High Efficiency Production of Urea from Electrochemical Coreduction of Carbon Dioxide and Nitrite at Carbon Supported Iron(III) Tetrasulfophthalocyanine Under Ambient Conditions

Electrochemical coreduction of carbon dioxide with nitrite can potentially be used to remove two serious pollutants from the environment while producing urea as an essential fertilizer and valuable fuel. However, efficiencies are currently much too low due to the high overpotentials required and/or low faradaic efficiency (FE) for urea formation. Although metal phthalocyanine catalysts can provide reasonably high FE (≤42%), high negative potentials (≤−0.75 V vs RHE) are required. Here it is shown that a water soluble, sulfonated iron(III) phthalocyanine can produce a higher FE for urea (54%) at +0.053 V vs RHE, with 25% coproduction of ammonia. Urea production was measured by the diacetyl monoxime (DAM) method, and verified by liquid chromatography–mass spectrometry. Electrodes prepared with a carbon black support and Nafion binder exhibited good stability in the 0.1 M NaHCO3 electrolyte.

, 21 and its water solubility makes it more versatile than unsubstituted iron phthalocyanine for incorporation into electrodes and control of the catalytic environment. We demonstrate here that it is highly effective for coreduction of NO 2 − and CO 2 to produce urea. A maximum faradaic efficiency of 54% was obtained at +0.053 V vs RHE.
Preparation of carbon supported FeTSPc electrodes.-Carbon supported FeTSPc catalysts (3% FeTSPc/C and 10% FeTSPc/C) were prepared by depositing FeTSPc on Vulcan carbon black at mass loadings of 3% and 10%. FeTSPc in water (50 mg ml −1 ) was added to 10 mg of carbon black dispersed in a mixture of 150 μL of H 2 O, 75 μL of 2-propanol and 75 μL of 5% Nafion TM solution, and the resulting ink was sonicated for 1 h. The ink was deposited onto a 1.0 cm 2 disc of CFP with a micropipette to give a FeTSPc/C loading of 1.5 mg cm −2 , and Nafion loadings of 0.44 and 0.42 mg cm −2 for the 3% FeTSPc/C and 10% FeTSPc/C electrodes, respectively. Product analysis.-Ammonia in electrolysis solutions was measured by the salicylate method. 22 Samples (0.5 ml) were mixed with solutions prepared with 5 g of salicylic acid and 5 g of sodium citrate in 100 ml of 1 M NaOH (2 ml), 7 ml of sodium hypochlorite solution in 100 ml of water (1 ml), and 1 g of sodium nitroprusside in 100 ml of water (0.2 ml), and diluted to 10 ml with water. After 2 h in the dark at ambient temperature for 2 h, the absorbance of each solution was measured at 655 nm against a reagent blank, using an Agilent Cary 100 UV-vis spectrophotometer. The calibration plot is shown as Fig. S1 in the Supplementary Material. Urea was measured by the diacetyl monoxime (DAM) method. 23,24 In order to avoid interference from NO 2 − , 24 standards and samples were first deionized with a mixed bed ion exchange resin (AG ® 501-X8; Bio-Rad) using a conical 0.8 × 4 cm polypropylene column (Poly-Prep ® Chromatography Columns; Bio-Rad). 13 DAM (0.225 ml; 50.0 g L −1 ), TSC (0.0375 ml; 2.0 g L −1 ), Fe 2 (SO 4 ) 3 (0.0375 ml; 600 mg L −1 in 5% H 2 SO 4 ) and 50% H 2 SO 4 (2 ml) were added sequentially to 2 ml aliquots of the deionized solutions with dilution to a final volume of 5 ml. Following 30 min at 90°C, and cooling to ambient temperature, the absorbance of the resulting red compound was measured at 520 nm. The calibration curve is shown in Fig. S2.
To confirm accuracy, 24 urea concentrations were also determined by liquid chromatography-mass spectrometry (LC-MS) 25,26 using a SCIEX triple TOF TM 5600 mass spectrometer coupled with an Agilent 1100 Series HPLC system with a Poroshell 120 HILIC, 4.6 × 100 mm, 4 μm column (Agilent Technologies) with a KrudKatcher ULTRA HPLC in-line filter (2.0 μm, 0.004 in ID, Phenomenex). Deionized samples (as described for the DAM method) were diluted with acetonitrile in a 1:9 volume ratio and eluted using a mobile phase of acetonitrile containing 10% HPLCgrade water, 0.5 mM ammonium acetate and 0.2 mM acetic acid, at a flow rate of 0.5 ml min −1 . The injection volume was 10 μL and the column temperature was maintained at 30°C. A calibration curve ( Fig. S3) was prepared by plotting the integrated peak area of mass 61.03 m z −1 against the urea standard concentration.

Results and Discussion
Coreduction of NO 2 − and CO 2 at FeTSPc/C electrodes.- Figure 1 shows cyclic voltammograms of electrodes prepared with the 10% FeTSPc/C catalyst in 0.1 M NaHCO 3 with and without 5 mM NaNO 2 , under N 2 and CO 2 . Under N 2 in the absence of NaNO 2 , three main redox waves are observed at formal potentials of +0.92 V (ΔE p = 0.07 V) for the Fe(+3/+2) couple, and +0.06 V (ΔE p = 0.02 V) and −0.11 (ΔE p = 0.15 V) for the Fe(+2/+1)/TSPc (−2/−3) couples, which can not be unambiguously assigned. 27 The low peak separation seen for the Fe(+3/+2) couple is characteristic of adsorbed FeTSPc. 28,29 Addition of NaNO 2 (under N 2 ) did not influence the Fe(+3/+2) wave significantly, but resulted in an enhanced cathodic peak at ca. +0.0 V due to the reduction of NO 2 − . The proximity of this peak to the wave at +0.06 V strongly suggests that the process is catalyzed by the Fe(+1) state, although there is likely involvement of the TSPc (−3) state. 27 Under CO 2 , in the absence of NO 2 − , the waves for all three couples became indistinct and shifted to higher potentials, presumably due to the decrease in pH from 8.30 to 6.75 due to the formation of H 2 CO 3 . An irreversible oxidation peak appeared at +0.78 V on the reverse scan, indicating that an oxidizable species was formed at low potentials, although there was no enhancement of the cathodic current to suggest that there was significant reduction of CO 2 .
In the presence of both NO 2 − and CO 2 , a large increase in the cathodic wave was observed with an onset potential of ca. +0.08 V. The cathodic peak at −0.075 V was more than 2.5 times higher than the peak observed at 0 V for NO 2 − reduction. This indicates that there was substantial coreduction of NO 2 − and CO 2 in addition to their individual reduction processes. Shibata and Furuya have reported that, in addition to CO and NH 3 (Eqs. 1 and 2), there is a significant amount of urea produced from the coreduction of NO 2 − and CO 2 (Eq. 3) at metallophthalocyanine catalysts. 12 In order to measure NH 3 and urea production, electrolysis of 5 mM NaNO 2 under CO 2 was performed in 0.1 M NaHCO 3 at various constant potentials, using the 3% FeTSPc/C catalyst. Figure 2 shows the current vs time data. It can be seen that during electrolysis, the current was quite steady at potentials ⩾−0.147 V while it decayed over time at lower potentials. The highest average current of −0.55 mA was observed at −0.147 V, which corresponds approximately to the peak seen at −0.075 V in cyclic voltammetry.   Urea and ammonia production during electrolysis was measured using the DAM (Fig. S4) and salicylate (Fig. S5) methods, respectively. Additionally, LC-MS was used to verify the urea concentrations, since the DAM method can be unreliable. 24 The results are shown in Table I. Urea concentrations from the DAM and LC-MS methods were averaged for calculation of urea production rates and faradaic efficiencies.
The maximum amount of urea (50.7 μM) was produced at +0.053 V and there was a gradual decrease in urea production as the potential was decreased (increasing overpotential). NH 3 production initially increased as the potential was decreased, to a peak at −0.147 V, and then decreased. The maximum rate of urea production (0.381 μmol h −1 ) also occurred at +0.053 V, while the maximum rate for NH 3 (1.39 μmol h −1 ) occurred at −0.147 V. The maximum catalyst turnover number for urea formation was 20.5 (over 2 h) at a turnover frequency of 2.85 × 10 −3 s −1 . Figure 3 shows faradaic efficiencies for production of urea and NH 3 . The maximum faradaic efficiency for urea (53.6%) was observed at 0.053 V, and 79% of the charge passed at this potential produced urea or NH 3 (25.4%). The remaining charge would be expected to produce mainly CO. 20 As the potential was decreased, the yield of urea decreased and then increased to a peak at −0.347 V. In contrast, the faradaic efficiency for NH 3 increased to a peak of 59.5% at −0.247 V as the potential was decreased, and then decreased. The resulting decrease in the urea:NH 3 mole ratio (Table I), shows that the high currents seen in cyclic voltammetry at potentials below ca. 0 V were due to enhanced reduction of NO 2 − rather than coreduction of CO 2 to produce urea. This is presumably due to the lower pH when CO 2 is present.
Control experiments were performed to determine whether there were other sources of urea, such as contamination and electrochemistry at the carbon support materials (carbon black and carbon fiber paper). Table II shows the concentrations of urea measured at the open circuit potential (OCP), and in the absence of CO 2 or FeTSPc. No urea was detected following 2 h at OCP in 0.1 M NaHCO 3 containing 5 mM NaNO 2 under CO 2 , demonstrating that urea contamination was negligible. However, 1.0 μM urea was detected following electrolysis at 0.053 V in the absence of CO 2 (i.e.  Table I.    under N 2 ), indicating that coreduction of the NaHCO 3 electrolyte with NO 2 − produces small amounts of urea. No urea was detected at 0.053 V in the absence of the FeTSPc catalyst, indicating that the carbon support materials (CB and CFP) did not produce urea at this potential. However, significant urea production (2.5 μM) was observed at −0.447 V, corresponding to 8.9% of the amount measured in the presence of the FeTSPc catalyst. The ability of carbon to produce urea from NO 2 − and CO 2 has recently been demonstrated with carbon nanotubes. 30 Effect of FeTSPc loading and electrode stability.-The 10% FeTSPc/C catalyst was used to increase the urea formation rate and also used to evaluate stability. Figure 4 shows 30 potential cycles in 0.1 M NaHCO 3 under N 2 . The stability of the electrode was good, with only gradual changes in the large background current due to the carbon support. Notably, the wave for the Fe(+3/+2) couple at +0.92 V was stable, indicating that there was not electrochemically significant leaching of the FeTSPc. However, a slight blue coloration of the electrolyte showed that there was some loss of the catalyst. Figure 5 shows current vs time curves for two consecutive 2 h electrolyses of NO 2 − under CO 2 at 0.053 V, and Table III shows the average current, and concentrations, faradaic efficiencies, and production rates of urea and ammonia. Comparison with the data in Table I shows that increasing the FeTSPc loading from 3% to 10% increased the average current by 123% and also increased urea and ammonia production. It is worth noting that the rate of urea production increased by 120% (to 0.86 μmol h −1 ) without a significant decrease (∼1%) in the faradaic efficiency. The average current during the second electrolysis (i.e. for the same electrode with fresh electrolyte) was 6.3% lower than in the first, and there was a decrease in the yield of urea from 52.6% to 47.7%, suggesting that there was some instability of the electrode.
Cyclic voltammograms recorded before electrolysis and after 2 h and 4 h of electrolysis at 0.053 V (Fig. 6), show that the reduction current at potentials below 0.1 V increased notably after the first electrolysis, and that the activity of the electrode was still slightly higher than the initial activity after the second electrolysis. These results indicate that the 10% FeTSPc electrode was quite durable for an extended period of electrolysis. However, blue coloration of the electrolyte during the first electrolysis indicated that there was some loss of FeTSPc from the electrode. Quantification of the FeTSPc in the electrolyte by electronic absorption spectrometry after each electrolysis revealed that there was a 14% loss of FeTSPc from the electrode during the 1st electrolysis. However, no FeTSPc was detected in the electrolyte after the 2nd electrolysis, indicating that the remaining FeTSPc was strongly bound or adsorbed. This implies that the electrodes should be stable for longer term use, once loosely bound, excess FeTSPc has been lost.
Discussion.-The maximum faradaic yield of urea reported here, of 53.6% at 0.053 V vs RHE, is more than double the yield of ca. 25% at −0.75 V vs RHE reported by Shibata and Furuya for unsubstituted Fe phthalocyanine (FePc), 12 and was obtained at a much lower overpotential. It is also much higher than a yeild of 5.8% at −0.047 V vs RHE recently reported for FePc. 13 These differences presumably arise in part from the effects of the hydrophilic and electron withdrawing sulfonyl substituents on solubility and the electronic structure of the Fe center and Pc π system. 19 Shibata and Furuya demonstrated that the yields of urea obtained with MPc catalysts depend on their efficiencies for reduction to both CO 2 to CO, and NO 2 − to NH 3 . 12 Their proposal that urea was formed by coupling of asdorbed -CO and -NH 2 species (Langmuir−Hinshelwood mechanism) 31 has received wide acceptance in more recent work on other types of catalyst. 5,11,17 Phthalocyanine complexes are well known for their ability to catalyse the reduction of CO 2 to CO, 19 can reduce NO 2 − to NH 3 , 12 and have two free axial sites for binding -CO and −NH 2 intermediates. However, coupling of −CO and −NH 2 species would require separate PcM−CO and PcM−NH 2 species in a cofacial arrangement. The solubility of FeTSPc would increase the probability of this occuring, relative to FePc adsorbed on a carbon surface. Alternatively, one of the adsorbed −CO or −NH 2 species could form on the carbon black support. 30 The observation in Table II that urea was formed in the absence of FeTSPc indicates that both urea precursors can be formed on the carbon support. In addition, urea formation could occur at a single Fe site by reaction of a Fe−NH 2 species with CO 2 (Eley−Rideal mechanism). 31 Under the conditions employed here, the sulfonyl substituents shift the formal potential for the Fe(3+/2+) couple from +0.89 V for FePc (Fig. S6) to +0.92 V, while the second reduction wave shifts from 0 V to +0.06 V. Since the second reduction is responsible for urea production, the high activity and efficiency provided by FeTSPc can be attributed in part to the electron withdrawing effect of the sulfonyl substituents. In addition, the hydrophilicity of FeTSPc would make it more mobile within the catalyst layer than FePc, and increase the probability and rate for formation of adjacent −CO and −NH 2 species. However, this can also increase diffusion into the bulk solution and require methods for confinement or tethering of the FeTSPc catalyst.

Conclusions
Carbon supported FeTSPc is a highly active catalyst for NO 2 − and CO 2 coreduction to produce urea at high potentials (low overpotentials). The 53.6% faradaic yield of urea at 0.053 V vs RHE is comparable to the highest yield that has been reported (ca. 55% for a Cd catalyst at −1.0 V vs RHE 10 ), and the highest potential at which urea formation has been observed. Although some initial leaching of the water soluble FeTSPc catalyst was observed, high electrochemical activity and urea production was maintained, and longer term stability was good. However, further work on electrode design and composition is needed to improve the rate of urea production and ensure durability.

Acknowledgments
This project is funded in part by the Government of Canada/Ce projet est financé en partie par le gouvernement du Canada, and by Memorial University. Funding was provided by NRCan's Office of Energy Research and Development (OERD) through Carbon Capture Storage and Utilization (CCUS) program, NRC's Materials for Clean Fuels Challenge program. We thank Dr. Stefana Egli of Memorial University's Core Research Equipment and Instrument Training Network (CREAIT) for LC-MS analyses, and Nicholas Ryan (Department of Chemistry) for assistance with the spectrophotometric measurements.