50th International Conference on Environmental Systems ICES-2021-292 12-15 July 2021

Experimental Proof of Concept of a Cold Trap as a Purification Step for Lunar Processing

Jordan B. Holquist1, Sean Gellenbeck2, Chad E. Bower3, and Philipp Tewes4 Paragon Space Development Corporation, Tucson, AZ 85714, USA

Since the observation of direct evidence of water-ice in the permanently shadowed regions (PSR) on the lunar surface, in-situ resource utilization (ISRU) has been proposed for processing the regolith-bound water-ice to provide fresh water, breathable , and liquid rocket propellant for lunar exploration missions. One possible method of extraction is the sublimation and vapor transport of the water from the regolith to a collection and processing system. However, the water-ice is found concurrently with other typically volatile species that can sublimate with water vapor and that would contaminate and degrade downstream processing systems. Paragon Space Development Corporation® is developing the ISRU- derived water purification and Hydrogen Oxygen Production (IHOP) system to collect, purify, and process water-ice from PSRs on the lunar surface. A critical component of this system concept is a cold trap that selectively deposits water-ice from a saturated water vapor stream while rejecting the contaminant volatiles. This paper presents results, analysis, and discussion of an experimental proof of concept demonstration wherein a warm process gas containing water vapor and volatile contaminant components (H2, CO, H2S, NH3, SO2, C2H4, CH4, CO2, and CH3OH) was passed into an evacuated and actively chilled thick-walled glass bottle. Water-ice samples were collected from each of eight test batches with varying contaminant components present at concentrations with respect to water vapor matching their observed proportionality from the LCROSS mission results. Each water sample was analyzed for impurities in both the liquid water and the gaseous headspace of the sample. Results indicate that retained contaminants have concentrations at or below the Henry’s Law estimates made in ICES-2020-71, demonstrating proof of concept of a freeze purification step for lunar water processing by using a cold trap for water collection.

Nomenclature

B.C. = Best Case H2O = Water CH4 = Methane H2S = Hydrogen C2H4 = Ethylene ICICLE = ISRU Collector of Ice in a Cold CH3OH = Methanol Environment CO = Carbon monoxide ICPMS = Inductively Coupled Plasma Mass CO2 = Spectrometry COTS = Commercial Off-The-Shelf IHOP = ISRU-derived water purification and CTE = Cold Trap Experiment Hydrogen Oxygen Production CTTS = Contaminant Tolerant Test Stand ISRU = In-Situ Resource Utilization DI = De-ionized water IWP = Ionomer-membrane Water Processing GCMS = Gas Chromatography Mass JSC = Johnson Space Center Spectrometry LCROSS = Lunar Crater Observation and Sensing H-L = Henry’s Law Satellite H2 = Hydrogen MFC = Mass Flow Controller Hg = Mercury N2 = HOPA = Hydrogen Oxygen Production NASA = National Aeronautics and Space Assembly Administration

1 Ph.D., Aerospace Engineer, Paragon Space Development Corporation, [email protected] 2 Aerospace Engineer, Paragon Space Development Corporation, [email protected] 3 Thermal Analyst, Paragon Space Development Corporation, [email protected] 4 Ph.D., Aerospace Engineer, Paragon Space Development Corporation, [email protected]

Copyright © 2021 Paragon Space Development Corporation NH3 = O2 = Oxygen SO4 = ppb = Parts per billion (molar basis) TIC = Total Inorganic Carbon ppm = Parts per million (molar basis) TOC = Total Organic Carbon PSR = Permanently Shadowed Region TRL = Technology Readiness Level RH = Relative Humidity W.C. = Worst Case SBIR = Small Business Innovative Research WIPE = Water, ISRU-derived, Purification SO2 = dioxide Equipment I. Introduction ASA has identified a critical need for the design, fabrication, and testing of in-situ resource utilization (ISRU) N components to produce purified water, oxygen, and hydrogen on the Moon and/or Mars from regolith-based resources. Extended stays on the Lunar or Martian surface will require a readily available source of purified water. Once the water is purified, it can be used as a source of oxygen, (both as breathable air for habitat crew and as propellant oxidizer), and hydrogen as propellant fuel. In the near term, NASA has specifically identified the need for development and testing of critical components for the extraction and purification of water from ice that exists at the lunar poles in permanently, or near-permanently, shadowed regions (PSRs) on the Moon. Purification and electrolysis of in-situ lunar water has never been done before. It presents unique challenges related to the hazardous, corrosive, toxic, and flammable gases present with the lunar water and the lunar polar environment; as well as the typical constraints of system launched to the lunar surface (mass, volume, power, autonomy, robustness, reliability, and lifetime). This technology development is vital to allow humans to achieve a sustainable presence on the Moon. Paragon Space Development Corporation (Paragon) and our partner Giner, Inc. (Giner) are pursuing development and testing of key components in the ISRU-derived water purification and Hydrogen Oxygen Production (IHOP) subsystem (see Figure 1) as well as advancement of the subsystem architecture. Paragon is developing an innovative, contaminant robust subsystem that removes acidic and water soluble contaminants found within ISRU-derived water on the Moon that could corrode systems, degrade electrolyzer (or other downstream system) performance, or present serious toxicity issues to humans. In the IHOP subsystem shown in Figure 1, an H2O Cold Trap Assembly (Cold Trap) and Paragon’s Nafion-based Ionomer-membrane Water Processing (IWP) technology provide broadband contaminant filtration, while an ammonia (NH3) scrubber and water polisher are optimized for specific contaminant and final trace contaminant removal, respectively. The purified water is then electrolyzed using a static feed water electrolyzer to produce H2 and O2 propellant. This closely-coupled architecture of water purification and propellant production allows NASA maximum flexibility to adapt IHOP to a variety of upstream and downstream ISRU subsystems for applications on either the Moon or Mars. Our technical solution builds on both prior executed contracts as well as internal research and development.

Figure 1. IHOP Subsystem Block Diagram with IHOP Contract Development Level-of-Effort Labelling

2 International Conference on Environmental Systems II. Background NASA’s LCROSS mission measured the presence of water-ice in Table 1. “Updated” Lunar Volatiles the PSRs on the Moon’s South Pole at 5.6 wt% ± 2.9 wt% with respect 2,3 to regolith mass. In addition, the LCROSS mission detected the from LCROSS Data presence of other chemical species shown in Table 1. Many extraction methods of the lunar ice have been proposed in recent years and are Conc. (mol% Conc. undergoing active investigation. Cold trapping of water vapor and the relative to (wt% in cleanup of collected water are integral components to these Compound total H2O) regolith) architectures, but the technology to accomplish these tasks in the lunar H2O 100 5.60 PSR environment remains critically under-developed. Current state- H2 225 1.40 of-the-art hardware does not provide the capability to collect and CO 8.09 0.70 purify water from various lunar mining architectures. This research is H2S 7.30 0.77 aimed at closing that capability gap through the development of a Cold NH3 2.66 0.14 Trap to collect volatilized (sublimated) lunar water while purifying it SO2 1.40 0.28 of the co-located contaminants described in Table 1. C H 1.37 0.12 In the course of the execution of the IHOP contract, Paragon 2 4 CO 0.94 0.13 identified the need to define the expected contaminant load being 2 CH OH 0.67 0.07 provided to downstream IHOP assemblies from the H O Cold Trap 3 2 Hg 0.36 0.22 Assembly shown on the left of Figure 1. As part of this definition effort, Paragon developed a contaminant retention model based on CH4 0.28 0.014 Henry’s Law temperature-dependant of gasses in water, NOTE: These values are updated from the original described in Ref. 1. The model predictions of contaminant retention proposal and early reporting values. They are now from Ref. 1 are shown graphically in Figure 2. However, because a referred to as the “Updated” as per discussions with Cold Trap operating on the Moon will collect the water from the NASA customers and the LCROSS Principal process flow via deposition (the phase change of water from vapor to Investigator, Anthony Colaprete (JSC) solid, the opposite of sublimation), the Henry’s Law solubility analysis, while useful as an approximation, does not represent realistic performance. The present article describes the experiments conducted at Paragon with a simulated Cold Trap operated with relevant wall temperatures, flow pressures, water content, and contaminant Figure 2. First order estimate of the lunar water purification concentrations to study contaminant retention performance of the IHOP Cold Trap1 with deposited water vapor.

III. Methods & Apparatus

A. Test Overview & Description A contaminant tolerant test stand (CTTS) was developed for the IHOP program and configured to initially support the Cold Trap Experiment. The general overview of the test set up and procedures were as follows: 1) Generate pressure and temperature conditions in a Cold Trap bottle that allow for deposition of water and minimize the condensation, deposition, or solvation of contaminant gases; 2) Generate process gas conditions for gases flowing into the Cold Trap that match contaminant ratios with respect to water similar to those expected on the Moon; 3) Collect water samples; 4) Prepare liquid and gas samples for shipment to JSC for analysis. Initial testing included ring-out tests with inert gases (only nitrogen and water vapor) to check functionality, leak- tightness, and performance of all aspects of the test stand. These aspects were re-checked between each test batch to ensure no leaks developed in set-up, test, or post-test operations.

3 International Conference on Environmental Systems The test campaign generated 8 liquid samples and 6 gaseous samples:  1 liquid sample and 1 gas sample as a baseline with only N2 and water present  1 liquid sample and 1 gas sample with N2, methanol, and water present  1 liquid sample with N2, SO2, and water present  1 liquid sample with N2, NH3, and water present  1 liquid sample and 1 gas sample with N2, H2, H2S, CO, CO2, C2H4, CH4, and water present Figure 3. The Test Article Section of the Cold Trap  3 liquid samples and 3 gas samples with Experiment (CTE) (gas mixing not shown) each above the above components Gas samples were not taken from the tests with only SO2 and NH3 as analytical methods for their analysis were not developed by the analysis lab.

B. Test Article Description The test article, the Cold Trap bottle, is a Duran® PressurePlus Bottle, 500 mL, with a GL45 threaded cap. A Dreschel head impinger is attached to the bottle to direct incoming flow downward into the bottle. The interior tube of the impinger is for out-going flow and a larger diameter, surrounding tube provides the flow inlet, forcing flow to turn 180o inside the Cold Trap bottle. The bottle is rated “ resistance guaranteed to 1 bar vacuum” (i.e. rough vacuum) and “pressure resistance guaranteed to +1.5 bar gauge,” and made of borosilicate 3.3 glass with excellent chemical resistance. It is thermal shock resistant to a temperature difference of 30oC under vacuum or pressure, with maximum operating temperature of 140oC. A picture of the test article and concept of operations is shown in Figure 4. The dimensions of the test article are 8.6 cm diameter and a base-to-lip height of 17.6 cm. To mimic the how the Cold Trap would operate on the Figure 4. Duran® PressurePlus+ Laboratory Moon and to maintain sample quality, the water must be Bottle interfaced with the Dreschel head (left), deposited (i.e., not first condensed, then frozen) and must concept of operation (right) not be allowed to thaw before completion of a test. The Cold Trap bottle is used as both the chilled deposition surface and the liquid sample collection bottle, eliminating the need to melt the sample of deposited water-ice to move it into a sample vial.

C. Test Bed Description & Layout The CTTS is designed to provide gas and water vapor to the CTE test article at set flow rates, concentrations, temperatures, and pressures. The CTE configuration of the CTTS is specific to maintaining low pressure within the Cold Trap and sub-zero temperatures at the external wall of the Cold Trap while measuring both the pressure and temperature at the Cold Trap. To generate the collected water sample, a humidified stream of N2 is mixed with a gas flow from a cylinder(s) containing contaminant gas(es) to generate a humidified and diluted stream of mixed gases to the sub-zero cooled sample bottle via a gas impinger tube. The gas stream is metered into the Cold Trap bottle at a low flow rate to maintain vacuum pressure provided by a chemically resistant . The vacuum pressure and Cold Trap temperature are set such that water will preferentially deposit (not condense) on the interior surface of the sample bottle while minimizing the condensation of other species present in the gas stream.

4 International Conference on Environmental Systems D. Test Conditions & Test Matrix The operating conditions for the Cold Trap Experiment were designed to fall within a pressure and temperature range where water vapor is expected to deposited and, based on phase equilibrium diagrams, the other components will mostly remain uncondensed and allowed to be exhausted from the Cold Trap. The operational envelope of the Cold Trap Experiment, compared with the expected operational envelope of the lunar Cold Trap are compared in Figure 5. Similarly, in order to maintain similarity between conditions expected of the Lunar Cold Trap with those in the Cold Trap Experiment, the concentration of each contaminant was scaled up to be have the same proportionality with respect to water vapor as expected from the updated LCROSS values (except for hydrogen, which would have caused detonation concerns during Figure 5. Comparison of the Lunar Cold Trap and Cold testing). The tested concentrations are therefore Trap Experiment operational envelopes relative the much higher than in the lunar case because this single-component phase equilibrium diagrams of water experiment has a much higher water vapor partial and the lunar volatiles pressure than expected in the lunar Cold Trap. The predicted retained contaminants the collected in the water samples are presented in the right-most two columns of Table 2 (calculated on the basis of Henry’s Law solubility of each gas in liquid water at the temperature listed in parentheses, i.e., 298K or 273K, see Ref. 1) and shown as parts per million (ppm) concentrations. These explicit set points and conditions used for controlling parameters in the CTTS for each test batch is described below in Table 3. The values were subject to fine-tune adjustment during the test and revision prior to each test batch based on information learned during the previous test batches. The gases are grouped per their separate cylinders. The in-going process gas set point was 46oC with an RH of 66% dewpoint 38oC) measured by a Vaisala HMT 337 humidity temperature sensor. The coolant bath temperature was controlled to a -25oC set point with a Julbo recirculating Table 2. Cold Trap Inlet Gas Concentration chiller operating with a silicone oil working fluid. The and Estimated Retained Concentration in the upstream pressure (Pup) was measured by an Omega PX409 Water Sample, with Lower and Upper Ends of 0-30 psia pressure sensor and the downstream pressure (P- Analytical Assumptions ) was measured by an Omega PX209 0-15 psia pressure down Expected Expected sensor. Mass flow rates of the gases in the upstream mixing Conc. in Conc. in were controlled with ALICAT MCS series mass flow Mixed Sample - low Sample - high controllers and the Cold Trap flowmeter was a Cold Parmer Gas Conc. case (298K) case (273K) GH-03216-85 flow rotameter. The Cold Trap flowmeter and Formula (vol%) (ppm) (ppm) vacuum pump were used to maintain set point pressures of 92 kPa upstream and 2.6 kPa downstream of the Cold Trap.The CO 0.58 0.09 0.14 temperature, pressure, flow rate, and humidity data were H2 0.16 0.02 0.02 collected to evaluate the process conditions under which H2S 0.52 8.77 16.71 water is collected. NH3 0.19 1884.97 6842.90 Water-ice samples were collected at Paragon during the SO2 0.10 21.86 53.24 Cold Trap Experiment to test the water purification C2H4 0.10 0.10 0.19 performance of a Cold Trap when exposed to a humidified CH4 0.02 0.005 0.01 process gas containing nitrogen and either single CH3OH 0.05 1609.45 8979.59 contaminants or a mixture of contaminants from lunar water. CO2 0.07 0.37 0.78 Each test batch was conducted with a target duration of at least 600 minutes (10 hours) to maximize sample collection volume for analysis (minimum 100 mL required). Dry sample collection bottles with caps on were weighed before testing, then again after testing was complete.

5 International Conference on Environmental Systems Collected samples were sealed with mostly N in the 2 Table 3. Test Matrix for the Cold headspace (backfilled) and allowed to thaw and Trap Experiment Test Batches equilibrate to room ambient temperature (~19oC). The gas headspace was then sampled through a gas transfer Test Conc. Gas MFC Flow Rate apparatus at Paragon and a quick-disconnect grab Batch (vol%) sampling with a NASA-provided evacuated gas canister for each sample. After the headspace transfer, the bottle N2 1.12 92.8 CTE_A 1 was backfilled with N2 to re-pressurize it to ambient H2O (SLPM) 7.2 pressure (13.4 psia). The gas in the headspace that was N2 1.02 92.0 sampled is thus assumed to be only from capture with 1 H2O (SLPM) 6.55 the collected water. CO 0.57 Because liquid sample collection volumes were all H2S 0.52 below 100 mL, dilution was performed at Paragon with CTE_B H2 100 0.16 a known amount of de-ionized (DI) water (>1 MΩ-cm 2 C H (SCCM) 0.10 resistivity, same as used in the humidifier). The dilution 2 4 CO 0.07 was performed under 13.4 psia atmosphere at room 2 CH 0.02 temperature, keeping the amount of time any water was 4 N2 1.10 92.83 exposed to atmosphere as low as possible. This dilution 1 factor is accounted for in all liquid sample analysis CTE_C H2O (SLPM) 7.07 results, as emphasized by the reported “Molar Conc. SO2 3 25 (SCCM) 0.10 N2 1.10 92.74 Original Sample (ppm)” label. While contact with 1 atmospheric CO2 was minimized with N2 blanketing in CTE_D H2O (SLPM) 7.07 the container, it is expected that atmospheric CO2 was NH3 5 25 (SCCM) 0.19 present in the liquid sample bottle head spaces due to N2 1.12 92.78 1 lack of stringent inert atmosphere control. CTE_E H2O (SLPM) 7.17 Sample bottles were sent to the Toxicology and CH3OH 1 3 (SCCM) 0.05 Environmental Sciences Laboratory at NASA Johnson N2 1.12 82.9 Space Center for analysis. The reporting limit for each 1 H2O (SLPM) 6.4 liquid sample analysis test/compound varied depending CO 0.57 on the Test Batch due to further dilution by the sample H S 0.52 analysis lab in order to protect equipment. These 2 H2 100 0.16 dilutions are already accounted for in reported values. 2 C H (SCCM) 0.10 The reporting limit at the equipment for each liquid 2 4 sample analysis is reported in Table 4. The equipment CTE_F-1 CO2 0.07 detection limit for each gas sample analysis CH4 0.02 test/compound was: methane – 3.9 ppm, carbon SO2 3 25 (SCCM) 0.10 100 monoxide – 0.4 ppm, hydrogen – 1.7 ppm, carbon N2 4 8.9 dioxide – 390 ppm, – 0.4 ppm. (SCCM) However, these limits have minor deviations as reported NH3 5 25 (SCCM) 0.19 in Table 5 due to corrections accounting for variations in CH3OH 1 3 (SCCM) 0.05 pressure at time of collection and total amount of CTE_F-2 Repeat Above collected sample. CTE_F-3 Repeat Above NOTE: All values are SET POINTS IV. Results The presented “Total Conc. Captured (ppm)” is a calculated on a molar basis from the total gaseous component detected and the assumed equilibrium concentration of the component in liquid water (in accordance with Henry’s Law equilibrium of the component and water), then corrected for the dilution factor. The values of “Molar Conc. Original Sample (ppm)” are calculated on a molar basis to compare sample analysis results directly with the pre-test estimates of the expected component concertation retained with water based on the Henry’s Law analysis referenced Table 2. The relevant averaged data and summarized measured ranges for each test batch are shown in Table 5, along with the processed results from both liquid and gas sample analysis of the collected sample from each test batch. The following is an accounting of the off-nominal test operations (if any occurred) from each Test Batch and the possible issues that may confound some results.

6 International Conference on Environmental Systems

Table 4. Reporting Limits of Liquid Sample Analytical Methods Cold Trap Experiment (CTE) Test Batch Measurement units A B C D E F1 F2 F3 Sulfate (Anions IC) mg/L 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 Total N (Cations IC) mg/L 0.25 0.25 0.25 5 5 12.5 2.5 2.5 Total S (Total S ICPMS) mg/L 0.4 0.4 0.4 0.4 0.4 0.4 0.4 0.4 Total Inorganic Carbon (TIC) mg/L 0.6 0.6 0.6 30 30 30 30 30 Total Organic Carbon (TOC) mg/L 0.4 0.4 0.4 20 20 20 20 20 Methanol (Alcohols & GCMS) mg/L 0.4 20 20 20 2 20 20 200

1. Test Batch CTE_A The sample in bottle B3 is a combination of two tests (CTE_Av1 and CTE_Av2), both with the same conditions for CTE_A, in addition to dilution to reach 100 mL described above. From the mixing and dilution this sample likely contained atmospheric CO2 in the liquid sample headspace. Other Test Batches received diluent water after gas headspace sampling and no other Test Batch liquid samples were a mixture of collected samples. 2. Test Batch CTE_B Test Batch CTE_B proceeded nominally and no caveats on the data exist. 3. Test Batch CTE_C Test Batch CTE_C proceeded nominally and no caveats on the data exist. 4. Test Batch CTE_D During the pre-test set-up of conditions and flow rates for Test Batch CTE_D, one test anomaly occurred. A gas washing bottle containing 1M acetic acid was used on the exhaust of the CTTS in order to remove ammonia (through an acid-base reaction) from the gas being exhausted into the fume hood. During the N2 purge step (to remove any ambient air from the CTTS) the wrong valve was opened for ~1 second, exposing the gas washing bottle to vacuum conditions present within the test stand (also a simultaneous pre-test set up step). This error led to an estimated (visually) 1-2 mL of acetic acid to backflow into the evacuated section of tubing between the three-way valve, the Pup purge valve, and the flowmeter of Figure 3. Because the acetic acid would absorb the ammonia gas used for CTE_D, the pre-test set up was paused in order to isolate and disconnect the affected tubing and flush it with DI water. Once the tubing was re-assembled, the CTTS was again purged with N2 to remove any acetic acid that may have remained to the extent possible without flowing any of the possibly contaminated vapor into the Cold Trap Bottle. Unfortunately, this ad-hoc cleaning procedure did not remove all of the acetic acid present, as is evidenced by the TOC measurement for CTE_D (where otherwise there should have been no components present that would have generated a TOC signal). The analysis and implications of this test anomaly are discussed further in the Discussion section. 5. Test Batch CTE_E Test Batch CTE_E supplied methanol as the sole intended contaminant, yet the methanol signal was below detection limits and the TOC signal was present (Table 5, both methanol and acetic acid should generate a TOC signal). Because each test was conducted one after the other, and as a result of the anomaly from CTE_D described above, it appears evident from the liquid sample analysis results that some additional amount of acetic acid still remained in the CTTS in between tests. The analysis and implications of this test anomaly are discussed further in the Discussion section. 6. Test Batch CTE_F1 Test Batch CTE_F1 was the first test batch to supply the Cold Trap Bottle with all contaminants at once. Similar to the discussion from CTE_E, the methanol signal was below detection limits and the TOC signal was present (Table 5, both methanol and acetic acid should generate a TOC signal). While CO2 from cylinder X2 could have increased the TOC signal, results from CTE_B indicate that the CO2 concentration supplied by cylinder X2 was insufficient to raise the TOC signal significantly above the baseline measurement from CTE_A. It is therefore expected that some additional amount of acetic acid still remained in the CTTS in between tests, despite now multiple N2 purge operations. A second anomaly observed during Test Batch CTE_F1 was the increase of the temperature of the coolant bath during the test. The test was halted early in an attempt to preserve quality of the sample: if the Cold Trap Bottle temperature were allowed to rise during the test, the already-collected water-ice could melt or the incoming water vapor could condense, generating liquid water in the sample container. This liquid water could, in turn, absorb the contaminant vapors and artificially increase their concentration in the sample through solubility equilibrium and/or aqueous reactions. The value of “End Coolant Bath Temp. (C)” of column CTE_F1 in Table 5 is highlighted to show

7 International Conference on Environmental Systems that the temperature at the end of the test was −12.3oC. The analysis and implications of this test anomaly are discussed further in the Discussion section. Lastly, formation occurred in the salt trap in Figure 3. White, crystalline particles were clearly visible on the salt trap walls during and at the conclusion of CTE_F1. This salt formation was anticipated and the intent of the salt trap was to allow collection of the salts upstream of tubing and valves that could be clogged by the salt if allowed to blow through the CTTS. The salt is anticipated to be due to the mixing of NH3 with SO2 and H2S to form salts (sulfide, sulfite, and sulfate). The analysis and implications of this test anomaly are discussed further in the Discussion section. 7. Test Batch CTE_F2 As resolution for the second test anomaly observed in Test Batch CTE_F1, the recirculating chiller used to cool the Cold Trap Bottle coolant bath was heated up to 95oC and held there for c.a. 2 hours to bake out the presumed absorbed water vapor that had contaminated the silicone oil coolant. The absorbed water had both increased viscosity and froze into chunks of ice within the oil, causing flow blockages and pumping issues that led to the rise in coolant bath temperature observed in CTE_F1. Heating the coolant caused the chunks of ice to disappear and visible water vapor to bake out. Thus, the previous issue appeared to have been completely mitigated prior to the start of CTE_F2. With the additional time used between CTE_F1 and CTE_F2 to address the coolant temperature issue described above, the CTTS tubing exposed to contaminant flow was more thoroughly disassembled, cleaned with DI water, and reassembled prior to starting CTE_F2 as a further precaution to improve or preserve quality of the collected sample. This appears to have been advantageous, as the TOC signal of CTE_F2 in Table 5 show a marked decrease compared with the previous tests. Similar to CTE_F1, salt formation occurred in the salt trap. The analysis and implications of this test anomaly are discussed further in the Discussion section. 8. Test Batch CTE_F3 The coolant bath temperature rise anomaly that occurred during Test Batch CTE_F1 was again observed during CTE_F3. The summary data in the CTE_F3 column of Table 5 show that the “End Coolant Bath Temp. (C)” reached −9.8oC prior to the test being prematurely ended after only 419 minutes of operation. The analysis and implications of this test anomaly are discussed further in the Discussion section. Similar to CTE_F1 and CTE_F2, salt formation occurred in the salt trap.

Table 5. Summarized Results of each Test Batch for the Cold Trap Experiment Test Batch Description Legend CTE_A H2O, N2 (background) >= 10x above prediction CTE_B CO, H2S, H2, C2H4, CO2, CH4, H2O, N2 >= 1x above prediction CTE_C SO2, H2O, N2 <= 1x below prediction CTE_D NH3, H2O, N2 <= 10x below prediction CTE_E CH3OH (MeOH), H2O, N2 below detection limit, upper limit listed CTE_F1 CO, H2S, H2, C2H4, CO2, CH4, SO2, NH3, CH3OH, H2O, N2 predicted concentration CTE_F2 CO, H2S, H2, C2H4, CO2, CH4, SO2, NH3, CH3OH, H2O, N2 test or lab anamoly CTE_F3 CO, H2S, H2, C2H4, CO2, CH4, SO2, NH3, CH3OH, H2O, N2 N/T = not tested; N/A = not applicable

8 International Conference on Environmental Systems

V. Discussion Water collection efficiency was higher for tests with lower coolant bath temperatures (except for CTE_A, where issues with flow blockages occurred). Despite the un-optimized cold trap design, the sub-cooled wall temperatures (coolant bath temps in Table 5) collected water with good efficiency when wall temperature remained below −20oC. The remainder of the water vapor was lost to vacuum. At these pressures and flow rates, even a poor cold trap design works well enough to collect appreciable amounts of water at a -20oC sub-cooled wall. Future cold trap designs need to take into account the optimization that can occur with flow path design and increased surface area. It may be possible that as water collection efficiency increases so too does contaminant retention, but these results do not indicate strong accumulation. Observations of collected water samples appear to show vapor deposited ice (or at least rapidly condensed and frozen ice) due to the lack of pooled water, high roughness, opacity, and dissimilarity to amorphous ice. This can be seen in the post-test photos shown in Figure 6 (CTE_B, CTE_C, CTE_F2). The test pictures from the Test Batches that had the fewest (or no) anomalies provide us with reasonable confidence that this was water deposition or rapid condensation & freezing, providing thermophysical process similarity to the expected performance of a lunar water cold trap. Whereas pictures from Test Batches that had coolant bath temperature anomalies (CTE_E, CTE_F1, CTE_F3) do not appear to have formed ice similarly to or as well as those shown in the top row of Figure 6. For completeness, photos of the baseline/background tests of water collection (CTE_Av1 and CTE_Av2) as well as of CTE_D are shown in Figure 6 as well. Analysis of the sample analysis results in tandem with an interpretation of the validity of the results of each test batch allow the generation of Table 6 as a summary of the Cold Trap Experiment results. Table 6 can also serves as a rough validation of the use of the Henry’s Law analysis model as a conservative estimate of contaminant retention. These experiment-to-model ratios can then be applied as an input to the predicted ranges of contaminant concentration downstream of a lunar Cold Trap. The worst case and best case experimental result estimates are chosen from the tests listed adjacent to the estimates on the bases provided below. Upon review of the test results shown in Table 5 and in light of the anomalies that occurred during the Cold Trap Experiment, we believe that the Test Batches CTE_B, CTE_C, and CTE_F2 in general provide the most reasonable estimates of anticipated performance of a Cold Trap for water vapor purification through freeze distillation with the provided contaminants. For ammonia, CTE_F1 showed the highest concentration of retained ammonia, but CTE_F3 has the highest likelihood of not being affected by the acetic acid contamination event (see the below discussion on test anomaly effects).

9 International Conference on Environmental Systems From Table 6 it is clear that a bulk amount of the lunar water contaminants detected by LCROSS were filtered from the collected water during the freeze distillation process that occurred in the Cold Trap, demonstrating proof of concept and justifying further development of this process into a lunar ISRU technology.

Table 6. Cold Trap Experiment Sample Analysis Results Comparison to Henry’s Law Analysis Model with Worst and Best Case Experimental Result Estimates Experimental Inputs Model Worst Case Experimental Result Best Case Experimental Result Tested H-Law Exp. Exp.-to- Exp. Exp.-to- Conc. w.r.t 273K Conc. Result Model From Result Model From Formula H2O (ppm) Est. (ppm) Est. (ppm) (ratio) Test: Est. (ppm) (ratio) Test: CO 80900 0.14 0.72 5.09 CTE_B 0.06 0.43 CTE_F2 H2 22500 0.02 0.07 2.73 CTE_B 0.03 1.35 CTE_F2* H2S 73000 16.75 5.62 0.34 CTE_F2 1.16 0.07 CTE_B NH3 26600 6862.46 2380.90 0.35 CTE_F1 450.49 0.07 CTE_F3 SO2 14000 53.40 57.05 1.07 CTE_F2 1.84 0.03 CTE_C C2H4 13700 0.19 1.57 8.22 CTE_F2* 0.19 1.00 Est.** CH4 2800 0.01 0.07 8.22 CTE_F2* 0.01 1.00 Est.** CH3OH 6700 9005.25 3834.22 0.43 CTE_F3 54.00 0.01 CTE_F2* CO2 9400 0.78 7.38 9.45 CTE_B* 0.78 1.00 Est.** Updated LCROSS #'s *Det. Lim. *Det. Lim., **Model used

E. Lunar Cold Trap Estimates This application of the CTE results allows for a rough prediction of the ranges of contaminant concentration downstream of a lunar Cold Trap. The worst case (W.C.) and best case (B.C.) estimates are the highest and lowest concentrations as described above in Table 6. These predictive estimates are shown in Table 7 and graphically in the series of charts in Figure 7. The experiment-to-model ratios from Table 6 are multiplied with the Henry’s Law analysis model results for each contaminant in the lunar case: where incoming water vapor to the lunar Cold Trap comes from a capture tent at a partial pressure of 2.6 Pa and partial pressures of contaminants are proportional (based on LCROSS estimates) to the water vapor partial pressure.

F. Discussion of Test Anomaly Effects Over the course of the Cold Trap Experiment a few off-nominal test operations occurred that may confound some results. The following is the discussion of each of those anomalies as they relate to interpretation of the results from each Test Batch. 1. Atmospheric Gases The atmospheric CO2 present in the collected samples is most evident in CTE_A. The pH of DI water is nominally 7. Contact of DI water with atmospheric CO2 produces carbonic acid which can reduce water’s pH to 5.5. The pH of the liquid water sample from CTE_A is 5.71. The air quality analysis report shows that the CO2 levels were below 400 ppm (atmospheric ambient CO2 concentration), so it appears that gas headspace quality was preserved (i.e., more nitrogen present than air as the inert gas blanket from test shutdown) to some extent for gas samples. Because this is the background water test, and Figure 6. Post-Test Cold Trap Photos because similar procedures were followed for each subsequent test (dilution in ambient atmosphere occurred with all other tests, but no other samples were mixed from two separate days of testing), it is assumed that all liquid sample bottles contained some CO2 in the headspace and in equilibrium with the liquid as a result of the atmospheric CO2 contamination.

10 International Conference on Environmental Systems 2. Acetic Acid Table 7. Predictive Estimates of Contaminant Retention in Lunar Cold Contamination Trap (Based on a Capture Tent ISRU Water Extraction Architecture) The acetic acid contamination event of the Worst Case Lunar Best Case Lunar CTTS affected Test Batches Lunar Inputs Model Est. Est. CTE_D, CTE_E, CTE_F1, and Conc. H-Law Exp.- Lunar Lunar CTE_F2. Acetic acid is an w.r.t 273K to- Cold Trap Exp.-to- Cold Trap organic acid (formula: H2O Conc. Est. Model Outlet Est. Model Outlet Est. CH3COOH). Because of its Formula (ppm) (ppb) (ratio)* (ppb) (ratio)* (ppb) organic nature, it is detectable in CO 80900 0.05 5.09 0.3 0.43 0.02 total organic carbon (TOC) H2 22500 1 2.73 2.7 1.35 1.3 liquid sample analysis H2S 73000 7 0.34 2.2 0.07 0.5 measurements. Its presence in NH3 26600 2677 0.35 929 0.07 176 the sample analysis results for SO2 14000 21 1.07 22 0.03 0.7 each of the aforementioned Test C2H4 13700 0.1 8.22 0.6 1.00 0.1 Batches is indicated by 1) a TOC CH4 2800 0.003 8.22 0.03 1.00 0.003 signal from CTE_D (when no CH3OH 6700 3512 0.43 1495 0.01 21 organic carbon was supposed to CO2 9400 0.3 9.45 3 1.00 0.3 be present), and 2) a TOC signal much greater than a methanol signal for CTE_E, CTE_F1, and CTE_F2. By CTE_F3, the methanol signal approximately matches the TOC signal, indicating that methanol is the primary contributor to the TOC signal (low to no acetic acid present). For CTE_D, CTE_E, and CTE_F1, this acetic acid anomaly may have affected the Ammonium (as Total N) measurement in two ways: 1) acetic acid upstream of the Cold Trap Bottle in the lines of the CTTS may have absorbed ammonia to form , depressing the actual concentration of ammonia gas that reached the Cold Trap Bottle, and 2) acetic acid, which has a vapor pressure of 15.7 mmHg at 25oC (water has a vapor pressure of 23.8 mmHg at 25oC), may have condensed in the Cold Trap Bottle allowing for its reaction with ammonia there and trapping ammonia that may otherwise not have condensed or collected there. A further complication is that upstream ammonium acetate that may have formed in one Test Batch may have still been present in the CTTS during the next Test Batch (despite DI water rinsing and N2 purges before and after each test). As ammonium acetate has a vapor pressure of 13.9 mmHg at 25oC, previously formed ammonium acetate may have vaporized and condensed in the Cold Trap Bottle of the next test batch. This is evidenced by the TOC and Ammonium signals present during CTE_E, where only methanol should have been present. These issues lead us to have lower confidence in the TOC and Ammonium measurements from CTE_D, CTE_E, and CTE_F1. Between CTE_F1 and CTE_F2, a more thorough cleaning of the CTTS was performed due to required downtime between tests to resolve the coolant bath temperature issue that allowed for complete leak checking of the CTTS after cleaning. As the TOC levels dropped considerably in CTE_F2 from CTE_F1, this cleaning appears to have removed a significant amount of remaining acetic acid or accrued ammonium acetate. The near matching measurements of TOC and methanol during CTE_F3 lead us to believe that the acetic acid test anomaly was no Figure 7. Summary of Henry's Law (H.L.) Estimates, Worst Case longer affecting the Ammonium (W.C.) and Best Case (B.C.) Experimental Molar Contaminant measurement, thus giving higher Retention Concentrations Converted to Estimates at Lunar Conditions with respect to Theoretically Collected Water

11 International Conference on Environmental Systems confidence in the Ammonium measurement of CTE_F3 than all other CTE Test Batches. 3. Coolant Bath Temperature Rise The coolant bath temperature rise anomaly appears to have most significantly affected Test Batches CTE_F1 and CTE_F3, and to a minor extent, CTE_E. The primary impact was the early termination of each of these tests in order to preserve the quality of the collected water sample, resulting in lower total amounts of water collected, higher dilution factors, and shorter exposure of the accumulated water-ice to the contaminant flow. The secondary, though also important, impact of this anomaly its effect on the quality of the collected water sample with regards to contaminant concentrations. As temperatures in the coolant bath rose, so too did the effective temperature inside the Cold Trap Bottle. Because glass and ice both act as thermal insulators, it is likely that the temperature inside the Cold Trap Bottle was higher than the actual coolant bath temperature. As this temperature rises, the likelihood of water condensation and water- ice melting increases. The contaminants are far more likely to dissolve and accumulate in liquid water, and possibly even react with one another, further increasing likelihood of accumulation. Across CTE_F1, CTE_F2, and CTE_F3, this is readily apparent: both CTE_F1 and CTE_F3 had much higher concentrations of almost all contaminants compared with CTE_F2. For this reason, we generally have higher confidence in the results shown by CTE_F2 compared with those from CTE_F1 and CTE_F3. The notable exceptions are for ammonium and methanol measurements from CTE_F3 (see above section on acetic acid contamination). 4. Total Sulfur Measurement for CTE_F1 The total sulfur measurement of CTE_F1 appears incongruous with the sulfate measurement for the same test. In communication with the Water Quality Analysis Laboratory at JSC, if sulfate is present in the sample, one expects it will show up in the total sulfur measurement as well. Their explanation is that while the detection limit for total sulfur is as stated, sometimes when the measurement is close to the limit, the signal may not show up or gets filtered out of the data. 5. Salt Trap Precipitate Accumulation A “salt trap” was included in the CTTS to mitigate the issue of reactive gases (namely NH3, SO2, and H2S) generating solid precipitate that could clog valves and accumulate in tubing or the Cold Trap Bottle. It was also included to determine how much of each reactant precipitated into salt over the course of the mixed gas tests. At the end of the CTE test campaign, the salt trap was removed and weighed with mass compared to its empty baseline. In total, 8.5g was collected in the salt trap over the course of CTE_F1 through CTE_F3. Over the course of these three tests, 13.62g of H2S, 4.96g of SO2, and 2.49g of NH3 were supplied to the CTTS from the gas cylinders. The salts that may have formed include ammonium sulfite (NH4)2SO3, ammonium sulfate (NH4)2SO4, ammonium hydrosulfide (NH4), HS, and ammonium sulfide (NH4)2S. Of these salts, the sulfite and sulfate are far more stable over time and temperature than the . Analysis of the collected salt mass and supplied gas masses reveals that hydrogen sulfide reacting with NH3 would have generated 7.5g of salt mass, whereas reacting with NH3 would have generated 8.5g of salt mass. These two factors combined lead us to believe that the collected salt was mainly based on the reaction between NH3 and SO2. In the absence of oxygen, NH3 and SO2 gases will react to produce ammonium sulfite, not the sulfate. The effect that the salt reactions had on the Cold Trap Experiment results is general reduction in the concentration of the NH3, SO2, and H2S contaminants that reached the Cold Trap Bottle during tests CTE_F1 through CTE_F3. Therefore, CTE_B and CTE_C are expected to have a more representative sample analysis result of the sulfur compound presence in the retentate of the cold trap if no ammonia were present. However, as mass of salt accumulated in the salt trap over each CTE_F1-3 test, it is possible that the lightweight, powdery salt was blown into the Cold Trap Bottle on either the same or subsequent Test Batches in that subset of Test Batches. This could explain the large total sulfur and sulfate measurements in Test Batch CTE_F3 to some extent. Three further questions regarding these salts remain: 1) Are the NH3, SO2, and H2S measured by LCROSS actually present on the Moon as pure ices, or are they present as salts? 2) If the NH3, SO2, and H2S are present as pure ices and volatilize with water-ice, would they react and precipitate out of the “flow” before they even get to the Cold Trap on the lunar surface? 3) Since most of the salts in the Cold Trap Experiment accumulated in the higher pressure portion of the CTTS, would their reaction and precipitation be just as likely in a less dense (i.e. lower pressure) “flow” as expected in a lunar system? These questions cannot be answered by this study, but they are important to consider in future efforts.

12 International Conference on Environmental Systems VI. Conclusions These goals of this experiment accomplished, as demonstrated by the results of Table 5 and Table 6; estimations shown in Table 7 and Figure 7; photos in Figure 6. The general takeaways are:

 Hydrocarbons (methane and ethylene) are not present at concentrations sufficient to be retained with water during deposition enough to exceed detection limits. These limits are higher than the Henry’s Law analysis estimation.  Carbon dioxide is not present at a concentration sufficient to be retained with water during ice deposition enough to exceed detection thresholds as a gaseous species. These thresholds are higher than the Henry’s Law analysis estimation. It is likely present at its equilibrium dissolved species in the liquid water due to the diluent water added to the liquid samples prior to analysis.  Methanol appears to be poorly retained in a Cold Trap at these conditions, providing for significant reduction in its concentration from the input contaminant stream to the collected water sample, falling below the Henry’s Law analysis estimation. The captured methanol in CTE_F3 (3834 ppm in 11.4 mL of water, or 143 mg total), while likely present as a liquid, may not have been enough volume to absorb significant amounts of other contaminants.  Hydrogen and carbon monoxide are both small molecules that appear to have a variability within ± one order of magnitude of their Henry’s Law analysis estimations for contaminant retention with deposited water ice. Due to anomalies with CTE_F1 and CTE_F2, and due to no expected cause of retention for these molecules other than gas bubble or physical solubility trapping, the Henry’s Law analysis estimation is recommended as the best estimate for retention for both H2 and CO.  Due to the test anomalies, it is difficult to state definitively how ammonia can be expected to be retained with water during water vapor deposition. The best estimates appear to be from CTE_F2 and CTE_F3 where the acetic acid anomaly appears to have been largely resolved, though ammonium-sulfur salt generation upstream of the Cold Trap likely contributes to a low ammonia concentration reaching the Cold Trap in those tests.  On their own, sulfur compounds (sulfur dioxide and hydrogen sulfide) are poorly retained with water during water vapor deposition. The introduction of either ammonia or liquid water (resulting from higher coolant bath temperatures) appear to increase their retention.  Ammonia and sulfur containing compounds will react with one another when brought into contact in the gas phase with or without oxygen present. They will form salt that is soluble in both water and methanol.

VII. Future Work While this Cold Trap Experiment represents a proof of concept of water purification via “freeze distillation” in a Cold Trap with a controlled wall temperature, more rigorous experiments and measurements conducted with these contaminants, water vapor, and lunar surface system internal conditions (in a laboratory environment) could reveal better metrics through which to tailor the design of such a Cold Trap and estimate its performance with respect to water purification. Other aspects of Cold Trap design must be considered, such as optimization of water collection efficiency through internal geometry design; means of isothermalizing the interior wall temperature despite high heat rejection loads with temporal and spatial variations; interfaces of the Cold Trap; and overall thermal control system architecture that allows for heat management in both collection and evacuation (heating) modes. These are being developed through Paragon’s ISRU Collector of Ice in a Cold Environment (ICICLE) SBIR program. As it relates to the scope of work in Paragon’s IHOP component development effort, these results indicate that the Henry’s Law analysis model serves as a viable conservative estimate (by up to 1-2 orders of magnitude in the case of some contaminants) for outlet concentrations of contaminants in the interfacing flow to a downstream component. This model and these results can thus be used as the input estimate for contaminant concentrations with respect to water vapor being provided to Paragon’s IWP bundle for the next major water purification step in the IHOP subsystem. Within the IHOP subsystem, the water delivered from this Cold Trap would be further purified, then delivered to a water electrolyzer for electrolysis into H2 and O2 (see Figure 1 for a subsystem process flow depiction). Further work on developing the IWP bundle is presented in a companion ICES-2021-295 paper4. Component development and testing, system design and analysis, and assembly integrated testing continue on the IHOP contract at Paragon.

Acknowledgments The authors would like to thank James Schuler for his assistance in running long duration tests. The authors would also like to thank Ed Hudson and Steve Beck of the Toxicology and Environmental Sciences Laboratory at JSC. Gracious acknowledgment of support for the sample analyses is extended to Aaron Paz of JSC. Thanks also to David Eisenman for his technical support. This work was supported in part by NASA contract 80HQTR19C0018.

13 International Conference on Environmental Systems References 1Holquist, J. B., Pasadilla, P., Bower, C., Cognata, T., Tewes, P., and Kelsey, L., “Analysis of a Cold Trap as a Purification Step for Lunar Water Processing,” International Conference on Environmental Systems, ICES-2020-71, 2020. 2Colaprete, A., et al., “Detection of water in the LCROSS ejecta plume,” Science, Vol. 330, No. 6003, 2010. pp. 463-468. 3Gladstone, G. R., et al. “LRO-LAMP observations of the LCROSS impact plume,” Science, Vol. 330, No. 6003, 2010. Pp.472- 476. 4Holquist, J. B., Gellenbeck, S., Bower, C., and Tewes, P., “Demonstration of Paragon’s Ionomer-membrane Water Processing (IWP) Technology as a Purification Step for Lunar Water Processing,” 50th International Conference on Environmental Systems, ICES-2021-295, 12-15 July 2021.

14 International Conference on Environmental Systems