DOI: 10.1002/cssc.

201100220

The Electrochemical Reduction of Carbon Dioxide to

Formate/Formic Acid: Engineering and Economic
Feasibility
Arun S. Agarwal, Yumei Zhai, Davion Hill, and Narasi Sridhar*[a]

The engineering and economic feasibility of large-scale electro- formic acid at a gas/solid/liquid interface, using a flow-through
chemical reduction of carbon dioxide to formate salts and reactor. The overall economics and energy consumption of the
formic acid is the focus of this Full Paper. In our study we in- process are evaluated through a value chain analysis. The sen-
vestigated the long-term performance of tin and other propri- sitivity of the net present value of the process to various pro-
etary catalysts in the reduction of carbon dioxide to formate/ cess parameters is examined.

Introduction

There are essentially four approaches for reducing emissions of Much greater reductions are possible through improvements
CO2 into the atmosphere: (1) improving of energy efficiency; and wider adoption of these technologies. Additionally, such
(2) using non-carbon (or lower-carbon) energy sources (e.g., technologies may provide a method for storage of energy
solar, wind, nuclear); (3) CO2 capture and geologic sequestra- from renewable sources. This paper discusses one such path-
tion (CCS); and (4) CO2 utilization. Improving energy efficiency way—the electrochemical reduction of CO2 to formic acid or
provides the greatest return on investment and is being vigo- formate salts. The electrochemical method has several advan-
rously pursued in most commercial and consumer applications. tages: (1) it is a room-temperature process so that no source
Renewable energy sources, such as solar and wind, are attrac- of heat is required; (2) if the supporting electrolytes are fully
tive but have temporal variations that require energy manage- recycled and the anode reactions can be performed using
ment. Nuclear energy provides a highly desirable constant waste water, the overall chemical consumption can be mini-
base load of electrical energy, but is attended by concerns mized to just water or waste water; (3) the source of electricity
about safety, security, and long-term waste disposal. Shale gas used to drive the process can be tailored not to generate any
and other unconventional natural gas sources are currently new CO2. Such sources include solar, wind, hydroelectric, geo-
touted as large-scale, lower-carbon fossil resources, but there thermal, tidal, and thermoelectric processes. Therefore, this
are continuing concerns about the environmental footprint as- method can also be used as a renewable electricity storage
sociated with their extraction. Geologic media, such as saline mechanism; and (4) the electrochemical reaction systems are
aquifers, can be used to sequester large quantities of CO2, but compact, modular, on-demand, and easy for scale-up applica-
geologic storage is not economical unless subsidized by gov- tion.
ernments or if a sufficiently large economic penalty exists for The electrochemical conversion of CO2 can be customized to
CO2 release. Some regions do not have the appropriate geo- provide a preferred product by selecting suitable electrocata-
logic medium for long-term CO2 storage. In addition, the long- lysts, electrolytes, and applied potential. CO2 can be converted
term environmental effects of geologic storage are not suffi- electrochemically to several products.[5–10]
ciently understood to engender confidence from local com- The specific energy consumption, E (expressed in MWh t1),
munities near storage sites. In the fourth option, rather than for any electrochemical process is given by Equation (1):
consider it as a waste product, CO2 is utilized for a variety of
purposes, including conversion to other useful chemicals. How- Vð1 þ aÞn
E¼ ð1Þ
ever, since CO2 is a stable molecule, energy is required to con- 1:641FE
vert it. Thus, the development of CO2 utilization processes in-
volves minimization of energy required for its conversion to a Herein, V is the cell voltage, a is a factor representing the
higher energy state and the utilization of energy sources that auxiliary energy required for the process (e.g., pumps, separa-
do not generate net additional CO2 in the process. tors), n is the number of electrons in the reaction, and FE is
CO2 can be utilized[1–3] as a storage medium for renewable the Faradaic efficiency. The specific energy consumption for a
energy, as a feedstock for various chemicals, and as a solvent
or working fluid. Together, various utilization technologies [a] Dr. A. S. Agarwal, Dr. Y. Zhai, Dr. D. Hill, Dr. N. Sridhar
Research & Innovation, Det Norske Veritas (USA), Inc.
have the potential, directly and through avoidance of fossil
5777, Frantz Road, Dublin, OH 43017 (USA)
fuel, to reduce at least 4 gigatons per year (Gt a1) of CO2 or Fax: (+ 1) 614 7611633
approximately 10 % of total annual CO2 emitted currently.[4] E-mail: narasi.sridhar@dnv.com

ChemSusChem 2011, 4, 1301 – 1310  2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 1301
 N. Sridhar et al.

duced at the cathode. Oxygen evolution or other oxidation re-

actions take place at the anode.
Typical requirements for commercial electrochemical pro-
cesses are continuous (or on-demand) operation, high product
selectivity and throughput, and long-term stable operation.
The CO2 conversion processes also require that the energy use
does not generate additional CO2. The major components for
technology assessment can be broadly divided into: (1) cata-
lyst development, (2) continuous processing in reactors, and
(3) value chain analysis for economic feasibility. The selectivity
of the catalyst, given by its Faradaic efficiency, determines in
part the specific energy consumption [Eq. (1)]. To reduce the
Figure 1. Specific energy consumption as a function of the Faradaic efficien- reactor size , it is also required to have high reactivity or prod-
cy for various products. Legend: Methane (-&-), ethylene (-^-), methanol uct generation rate (determined by the current density on the
(-&-), formic acid/formate (-*-), carbon monoxide (-~-). The thicker lines cathode). For constant selectivity and electrochemical polariza-
with data points indicate current experimental results achieved by various
studies.[9–21] tion behavior (dependent on the catalyst material and electro-
lyte conditions), the electrode reactivity can be significantly im-
proved by increasing the active area per superficial area of the
number of end products is shown in Figure 1. The lines indi- electrode. There are limited efforts in developing and operat-
cate theoretically calculated values whereas the data points ing near-commercial-size reactors to solve issues related to
are experimental values reported for these end products.[9–21] long-term stable continuous operation. Several literature stud-
For the data points, the total cell voltage is estimated based ies have focused on the mechanism of CO2 reduction and the
on our experimental data because many literature studies only effects of catalyst properties on selectivity.[4–10]
reported half-cell potentials. In this work, formate salt or The degradation of catalyst with time has been investigated
formic acid is chosen because it is possible to produce them over a relatively short time period by several investigators.[5, 6]
with relatively low energy and high Faradaic efficiencies. Although the economic feasibility of a commercial electro-
Figure 2 shows a basic configuration for continuous electro- chemical reactor producing formic acid was explored by Li and
chemical reduction of CO2. It consists of a cathode and an Oloman,[22–25] a more systematic modeling of the entire value
anode across which an electrical current or voltage is applied. chain of CO2 utilization has not been performed. Such a value
chain analysis is necessary to determine the sensitivity of the
entire process to operating parameters, feedstock, and product
price, chemicals, and energy source. In this work, a multi-scale
approach is adopted to address these issues, as depicted in
Figure 3.

Figure 2. Schematic representation of the electrochemical process to con-

vert carbon dioxide into formate/formic acid.

The two electrodes are placed in chambers separated by an

ion exchange membrane, which disallows bulk mixing of the
solutions flowing in each of the two chambers, while allowing
ions to move across and maintain electrical contact in the solu-
tion. A suitable electrolyte along with CO2 is introduced in the
cathode chamber, where it comes into contact with the cath-
ode and the dissolved CO2 in the solution is electrochemically
reduced to the products. This reaction is completed by the
complementary oxidation reaction occurring at the anode
chamber. Tin has been employed as a highly selective electro-
catalyst for the conversion of aqueous CO2 to formate salt/ Figure 3. Approach and assessment criteria for the electrochemical conver-
formic acid. Some by-products (H2 and CO) may also be pro- sion of carbon dioxide into formate/formic acid.

1302 www.chemsuschem.org  2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim ChemSusChem 2011, 4, 1301 – 1310
Electrochemical Reduction of Carbon Dioxide to Formate/Formic Acid

Results and Discussion

Cathode catalyst performance in half cells

The FE of the electrochemical CO2 reduction process is affected
by cathode catalyst composition as well as the degree of po-
larization, electrode structure, and supporting electrolyte.
Figure 4 illustrates a series of curves showing the short-time FE
of a variety of cathodic materials obtained in a GLS half cell.
Potentiostatic polarization at each potential was applied for
2 h. In addition to pure Sn, three modified catalysts (alloys of
Sn), were also tested in the same conditions for comparison.
For pure Sn, there exists an optimized potential window
(Eopt) within which the highest FE of formate production was
obtained for each cathodic catalyst. At a potential more posi-
tive than Eopt, the FE was lower due to insufficient degree of
polarization for CO2 reduction, perhaps due to insufficient ki- Figure 5. Long-term (54 h) test of two Sn-based cathodic catalysts in a GLS
half cell. Legend: pure Sn FE (-*-), pure Sn current (b), pure Sn pH
netics of adsorbed CO2 radical formation or its subsequent hy-
(–*–), Sn alloy 1 FE (-&-), Sn alloy 1 current (a), Sn alloy 1 pH (–&–).
drogenation. At a very negative potential, FE was also de-
creased due to an increase of proton reduction; a competitive
process to the formate production. In the case of pure Sn, a Figure 5 illustrates the longer-term (greater than 54 h) per-
narrow Eopt window existed around 1.9 V (vs. SCE). The Sn- formance of selected cathodic catalysts in a GLS half-cell. Be-
alloy catalysts in general provided wider ranges of Eopt than cause this was a semi-continuous operation, where the catho-
pure Sn with similarly high FEs. The wider range of Eopt of the lyte was batched into the apparatus and the CO2 was fed con-
Sn-alloy catalysts allows more reliable or flexible operation for tinuously, the FE, current (I), and pH of the catholyte were not
high cathode performance. In a commercial electrochemical constant with time due to the accumulation of the reaction
process, three-dimensional (3D) electrodes rather than flat products. The FE in Figure 5 shows that pure Sn and one of
sheet electrodes are often used to increase the current densi- the Sn-alloy catalysts exhibited high FE on a GLS electrode for
ties or the total reaction rate for a fixed electrode/reactor size. more than 20 h. However, the FE of both catalysts suddenly
For 3D electrodes, it is hard to implement a uniform polariza- dropped to about 20 % at a longer polarization time. Although
tion over the whole electrode area. Thus the more distributed both the currents and pH started to decrease with time
Eopt can help to achieve high overall FE in non-uniformly polar- around 20 h (peak FE times), it is unlikely that the slow change
ized 3D cathodes. In particular, the Sn-alloy catalysts gave of currents or pH account for the sharp FE decrease. On the
higher FEs at less polarized conditions (less negative applied other hand, FE of both catalysts gradually increased back to
potentials) than the pure Sn. This can lower the energy con- high values at even longer time of the electrolysis whereas the
sumption of the process by reducing full-cell voltage. pH continued to drop while currents started to increase.
High FE at high current densities at the cathode over ex-
tended times is desirable for any commercial electrochemical
reduction of CO2 process. Although the 54 h cited in this paper
represents a significant improvement over the data published
in the literature (4 h, maximum), much longer test times (2
years or more) are required for an industrially viable process.
Accelerated laboratory testing is required to ascertain longer-
term catalyst performance in the laboratory prior to commer-
cial implementation, necessitating a better understanding of
the degradation mechanisms. It has been noted in previous
studies by other researchers that black deposits on metal elec-
trodes, including pure Sn[22–24] and Cu[18] electrodes, may occur
after certain runtimes of electrochemical reduction of CO2. In
runtimes up to 20 h on pure Sn coil-form electrodes in a GLS
half-cell, no black deposits were found. However, different
amounts of black deposits (spots or a fully covered coating)
were generally found on the electrodes of pure Sn and the Sn-
alloy catalysts after more than 20 h. Energy-dispersive X-ray
(EDX) spectroscopy in a scanning electron microscopy instru-
Figure 4. Short-term (2 h) Faradaic efficiency of a pure Sn and various pro-
prietary cathodic catalysts in a GLS half cell. Legend: Sn alloy 1 (-*-), Sn ment and Raman microscopy analysis of the black deposits in-
alloy 2 (-~-), Sn alloy 3 (-^-), pure Sn (–&–). dicated they were likely a graphitic form of carbon. The carbon

ChemSusChem 2011, 4, 1301 – 1310  2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chemsuschem.org 1303
 N. Sridhar et al.

formation is believed to result from the further reduction of

formate on the catalysts surface if the formate product stays in
contact with the catalyst for too long. The GLS cathode with
gas-through feature can facilitate the removal of the formate
products from the catalysts by gas bubbling. This may explain
the longer high-FE time reported in this study, even in semi-
continuous operated half-cells.
The time-varying behavior of the FE with time shown in
Figure 5 appears to be typical for these kinds of catalysts, The
surface of the cathodes polarized for long time periods can
vary from very dark (with carbon deposits) to slightly gray, de-
pending on when the experiment was stopped; typically the
color was much lighter when the experiment was stopped at
Figure 6. Effect of cathodic potential on the superficial current density of
lower FE such as 30 h in Figure 5. We hypothesize that the au-
the Sn electrodeposited cathode. Legend: Current density for electrodepos-
tomatic recovery of FE after some time of ‘deactivation’ was ited Sn on CFP electrode (-*-), current density on pure Sn solid electrode
caused by the removal of the carbon layer at low FE where (a).
most of the current was used to generate large amount of H2
gas, which may physically remove the loosely coated carbon
layer. However, the natural removal process took several to The average current density was used to obtain the Ohmic po-
dozens of hours depending on catalyst composition. Therefore tential drop. The presence of gas bubbles, which would in-
it is believed the deep cathode polarization (DCP) pulse tech- crease the Ohmic drop, was not considered in this calculation.
nique can speed up the removal process by increasing the An Ag/AgCl reference electrode wire was placed close to the
cathodic potential for a short time, which changes the main re- membrane and the potential of the cathode was held at differ-
action from formate production to fast H2 production. Anodic ent constant values in the range noted above with respect to
pulse (AP) polarization can also remove the carbon deposit by this reference electrode. The actual cathode potential, noted in
removing a thin layer of the catalyst itself. Further research Figure 6, was obtained by subtracting the calculated IR drop
into the degradation mechanisms of pure Sn catalyst is being from the applied cathode potential. The other conditions were
conducted. maintained as delineated in the Experimental section.
Different polarization treatment methods, including deep The test results presented in Figure 6 were performed for
cathodic polarization (DCP), anodic polarization (AP), and com- short times (ca. 30 min under constant conditions). Hence,
binations thereof, were utilized to remove the black deposits knowing how the electrodes will behave under prolonged op-
on long-run-time coil-form cathodes. The polarization treat- eration is imperative. Although the rate of production of for-
ments fully removed the black deposits from the cathodes. Re- mate product is higher with the Sn-electrodeposited electro-
testing of the treated cathodes in GLS half cells showed full FE des, if substantial electrode degradation does occur over time,
recovery (back to above 80 %) by all the three types of polari- then their applicability would be significantly undermined. In
zation, among which the combination of AP and DCP con- Figure 7, the operation of these Sn-electrodeposited high-sur-
sumed minimum electricity for effective removal of black de- face-area electrodes under optimal conditions (cathode held at
posits. about 2.0 VSCE) is indicated for a period of 4 days. Catholyte
samples were evaluated periodically to obtain the formate
product concentration and the current and potential values.
Full flow cell Performance
The rest of the parameters were monitored during this period.
The impact of the high-surface-area tin electrodes on the cur-
rent density, and hence the overall reactivity of the electrode,
was tested as a function of cathode potential. As indicated in
Figure 6, there is an approximately five-fold increase of the su-
perficial current density (current divided by the superficial elec-
trode area, not considering the actual particle area) as com-
pared to pure tin solid cathode tested in half cells as the cath-
ode potential is increased to 2.2 VSCE. Although the current
increases beyond this cathodic potential, the selectivity (%FE)
for formate generation falls much below 60 %, indicating that
the optimum potential window for high FE (60–70 %, not
shown here) exists between 1.8 to 2.2 VSCE. The Ohmic re-
sistance or IR drop correction was applied based on the thick-
Figure 7. Performance of electrodeposited Sn cathode with respect to cur-
ness of the catholyte channel and the conductivity of the cath-
rent density and Faradaic efficiency as a function of time over four days
olyte. Based on the electrolyte conductivity (2 m KCl) and the under optimal processing conditions. Legend: Faradaic efficiency (-^-), cur-
channel thickness, the resistance in the gap was calculated. rent density (-&-).

1304 www.chemsuschem.org  2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim ChemSusChem 2011, 4, 1301 – 1310
Electrochemical Reduction of Carbon Dioxide to Formate/Formic Acid

As shown in Figure 7, the current density and the Faradaic effi-

ciency of the process remained fairly constant over this period.
This suggests that the Sn-electrodeposited electrodes can be
used in relatively long-term commercial operations. This opera-
tion time was found to be 20 times that reported in literature
for this system[22–25] under high-current conditions, where sig-
nificant loss of powdered tin catalyst was stated as the prime
cause for significant decrease in current and hence the produc-
tion of formate after 4 h.
Initial examination of used tin electrodeposited cathode
samples by SEM and EDS did not show significant physical de-
formation or loss of tin deposits, although it was difficult to
Figure 8. Possible CO2 mitigation pathways compared in the value chain
differentiate between the fresh and used samples due to the
analysis. The boundaries of the economic analyses include all the major
amorphous nature of the carbon fibre paper on which tin was steps shown to the right and below the dashed lines.
electrodeposited. Tests are currently underway to study possi-
ble electrode degradation under more extended times of oper-
ation. product (formic acid or formate salt solution); and (3) capture,
pipeline transportation, and geologic storage. Each step in
Figure 8 represents energy- and materials-related expenses or
Value chain analysis
revenues. The expenses can be both capital (CAPEX) and op-
Economic, energy, and net CO2 emission calculations can be erational (OPEX). Comparisons are made using net present
performed for a system using a variety of boundary conditions. value as the metric (net present value calculations, as opposed
Life cycle analysis (LCA) examines the net CO2 emitted by any to simple return-on-investment ones, reflect the time value of
product during its entire life cycle. A value chain analysis is a money). A payback period calculation is also made to identify
gate-to-gate analysis, which assumes a certain boundary for the return on capital invested.
the delivery of raw materials and finished product of a process. In the following case studies, the process is compared
The process analysis only considers the specific process. Most against a carbon capture and storage scenario for a 1 ton CO2
of these analyses are “deterministic”, that is, they consider per day (tpd) emission source with a storage site located be-
single-valued inputs and calculate a single number as output tween 100 and 300 km away. The cost of the CCS process is in
(e.g., energy per ton of product, net CO2, net present value). the construction of the pipeline to the storage site (ranging
Oloman and Li performed a simple return-on-investment analy- between $55–200 tpd1 km1), the drilling of the wells (averag-
sis without considering in detail the supporting systems.[25] The ing between $600 000 to $2 million per drilling pattern), requir-
analysis in the present paper incorporates supporting systems ing between 0.02–0.05 kWh t1 km1 of pumping and compres-
and infrastructure that are required to create a viable and prac- sion energy during transport.[26] There is an obvious advantage
tical CO2 conversion facility, including stages for capture and to CCS for higher tonnage per day that makes the drilling of
separation of CO2. When the sum of these investments is in- the wells disadvantageous for a 1 tpd scale, however, this anal-
cluded the capital costs are somewhat higher than the previ- ysis highlights that utilization and CCS may have different rele-
ous analysis.[25] In addition, Oloman and Li assumed a CO2 vance to different emission sources.
price up to $1000 t1, which is much higher than the present In these studies a CO2 conversion plant is rated by its daily
market value. The present analysis assumes that the price of capacity, such as 1 tpd or 10 tpd. Therefore the capital expen-
CO2 is zero, although other values may be considered. Finally, ditures (CAPEX) are related to the capacity and are in units of
the previous studies did not fully account for consumable $106 tpd1. Costs can be reduced when the production rates
chemicals in the reaction and only considered energy costs. As are considerably increased, as indicated by previous experience
shown here, chemical consumption and electrode lifetime are from large-scale electrochemical hydrogen production.[27] How-
important factors. The time value of money is also considered ever, in other commercial electrochemical processes, the bene-
in the present analysis. fits of scale-up are considered to be much smaller. Therefore,
The value chain analysis reported here is a probabilistic anal- only modest cost savings are assumed due to scale-up in the
ysis that recognizes the fact that considerable uncertainties present analysis. The studies to follow are for an example
exist in many of the input parameters in the analysis. It uses a 1 tpd plant and draw from industrial examples as well as the
Monte Carlo-based computational engine (@Risk, Palisade laboratory data presented here.
Corp.) that uses triangular probability distributions of the Two different process scenarios are analyzed in this paper:
inputs to the model. The output of the model is then a proba- Scenario 1: In this case, a consumable chemical (in the form
bility distribution based on the corresponding values of the of NaOH) is used on the anode side to reduce the total cell
inputs. In this paper, we compare the value chains for three voltage and, thus, energy consumption.
pathways (Figure 8): (1) direct emission of CO2, which may or Scenario 2: In this case, consumable chemicals are mini-
may not have a direct cost depending on the regulations, or mized by the use of waste water at the anode and other meth-
other market forces; (2) capture and conversion to a useful ods of chemical recycling, which results in greater cell voltage

ChemSusChem 2011, 4, 1301 – 1310  2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chemsuschem.org 1305
 N. Sridhar et al.

and energy consumption. Under this assumption, alkaline

Table 2. Operational expenditures for Scenario 1, using consumable chemi-
waste water would be available from an industrial source. For cals to reduce electrical energy.
economic analysis, the details of the waste water source or
chemistry are not analyzed, but an assumption is made that Parameter[a] Minimum Most likely Maximum
the use of alkaline waste water eliminates the need for pur- Conversion electricity consumption [kWh t 1]
4630 5210 6070
chasing NaOH. CAPEX conversion plant [106 $ (tpd)1] 0.60 0.70 1.40
Electrode cost [$ (tpd)1] 1968 2361 4722
Both scenarios share common inputs, shown in Table 1.
Product price [$ t1] 800 1200 1300
These general inputs describe the capital cost, operational H2O production [t tconverted1] 0.00 0.27 0.27
cost, and scalable input factors that are common to most NaOH consumption [t tconverted1] 1.82 1.82 1.82
value chain models for this process. In 2009, the average in- HCl consumption [t tconverted1] 1.66 1.66 1.66
NaCl production [t tconverted1] 0.00 2.66 2.66
dustrial electricity rate in the USA was $0.068 kWh1 including
O2 production [t tconverted1] 0 0.45 0.45
state and local taxes, demand charges, and all other end-use H2 production [t tconverted1] 0 0.01 0.01
costs.[28, 29] It is understood that the cost of electricity would HCOOH production [t tconverted1] 0.8 1 1.05
depend on the source, with solar power at present costing sig- [a] kWh t1: kWh of electricity per ton, t tconverted1: ton per ton of CO2 convert-
nificantly more. However, these costs are expected to go down ed.
significantly in the future. Materials costs were sourced from
chemical market reporting sources.[29] Some chemical plants
are expected to have a 10 year lifetime (with opportunities for
life extension), however in other industrial sectors such as
Table 3. Operational expenditures for Scenario 2, using waste water and reus-
power generation or petrochemical refining, facilities are ex- able chemicals with a higher electrical energy use.
pected to last 20–30 years. In this analysis, it is assumed that
the electrochemical facility is paired with a long-life industrial Input Minimum Most likely Maximum
1]
CO2 emitter and is using a fraction of its emissions to generate Conversion electricity consumption [kWh t 6170 6940 7810
products, and therefore its life (including life extensions) is CAPEX conversion plant [106 $ (tpd)1] 0.60 0.70 1.40
Electrode cost [$ (tpd)1] 1968 2361 4722
comparable to the host plant. The electrical energy inputs for Product price [$ t1] 800 1200 1300
this analysis are derived from our current laboratory reactor ex- H2O consumption [t tconverted1] 0.27 0.54 0.82
perience. NaOH consumption [t tconverted1] 0 0 0
The operating expenditures (OPEX) for the process under H2SO4 consumption [t tconverted1] 0.001 0.002 0.003
NaCl consumption [t tconverted1] 0.001 0.002 0.003
Scenario 1 are detailed in Table 2. The operational expenditures O2 production [t tconverted1] 0.45 0.45 0.45
under Scenario 2 are described in Table 3. The revenue stream H2 production [t tconverted1] 0.01 0.01 0.01
from the production of O2 is included in the cash flow analysis. HCOOH production [ t tconverted1] 1.05 1.05 1.05
For Scenario 1, using the inputs from Table 2, the mean net
present value of the investment in electrochemical conversion

of CO2 to formic acid is slightly less negative than an invest-

ment in carbon capture and storage in this case (Figure 9) that
Table 1. General inputs for Scenarios 1 and 2, not specific to the product
or electrochemical reaction.
is, the net cost of CO2 conversion process over a 25 year life of
a plant is less than that of a comparable CCS operation. The
Parameter[a] Minimum Most likely Maximum fact that both are negative indicates that neither process is
Emissions source output [tpd] 0.8 1 1.2 profitable by the 25th year of operation. The up-front capital in-
Electricity price [$ (kWh)1] 0.06 0.07 0.08 vestment (mean) is also less for electrochemical conversion,
CAPEX MEA plant [106 $ (tpd)1] 0.6 0.7 1.4 amounting to about $2.9 million vs. $4.7 million for the CCS
CAPEX diversion and 0.10 0.15 0.20
delivery equipment [106 $ (tpd)1] process (Figure 10).
CAPEX product purification 0.75 1.00 1.20
and storage [106 $ (tpd)1]
CO2 price [$ t1] 0 0 0
NaOH price [$ t1] 150 165 180
NaCl price [$ t1] 75 101 130
O2 price [$ t1] 50 95 140
HCl price [$ t1] 72 150 300
H2SO4 price [$ t1] 72 150 300
H2 price [$ t1] 2300 2700 5000
H2O price [$ t1] 1 2 3
Nominal discount factor 0.03 0.04 0.05
Financing interest rate 0.04 0.05 0.06
Electrode lifetime [year] 0.4 0.5 0.8
Plant lifetime [year] 25
Figure 9. The net present value of the CO2 conversion process is less nega-
[a] tpd: ton per day, $ (kWh)1: USD per kWh of electricity, $ t1: USD per
tive than the NPV of an investment in carbon capture and storage (Scenario
ton.
1).

1306 www.chemsuschem.org  2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim ChemSusChem 2011, 4, 1301 – 1310
Electrochemical Reduction of Carbon Dioxide to Formate/Formic Acid

Figure 10. The capital investment in the electrochemical conversion process

is less than carbon capture and storage (Scenario 1).

In Figure 11 the mean OPEX (on a per-ton basis) is roughly Figure 13. Gross revenue per avoided ton of CO2 emissions for electrochemi-
three times greater than for a carbon capture and storage cal conversion and CCS (Scenario 1).
system of comparable capacity , more than half of which is ac-

Figure 14. Energy consumed per ton of conversion (Scenario 1) with and
without an amine capture step compared to the energy required per ton
captured for CCS.
Figure 11. Total OPEX per ton for conversion (Scenario 1) and CCS.

The cash flow, and therefore net present value, of the invest-
counted for by chemical consumption (Figure 12), but there is ment is driven by a number of factors that can be captured by
a mean > $1000 per ton of revenue from the electrochemical performing a sensitivity analysis on the net present value. In
process (Figure 13). By taking the difference between the this case, the net present value is the sum of all discounted
mean values for OPEX and revenue, it can be seen that the cash flows up to the final year of operation minus the original
electrochemical process offers the potential for a net operating 20 % down-payment to finance the capital to build the plant.
profit per ton of CO2 converted. Each of the inputs indicated in the tables factors into CAPEX,
As expected, energy is a large contributor to the operational OPEX, and revenues (if any) based on the plant capacity (in
costs of a conversion process, as are consumable chemicals. tpd), materials and energy consumption rates (in $ tconverted or
The total process energy consumption is shown in Figure 14, kWh tconverted), and plant unit costs (in 106 $ tpd1). The three
relative to CCS. In addition, the energy consumption of the largest factors affecting the net present value of the invest-
process without a CO2 capture step (such as MEA) is included ment for Scenario 1 are shown in Figure 15. The capital ex-
to illustrate that the MEA capture step contributes only about pense (the estimated payment on the financing of that capital)
4–5 % more energy to the process. of the process is the strongest (negative) driver for the net

Figure 12. The majority of the OPEX is in the consumption of chemicals Figure 15. The three highest sensitivity factors for the investment in the
(Scenario 1). conversion process (Scenario 1).

ChemSusChem 2011, 4, 1301 – 1310  2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chemsuschem.org 1307
 N. Sridhar et al.

present value of the process, but most notable is the cost of

NaOH as a strong negative driver. Energy consumption (such
as kWh t1 of conversion or the electricity price) is therefore
less important than chemical consumption in this case, mean-
ing that the cost of consumable chemicals is the main factor in
OPEX hurting the financial performance. The sensitivity analysis
confirms that the chemicals cost in Figure 12 is a concern, as it
is 1.3 times more than the energy costs. For this scenario, the
sensitivity analysis therefore prioritizes a reduction in consuma-
ble chemicals in order to improve the net present value of the
investment.
Given the findings from Figure 15, the impetus to use the
process in Scenario 2 (Table 3) becomes clear, even though the
Figure 17. The sensitivity analysis for Scenario 2 illustrates that chemical
energy consumption will be higher (7838 kWh t1 vs. only
costs are no longer a major negative driver.
6450 kWh t1 for Scenario 1). In Scenario 2, the CAPEX for the
plant will remain the same, but the OPEX is actually reduced
from the former $1065 t1 to $573 t1, increasing net operating major role in the economic feasibility. Value added process ad-
profit to a mean of $582 t1. The net present value is now ap- ditions are possible but need to be investigated to enhance
proximately 50 % less negative than the CCS investment, and the viability of electrochemical conversion of CO2 to formic
the factors that affect the OPEX are shown in Figure 16. Since acid.

Utility of Formate/Formic Acid

Currently, the global demand for formic acid is low (< 0.5 
106 t a1).[30] This can be attributed to the fact that it is relatively
expensive as currently produced. The traditional uses of formic
acid have been in the leather tanning and animal feed mar-
kets.[30] However, as the volume of formic acid increases and its
price decreases, more applications will be found. Formic acid
can also be used to replace mineral acids such as HCl and
H2SO4 in steel pickling as it is biodegradable and would not re-
quire recycling. Some pharmaceutical as well as biofuel pro-
duction processes may also use formic acid in the future.
Formic acid represents a good storage media for hydrogen
gas (demand for H2 in 2004 was 50  106 t), as the former can
be easily decomposed to H2 and CO2.[31] At standard tempera-
Figure 16. The three dominant factors for Scenario 2.
ture and pressure, formic acid stores 580 times more H2 than
the same volume of hydrogen gas.[31] Also, storage and trans-
portation of pressurized hydrogen is more expensive than
chemicals are reduced at the expense of electrical energy, the formic acid. This indirect production of H2 through CO2 recy-
latter shows up as the dominant factor in the OPEX costs. The cling is attractive because typical steam methane reforming for
sensitivity analysis (Figure 17) illustrates that chemical costs are H2 uses natural gas and produces twice the amount of CO2.
now less of a concern, and the process has more potential to Formic acid can also be selectively converted to synthesis gas
be profitable as the product price is a stronger positive driver (CO+H2, in varying compositions), which is used to produce
ranking in the top three regression coefficients. These coeffi- several bulk chemicals such as methanol (38  106 t demand in
cients rank similarly for year 10 and year 25 in this scenario, 2006) and acetic acid (5  106 t in 2005). Thus, conversion of
though the process has a better chance of becoming profita- large amounts of CO2 to formic acid can be readily employed
ble beyond year 10. in production of a fraction of these bulk chemicals, a market
The value chain analysis demonstrates that the economic that has not been previously explored.
feasibility of the process is dependent on factors such as Another use of formic acid is in the production of various
chemical consumption, Faradaic efficiency, and voltage (and formate salts. Formate solutions are used principally in Europe
therefore energy consumption). In our analysis we have dem- as airport de- and anti-icing agents, whereas airports in the
onstrated that increased lifetime reduces electrode costs to a USA use a combination of formate salts and acetate solutions.
minor driver in the sensitivity analysis, but without sustained A typical airport uses as much as 2000 t of formate solution at
lifetime and low cost, electrode costs have a strong effect on 50 % concentration in a year. Formate has proven to be an ef-
the process. Also, despite positive cash flow per ton of CO2 fective, non-corrosive, environmentally friendly deicing salt,
converted, the financing of the total plant CAPEX plays a but is used in small amounts because it is expensive. Electro-

1308 www.chemsuschem.org  2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim ChemSusChem 2011, 4, 1301 – 1310
Electrochemical Reduction of Carbon Dioxide to Formate/Formic Acid

chemical reduction of CO2 yields a formate solution readily and

does not need further acidification to generate formic acid. It
can be produced on-site and can be an economical alternative
to conventional approaches. Formate salts are also used in oil-
well completion to prevent the hydrocarbons from entering
the annulus between production tubing and casing.

Conclusions
The utilization of CO2 requires detailed consideration of energy
and chemical use through a value-chain analysis. In this paper
we considered in detail an electrochemical process, ECFORM,
for the reduction of CO2 to formic acid or formate salt. This
work has accomplished the improvement of three factors relat-
ing to the economic success of the process: (1) a reduction of
electrical energy use, (2) a lengthening of catalyst lifetime, and
(3) the identification of reaction routes to reduce the consump- Figure 18. Schematic of the gas/liquid/solid (GLS) interface in a) a half-cell
and c) a full-cell (c). In the full-cell operation, Sn electrodeposited carbon
tion of chemicals. Faradaic efficiency, current density, and volt- fiber paper (CFP) cathode as in (b) was used.
age determine the energy and electrode costs for the large-
scale process. Chemical consumption is a major consideration
The superficial electrode surface area was 5 or 10 cm2. CO2 gas,
for large-scale implementation of CO2 conversion to formate or catholyte, and anolyte were fed to the three-chamber cell from the
formic acid, and there are pathways that minimize chemical bottom and withdrawn from the top of the reactor. The flow rates
consumption. If the CO2 price is ignored, the net present value of CO2 gas, liquid catholyte and anolyte in the cell were regulated.
of the electrochemical conversion of CO2 to formate or formic The pressures in the three chambers of the reactor were controlled
acid is negative over a 10 year lifespan. This is mainly dominat- using the back-pressure regulators. The CO2 gas chamber pressure
ed by the financing of the capital equipment for the process. was always maintained higher than the catholyte flow chamber
The process can be operationally profitable, however, even if pressure, to force the gas across the porous electrode to the cath-
olyte. The gas–liquid mixed outlet streams from catholyte and ano-
there is no price for carbon. The operational profitability of the
lyte compartments were separated in a tank where the gasses
process will vary for different end uses of formic acid or for- were allowed to escape in to the hood and the liquid spent elec-
mate salt, which may include hydrogen storage, syngas pro- trolytes were collected. The GLS cathode/catholyte interface in this
duction, chemical feedstock, deicing solutions, and oil and gas reactor setup facilitated the shortest liquid diffusion path between
drilling completion fluid markets. Further work will be needed the gas and solid catalysts as well as efficient removal of products
in the areas of reducing consumable costs, increasing catalyst and byproducts from the catalyst surface. A constant cell potential
life time, and improved reactor designs to enhance the overall was applied across the anode and cathode and the current flowing
attractiveness of the process. through the cell was recorded.
Figure 19 shows the schematic and the assembled views of the full
cell. The distance between the cathode and anode was about 7–
Experimental Section 8 mm. The flow rate of the catholyte, CO2 pre-saturated 2 m KCl so-
lution (pH 3.8), was 3 mL min1 and the flow rate of the anolyte,
A gas–liquid–solid (GLS) half-cell was used to evaluate the electro- 1 m KOH solution (pH 14), was 40 mL min1. The pure gaseous CO2
catalytic performance of different Sn-based catalysts in CO2 reduc- flow rate was 550 mL min1. The CO2 gas chamber pressure was
tion to formate salts. Figure 18 a illustrates the cathode configura- 1.4–4.8 kPa (0.2–0.7 psi) above the catholyte chamber. The anolyte
tion in such a half-cell. The coil-form cathode was placed on top of and catholyte chamber pressures were maintained to be almost
a glass filter which acted as a gas distributer. The other side of the equal (the pressure difference was 0–1.4 kPa). A Pt-coated Nb
cathode was exposed to the catholyte of 400 mL 2 m KCl. Pure CO2 mesh was employed as high-surface-area anode for all the tests in
gas was purged from the bottom of the cathode at a constant the full-cell reactor. An Ag/AgCl reference electrode wire was
flow rate of 75 mL min1 throughout the testing time while the placed close to the cathode to measure the cathode potential.
catholyte remained the same (semi-continuous operation). The su- The formate solution concentration was analyzed with a Dionex
perficial surface area of the cathode was about 5 cm2. The half-cell ion chromatograph (Dionex Corp., Sunnyvale, CA). Cathode elec-
studies were considered to be semi-continuous experiments be- trode samples were also analyzed for degradation over time by
cause the CO2 was purged continuously, but the electrolyte was in- using optical and scanning electron microscopy instruments.
troduced once during the experimental time period.
In an attempt to increase the current density or the reaction rate
per superficial cathode surface area, cathodes were made by elec- Acknowledgements
trodeposition of Sn onto carbon fiber papers (CFP) with thickness
about 375 mm. As shown in Figure 18 b, after the electrodeposition,
the fibers in the CFP were uniformly covered with nanoparticles of The authors acknowledge the support of Amnoyporn Kelley,
Sn. Continuous electrochemical reduction of CO2 on the Sn-depos- Edward Rode, Kyle Hilgefort, Vince Arnett in various aspects of
ited CFP cathodes was performed in a full electrochemical reactor this study. The work has been performed as a part of the strate-
with GLS cathode/catholyte interface as illustrated in Figure 18 c. gic research funding of Det Norske Veritas AS.

ChemSusChem 2011, 4, 1301 – 1310  2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chemsuschem.org 1309
 N. Sridhar et al.

[7] C. M. Snchez-Snchez, V. Montiel, D. A. Tryk, A. Aldaz, A. Fujishima,

Pure Appl. Chem. 2001, 73, 1917 – 1927.
[8] M. Jitaru, D. A. Lowy, M. Toma, B. C. Toma, L. Oniciu, J. Appl. Electrochem.
1997, 27, 875 – 889.
[9] M. Todoroki, K. Hara, A. Kudo, T. Sakata, J. Electroanal. Chem. 1995, 394,
199 – 203.
[10] T. Yamamoto, D. A. Tryk, A. Fujishima, H. Ohata, Electrochim. Acta 2002,
47, 3327 – 3334.
[11] Y. Hori, H. Ito, K. Okano, K. Nagasu, S. Sato, Electrochim. Acta 2003, 48,
2651 – 2657.
[12] M. N. Mahmood, D. Masheder, C. J. Harty, J. Appl. Electrochem. 1987, 17,
1159 – 1170.
[13] B. Innocent, D. Liaigre, D. Pasquier, F. Ropital, J. M. Leger, K. B. Kokoh, J.
Appl. Electrochem. 2009, 39, 227 – 232.
[14] K. Hara, T. Sakata, Bull. Chem. Soc. Jpn. 1997, 70, 571 – 576.
[15] K. Hara, A. Kudo, T. Sakata, J. Electroanal. Chem. 1995, 391, 141 – 147.
[16] D. P. Summers, S. Leach, K. W. Frese Jr, J. Electroanal. Chem. 1986, 205,
219 – 232.
[17] J. Qu, X. Zhang, Y. Wang, C. Xie, Electrochim. Acta 2005, 50, 3576 – 3580.
[18] K. Ogura, in Energy Technology Perspectives: Conservation, Carbon Diox-
ide Reduction and Production from Alternative Sources (Eds.: N. R. Neela-
meggham, R. G. Reddy, C. K. Belt, E. E. Vidal), Wiley, Hoboken 2009, pp.
25 – 38.
[19] K. Ogura, R. Oohara, Y. Kudo, J. Electrochem. Soc. 2005, 152, D213-D219.
[20] R. L. Cook, R. C. MacDuff, A. F. Sammells, J. Electrochem. Soc. 1990, 137,
607 – 608.
[21] K. Hara, T. Sakata, J. Electrochem. Soc. 1997, 144, 539 – 545.
[22] H. Li, C. Oloman, J. Appl. Electrochem. 2007, 37, 1107 – 1117.
[23] H. Li, C. Oloman, J. Appl. Electrochem. 2006, 36, 1105 – 1115.
Figure 19. a) Expanded view, and b) assembled reactor employed in the full- [24] H. Li, C. Oloman, J. Appl. Electrochem. 2005, 35, 955 – 965.
cell studies. EG: end gasket, EFG: end-frame gasket, MG: membrane gasket, [25] C. Oloman, H. Li, ChemSusChem 2008, 1, 385 – 391.
I/O: input and output ports, PTFE: polytetrafluoroethylene. [26] Evaluation of Fossil Fuel Power Plants with CO2 Recovery, US Department
of Energy, Washington 2002. Available from http://www.netl.doe.gov/
technologies/carbon_seq/refshelf/analysis/pubs/Evaluation_of > _Fossil_
Keywords: electrochemistry · hydrogen · carbon dioxide · Fuel_Power_Plants_with_CO2_Recovery.pdf (accessed August 2011).
[27] DOE H2A Analysis, US Department of Energy, Washington 2011. See
formic acid · renewable resources http://www.hydrogen.energy.gov/h2a_analysis.html (accessed August
2011).
[1] M. M. Halmann, M. Steinberg, Greenhouse Gas Carbon Dioxide Mitiga- [28] R. Williams, A. Bloom, US Patent 4160816, 1979.
tion: Science and Technology, CRC Press, Boca Raton 1998. [29] Electricity Prices for Industry, US Energy Information Administration,
[2] Carbon Dioxide as a Chemical Feedstock (Ed: M. Aresta), Wiley-VCH, Washington 2010. Available from http://205.254.135.24/emeu/
Weinheim 2010. international/elecprii.html (accessed August 2011).
[3] G. A. Olah, A. Goeppert, G. K. Surya Prakash, Beyond Oil and Gas: The [30] Dunia Formic Acid Survey, Dunia Frontier Consultants, Washington 2008.
Methanol Economy, 2nd Ed., Wiley-VCH, Weinheim 2009. Available from http://www.dfcinternational.com/files/DuniaFormicAcid-
[4] N. Sridhar, D. M. Hill, A. Agarwal, Y. Zhai, E. Hektor, Carbon dioxide Uti- Survey15June2008.pdf (accessed August 2011).
lization: Electrochemical Conversion of CO2—Opportunities and Challeng- [31] C. Fellay, N. Yan, P. J. Dyson, G. Laurenczy, Chem. Eur. J. 2009, 15, 3752 –
es, Det Norske Veritas, Hovik 2011. Available from http://www.dnv.com/ 3760.
resources/position_papers/co2utilisation.asp (accessed August 2011).
[5] M. Gattrell, N. Gupta, A. Co, J. Electroanal. Chem. 2006, 594, 1 – 19. Received: May 3, 2011
[6] R. P. S. Chaplin, A. A. Wragg, J. Appl. Electrochem. 2003, 33, 1107 – 1123. Revised: July 16, 2011

1310 www.chemsuschem.org  2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim ChemSusChem 2011, 4, 1301 – 1310