Hydrogen Production and Utilization in a Membrane Reactor

Thermochemical hydrogenation reactions are used in ~20% of all chemical synthesis¹. These reactions require large quantities of H₂ gas, which are usually derived from fossil fuels, temperatures between 150 °C and 600 °C, and pressures up to 200 atm². Electrochemical hydrogenation is an appealing way to bypass these requirements and to drive hydrogenation reactions using water and renewable electricity³. For conventional electrochemical hydrogenation, an unsaturated feedstock is dissolved in a protic electrolyte in an electrochemical cell. When a potential is applied to the cell, water oxidation occurs at the anode, while hydrogenation occurs at the cathode. In this reaction setup, both electrochemical water oxidation and chemical hydrogenation occur in the same reaction environment. The organic substrate is dissolved in a protic electrolyte to enable both electrochemical water splitting and hydrogenation of the feedstock. The proximity of these reactions can lead to byproduct formation and electrode fouling when the reactant is susceptible to nucleophilic attack or if the reactant concentration is too high (>0.25 M)⁴.

These challenges prompted our group to explore alternative ways to electrochemically drive hydrogenation reactions^5,6,7. This search resulted in the use of a Pd membrane, which is conventionally used in hydrogen gas separation⁸. We use it as an electrode for water electrolysis on the electrochemical reactor side. This novel application of a palladium membrane enables the physical separation of the site of electrochemical water oxidation from the site of chemical hydrogenation. The resulting reactor configuration has two compartments: 1) an electrochemical compartment for hydrogen production; and 2) a chemical compartment for hydrogenation (Figure 1). Protons are generated in the electrochemical compartment by applying a potential across the Pt anode and the Pd membrane, which also serves as the cathode. These protons then migrate to the Pd membrane, where they are reduced to surface-adsorbed hydrogen atoms. The electrochemical compartment can be subdivided to include an optional cation exchange membrane to facilitate this proton migration. The surface-adsorbed hydrogen atoms permeate through the interstitial octahedral sites of the Pd fcc lattice⁹ and emerge on the opposite face of the membrane in the hydrogenation compartment, where they react with the unsaturated bonds of a given feedstock to form hydrogenated products^{7,10,11,12,13,14,15,16}. The Pd in the membrane reactor, therefore, acts as (i) a hydrogen-selective membrane, (ii) a cathode, and (iii) a catalyst for hydrogenation.

Figure 1: Hydrogenation in a membrane reactor. Water oxidation at the anode produces protons, which are reduced on the palladium cathode. H permeates through the Pd membrane and reacts with propiophenone to form propylbenzene. Hydrogen evolution is a competing reaction that can occur on either side of the palladium membrane. For atmospheric mass-spectrometry, no chemical feedstock is used, necessitating H to leave the reactor in the form of H₂ gas in either the electrochemical or hydrogenation compartments. Please click here to view a larger version of this figure.

The membrane reactor is assembled by sandwiching a Pd membrane between the anode and cathode compartments of an electrochemical H-cell¹². Chemical-resistant O-rings are used to secure the membrane into place and ensure a leak-free seal. The electrochemical compartment of the membrane reactor contains a hydrogen-rich aqueous solution. In this study, we use 1 M H₂SO₄ and an anode that consists of a Pt wire enveloped in a 5 cm² piece of platinum mesh. The anode is submerged in the electrolyte solution through a hole in the top of the electrochemical compartment. The chemical hydrogenation compartment contains a solvent and hydrogenation feedstock^{7,10,11,12,16,17}. The hole at the top of the H-cell compartment is used for sampling. The experiments shown here use 0.01 M propiophenone in ethanol as the hydrogenation feed. However, the starting material (and concentration) can be varied to fit the experimental needs. For instance, a starting material that contains a long hydrocarbon chain and an alkyne functional group may be dissolved in pentane to improve solubility¹¹. The applied current for the reaction can be between 5 mA/cm² and 300 mA/cm². All reactions are carried out under ambient temperature and pressure.

Atmospheric mass spectrometry (atm-MS) is used to measure the percent of hydrogen in the electrochemical compartment that permeates to the hydrogenation compartment^11,12. This measurement is important to understand the energy inputs required for the membrane reactor, because it reveals the maximum possible hydrogen utilization (i.e., how much of the hydrogen being produced can actually be used for hydrogenation reactions). Hydrogen permeation through the Pd membrane is calculated by measuring the amount of H₂ that evolves from both the electrochemical and hydrogenation compartments^11,12. A permeation value of 100% means that all the hydrogen produced in the electrochemical compartment is transported through the Pd membrane to the hydrogenation compartment and then subsequently combines to form hydrogen gas. A permeation value of <100% means that hydrogen evolution occurs in the electrochemical compartment prior to permeating through the membrane. As H₂ is produced from either the electrochemical or hydrogenation compartment, it enters the instrument and is ionized to H₂⁺. The quadrupole selects fragments of m/z = +2, and the corresponding charge is measured by the detector. The plot obtained by this technique is the ionic charge over time. The ionic charge is measured for the hydrogenation compartment first, and when the signal stabilizes, the channels are changed to measure the electrochemical compartment. Hydrogen permeation is calculated by dividing the average ionic charge in the hydrogenation compartment by the total ionic charge measured in the reactor (Equation 1)^11,12. To calculate hydrogen permeation, H₂ from the hydrogenation and electrochemical compartments are measured separately using atm-MS.

(Eq. 1)

Gas chromatography mass spectrometry (GC-MS) is used to monitor the progress of the hydrogenation reaction^12,14,15,16. To collect data for the example, the hydrogenation compartment of the reactor is filled with 0.01 M propiophenone in ethanol. By applying a potential across the Pt anode and the Pd cathode, reactive hydrogen is supplied to the hydrogenation compartment. The reactive hydrogen atoms then hydrogenate the unsaturated feedstock, and the products are quantified using GC-MS, where the sample is fragmented and ionized. By analyzing the mass of these fragments, the composition of the hydrogenation solution can be determined, and reaction rates can be calculated^12,14,15,16.

1. Pd rolling

1. Clean the Pd wafer bar with a mixture of hexanes using a cotton cloth.
   CAUTION: Hexane is flammable, a health hazard, an irritant, and environmentally damaging. Work under proper ventilation (i.e., a snorkel or a fume hood).
2. Roll the Pd wafer using a manual roller until reaching a thickness of ≤150 µm, as determined by a digital micrometer.
3. Roll the Pd using an automatic roller to a thickness of 25 µm, as determined by a digital micrometer. Then, cut the resulting Pd into the desired dimensions (e.g., 3.5 cm x 3.5 cm).

2. Pd annealing

1. Load the rolled Pd foils into a muffle oven with an N₂ atmosphere.
2. Heat the Pd foils starting at 25 °C, and ramp the temperature to 850 °C at a rate of 60 °C/h. Hold the temperature at 850 °C for 1.5 h, and then cool the oven to room temperature at a rate of 60 °C/h.

3. Pd cleaning

1. Prepare a cleaning solution by combining 10 mL of nitric acid, 20 mL of 30% (v/v) hydrogen peroxide, and 10 mL of deionized water.
   CAUTION: Nitric acid is corrosive, an oxidant, and toxic. Hydrogen peroxide is corrosive, an oxidant, and harmful.
2. Submerge the annealed Pd foils in the cleaning solution until the vigorous bubbling subsides or the solution turns yellow (20-30 min).
3. Rinse the Pd foils twice with DI water and once with isopropyl alcohol, and then dry with air.

4. Reactor assembly (Figure 2, left to right)

1. Assemble the reactor by sandwiching the Pd membrane between two halves of an electrochemical H-cell.
2. Place a chemical-resistant gasket between the left-hand side of the cell and the Pd membrane.
3. Place an additional chemical-resistant gasket between the Pd membrane and the right- hand side of the electrochemical cell.
4. Seal the resulting cell configuration with a clip.

Figure 2: An image of the H-cell assembly. The electrochemical compartment contains 1 M H₂SO₄ electrolyte; this is where water oxidation occurs. The palladium membrane separates the two halves of the H-cell, and the gaskets provide a leak-free seal. The hydrogenation compartment contains 0.01 M propiophenone in ethanol (EtOH). Please click here to view a larger version of this figure.

5. Pd electrodeposition

1. Prepare an electroplating solution by dissolving PdCl₂ into 1 M HCl to reach a concentration of 15.9 mM.
   CAUTION: PdCl₂ is harmful and corrosive. HCl is corrosive and an irritant.
2. Assemble the reactor using a clean Pd foil from step 3.
3. Fill the electrochemical compartment of the reactor with 24 mL of the prepared electroplating solution, and leave the hydrogenation compartment empty.
4. Place a Pt mesh anode and an Ag/AgCl reference electrode into the solution in the electrochemical compartment.
5. Connect the electrodes to a potentiostat, and apply a potential of -0.2 V versus Ag/AgCl to the Pd foil until a charge of 15 C has passed.
6. Disassemble the reactor, rinse the resulting Pd membrane twice with deionized water and once with isopropyl alcohol, and then dry the membrane under a stream of air or N₂. The Pd membrane will now have a visible deposition of Pd black on the surface that was exposed to the electroplating solution.

6. Atm-MS reactor setup

1. Assemble the reactor as indicated in step 4. Fill the electrochemical compartment with 1 M H₂SO₄, and fill the hydrogenation compartment with ethanol. Do not add hydrogenation feedstock.
   CAUTION: H₂SO₄ is harmful and corrosive. Ethanol is flammable, harmful, and a health hazard.
2. Submerge a Pt counter electrode in the electrolyte. Connect the Pt counter electrode and the Pd membrane to a power supply using alligator clips. Connect the Pt counter electrode as the positive electrode and the Pd membrane as the negative electrode.
3. Apply a constant current of 25 mA.

7. Atm-MS instrument setup

1. Turn on the power switch on the back of the atm-MS unit, just below the power cord.
2. Press the pump button on the front to turn on the pump (it will light up blue when on). Then, turn on the bake jacket (green round switch, the light will come on).
3. Turn on the capillary channel to be used (red round switches next to the channels, the light will come on). Check that the channel used is turned on by feeling the tube being heated.
   NOTE: The green LED next to "vac ok" will come on within a few minutes of turning the pump on. To turn off the system after finishing the experiments, turn off all the switches that have been turned on.
4. Connect the hydrogenation cell outlet to the atm-MS capillaries. This connection must be airtight.

8. Atm-MS software setup

1. Click on the Service desktop shortcut. Go to Setup | SEM/Emission Control, and check the boxes for both SEM and emissions. Press OK. Close the service window.
2. Click on the Measure shortcut, and go to Sequence | Execute.
3. Fill in the following parameters: Measurements = 30, Purge time = 30 s. Press File Manager, and create a folder to save the output data in. These settings will take 30 measurements with a purge time of 30 s between each measurement set; this can be changed if needed.
4. The MID measurements file will then open. Select File Management, and open a program to measure the mass spectrometry signal for m/z = 2. This signal corresponds to the ionic current from H²⁺, the ionized form of hydrogen gas.
5. Press OK to start the program. Do not close the measurement window as this will stop the instrument running.
6. After the signal has stabilized (1-3 h), disconnect the atm-MS capillary from the hydrogenation compartment, and connect it to the electrochemical compartment.
7. Save the data, and terminate the experiment when the signal for the electrochemical compartment has stabilized (about 30 min).
8. Calculate the percentage of hydrogen permeation through the Pd membrane using Equation 1.

9. Electrochemical hydrogenation

1. Assemble the reactor according to step 4.
2. Fill the electrochemical compartment with 24 mL of 1 M H₂SO₄.
3. Insert a Pt counter electrode into the electrochemical compartment through the counter electrode opening. Connect the Pt counter electrode to the positive terminal of a power supply, and connect the Pd membrane to the negative terminal via Cu tape.
4. Apply a galvanostatic current of 25 mA (corresponds to 40 mA/cm²) across the cell for 15 min. The voltage will read between 3 V and 5 V.
5. After 15 min has passed, fill the chemical compartment with 24 mL of reaction solution (e.g., 0.01 M propiophenone in ethanol). Maintain the galvanostatic current during reactant addition.
   NOTE: Sample the initial reaction solution before it is added to the reactor. See step 9.6.
   CAUTION: Propiophenone is harmful.
6. Sample the chemical compartment periodically (e.g., every 15 min) by taking 30 µL of reaction solution from the chemical compartment using a micropipette and dissolving the sample in 1 mL of dichloromethane. Store the samples in GC-MS vials until the reaction is complete.
   ​CAUTION: Dichloromethane is harmful and a health hazard.

10. Gas chromatography-mass spectrometry

1. Load the sample vials into the autosampling tray.
2. Launch the GC-MS software by clicking on the green Masshunter icon.
3. Click on Sequence | Edit Sequence to open the sequence editing window. Fill in the desired sample names, vial (position in the autosampling tray), method path, method file, data path, and data file into the chart. Set the sample type to "sample" and the dilution to 1, and ensure that the data file matches the sample name.
4. Adjust the method by clicking on Method | Edit Entire Method.
   1. Ensure both Method information and Instrument acquisition are checked. Click on OK. Add Method comments (if desired).
   2. Ensure Data Acquisition and Data analysis are checked. Leave every other field blank. Click on OK.
   3. Ensure that the sample inlet is set to GC and the injection source is set to GC ALS. Check the Use MS box. Ensure that the inlet location is set to Front and the MS is connected to Front. Click on OK.
5. Under the Inlet tab, ensure the heater temperature is set to 250 °C. Set the pressure to 7.2 psi and the He flow to 23.1 mL/min.
6. Under the Oven tab, set the initial temperature to 50 °C, and hold for 1 min. Next, set the ramp rate to 25 °C/min and the temperature to 200 °C, and hold for 0 min. Click on OK.
7. Ensure none of the display signals are checked. Click on OK.
8. Set the solvent delay to 2.50 min. Click on OK.
9. Ensure that the selected monitors include the following: GC oven temperature, GC inlet F temperature, GC inlet F pressure, GC column 2 flow calc, MS EM volts, MS MS source, MS MS quad. Click on OK.
10. Save the method under the desired method name.
11. Start the sequence by clicking on Sequence | Start Sequence | Run Sequence.
12. When the sequence is complete, view the data by opening the Masshunter software and clicking on the file name that was programmed in step 10.3.
13. Identify the product peaks by clicking on Spectrum | Library search report to compare the acquired mass spectra to the NIST database.
14. Calculate the relative composition of starting materials and products by using Equation 2.
      (Eq. 2)
    where A is the chemical component of interest, and n is the number of components measured by GC-MS. An example is as follows:
    

Atm-MS is used to measure the ionic current of the hydrogen that is produced in the membrane reactor. We can use these measurements to quantify how much hydrogen permeates through the Pd membrane during electrolysis. First, the hydrogen evolving from the hydrogenation compartment is measured (Figure 3, left of the dotted lines). When the signal reaches a steady state, the channel is switched to the electrochemical compartment. The H₂ gas evolving from the electrochemical compartment is then measured until the signal is stable (Figure 3, right-hand side of the dotted line). The total hydrogen permeation is calculated from the average current on the hydrogenation side divided by the total average current (electrochemical + hydrogenation compartments, Equation 1). Figure 3A shows 73% hydrogen permeation. An average ionic current of 27 pA was measured in the hydrogenation compartment, while the electrochemical compartment had an average current of 10 pA. In contrast, Figure 3B shows a membrane that is very poor at permeating hydrogen; its hydrogen permeability was less than 1%.

Figure 3: Representative data of atmospheric-mass spectrometry. The plots show the ionic current corresponding to m/z = 2 versus the cycle number, where one cycle is 5 s. For both datasets, the left side of the plot (as shown by a left arrow from a dotted line) represents the hydrogen current evolving from the electrochemical compartment. The right-hand side of the plots (as shown by a right arrow from a dotted line) represents the hydrogen signal coming from the hydrogenation compartment. (A) Hydrogen permeation data for a Pd membrane where ~70% of the hydrogen evolves to the hydrogenation compartment. (B) Hydrogen permeation data for a membrane where <1% of hydrogen evolves to the hydrogenation compartment. Please click here to view a larger version of this figure.

GC-MS enables the identification and quantification of the species present during hydrogenation reactions. We show two examples of representative results for hydrogenation experiments in Figure 4. Figure 4A,C,E represents a scenario in which the Pd membrane is under an electrochemical bias (and, therefore, acts as a cathode). Figure 4B,D,F represents a scenario in which the Pd membrane is not under an electrochemical bias, and a separate Pd cathode (of the same surface area) completes the electrochemical circuit. In the first example (Figure 4C), a sharp peak at retention time (RT) 5.6 min was observed. This peak corresponds to the starting material, propiophenone (PP). As the reaction progressed, peaks at RT 5.5 min and RT 4.2 min began to form, while the PP peak diminished. These forming peaks represent 1-phenyl-1-propanol (PA) and propylbenzene (PB), respectively. We could achieve 100% selectivity toward PB in this example. In the second example (Figure 4D), the PP peak did not diminish over time, and no product peaks emerged. Moreover, this chromatogram displayed an unexpected peak at RT 2.9 min, which was attributed to an impurity.

Figure 4: Representative results for GC-MS. These results show the hydrogenation of propiophenone (PP) to 1-phenyl-1-propanol (PA) to propylbenzene (PB). Legend: PP is orange, PA is grey, and PB is blue. (A) Schematic of a hydrogenation experiment where electrochemical bias was applied to the Pd membrane.(B) Schematic of a hydrogenation experimental setup where there was no electrochemical bias on the Pd membrane. (C) GC-MS results for a 4 h hydrogenation experiment where electrochemical bias was applied to the Pd membrane. (D) GC-MS results for a 4 h hydrogenation experiment where electrochemical bias was not applied to the Pd membrane. (E) Kinetic profile of the hydrogenation solution over time during a hydrogenation reaction where electrochemical bias was applied to the Pd membrane. (F) Kinetic profile of the hydrogenation solution over time during a hydrogenation reaction where electrochemical bias was not applied to the Pd membrane. Please click here to view a larger version of this figure.

The Pd membrane enables hydrogen permeation and chemical hydrogenation. The preparation of this membrane is, therefore, important to the efficacy of the membrane reactor. The Pd membrane size, crystallography, and surface are tuned to improve the experimental results. Although Pd metal can evolve hydrogen at any thickness, the Pd membranes are rolled to 25 µm. This standardization of membrane thickness ensures that the time it takes for hydrogen to permeate through the membrane is constant for all the experiments. Moreover, the thinner the membrane, the more susceptible it is to pinhole formation. The Pd membranes are cut to be 30%-40% larger than the dimensions of the gasket used to hold it in place. Over multiple uses, the Pd membrane succumbs to hydrogen embrittlement by pinholing or tearing. This phenomenon makes the membrane crinkle and causes the membrane geometric surface area to shrink. The Pd used for the membranes must be annealed once it is prepared to the desired dimensions. This step aims to reduce crystalline imperfections, resulting in better hydrogen transport through the membrane. The surface of the Pd membrane is tuned to enhance the catalytic activity. The electrodeposition of Pd black enables a 250-fold increase in the catalytic surface area¹¹. This increase in the surface area enables hydrogenation reactions to occur in the span of hours rather than days. Once a Pd membrane has been used for five hydrogenation reactions, it must be re-cleaned and re-electrodeposited. This cycle can be repeated until the membrane shows signs of irreversible damage (pinholes, cracks, or low/no hydrogenation activity).

Several issues may arise if the membrane reactor is not assembled with proper care. The problems that emerge are leaky cells and punctured Pd membranes. To prevent leaking and damage to the membrane, the Pd membrane is sandwiched between two gaskets. When the cell is tightened, the gaskets are gently compressed against either side of the Pd membrane so that no fluid can escape around the edges. The gaskets also prevent direct contact of the cell walls with the Pd membrane and reduce the potential for physical damage.

The success of a hydrogenation reaction in a membrane reactor depends on numerous factors. The two primary considerations are the availability of hydrogen and its reactivity. The availability of hydrogen is dependent on, but not limited to, the current density, hydrogenation catalysts, and solvents. The current density is directly related to the number of protons reduced on the electrochemical side of the Pd membrane, where higher current densities result in more hydrogen production. However, this does not always correlate to the amount of hydrogen emerging in the hydrogenation compartment. H₂ evolution is a competing reaction that can occur on both the electrochemical and chemical hydrogenation sides of the Pd membrane. This reaction occurs when two hydrogen atoms recombine and desorb from the Pd membrane surface as H₂ gas. To minimize the amount of hydrogen loss via this process, appropriate catalysts and solvents must be used to maximize the hydrogen availability in the hydrogenation compartment. Pd black is a catalyst that is electrodeposited onto the hydrogenation side of the Pd foil, and it increases the hydrogenation rates. The Pd electrodeposition looks black, and this indicates a high surface area. If the Pd deposition is grey in color, it means that the catalyst was poorly deposited, which can be the cause of slow hydrogenation rates or poor hydrogenation product yields. To select an appropriate and efficient solvent, the solvent must be able to coordinate with the reactive hydrogen atom to prevent H-H recombination, but it cannot coordinate excessively such that the solvent prevents the unsaturated molecules from reaching the reactive hydrogen. To troubleshoot slow hydrogenation rates, one should try increasing the current density, verifying successful Pd black deposition, and/or using a different solvent.

Atm-MS enables the quantification of gasses at very low detection limits. The mass spectrometer ionizes gaseous samples and then separates and quantifies the fragments using a quadrupole. Hydrogen gas is quantified by selectively plotting an m/z ratio of 2. This ratio represents the H₂⁺ fragment, where the mass is 2 amu and the charge is +1. Therefore, atm-MS measures the ionic current corresponding to the ionized hydrogen. Each side of the membrane reactor must have only one opening that can be connected to the atm-MS channel. It is important that there are no gas leaks to ensure all the gas formed is measured by the instrument. To ensure that all the electrochemically formed H₂ is measured, it is critical that during reactor assembly, the hydrogenation compartment is filled with the selected saturated solvent only, as well as that there is no unsaturated substrate present. If the hydrogen reacts with a substrate in the hydrogenation compartment, then that hydrogen will not be measured by atm-MS, and the permeation will be underestimated. When the membrane reactor is first connected to the atm-MS and potential is applied to the membrane reactor, it takes several hours for the H₂⁺ ionic current to stabilize. It is best practice to connect the hydrogenation compartment to the atm-MS first and then to change the connection to the electrochemical compartment. This is because it takes longer for the H₂ gas evolving in the hydrogenation compartment to equilibrate than the H₂ gas from the electrochemical compartment.

The membrane reactor was designed to carry out hydrogenation reactions under ambient conditions using only electricity and water. The progress of these hydrogenation reactions can be monitored using GC-MS. Figure 4 depicts the resulting chromatograms and kinetic profiles for the hydrogenation of propiophenone in a membrane reactor under an electrochemical bias (Figure 4A,C,E) and without an electrochemical bias (Figure 4B,D,F). When the Pd membrane is under an electrochemical bias (Figure 4A), hydrogen atoms are reduced in the electrochemical compartment and permeate through the Pd membrane^7,10,12. The hydrogen emerges into the hydrogenation chamber with an effective pressure that is proportional to the applied potential¹⁸. Unsaturated bonds in the hydrogenation compartment then react with this hydrogen to form saturated products. The progress of the reaction can be monitored by taking periodic samples and analyzing them using GC-MS. A typical chromatogram (Figure 4C) will have only one peak that corresponds to the selected starting material at T = 0. As the reaction progresses, the starting material peak will lose intensity, while peaks corresponding to the hydrogenated products will form and increase in intensity. The peak intensities at different time points can then be turned into a kinetic plot (Figure 4E) by calculating the relative compositions using Equation 2. To assign the product peaks, two methods can be employed: 1) database search and match; and/or 2) comparison to a standard solution. The first method involves the comparison of the measured m/z ratios (measured by MS) to a database of standard mass spectra (e.g., the National Institute of Standards and Technology) to find the best match. This method is usually built-in in GC-MS software and can be executed automatically. The second method involves running standard solutions of all the expected hydrogenation products and recording the retention times of each one. Ideally, both methods are employed to confirm experimental results. If an unexpected peak arises in the GC chromatogram (Figure 4D), it is likely due to contamination or byproduct formation. When the Pd membrane is not under an electrochemical bias (Figure 4B), there is no hydrogenation reaction. The resulting chromatogram does not display product peaks over time (Figure 4D). A similar result may arise if the catalytic properties of the Pd membrane are not tuned for a particular hydrogenation chemistry, or if the applied potential to the Pd membrane is too low. To troubleshoot this situation, different secondary catalysts can be deposited on top of the electrodeposited Pd layer¹⁹, or a greater potential can be applied to the Pd membrane.