Performance of an Electrothermal Swing Adsorption System with

Jun 21, 2013 - (8) demonstrated the proof of concept for this technology, and Johnsen et al.(9) described the control of electrothermal heating of the...
0 downloads 0 Views 2MB Size
Download PDF
Article pubs.acs.org/est

Performance of an Electrothermal Swing Adsorption System with Postdesorption Liquefaction for Organic Gas Capture and Recovery Kaitlin E. Mallouk and Mark J. Rood* Department of Civil and Environmental Engineering, University of Illinois, Urbana, Illinois 61801, United States ABSTRACT: The use of adsorption on activated carbon fiber cloth (ACFC) followed by electrothermal swing adsorption (ESA) and postdesorption pressure and temperature control allows organic gases with boiling points below 0 °C to be captured from air streams and recovered as liquids. This technology has the potential to be a more sustainable abatement technique when compared to thermal oxidation. In this paper, we determine the process performance and energy requirements of a gas recovery system (GRS) using ACFC-ESA for three adsorbates with relative pressures between 8.3 × 10−5 and 3.4 × 10−3 and boiling points as low as −26.3 °C. The GRS is able to capture > 99% of the organic gas from the feed air stream, which is comparable to destruction efficiencies for thermal oxidizers. The energy used per liquid mole recovered ranges from 920 to 52 000 kJ/mol and is a function of relative pressure of the adsorbate in the feed gas. Quantifying the performance of the bench-scale gas recovery system in terms of its ability to remove organic gases from the adsorption stream and the energy required to liquefy the recovered organic gases is a critical step in developing new technologies to allow manufacturing to occur in a more sustainable manner. To our knowledge, this is the first time an ACFC-ESA system has been used to capture, recover, and liquefy organic compounds with vapor pressures as low as 8.3 × 10−5 and the first time such a system has been analyzed for process performance and energy consumption.

■

concept for this technology, and Johnsen et al.9 described the control of electrothermal heating of the ACFC. This benchscale system uses adsorption on activated carbon fiber cloth (ACFC) with electrothermal swing adsorption (ACFC-ESA) to generate a low-flow, high-concentration organic gas stream during regeneration cycles. The organic gas is then condensed so that the organic compound can be reused in its liquid form. The condensation apparatus provides compression and cooling that allows for pressure and temperature control of the desorbed gas stream to condense the adsorbate. ACFC-ESA has been used for capturing low-concentration organic vapors (e.g., toluene, methyl ethyl ketone, and acetone) for over 30 years.10−12 ACFC-ESA has also been implemented for capture and liquid recovery for reuse of low-concentration organic vapors at the bench and pilot scale.13−16 Extensive characterization of the vapor-recovery system has occurred, including material and energy balances13,16 and cost estimates.14,17 Additionally, Dombrowski et al.15 characterized the energy used by an ACFC-ESA system per liquid mole of organic vapor recovered. In that analysis, there was no postdesorption treatment of the gas stream and none of the desorption gas was recycled to an adsorbing vessel. Dombrowski et al.15 determined that the energy required to recover organic vapors as liquid ranged from 300 to 4700 kJ/ mol. Ramirez et al.18 characterized the energy to recover methyl

INTRODUCTION Several industrial processes (e.g., packaging manufacturing and oil refining)1 produce gas streams containing low-concentration (300 m3/min) and recovering the captured organic gases at reusable concentrations/phases. Available methods for recovering organic gases include absorption, condensation, membrane separation, and adsorption.3−6 Adsorption is an optimal recovery technique for lowconcentration organic gases because it can have >90% removal efficiency for organic gases with molecular weights >45 g/mol and concentrations >500 ppmv.7 Previous research has described the development of a gas recovery system (GRS) to capture, recover, and condense dilute organic gases. Mallouk et al.8 demonstrated the proof of © XXXX American Chemical Society

Received: February 4, 2013 Revised: May 8, 2013 Accepted: June 6, 2013

A

dx.doi.org/10.1021/es4005703 | Environ. Sci. Technol. XXXX, XXX, XXX−XXX

Environmental Science & Technology

Article

Figure 1. Experimental apparatus. Solid and dashed lines represent the flow paths for adsorption and desorption cycles, respectively.

of ACFC (Kynol ACC5092-15). Each ACFC cartridge had 20 layers of ACFC and was 25 cm from top to bottom with a 5 cm outer diameter and 2 cm inner diameter. Temperature of the ACFC was measured by use of 0.16 cm (0.062 in) diameter type K thermocouples (Omega, Inc.). During the adsorption cycle, pressurized house air passed through a pretreatment system that consisted of a high-efficiency particulate air (HEPA) filter, silica gel, and ACFC to remove particulate matter, water vapor, and organic compounds, respectively. The air flow rate ranged from 10 to 50 standard liters per minute (SLPM; T = 273 K, P = 1 atm) depending on the adsorbate and was controlled with a mass flow controller (Aalborg, model GFC571S). This is equivalent to a gas velocity through the outer cross section of the cartridge of 4.2 × 10−3 to 2.1 × 10−2 m/s.18 The air was then combined with the organic gas, which was obtained from a pressurized cylinder (purities as follows: nbutane, 95%; isobutane, 98.3%; R134A, 99.5%) at a controlled flow rate. All gas flow rates were controlled with mass flow controllers (n-butane, isobutane, and R134A, Alicat Scientific Model MC-200SCCM-D5/M; nitrogen, Tylan Inc. FC-280). The organic gas concentration was set to a given relative pressure (Pi/Pi,s), where Pi is the partial pressure of compound i and Pi,s is the saturation vapor pressure of compound i at the temperature of the experiment. The organic gas and air stream entered the side of one vessel and passed through the ACFC cartridge from the outside to the inside of the cartridge. The empty bed contact time determined from the envelope volume of the ACFC ranged from 0.15 to 0.78 s. The organic gas concentration was monitored downstream of the adsorption vessel with either a photoionization detector (PID; RAE Systems, Inc. PDM-10A) or flame ionization detector (FID; MSA/Baseline Inc. series 8800). The PID and FID were calibrated with the relevant compounds used to characterize the GRS. The desorption cycle of one vessel occurred concurrently with the adsorption cycle of the other vessel and consisted of six steps:

ethyl ketone (MEK) via an ACFC-ESA bench-scale system, including recycle of the desorption gas, and found that the system used 335 kJ/mol of liquid MEK recovered. This research builds on the work by Dombrowski et al.15 and Ramirez et al.18 and extends the performance analysis of ACFC-ESA to organic gases. This is in contrast to previous research, which was limited to higher boiling point organic vapors. The objectives of this research are as follows: 1. Demonstrate for the first time that ACFC-ESA can capture and recover organic gases over a wide range of boiling points (−26.3 to −0.5 °C) and relative pressures (8.4 × 10−5 to 3.4 × 10−3). 2. Determine the system capture efficiency and effective adsorption capacity for several organic gases. 3. Determine the energy requirements for specific system components and the overall system to capture, recover, and liquefy organic gases. 4. Determine the mass distribution of the organic vapors of the system. This is the first time the energy required to recover organic gases as liquids via ACFC-ESA has been published. Quantifying the performance of the bench-scale gas recovery system in terms of its ability to remove organic gases from the adsorption stream and the energy required to liquefy the recovered organic gases is a critical step in developing new technologies to allow manufacturing to occur in a more sustainable manner.

■

EXPERIMENTAL APPARATUS AND METHODS Adsorption/Desorption Cycling Apparatus and Methods. The GRS used to capture organic gases consists of a gas generation system, two ACFC adsorption vessels with electrothermal regeneration capability, a postdesorption pressure and temperature control system, and a data acquisition and control system (Figure 1). Each adsorption/desorption vessel provided 1.43 L of empty internal volume and held a vertical annular cartridge with 115 g B

dx.doi.org/10.1021/es4005703 | Environ. Sci. Technol. XXXX, XXX, XXX−XXX

Environmental Science & Technology

Article

1. Nitrogen flowing at 3.5 SLPM (gas velocity through inner cross-sectional area of the cartridge = 3.7 × 10−3 m/s) entered the top of the desorption vessel and then passed through the ACFC from the inside to the outside of the cartridge. This was done to purge oxygen from the vessel prior to heating. The flow rate was chosen to clear >2 column volumes in 1 min, which earlier tests showed resulted in a 90% reduction in oxygen concentration in the vessel. Duration of step 1 = 1 min. 2. The nitrogen flow rate was reduced to 0.5 SLPM (gas velocity through inner cross-sectional area of the cartridge = 5.3 × 10−4 m/s) to minimize the amount of carrier gas used during a desorption cycle. This change allows for concentrations of organic gas on the order of 50% to be generated during desorption. The ACFC began heating to a set point of 150 °C. The value of 150 °C was chosen on the basis of previous experiments that examined the energy used by the system at 100, 150, 175, and 200 °C, which found that an ACFC temperature of 150 °C regenerated the carbon for the lowest energy cost.19 ACFC heating required controlling the voltage applied to the cloth with a silicon controlled rectifier (SCR, Robicon, model 440 102.10). As the organic gas was desorbed from the ACFC in this step, it was recycled to the adsorbing vessel. Duration of step 2 = 1 min. 3. ACFC heating continued and the desorbing organic gas entered a compression module consisting of a compressor (Air Dimensions, Inc. R272-BT-EA1) capable of producing pressures up to 13.1 bar gauge (190 psig) and a custom glass pressure vessel (0.5 L for isobutane at a relative pressure of 6.7 × 10−4 and 0.1 L for n-butane, R134A, and higher relative pressures of isobutane). The pressure vessel was submerged in a polycarbonate vessel containing 16 L of a 70/30 glycol/ water mixture. The glycol/water mixture was chilled by a cryogenic cooler (FTS Systmes FC100 Immersion Cooler) which maintained the internal temperature of the pressure vessel at ≤0 °C. Pressure in the pressure vessel was controlled to 10.3 bar gauge (150 psig), with exhaust from the vessel being returned to the adsorbing vessel. This pressure was chosen because it allows the compressor to operate without overheating and it is consistent with earlier experiments.8 Total pressure was monitored with a pressure transducer (Dwyer, IS626-12GH-P1-E1-S1), and temperature was measured with a 0.16 cm (0.062 in) diameter type K thermocouple (Omega, Inc.). Duration of step 3 = 2−7 min. 4. The nitrogen flow rate was increased to 3.5 SLPM and the ACFC heating continued with the desorbed gas being recycled to the adsorbing vessel. This step cleared the ACFC of any remaining organic gas. Duration of step 4 = 0.5 min. 5. Heating was stopped and the vessel was exposed to 3.5 SLPM of nitrogen. This was done to clear any remaining gas-phase organic gas out of the desorbing vessel. Duration of step 1 = 2 min. 6. The nitrogen flow rate was reduced to 0.5 SLPM, and once the temperature of the ACFC reached 60 °C, nitrogen flow to that vessel was stopped. Reducing the temperature to 60 °C allowed the ACFC to have sufficient adsorption capacity for the next adsorption cycle to achieve steady-state conditions.

The desorption process time (steps 1−5) ranged from 6.5 to 11.5 min depending on the experiment. During step 3 when the compression module was used, the pressure in the desorption system was maintained at 1.05 bar absolute (15.2 psia) by controlling the flow through the compressor via a variable-area solenoid valve (AscoValve). This pressure maintenance was done to ensure that the desorbing vessel was slightly pressurized so that O2 was prevented from entering the system. The amount of liquid organic compound collected was determined by visually noting the liquid level in the pressure vessel. The GRS is fully automated.8,9 The control system consists of National Instruments Fieldpoint hardware connected to a personal computer with LabView 6.1 software. The control system controls heating cycles and timing of the cycles, and it records values for gas concentrations, ACFC and gas temperatures, pressures, and energy consumption. The LabView program maintains a user-defined ACFC temperature profile during desorption by controlling the power applied to the ACFC with a feedback controller. The program also determines when the ACFC that is undergoing an adsorption cycle is saturated, on the basis of user-defined parameters, and automatically switches that vessel to a desorption cycle while simultaneously beginning the reciprocal adsorption cycle. This ensures that the organic gas/air stream is treated continuously by adsorption. Adsorbates. Three adsorbates were tested as part of this research: n-butane, isobutane, and 1,1,1,2-tetrafluoroethane (R134A) (Table 1). These adsorbates were chosen due to Table 1. Properties of Adsorbates of Interest characteristic class mol formula mol mass (g/mol) boiling point (°C) liquid density at 0 °C20 (kg/m3) saturation vapor pressure at 20 °C20 (kPa)

nbutane

isobutane

1,1,1,2tetrafluoroethane (R134A)

alkane C4H10 58.12 −0.5 600

alkane i-C4H10 58.12 −11.7 600

haloalkane CH2FCF3 102.03 −26.3 1300

207

302

577

their range of low boiling points (−26.3 to −0.5 °C), wide range of densities and saturation vapor pressures, and relevance to the packaging manufacturing industry. We chose both nbutane and isobutane to isolate the effect of boiling point from other properties such as density. Process Performance Characterization. Benchmarks used to characterize the performance of the GRS during operation were gas capture efficiency, energy used per mole of recovered liquid, effective adsorption capacity, and mass distribution of adsorbate in the GRS. Capture Efficiency. Capture efficiency describes the amount of captured adsorbate relative to the amount fed into the vessel for a particular adsorption cycle. To determine the capture efficiency, we must first know the total mass adsorbed by the GRS. The mass of organic gas adsorbed at any time, ti, during adsorption is Mti (eq 1),13 where the subscript i represents the maximum ratio of outlet to inlet gas concentration. C

dx.doi.org/10.1021/es4005703 | Environ. Sci. Technol. XXXX, XXX, XXX−XXX

Environmental Science & Technology ⎡ Ptot(M w )Q ⎤ g ⎥ M ti = ⎢ RT ⎢⎣ ⎥⎦

t = ti

∫t =0

Article tcomp

⎛ C in Cout ⎞ − ⎜ ⎟ dt 1 − Cout ⎠ ⎝ 1 − C in

Ecomp =

where Icomp and Vcomp are the true RMS current (amperes) and voltage (volts) used by the compressor, respectively, and tcomp is the duration of the compression process in seconds. The energy required for cooling the desorption gas stream can be calculated by one of several methods. The first method is to consider the total power consumed by the cryogenic cooler that is used to chill the glycol/water mixture surrounding the pressure vessel (actual conditions for bench scale). The second method is to consider the total energy required to maintain the glycol/water mixture at a given temperature, including heat lost to the ambient atmosphere and the cooling requirement to cool and condense the nitrogen/adsorbate desorption gas stream (100% efficiency for cooling with 16 L glycol/water bath). The third method is to consider only the cooling requirement to chill and condense the nitrogen/ adsorbate desorption gas stream for an adiabatic device. However, analysis of the cooling energy determined by the third method showed that the cooling component is 2−3% of the total energy requirement (20 kJ/mol compared to 700− 1200 kJ/mol for heating.19 Additionally, the cooling energy used in the laboratory setup cannot be readily compared to what might be used in an industrial setting, therefore, the energy to cool the desorption gas stream will not be considered in detail in the rest of this work, and the total energy used by the system will be the sum of electrothermal and compression energy (eq 5). The energy per liquid mole recovered (Etotal) was determined by normalizing the total energy used by that experiment to the total moles of liquid recovered during that experiment (eq 6):

where variables represent the following (units need only be internally consistent): Mti = mass of adsorbate desorbed at time ti Ptot = 1 atm Mw = molecular weight of the adsorbate Qg = total volumetric flow rate of the gas R = ideal gas law constant T = absolute temperature t = time ti = time when the outlet adsorbate concentration is i% of the inlet adsorbate concentration Cin = concentration of the adsorbate in the inlet gas stream (mole fraction) Cout = concentration of the adsorbate in the outlet gas stream (mole fraction) Once the amount of organic gas adsorbed is determined, it is compared to the amount of organic gas that was supplied to the GRS (eq 2): M t ,ads

(

PtotM w RT

)Q C t

g in ads

× 100 (2)

where Mt,ads = mass adsorbed when ti = tads and tads = duration of adsorption cycle. The minimum acceptable capture efficiency is set by emission permits and will vary by process. As an example, the emission requirements for isobutane for the packaging manufacturing industry requires 98% destruction of organic gases in thermal oxidizers.21 Breakthrough time and mass adsorbed were determined when the organic gas concentration at the vessel’s outlet reached 5% of the inlet concentration, corresponding to t5 and Mt5, respectively. Energy per Liquid Mole Recovered. The energy used per mole of liquid recovered was determined by considering three contributions to the total energy used by the experimental system: electrothermal energy used during regeneration cycles, energy to compress the gas stream after desorption, and energy to cool the desorption gas stream. Eelectrothermal (in joules) includes the energy required to heat the ACFC, adsorbed adsorbate, carrier gas, and fittings during desorption as well as the energy to desorb the adsorbate (isosteric heat of adsorption) and is calculated from eq 3:9,13,15

Etotal = Eelectrothermal + Ecomp energy per liquid mol recovered =

∑ IRMSVRMS(ti − ti− 1) i=1

(5)

Etotal ρi Vi /M w

(6)

where ρi is the density of the liquid adsorbate at the temperature inside the pressure vessel and Vi is the total volume of liquid adsorbate captured. Effective Adsorption Capacity. To determine the effective adsorption capacity of the ACFC (qeff), the total mass of adsorbate captured during operation (eq 1) is divided by the number of adsorption cycles and the mass of the ACFC and the number of adsorption cycles at steady-state operating conditions to determine the effective adsorption capacity.22 The maximum theoretical adsorption capacity of the ACFC (qequil) is determined with adsorption isotherms.23 Mass Distribution of Adsorbate in the GRS. The mass distribution of adsorbate in the GRS was determined at the end of each cycling experiment. Three possible locations of the adsorbate were considered: liquid phase (Mliq), vapor phase (Mvap) in equilibrium with the liquid phase, and mass adsorbed to the ACFC (MACFC). The amount in the liquid phase was determined by measuring the volume of liquid recovered and multiplying it by the liquid density of the adsorbate. The amount in the vapor phase was determined by considering the volume available for the vapor phase above the liquid phase in the pressure vessel and assuming that the adsorbate concentration in the volume was equivalent to the saturation partial pressure divided by the total pressure in the pressure

tdesorp

Eelectrothermal =

(4)

i=1

(1)

capture efficiency =

∑ IcompVcomp(ti − ti−1)

(3)

where IRMS and VRMS are the true RMS current (amperes) and voltage (volts), respectively, and tdesorp is the duration of the regeneration cycle. The energy required to compress the gas stream is determined directly from the electricity consumed by the compressor. The current consumed by the compressor was continuously monitored with an AC current transducer (Omega OM9-31382AHD1) and Ecomp (joules) was calculated from eq 4: D

dx.doi.org/10.1021/es4005703 | Environ. Sci. Technol. XXXX, XXX, XXX−XXX

Environmental Science & Technology

Article

vessel. The amount in the adsorbed phase was determined by difference (eq 7)15 MACFC = M tot − Mliq − M vap

(7)

where Mtot = total mass supplied to the system during the experiment. Effect of Heating/Compression Time on Energy Requirements and Mass Distribution. Tests using varying heating and compression times were conducted on isobutane to determine what heating and compression times to use for all three adsorbates so that system performance could be compared across relative pressures. All of the experiments used 40 SLPM air and 2000 ppmv isobutane (Pi/Pi,s = 6.7 × 10−4) as the adsorbing gas, and the total heating and compression time (step 3) varied from 2 to 7 min, with a total heating time (steps 2−4) of 3.5−8.5 min. The adsorption and desorption cycling was allowed to continue for 5−8 h so that operational steady-state equilibrium conditions were reached. Each experiment was run in triplicate. We chose to examine the effect of heating and compression time on the process performance because we expect a need to balance the amount of heating and compression that is used to recover the adsorbate and the amount of adsorbate recovered. For example, supplying too little heat to the ACFC during regeneration could result in much of the adsorbate being left in the adsorbed phase, leading to a reduction in the amount of adsorbate available for liquefaction. This is contrasted by operating the heating/compression module (step 3, above) for too long, which has diminishing returns, as there is less and less adsorbed compound to desorb and compress. Effect of Adsorbate Relative Pressure. Once the duration of step 3 in the desorption process that resulted in the lowest energy per liquid mole recovered for isobutane (Pi/ Pi,s = 6.7 × 10−4) was determined, the GRS was tested with varying relative pressures of isobutane, n-butane, and R134A ranging from 8.3 × 10−5 to 3.4 × 10−3 (500−10 000 ppmv). The duration of step 3 for all of these experiments was 4 min (total heating time in steps 2−4 was 5.5 min). At a minimum, each experiment was conducted in duplicate and lasted for 5 h to provide >5 complete adsorption, regeneration, and cooling cycles to achieve steady-state conditions. The system was considered to be at steady state after the first two adsorption cycles were complete because the adsorption cycle times had a standard deviation 99% for all experiments and all adsorbates. This result suggests that the GRS is an appropriate device for meeting air-quality permitting requirements. Effect of Heating/Compression Time on Energy Requirements and Mass Distribution. The total energy to recover liquid isobutane was determined for several heating/ compression times by use of eqs 5 and 6. Figure 2 shows the combined heating and compression energy per mole of liquid isobutane recovered for select heating and compression times (▲). It also shows the relative contribution of heating and

Figure 3. Mass distribution of isobutane as a function of heating time averaged over the duration of each experiment (>5 regeneration cycles). E

dx.doi.org/10.1021/es4005703 | Environ. Sci. Technol. XXXX, XXX, XXX−XXX

Environmental Science & Technology

Article

mg/g ACFC. Since the total amount of energy input per cycle is essentially equivalent (338 ± 16 kJ/cycle) for all tests with the same heating/compression time, the less adsorbate captured per cycle, the higher the energy required to recover that adsorbate. Because the effective ACFC adsorption capacity is an important variable in the energy requirements of the GRS, we also compared the effective adsorption capacity to the theoretical equilibrium capacity of the ACFC. When the average ACFC capacity for each of the three adsorbates during cycling experiments (qeff) is compared to the equilibrium capacity determined by the Dubinin−Radushkevich (DR) fit of the adsorption isotherms (qequil) or data from previous work24 (Figure 5), it is clear that the effective adsorption capacity for

diminishing returns on the reduction of mass in the adsorbed phase and the increase of mass in the liquid phase with every added minute of heating time, suggesting that heating time (i.e., energy input) and liquid recovery must be balanced for the most efficient operation of the GRS. The fraction of mass in the liquid phase at the end of each experiment ranged from 47.5% to 87% of the total inlet mass. These values are acceptable when compared to data from previous work. Dombrowski et al.15 reported fractional liquid recoveries of higher boiling point compounds between 11% and 81% for an earlier prototype of an ACFC-ESA system that did not include a postdesorption pressure and temperature control system and did not recycle desorbed adsorbate back to the adsorbing vessel. This prototype was being used to capture and recover organic vapors with boiling points >40 °C. Ramirez et al.18 reported a fractional liquid recovery of 90.6% for MEK in a bench-scale ACFC-ESA system that did not include a postdesorption pressure and temperature control system but did include desorbed gas recycle. Using the results from the isobutane screening experiments, we chose to conduct the experiments at varying relative pressures with a heating and compression time (step 3) of 4 min (total heating time in steps 2−4 was 5.5 min). This heating and compression time resulted in the lowest energy per liquid mole recovered for experiments with isobutane at a relative pressure of 6.7 × 10−4 and also had reasonable mass distribution of isobutane during the experiment compared to what was found by Dombrowski et al.15 Effect of Adsorbate Relative Pressure. The effect of adsorbate relative pressure on the energy required to recover liquefied organic gas is shown in Figure 4. Standard deviations

Figure 5. Effective adsorption capacity during GRS cycling experiments compared to equilibrium adsorption capacity determined from isotherm experiments. Error bars represent one standard deviation above and below the mean. The linear fit equation was forced through the origin and is y = 0.3611x; R2 = 0.92.

each of the adsorbates is lower than the equilibrium capacity. This is likely due to three causes: 1. ACFC in the adsorption vessels is not being used to its full capacity due to the nature of the geometry and configuration of the ACFC cartridges. 2. ACFC is not allowed to come to equilibrium with the adsorbate because breakthrough is considered to be 5% of the inlet adsorbate concentration. 3. The qequil value does not take into account the high concentration of adsorbate that is recycled to the adsorption vessel during regeneration, which reduces the amount of adsorbent available for the lowconcentration gas stream. This result suggests that one cannot rely on equilibrium adsorption isotherms to design ACFC-ESA systems and it is critical to run bench and pilot studies with representative adsorbates prior to designing a full-scale system. However, the linear fit of the data in Figure 5 shows that a reasonable estimate of the effective adsorption capacity during operation of the bench-scale GRS would be 36% of the equilibrium adsorption capacity predicted by adsorption isotherms. This work represents the first time an ACFC-ESA system has been used to capture, recover, and liquefy organic compounds with vapor pressures as low as 8.3 × 10−5 and the first time such a system has been characterized in terms of energy use and mass distribution in the system. The results from this work will act as a point of comparison for future system modifications, including the introduction of water vapor, and is an important step in improving the sustainability of manufacturing processes.

Figure 4. Average energy per liquid mole recovered as a function of relative pressure for the GRS The data fit line represents a power law fit to all of the data: y = 3.39x−0.97. The R2 value for this fit is 0.89.

were