Selecting the Correct Autosampler Vial and Insert to Improve Reproducibility and Accuracy in Mass Spectrometry Analyses

Abstract

Acquiring accurate and reproducible data is crucial for analytical chemists, however, sample storage and manipulation play a large role in achieving high-quality results. Autosampler vials and small-volume inserts are used for sample preparation and storage prior to mass spectrometry analyses, however, the utility, design, and material of these vary greatly. Here, we assessed differences in data reproducibility, sample loss, and physical differences among vials and inserts made of glass and polypropylene for combinations (vial/insert): glass/glass (GG), glass/polypropylene (GP), polypropylene/glass (PG), and polypropylene/polypropylene (PP). Six per- and polyfluoroalkyl substances (PFAS) were evaluated in 40:60 methanol:water with 3 mM ammonium acetate for 32 days to determine peak area reproducibility over time since PFAS do not degrade, therefore changes would most likely be due to adhesion and evaporation. On Day 1, the measured average peaks areas for the 6 PFAS were not statistically significant between any combinations, however, PP was the least reproducible (relative standard deviation >10%). All samples were then recapped and stored for one month at −20°C and reanalyzed to compare the peak areas between Day 1 to Day 32. All combinations showed >10% differences in peak area for at least 4 PFAS and both evaporation and adhesion were observed. While the greatest effects due to adhesion occurred in the glass inserts, the glass vials showed the least volume loss to evaporation. Thus, when adhesion and evaporation are considered together, the GP combination is recommended for PFAS analyses as the glass vial reduces evaporation rates, and the polypropylene insert minimizes adhesion.

Graphical Abstract

1. Introduction

The accuracy and reproducibility of biological and environmental analyses are quintessential for obtaining reliable data in clinical and toxicological studies. Analytical techniques including front-end separations like liquid chromatography (LC) and gas chromatography (GC) coupled with other back-end separations such as mass spectrometry (MS) have proven to be essential in these study types.^{1, 2} However, there are many essential steps for achieving reliable quantitative and qualitative data including sample handling, analyte extractions, instrumentation methods, and storage conditions.^3–5 In particular, autosampler vials and small-volume inserts are vital to LC, GC, and flow injection-based methods for sample introduction to the instrument and storage for later use.^{6, 7} It is therefore essential that the selected materials for a given study enable the collection of the most reliable results possible, while also minimizing sample loss and contamination. Materials, however, vary by vendor and in their composition, size, and utility, and each combination may or may not be the most appropriate for a given study. Typically, vials and inserts are made of either glass or plastic, such as polypropylene.⁷ Vial cap designs also vary and may be screw, snap, or crimp tops and composed of polypropylene, steel gold, silver aluminum, or combinations of silicone or natural red rubber with a polytetrafluoroethylene (PTFE) coating. This list is non-exhaustive and emphasizes how imperative it is to select the appropriate combination for a given study.

Per- and polyfluoroalkyl substances (PFAS) are a group of over 7 million man-made compounds containing fluorinated alkyl chains with a wide variety of chemical motifs.^8–10 PFAS are used as surfactants as they repel both oil and water, and have high thermal and chemical stability.⁸ However, growing concern regarding the environmental and biological accumulation of PFAS and their impact on health has prompted the Environmental Protection Agency to set regulatory drinking water limits for only six PFAS analytes: perfluorooctanoic acid (PFOA), perfluorononanoic acid (PFNA), perfluorohexane sulfonic acid (PFHxS), perfluorooctane sulfonic acid (PFOS), hexafluoropropylene oxide dimer acid (HFPA-DA, or GenX), and mixtures containing two or more of PFHxS, PFNA, GenX, and perfluorobutane sulfonic acid (PFBS).^11–15 Thus, while over 7 million analytes constitute as PFAS, these six were specifically selected for this study to monitor PFAS behavior during extended storage.^8–10 As research on PFAS is rapidly increasing due to the ubiquitous presence of these chemicals, analytical tools that are “PFAS-free” and resistant to PFAS adhesion are preferred.^{8, 9} Traditionally, glass vials with PTFE-coated caps have been considered problematic for PFAS analyses due to the fluorination of PTFE and evidence of PFAS adhering to glass.¹⁶ Researchers commonly opt for polypropylene materials for PFAS analyses instead.^11–15 However, current studies suggest that there is no significant difference between PFAS adhesion in polypropylene versus glass materials and PFAS recovery has been shown to not differ between materials when analyzed within 24 hours of preparation for 1 mL samples.^{7, 11, 15, 17} Current studies also report sample loss to evaporation, particularly in organic solvents, for polypropylene snap cap design vials recommended by EPA 1633.⁷ To our knowledge, no current research has examined the effects of vial designs on data collection following a month-long storage of small-volume samples. Thus, to further inform researchers of the appropriate autosampler vials selection for PFAS analyses, we have assessed differences in glass versus polypropylene vials and inserts for their performance of repeated measurements and longitudinal analyses of peak area changes. This is a specific application looking at vial and small-volume insert performance over time, where previous work has looked at the effects of within day studies, which is not always feasible. PFAS were selected as the analytes of interest as they are known to be resistant to breakdown and oxidation, and therefore the analyte peak abundance differences over time are not a reflection of chemical degradation, but of storage efficacy and autosampler performance.⁸ The findings of this study can therefore be used to inform analyses of a wide variety of analytes, however, the complexity of PFAS interactions with sample matrices may also impact performance.¹⁸ In addition to monitoring PFAS abundance changes, we also assessed sample loss by evaporation over time, as well as comparisons of physical design differences between vial and insert types. These comparisons will ultimately provide guidance to researchers on appropriate autosampler vial selection based on a given study and inform manufacturers of any design flaws to address.

2. Materials and Methods

2.1. Chemicals, Reagents, and Materials

Polypropylene vials (Agilent Technologies; Santa Clara, CA, part # 5182–0567), polypropylene inserts (Agilent Technologies; Santa Clara, CA, part # 5182–0549), glass vials (Agilent Technologies; Santa Clara, CA, part # 5182–0716)) and glass inserts (MicroSolv Technology Corporation, Leland, NC, part # 9502S-02N) were selected for this study, as these materials are compatible with our in-house Agilent 1290 Infinity II Ultra-High Performance Liquid Chromatography (UHPLC) system. Polypropylene snap caps (Agilent Technologies; Santa Clara, CA, part # 5182–0542) were paired with the polypropylene vials and PTFE/red silicone septa screw caps were paired with the glass vials (Agilent Technologies; Santa Clara, CA, part # 5185–5820). An isotopically labeled PFAS standard mix (MPFAC-HIF-ES) from Wellington Laboratories (Guelph, Canada) that contains PFAS at 250 to 5000 ng/mL in methanol was used to monitor vial/insert performance. Optima LC-MS grade methanol, water, and ammonium acetate were obtained from Fisher Scientific (Hampton, NH). The inner diameter (ID) of the vials and inserts, as well as the outer diameter (OD) of the inserts were measured using Snap-on 6” dial calipers (Kenosha, WI). Additional part numbers are available in Table S1.

2.2. PFAS LC-IMS-MS Analysis by Vial and Insert Combinations

A stock of diluted MPFAC-HIF-ES PFAS internal standard was prepared using an 8.33-fold dilution with 40:60 methanol:water and 3 mM ammonium acetate, as previous work from Boatman et. al. utilized buffered water and methanol as a reconstitution solvent post-extraction for complex samples.^{19, 20} In this work, we utilized this solvent system as a springboard for examining PFAS adhesion in the vials, however PFAS behavior may vary in different solvent systems.²¹ Additionally, while the following analyses were performed at concentrations 30 ng/mL (PFNA), 60 ng/mL (PFBS, PFHxS, PFOS, and PFOA), and 240 ng/mL (GenX) to account for loss and to ensure reliable detections without signal depreciation at lower levels, previous work has been conducted to determine the limits of quantification using the same platform coupling liquid chromatography (LC), ion mobility spectrometry (IMS), and mass spectrometry (MS) and reconstitution solvent in complex samples, which are included in Table S2.²⁰ For the preparation of each sample, 100 μL of this stock was pipetted with polypropylene pipette tips (Eppendorf, Hamburg, Germany, part # 022491539) and wasted to condition the tips prior to an addition of 100 μL stock to each insert as previous studies have documented challenges associated with pipetting organic solvents that are improved by a pre-rinse.^{22, 23} The vial/insert combinations prepared were glass vials with glass inserts (GG), glass vials with polypropylene inserts (GP), polypropylene vials with glass inserts (PG), and polypropylene vials with polypropylene inserts (PP). Each combination was prepared in triplicate. Glass vials were paired with screw caps with PTFE coatings and polypropylene vials were paired with polypropylene snap caps to adhere to product design intentions, however, it is worth noting that the screw versus snap cap design likely impact the seal to the vials as evaporation rates have already been proven to be higher with snap cap vials in a 24-hour study.^{7 24, 25} Samples were all analyzed with an Agilent 1290 Infinity II UHPLC system coupled to an Agilent 6560 IM-QTOF MS platform (Santa Clara, CA) with a method previously developed for PFAS analyses on the same day as preparation (Day 1) to minimize any initial loss and determine the reproducibility of the collected data prior to prolonged storage.^{9, 19, 26} Following Day 1 data collection, samples were recapped with new, unpunctured caps and stored at −20°C for one month (31 Days) and were reanalyzed using the same methods on Day 32. Samples were stored at −20°C as PFAS have been previously shown to not degrade at −20°C for up to 180 days.^{27, 28} The final volumes in each vial following the Day 1 and Day 32 analyses were recorded (μL) to determine percent volume lost to evaporation over time.

Prior to analysis, the Agilent ESI-L Low Concentration tune mix solution (Santa Clara, CA) was directly injected into the 6560 IM-QTOF for verification of mass accuracy and collision cross section (CCS) calibration. Double blanks (DB) containing neat 40:60 methanol:water with 3 mM ammonium acetate were injected in triplicate at a volume of 4 μL for chromatographic separation on an Agilent Zorbax Eclipse Plus^™ C18 column (Santa Clara, CA) (2.1 mm × 50 mm, 1.8 μm particle size) with an in-line 5 mm guard column. Samples were then injected in a randomized order with a DB injection between every eight sample injections to monitor carryover and instrument performance. Mobile phase A (MPA) was water with 5 mM ammonium acetate and mobile phase B (MPB) was 95:5 methanol:water with 5 mM ammonium acetate. A 16.5-minute gradient ramping MPB from 10% to 100% was applied at a flow rate of 0.4 mL/min (Table S3). Analytes were ionized in negative mode using the Agilent Jet Stream ESI source prior to drift tube IMS separation. Ion packets were pulsed through the drift tube using 4-bit multiplexing. Source conditions and IMS-MS settings are provided in Tables S4-5.²⁹

Agilent .d files were collected for each sample, containing the multidimensional LC, IMS, and MS information for further analysis. Each data file was demultiplexed using a PNNL PreProcessor (v4.1 2023.06.03) with a signal intensity threshold of 20 counts and 100% pulse coverage.^{19, 29} The Agilent ESI tune mix file was used to create a calibration curve for the single-field method and Agilent IM-MS Browser (v10.0) to relate known CCS values to observed drift times. The resulting single field CCS method function (ß and t_fix) was applied to all files to allow previously characterized PFAS using an in-house library containing characterized PFAS from the Wellington Laboratories (Guelph, Canada) MPFAC-HIF-ES internal standard mix.^{26, 30} All data files were then imported into and evaluated using Skyline-daily (MacCoss Lab Software, v24.1.1.398), an open source and vendor neutral software for MS data processing. Parameters including LC retention time index, CCS values within the resolving power window of 30, and m/z values with a mass error less than 10 ppm were manually assessed and compared to the library.^{26, 31, 32} Raw peak areas were normalized to the total ion current (TIC) of each sample for relative abundance values.

3. Results

The main goals of this study were to evaluate if four vial and insert material combinations provide better analyte reproducibility both day to day and over time. To understand potential differences, vial and insert parameters were evaluated along with peak area abundances for 6 PFAS analytes with established drinking water limits by the United States Environmental Protection Agency (EPA): perfluorobutanesulfonic acid (PFBS), perfluorohexanesulfonic acid (PFHxS), perfluorooctane sulfonic acid (PFOS), perfluorooctanoic acid (PFOA), perfluorononanoic acid (PFNA), and hexafluoropropylene oxide dimer acid (HFPO-DA or “GenX”).³³ These analytes are resistant to degradation due to their long half-lives, however they are surfactants and will adhere to surfaces, and hence serve as a model system for the vial comparison study demonstrated herein.¹⁷

3.1. Vial and Insert ID and OD Comparisons

To evaluate the effects vial and insert parameters have on the reproducibility of PFAS peak areas, first the inner and outer diameters (IDs and ODs) of four polypropylene and four glass inserts were measured, in addition to the IDs of four polypropylene and glass vials. The average glass insert had an OD of 5.63 ± 0.04 mm, while the average polypropylene insert OD was 5.61 ± 0.03 mm. Moreover, the average glass insert ID was 4.15 ± 0.09 mm, and the average polypropylene insert ID was 4.60 ± 0.01 mm (Figure 1A). Differences between the insert ODs and IDs were then assessed using independent Student’s t-tests with Benjamini-Hochberg p-value corrections to understand if major differences were present (Student’s t-test, Benjamini-Hochberg; p_adj < 0.05). While the insert ODs were not statistically significant, it is important to note that the manufacturer’s description of the polypropylene inserts claims a 5.6 mm OD, whereas the glass inserts manufacturer claims a 6 mm OD, so great differences could occur between batches. The insert IDs, however, were statistically significant and may impact instrument performance in favor of the polypropylene inserts as they have a wider target for the injection needle. Next, the vial IDs were evaluated. The average glass vial ID was found to be 5.81 ± 0.01 mm, and the average polypropylene vial ID was 6.58 ± 0.04 mm. These were also statistically significant as shown in Figure 1B, due to the 0.77 mm difference. Vial ODs were not measured as they were not suspected to impact the analyses. All measurements are provided in Table S6.

Figure 1.

Dimensional measurements were taken using calipers to determine the average (A) insert outer diameter (OD, mm) and (B) vial inner diameter (ID, mm) for each material: glass and polypropylene. The inserts were not statistically significant in OD based on p-value, but the vials were statistically significant (Student’s t-test, Benjamini-Hochberg; p_adj < 0.05). This information was used to determine the “wiggle room” of the inserts inside the vials.

The average vial ID and insert OD (mm) were next used to determine the gap distance between the insert and vial. This was designated as the “wiggle room” in which the inserts can misalign inside the vials and hinder the performance of an injection needle. The wiggle room values were calculated using Equation 1 for all vial/insert combinations (GG, GP, PG, and PP). This distance is the vial ID minus the insert OD, which was divided by 2 to account for the insert only tilting in one direction. Differences in the calculated wiggle rooms were then assessed using an Analysis of Variance (ANOVA) with a Tukey Honest Significant Difference (HSD) post hoc test to assess the various combinations using RStudio (v2024.12.0). A visual representation (Onshape, v1.190) of these vial/insert combinations is shown in Figure 2. The average wiggle room values were as follows for the different combinations: GG (0.07 ± 0.02 mm), GP (0.09 ± 0.02 mm), PG (0.48 ± 0.02 mm), and PP (0.49 ± 0.02 mm). Thus, a ~5-fold difference in wiggle room was observed between vial type, regardless of the insert material. Unsurprisingly, these wiggle rooms were statistically significant among the following combinations: PG vs GP, PP vs GP, PG vs GG, and PP vs GG as shown in Figure 3. These results indicate that the large vial ID differences may cause misalignment of the inserts inside the vials. It is common for the injection needle to operate based on its perceived depth (vial bottom sensing) with a 0 mm offset. This could mean any misalignment of the inserts within the vials may cause inconsistent data collection, particularly in the polypropylene vials with wider inner diameters.

Figure 2.

Visualization of the four combinations of vials and inserts: glass/glass (GG), glass/polypropylene (GP), polypropylene/glass (PG), and polypropylene/polypropylene (PP).

Figure 3.

The collected measurements of the insert outer diameters (mm) and vial inner diameters (mm) were used to determine the average “wiggle room” of the inserts inside the vials for each combination using the equation above. For vials of like material, the wiggle room was not statistically significant, however, between materials of vials were statistically significant (ANOVA, Tukey HSD; p_adj < 0.05). This indicates that the vial inner diameter is the primary factor in potential misalignment of the inserts. Combinations that are statistically significant from each other are noted with an *.

$$wiggleRoom=\frac{vialID\left(mm\right)-InsertOD\left(mm\right)}{2}$$ (1)

3.2. Volume Comparisons to Understand Evaporation and Losses

Evaporation of sample material can be detrimental to achieving accurate and reproducible results as it may alter analyte concentrations. To assess the degree to which each vial/insert combination is impacted by evaporation, each combination (GG, GP, PG, PP) in triplicate received an initial aliquot of 100 μL of isotopically labeled PFAS internal standard in 40:60 methanol:water and 3 mM ammonium acetate. Two LC-IMS-MS analyses were performed for each vial/insert combination (one on Day 1 and one on Day 32) using 4 μL injections each. Therefore, in total, 8 μL was used for instrument analysis, meaning the theoretical final volume should be 92 μL if no evaporation occurred during storage at −20°C over the one-month period. Following the Day 32 analysis, the experimental final volumes were determined by manually pipetting sample out at decreasing volumes down to 1 μL at a time until sample could no longer be reliably transferred. The experimental final volumes were recorded (μL) and the average percent lost was determined for each vial/insert combination (Table S7). In this analysis, the GG combination showed the least amount of volume loss at 2.2%, which may have simply occurred due to sample adhesion to the injection needle. GP showed the second lowest volume loss at 9.1%, followed by PG at 22.5% and PP at 34.1%, as shown in Figure 4. This elevated sample loss from the polypropylene vials indicates severe evaporation, likely due to a poor seal of the snap caps to the vials, which were the only option for the vial type used in this study. These results show that on average, glass vials with screw caps have less volume loss over time than polypropylene vials, and the GG combination had the least amount of sample loss over time due to a tight seal of the screw caps. While the utilization of internal standards helps to account for some concentration changes, uncontrolled sample evaporation presents many problems, including but not limited to, the loss of volatile compounds, disproportionate degrees of ion suppression by less volatile compounds, oxidation-induced structural changes, or uneven concentration of prepared external calibration standards. Thus, minimizing sample evaporation is critical as researchers frequently need to rerun samples weeks or months following their preparations or initial analyses, and volume loss impacts analyte concentrations.

Figure 4.

The theoretical initial volume for all vial and insert combinations was 92 μL, and the ideal percent volume lost over one month in storage at −20°C was 0%. The GG combination had the least loss at 2.2%, followed by GP 9.1%, PG 22.5%, and PP 34.1%.

3.3. Within Day PFAS Peak Area Comparisons

Since researchers often need to collect data on samples weeks or months after their initial preparation and analysis, it is crucial to define reasonable expectations on data reproducibility and reliability over extended time periods. To determine reproducibility of data collected over time, the peak areas for the PFAS evaluated on Day 1 were compared to those in the same samples analyzed a month later (Day 32). Each sample was recapped following the Day 1 analysis and stored at −20°C for the whole month. Furthermore, all collected raw peak areas were normalized to the total ion current (TIC), resulting in relative abundance values. The average relative abundances from the Day 1 analysis were processed for the six PFAS analytes: perfluorobutanesulfonic acid (PFBS), perfluorohexanesulfonic acid (PFHxS), perfluorooctane sulfonic acid (PFOS), perfluorooctanoic acid (PFOA), perfluorononanoic acid (PFNA), and hexafluoropropylene oxide dimer acid (HFPO-DA or “GenX”).³³ An ANOVA with a Tukey HSD post hoc test was performed for these six analytes to determine if any differences among vial/insert combinations existed prior to prolonged storage. Additionally, the relative standard deviation (RSD) for each vial/insert combination was calculated to determine the reproducibility on Day 1 and from Day 1 to Day 32. The relative abundances among the four different vial/insert combinations processed in triplicate were not statistically significant for the 6 PFAS analytes on the Day 1 analysis (Figure 5A). The lack of statistical significance for all six PFAS analytes indicates that when samples are prepared in triplicate, average relative abundances can be replicated with any combination when analyzed within a day of sample preparation, supporting previous findings that vial material does not impact quantitative PFAS analyses within 24 hours.⁷ We did however notice that the PP combination provided a noticeably wider range of relative abundances. Specifically, on Day 1, PP had RSDs between 17.1% (PFOS) and 31.7% (GenX), while the other three combinations had RSDs <10% for all analytes except GenX in the GP combination which was 12.1% (Figure 5B). Therefore, because the PP combination showed the most variation between injections, if the PP combination was selected for a given study, obtaining peak areas that are directly comparable between samples is less likely than the other three combinations. This may be due to misalignments of the inserts, although this was not observed for the PG combination, possibly due to differences in the flexibility of the polymer feet which is not easily measured. It is also worth noting that the average relative abundances inversely correspond to the RSD, meaning that as the signal intensity increased, so did the reproducibility. High reproducibility is crucial for obtaining reliable and accurate quantification values. Increased variability of peak area values may compromise the integrity of calibration curves and may result in inaccurate quantification for unknowns. This variability is especially problematic at the lower end of a calibration curve where trace-levels of PFAS at or near the limits of detection are more sensitive to matrix effects, ion suppression, and overall signal variation.

Figure 5.

A comparison of PFAS relative abundances of four combinations of vial and insert materials including glass/glass (GG, blue), glass/polypropylene (GP, green), polypropylene/glass (PG, yellow), and polypropylene/polypropylene (PP, orange) for six PFAS analytes in order by class and increasing chain length. (A) Samples were prepared in triplicate and analyzed utilizing LC-IMS-MS for Day 1 data collection. Relative abundances were not statistically significant among the combinations for all six PFAS. (B) The relative standard deviation (RSD) was calculated for the Day 1 relative abundances to determine the reproducibility of the signal for each combination. For the PP combination, all six PFAS had RSD values >10%. (C) Samples were reanalyzed after one month in storage at −20°C for a percent difference comparison between Day 1 and Day 32 relative abundances. Bars >0% indicate a greater signal on Day 32 and possible evaporation and bars <0% indicate a lower signal on Day 32 and possible analyte adhesion. A majority of analytes for all vial/insert combinations had a % difference greater than ±10%.

3.4. Between Day PFAS Peak Area Comparisons

Our next goal was to determine if data obtained after a month in storage replicated that obtained on Day 1. To make these measurements the 6 PFAS relative peak areas from Day 32 were compared to those from Day 1. Thus, percent differences greater than 0% indicate higher relative abundances on Day 32 and values less than 0% indicate lower relative abundances on Day 32. Uniformly, all combinations showed poor reproducibility over the month storage period with differences either greater or less than 10% for many of the analytes when compared to the Day 1 analyses. The directionality was, however, not consistent among all analytes (Figure 5C).

In most cases, the Day 1 relative abundances were greater than Day 32 except for PFOS and PFHxS in PP. The higher relative abundances likely indicate losses due to adhesion on the insert materials over time, which limits our ability to accurately analyze these analytes after prolonged storage without internal standards spiked in the samples. Within the perfluoroalkyl sulfonic acid (PFSA) class (PFBS (C4), PFHxS (C6), and PFOS (C8)), the effects of adhesion across all combinations decrease as the chain length increases, however the opposite trend is observable for the perfluoroalkyl carboxylic acid (PFCA) class (PFOA (C8) and PFNA (C9)), where the effects of adhesion are heightened for PFNA over PFOA. GenX (C6), which has a carboxylic acid headgroup, similar to PFOA and PFNA, along with an ether linkage also trends toward elevated adhesion over evaporation for all materials. These results suggest that for PFAS with sulfonic acid headgroups, shorter chains adhere more to both glass and polypropylene whereas PFAS with carboxylic acid headgroups and longer chains adhere more. Interestingly, the combinations with glass inserts (GG and PG) appear to be more impacted by adhesion and have larger percent differences versus the polypropylene inserts (GP and PP). This however is not surprising as PFAS are known to adhere to glass, although Folorunsho et. al. previously reported greater losses of long-chain PFAS due to adhesion to polypropylene over glass in 100% aqueous solutions.^{16, 23} Table S8 presents an overview of existing studies on the adsorption of PFAS in aqueous and organic materials for a variety of containers, container materials, temperatures, and durations. The higher relative abundances observed on the Day 32 for PFOS and PFHxS in PP illustrate that evaporation is greater than adhesion and the analytes become more concentrated. While the majority of analytes have higher relative abundances on Day 1, it is critical to state that the effects of adhesion and evaporation are not independent of one another. For example, while the GG combination appears to be more impacted by adhesion than the PG combination, this is not the case, as PFAS will adhere to glass the same despite the vial type. Instead, because it has been established that the PG combination also undergoes ~10x the evaporation over time compared to the GG due to the polypropylene snap cap design, the concentrating effects of evaporation lend to higher relative abundances on the Day 32 analysis. This trend can be observed across all PFAS and vial/insert combinations included in this study. Therefore, values closer to zero must not be immediately interpreted as ideal without further consideration, as the dual impacts of adhesion and evaporation play a large role in reproducibility of data over time and may be problematic for long term analyses. Here, we have established that the polypropylene snap cap vials induce more sample evaporation compared to the glass vials with screw caps and that the polypropylene inserts have less PFAS adhesion. Therefore, it is our recommendation that of the materials included in this study, PFAS researchers should opt for the glass vial with polypropylene insert (GP) combination or conduct similar experiments to determine the efficacy of other commercially available products.

4. Conclusions

In this study, different combinations of autosampler vials and inserts made of glass and polypropylene were assessed for differences in design and data collection performance. Here, we have shown that the polypropylene vials (with either insert) have significantly more wiggle room than the glass vials, indicating that the likelihood of the insert being misaligned in the vials and impacting data collection. Over a one-month period, the different combinations also varied greatly in the percent solvent volume lost to evaporation with the GG combination having the least loss and PP having the most loss (GG 2.2%, GP 9.1%, PG 22.5%, and PP 34.1%). The increased evaporation in the PP combination is likely due to a poor seal from the polypropylene snap cap design and may be resolved with screw cap designs. Six PFAS in 40:60 methanol:water with 3 mM ammonium acetate were also assessed to understand measurement reproducibility through time. The Day 1 analysis following sample preparation showed no differences in average relative abundances among each vial/insert combination. The PP combination, however, had the largest RSD, suggesting that PP may provide inconsistent measurements. The relative abundances comparisons between Day 1 and Day 32 also showed that while all insert materials are impacted by adhesion, glass inserts are more impacted than polypropylene despite their outer vial type. The weighted effects of wiggle room, evaporation rates, within day reproducibility, and impact of adhesion are provided in Table 1. Thus, when considering all of the factors assessed in this study, we recommend that due to the increased adhesion to glass inserts and increased evaporation and variability due to the polypropylene snap cap design, the GP, or glass vial with a screw cap design and polypropylene insert combination should be used for PFAS analyses.

Table 1.

Summary of Vial and Insert Performance Based on Dimensions, Evaporation Rates, Reproducibility of Peak Areas, and the Impact of Adhesion to the Inserts

Rank by Performance	Wiggle Room	Loss to Evaporation	Within Day Reproducibility of Peak Areas Based on Average %RSDs	Impact of Adhesion to the Inserts Only
1 (best)	GG	GG	GG	PP
2	GP	GP	PG	GP
3	PG	PG	GP	PG
4 (worst)	PP	PP	PP	GG

5. Limitations and Future Study

The results of this study point to using vial designs with screw caps paired with polypropylene small-volume inserts to minimize the effects of PFAS sample loss due to evaporation and adhesion over time. However, the scope of this study was limited to only four combinations of vials and inserts out of a large market of available consumables. While the polypropylene vials used in this study were limited to the snap cap design, future work should include an analysis of polypropylene autosampler vials with screw cap designs. Other small-volume inserts should also be considered, as both the glass and polypropylene inserts used in this study have flexible polymer feet. Tapered, flat bottom, conical and built-in inserts are all commercially available and should be tested similarly to this study for performance. Additionally, this study was limited to observing only 6 PFAS analytes, while over 7 million PFAS are known to exist.¹⁰ As PFAS vary in their structures and physicochemical properties, researchers studying PFAS and other environmentally and clinically relevant analytes should conduct their own analysis of autosampler vial and insert combinations performance, as other analytes and solvent systems may perform differently. Future studies may also consider storing extracted samples in colder conditions, like −80°C if feasible, to see if colder conditions further minimize sample loss. We also recommend that for long-term storage of extracted samples, internal standards should be used, and the mass of each sample should be obtained at room temperature prior to storage and again before analysis to approximate the volume of solvent lost due to evaporation. Finally, while a larger study analyzing more vial materials and designs, analytes, and solvent systems would help conclusively identify the optimal PFAS sample treatment, the results of this study demonstrate that seemingly small analytical decisions, like vial and insert selections, may have large impacts on collected data quality, and understanding the expected performance of selected consumables can help further contextualize final results.

Supplementary Material

Supplementary Materials

The Supporting Information is available free of charge on the website at DOI:

The data includes LC-IMS-MS methods and settings, consumables and part numbers, and vial and insert measurements.

Highlights.

• Autosampler vials with different physical dimensions impact analysis

• Snap cap designs for vials induce sample evaporation

• Glass inserts demonstrate more PFAS retention than polypropylene

• Screw cap design vials with polypropylene inserts are ideal for PFAS analyses

Data Availability Statement

All data are available in the main text or the supplementary materials. Skyline files containing LC-IMS-HRMS data are available at https://panoramaweb.org/VialComparisonFalls.url and raw data files are available at ftp://massive-ftp.ucsd.edu/v09/MSV000097749/.