, About Python TM, 2020 (2020) . [46] Anaconda Inc., Anaconda | Open Source, 2020 (2020) .
[47] E. Kendrick, A Mass Scale Based on CH2 = 14.0000 for High Resolution Mass Spectrometry of Organic Compounds, Anal. Chem. 35 (1963) 2146-2154.
[48] Q. Zheng, M. Morimoto, H. Sato, T. Fouquet, Resolution-enhanced Kendrick mass defect plots for the data processing of mass spectra from wood and coal hydrothermal extracts, Fuel 235 (2019) 944-953.
Page 34 of 38
[49] B.S. Crimmins, X. Xia, P.K. Hopke, T.M. Holsen, A targeted/non-targeted screening method for perfluoroalkyl carboxylic acids and sulfonates in whole fish using quadrupole time-of-flight mass spectrometry and MSe, Anal. Bioanal. Chem. 406 (2014) 1471-1480.
[50] K.A. Barzen-Hanson, S.C. Roberts, S. Choyke, K. Oetjen, A. McAlees, N. Riddell, R. McCrindle, P.L. Ferguson, C.P. Higgins, J.A. Field, Discovery of 40 Classes of Per- and Polyfluoroalkyl Substances in Historical Aqueous Film-Forming Foams (AFFFs) and AFFF-Impacted Groundwater, Environ. Sci. Technol. 51 (2017) 2047-2057.
[51] F. Xiao, R.A. Hanson, S.A. Golovko, M.Y. Golovko, W.A. Arnold, PFOA and PFOS Are Generated from Zwitterionic and Cationic Precursor Compounds During Water Disinfection with Chlorine or Ozone, Environ. Sci. Technol. Lett. 5 (2018) 382-388.
[52] J. Slobotnick, et. al., NORMAN Substance Database – NORMAN SusDat, 2020 (2020) . [53] Health Canada, Human Health Risk Assessment of Mercury in Fish and Health Benefits of Fish Consumption, (2007) .
[54] F. Sardenne, N. Bodin, L. Metral, A. Crottier, F. Le Grand, A. Bideau, B. Brisset, J. Bourjea, C. Saraux, S. Bonhommeau, V. Kerzerho, S. Bernard, T. Rouyer, Effects of extraction method and storage of dry tissue on marine lipids and fatty acids, Anal. Chim. Acta 1051 (2019) 82-93. [55] J.C. Garcia-Mesa, P. Montoro Leal, M.M. Lopez Guerrero, E. Vereda Alonso I.,
Simultaneous determination of noble metals, Sb and Hg by magnetic solid phase extraction on line ICP OES based on a new functionalized magnetic graphene oxide, Microchem. J. 150 (2019) 104141.
[56] M.M. Lopez Guerrero, E. Vereda Alonso, J.M. Cano Pavon, M.T. Siles Cordero, A. Garcia de Torres, Simultaneous determination of chemical vapour generation forming elements (As, Bi, Sb, Se, Sn, Cd, Pt, Pd, Hg) and non-chemical vapour forming elements (Cu, Cr, Mn, Zn, Co) by ICP-OES, J. Anal. At. Spectrom. 31 (2016) 975-984.
[57] J. McCord, M. Strynar, Identification of Per- and Polyfluoroalkyl Substances in the Cape Fear River by High Resolution Mass Spectrometry and Nontargeted Screening, Environ. Sci. Technol. 53 (2019) 4717-4727.
Page 35 of 38
Appendix A: Python script
# In[1]:
import pandas as pd import numpy as np import random
# In[2]:
EPA_list = pd.read_csv("PFAS list EPA.csv", delimiter=";") docname = "2020_09_14_AN23 FORMB"
Inputfile = (docname+".csv")
Outputfile = ("Output "+docname+".xlsx")
Sample_list = pd.read_csv(Inputfile, delimiter=";") sel_feat = []
mz = EPA_list["MONOISOTOPIC_MASS"]
samp = Sample_list["row m/z"] + 1.00784 #The addition of 1.00784 is the mass of a proton added because of the use of negative mode
RoundR = 12+2*19 R = 12.011+(2*18.998403) Mass = Sample_list["row m/z"] KMfrac = RoundR/R KMD = [] bind = np.arange(-400, 405, 10)/1000 bindata = -99*np.ones((len(mz), len(bind))) KMD2 = round(mz*KMfrac)-mz*KMfrac prob = []
# In[3]:
for a in range(np.size(bindata, 0)): for b in range(np.size(bindata, 1)):
if (KMD2[a] > bind[b]-0.005) and (KMD2[a] <= bind[b]+0.005): bindata[a, b]=KMD2[a] for i in range(np.size(bindata, 1)): tv1 = bindata[:, i] c = 0 for k in range(len(tv1)): if tv1[k] > -99.0: c = c+1 prob.append(c) # In[4]:
FalsePos_file = pd.read_csv("susdatabase.csv", delimiter=";") MonoMass = FalsePos_file["Monoiso_Mass"]
