The Use of Air Pollution Controls to Reduce the Gas-phase Emissions of Per- and Polyfluoroalkyl Substances from a Fluoropolymer Manufacturing Facility

Abstract

The efficacies of a thermal oxidizer and carbon adsorption beds used as air pollution control technologies at a fluoropolymer manufacturing facility were evaluated for nonpolar volatile fluorinated compounds (VFCs) using Other Test Method (OTM)-50. The target compounds for OTM-50 include industrial fluorocarbons, common products of incomplete combustion, reaction byproducts, and common refrigerants. The thermal oxidizer’s emissions were found to contain tetrafluoromethane, CF₄. Emissions from the carbon adsorption beds used to scrub the fugitive air emissions for two vinyl ether synthesis facilities were sampled and the analyses showed that the compounds with boiling points below 100 °C were not effectively adsorbed. This research shows that the facility’s thermal oxidizer is effective at destroying high concentration streams of perfluoroalkyl substances (PFAS) and that the carbon beds can reduce emissions for compounds with boiling points over 100 °C.

Synopsis

Evaluation of a thermal oxidizer and carbon adsorption beds used to reduce the air emissions of fluorinated chemicals from a fluoropolymer manufacturer.

Graphical Abstract

Introduction

Per- and polyfluoroalkyl substances (PFAS) and other fluorinated organic compounds used in the synthesis and manufacture of fluoropolymers can be emitted into the environment^1–13. Many of these compounds have negative impacts on human health and the environment. Currently, fluoropolymers are a useful component for many technologies and processes, including lithium- ion batteries, fuel cells, medical devices, and semiconductor manufacturing^{14, 15}. As demand grows for technologies that use these products, there will be an increase in production of certain fluoropolymers. This increased production necessitates that manufacturers develop processes and controls to produce the necessary polymers without the emissions of fluorochemicals³.

Manufacturers have developed and are continuing to develop new processing aids and methods in the attempt to reduce and ultimately eliminate the use of fluorinated polymer processing aids and surfactants³. However, these polymerization methods still use fluorinated monomers and still produce PFAS and fluorinated gaseous byproducts during polymerization^{16, 17}. To address this, companies are installing pollution control technologies to reduce or eliminate fluorochemical emissions that may occur during manufacturing¹⁸. Determining the performance of the pollution controls already installed is vital to helping to determine the best available technologies to curb emissions for the industry.

This study investigated the pollution control devices that began operation between 2018 and 2021 at a fluorochemical and fluoropolymer manufacturer to reduce the facility wide fluorinated organic compound air emissions by at least 99% compared to the 2017 emissions¹⁸. Here, Other Test Method (OTM) – 50¹⁹ was used to determine volatile fluorinated compound (VFC) emissions from the main pollution controls for air emissions. Results from compliance testing required by the air permit to evaluate emissions of the dominant fluorochemicals of concern are also presented. The aim was to refine emission inventories and to attempt to evaluate the control devices’ efficacies. Two main air pollution control technologies were evaluated: a thermal oxidizer (TO) used to destroy high concentration process vapor streams and carbon adsorption beds that are used to scrub indoor fugitive emissions.

Materials and Methods

The TO was designed and is permitted to destroy at least 99.99% of the high concentration PFAS influent and is fed process vapors from around the facility. The TO consists of a 10 million BTU per hour natural gas fired burner followed by a four-stage packed bed scrubber. The process feeds are delivered to a pipeline that feeds the TO burner where they are exposed to the flame with at least 30% excess hydrogen from the natural gas compared to fluorine. The furnace is kept at 1100 °C, about 120 °C higher than the minimum permitted requirement²⁰. To control the hydrofluoric acid produced during combustion, and other emissions, the emissions pass through the four-stage scrubber with one caustic stage and three water scrubbing stages. A pH greater than or equal to 7.1 is maintained to remove the HF.

The carbon bed adsorbers were installed to control the fugitive indoor air emissions of hexafluoropropene oxide – dimer acid (HFPO-DA) and hexafluoroprene oxide – dimer acid fluoride (HFPO-DAF) in the vinyl ether monomers manufacturing facilities (all the species are measured and reported as HFPO-DA). The vinyl ethers north building’s (VEN) carbon bed is 3 m x 2.4 m x 7.8 m and contains 6,360 kg of Evoqua VOCarb P60 carbon pellets (Pittsburgh, PA). VEN’s bed scrubs about 520 dry standard cubic meters per minute (dscm) of the air from the manufacturing rooms. The vinyl ethers south (VES) carbon bed is 3 m x 2.4 m x 8.4 m and contains 7,590 kg of the VOCarb P60 carbon pellets. The VES bed scrubs about 510 dscm of the facility’s manufacturing spaces’ indoor air.

Sampling and analysis of the VFCs was performed using OTM-50, or similar earlier draft version of OTM-50 described previously²¹, and is explained in more detail in the Supplemental Information (SI). Sampling at the TO outlet was performed using OTM-50’s water management system, which involves sampling through a 120 °C probe and filter with the filtered emissions passing through a set of water impingers to remove excess water and any remaining acid gases. The conditioned emissions were sampled into two evacuated 6 L Silonite stainless steel canister (Entech, Simi Valley, CA) in series with critical orifices (Entech, Simi Valley, CA) with flow rates of approximately 200 mL/min and 26 mL/min. The sampling duration was 25 minutes to sample between 4 to 5 L, a high volume (HV) sample, with the higher flow orifice and about 600 to 700 mL, a low volume (LV) sample, with the low flow orifice. The stack gas was pulled through the sampling train at 1 L/min using a dry meter box with a pump. The low volume canister was used as a dilution can to help verify the concentration of any compounds that may have been above the calibration range. The canisters were analyzed using the preconcentrator- GC/MS method described in OTM-50 including a library search (Wiley275.L) of the mass spectra of any unknown peaks. A summary of the quality metrics is shown in the SI Table S1, and the OTM-50 target list is shown in Table S2.

The carbon beds were sampled similarly but used OTM-50 without the water management system, see Figure S1 in the SI. The probe and filter were kept at 37 °C, approximately 10 °C above the duct temperature. A heated sample line at 37 °C was used to transport the sample from the filter to the sampling tee with the orifice with a flow rate of 100 mL/min. At VEN, an additional canister in series was sampled at using a 26 mL/min orifice. The canister with the 26 mL/min orifice at VEN sampled for 15 min to obtain about 390 mL of sample. The 100 mL/min canisters sampled for 48 min to collect about a 5 L sample.

The other fluorinated compounds measured for the compliance tests were sampled and analyzed by the company’s contractor according to the methods developed for the consent order¹⁸ and approved for the air permit²². The concentrations measured in the TO inlet and outlet emissions were determined using OTM-45²³, only quantifying HFPO-DA, and a version of Method 18²⁴ combined with a modified Method 8260 for analysis²⁵, for hexafluoropropene oxide (HFPO), carbonyl difluoride, heptafluoropropyl-1,2,2,2-tetrafluoroethyl ether (E1), and HFPO-DAF. The five compounds were the only fluorinated compounds that were measured according to the protocols approved in the consent order and air quality permit^{18, 22}. The HFPO-DA influent and effluent concentrations for the carbon beds were measured using OTM-45²³.

Results and Discussion

Thermal oxidizer

The TO was installed to mitigate the emissions of the high concentration fluorinated process streams by at least 99.99%. Regular testing is required to ensure this permit limit is met²². The compliance testing accounts for five different compounds that make up the main portion of the process stream, see Table S3 in the SI for the compliance measurements that were made the same day as the VFC testing. Results from the compliance test verified that on average 99.9988% of the measured PFAS were destroyed by the TO²⁰ on the days of the test. The destruction efficiency is promising but could be misleading, since poor incineration conditions can cause PFAS to produce fluorinated byproducts that may be emitted into the atmosphere²¹. At temperatures over 1000 °C, these byproducts are mainly compromised of nonpolar VFCs^{21, 26}, as found in the OTM-50 target list, see Table S2 in the SI. To help assess the efficacy of the TO as the facility’s primary pollution control device, the emissions were sampled at the stack to measure any VFCs that may be present. Due to the concentrations of the OTM-50 targeted compounds HFPO (4.95 g/m³) and E1 (7.23 mg/m³) at the inlet to the TO and other compounds not in OTM-50, the inherent risk of the pressurized sampling port, and the lack of appropriate sampling ports, OTM-50 was only used to sample at the stack of the TO.

The VFCs were sampled to bracket the three compliance sampling runs performed on the same days to allow for close to concurrent sampling and measurements. It was hypothesized that with the high influent concentration of fluorochemicals that tetrafluoromethane (CF₄) would be in the ppmv range, well above the calibration curve. For this reason, both an LV sample and an HV sample were taken to simplify the dilution procedure and to allow for the measurement of compounds at both trace levels and high concentrations. The HV sample was also run at a dilution in the lab to measure CF₄ and to give a verification of the concentration. The LV canisters for the first two runs leaked and the samples were invalid, and not included in the results shown in Table 1, but runs 3 and 4 provided duplicate measurements to the HV sample. The only VFC emitted from the TO that was measured consistently with confidence was CF₄, see Table 1. CF₄ averaged 8,730 μg/m³ with a 24 %RSD for the six samples. The other compounds measured were found near the detection limit in a few samples or were also found in the nitrogen blanks. The compounds not in the blank but near the detection limit are common refrigerants that likely were in the ultra zero air used to pressurize the canister, evidenced by the fact that the LV samples that used more air to pressurize the canister, ~500 mL sample pressurized to 18 L, had higher concentrations than the HV sample that went from ~5 L to 18 L during pressurization. The concentrations for these gases, listed in Table 1, are estimated values below the reporting limit but above the detection limit.

Table 1:

Thermal oxidizer volatile fluorinated compounds emissions

Volatile Fluorinated Compounds (μg/m3)	Pre-test N2	Post-test N2	Run 1 HV	Run 2 HV	Run 3 LV	Run 3 HV	Run 4 LV	Run 4 HV
Tetrafluoromethane	<3.73	<3.73	5416	6927	11030	9847	9622	9528
Fluoromethane (HFC-41)*	0.86	2.08	3.47	2.48	16.37	2.56	17.40	2.77
Chlorodifluoromethane (HCFC-22)**	<0.12	<0.12	0.69	0.68	<1.2	0.86	<1.2	<0.12
1,1,1,2-Tetrafluoroethane	<0.17	<0.17	0.47	0.60	<1.7	0.61	<1.7	<0.17
Decafluorobutane (Perfluorobutane)*	0.59	0.59	1.36	1.17	6.27	1.18	6.49	0.98
1,1,1,2,2,3,3-Heptafluoropropane	0.21	0.51	<0.19	<0.19	<1.9	<0.19	<1.9	<0.19
Trichlorofluoromethane (CFC-11)*	0.40	0.40	1.89	1.76	6.34	1.63	7.49	0.99
1,1,1,2,2,3,3,4,4-Nonafluorobutane *	0.46	0.46	1.26	0.87	7.25	1.09	7.50	1.36

^*Likely contaminant from the zero air used to pressurize the canister

^**common background refrigerant

italics – estimated concentration that is below the reporting limit but above the detection limit; < - below detection limit; bold – concentration was above the reporting limit for OTM-50

Carbon beds

The carbon beds at the facility’s vinyl ether synthesis buildings, VEN and VES, were installed to adsorb the indoor fugitive emissions and reduce the facility’s air emissions of HFPO-DA and HFPO-DAF¹⁸. The fugitive air emissions typically have low concentrations of the compounds, and the carbon beds are permitted to help achieve a facility wide reduction of 99% of annual air emissions of HFPO-DA compared to 2017. There is little in literature about carbon beds’ efficacy for the adsorption of gas phase PFAS, so the data from the compliance tests helps show that carbon beds may effectively remove about 90% of HFPO-DA from the fugitive emissions, helping to achieve the facility wide 99% reduction of HFPO-DA^{27, 28}. It is unknown how carbon performs with compounds that are less polar and have lower boiling points, though it is anticipated that low boiling point compounds are not adsorbed.

To determine what VFCs may be emitted and, if possible, their removal efficiencies (REs), OTM-50 was used to sample the carbon beds’ inlet and outlet air flows. VEN was tested concurrent with a compliance test, and the data is summarized in Table 2 (the complete OTM-50 data is in the SI, Table S4). These results indicate that as the boiling points approach 100 °C, the carbon begins to adsorb the compounds. For example, the RE for tetrafluoroethylene, with a boiling point of −76 °C, was essentially zero, but the RE for 2H-perfluoro-5-methyl-3,6- dioxanonane (E2), with a boiling point of 100 °C, was about 90%. Heptafluoropropyl-1,2,2,2- tetrafluoroethyl ether (E1), boiling point of 41 °C, seems to defy this trend, as more E1 is in the effluent than the influent. This may be caused by earlier processes that had higher emissions of E1 saturating the bed and the E1 was slowly being emitted from the bed over time or by changes in the gas composition like a change in humidity²⁹. Another possible explanation is that the HFPO-DA trapped in the bed degrades over time and forms E1^30–32. It is likely that the carbon bed did have compounds adsorbed, since the carbon bed was due to be replaced within a month of the test. More research is needed to better understand the phenomenon.

Table 2:

Vinyl Ethers North carbon bed’s emissions

Measured VFCs	Boiling point (°C)a	Inlet (μg/m3)	RSD (%)	Outlet (μg/m3)	RSD (%)	RE (%)	RSD (%)
Trifluoromethane (FC-23)	−83.3	21.36	25.7	15.93	9.9	25.4	87.0
Fluoromethane (HFC-41)	−78.3	0.19	18.2	0.21	14.5	−7.4	211.1
Difluoromethane (HFC-32)	−51.4	2.02	47.2	1.18	11.7	41.4	119.5
1,1,1,2,2-Pentafluoroethane (HFC-125)	−48.6	7.47	59.5	8.56	41.4	−14.6	155.0
Tetrafluoroethylene	−76.0	3.09	39.8	3.93	48.5	−27.4	157.3
Hexafluoropropene	−29.0	87.11	33.8	117.60	28.3	−35	98.7
Hexafluoropropene Oxide	−27.3	7393	6.4	6571	2.7	11.1	29.1
Chlorodifluoromethane (HCFC-22)	−40.8	1.22	29.9	1.14	30.5	7.0	58.6
1,1,1,2-Tetrafluoroethane	−26.4	0.68	0.9	0.59	10.7	13.0	65.4
Trichlorofluoromethane (CFC-11)	23.8	1.23	0.9	1.40	7.7	−14.3	60.9
Tetradecafluorohexane	57.9	<0.225	0	1.11	34.8	−394	43.6
Heptafluoropropyl 1,2,2,2-tetrafluoroethyl ether (E1)	41.0	21.00	26.8	195.03	11.7	−829	21.0
2H-Perfluoro-5-methyl-3,6-dioxanonane (E2)	100.0	19.04	29.9	1.97	41.9	89.7	11.0
Compliance test semivolatile PFAS b
2,3,3,3-tetrafluoro-2-[heptafluoropropoxy]propanoic acidb	180.0	404	8.7	8.31	3.6	97.9	0.26

^a.Boiling points found at https://comptox.epa.gov/dashboard/ or https://commonchemistry.cas.org/

^b.Semivolatile PFAS measurements were from the compliance test report from Alliance Technical Group²⁸

italics – estimated concentration that is below the reporting limit but above the detection limit; < - below detection limit; bold – concentration was above the reporting limit for OTM-50

The VES carbon bed was sampled about 5.5 months after the HFPO-DA compliance test and about five months after the carbon was replaced. The VES inlet and outlet measurements were lower than those of VEN, summarized in Table 3 and shown in full in Table S5, with HFPO-DA over two orders of magnitude lower. The only OTM-50 compounds above the reporting limit were tetrafluoroethylene and HFPO. The other compounds identified in the emissions have estimated concentrations that are above the detection limit but below the reporting limit. Some compounds observed were also present in the blanks and likely from the ultra zero air. No additional compounds were observed in the OTM-50’s unknowns analyses.

Table 3:

Vinyl Ethers South carbon bed’s emissions

Measured VFCs	Inlet (μg/m3)	RSD (%)	Outlet (μg/m3)	RSD (%)
Trifluoromethane (FC-23)a	1.80	31.1	1.64	39.4
1,1,1,2,2-Pentafluoroethane (HFC-125)	0.29	4.87	0.30	3.66
Fluoromethane	1.91	10.8	1.80	4.09
Tetrafluoroethylene	10.46	55.7	9.94	35.7
Hexafluoropropene	4.09	34.3	3.71	10.8
Hexafluoropropene Oxide	245.04	3.5	33.66	27.3
Chlorodifluoromethane (HCFC-22)	1.08	7.61	1.10	4.31
1,1,1,2-Tetrafluoroethane	0.75	31.1	0.71	12.2
Decafluorobutanea	0.73	4.1	0.70	11.4
1,1,1,2,2,3,3-Heptafluoropropane	<0.114	0	0.49	12.2
Trichlorofluoromethane (CFC-11)a	1.49	3.20	2.03	8.09
2H-Perfluoro-5-methyl-3,6-dioxanonane (E2)	11.67	68.8	3.96	142
Compliance test semivolatile PFAS b
2,3,3,3-tetrafluoro-2-[heptafluoropropoxy]propanoic acida	1.1	12	0.2	41

^a.Analyte was present in the blanks

^b.Semivolatile PFAS measurements were from the compliance test report from Alliance Technical Group.²⁷

italics – estimated concentration that is below the reporting limit but above the detection limit; < – below detection limit; bold – concentration was above the reporting limit for OTM-50

There is apparent removal of some compounds, but the emission rates are unknown and REs could not be calculated. Historically, the inlet flow is higher than the outlet flow²⁷. HFPO does show more of a change in concentration than implied by the VEN data above. This could be due to condensed water in the carbon bed, as it was 10–12 °C and raining at the time of sampling, potentially causing the hydrolysis of HFPO. HFPO is known to rapidly react with water to form a more polar compound, 3,3,3-trifluoro-2,2-dihydroxy propanoic acid³³, that would not be detected with OTM-50. The impact of moisture and other weather conditions can change adsorption characteristics²⁹ and are areas that need further research.

Environmental implications

The OTM-50 target list was developed from the identification of products of incomplete destruction found in samples from a variety of sources and common building blocks used for PFAS and fluoropolymer synthesis. This target list is particularly useful for evaluating the nonpolar VFC emissions from both the destructive and adsorptive pollution control technologies at a fluoropolymer manufacturer. There were some compounds with %RSDs over 30%, but these were usually due to the low estimated concentrations near the detection limit. Overall, OTM-50 provided valuable insight into the pollution controls and produced acceptable data.

Here, only a few VFCs were measured in the emissions, and no unknown VFC was apparent in the samples. The only VFC that was found in the TO emissions was CF₄. The carbon beds, while effective for some PFAS, do little to adsorb compounds with a boiling point under 100 °C. The carbon bed tests were performed during the production of one of many polymers that are produced at the facility, and the emissions could change depending on the polymer produced at the time of testing. More research is needed to determine the efficacies of carbon beds with other monomers and breakdown products from the production of the different polymers, investigating both VFCs and nontargeted polar compounds in the emissions. Due to the inconsistencies observed with the compounds, more research is needed to determine the carbon bed efficacies for HFPO and E1 under different conditions.

Supplementary Material

Supporting Information: Additional details for the sampling and analytical procedures, quality assurance metrics, targeted compound list, and additional data.