Comparison of CO 2 Photoreduction Systems : A Review

Carbon dioxide (CO2) emissions are a major contributor to the climate change equation, and thus strategies need to be developed in order to reduce increases in CO2 levels in the atmosphere. One of the most promising approaches is to convert CO2 into useful products in engineered processes. The photocatalytic reduction of CO2 into hydrocarbon fuels is a promising way to recycle CO2 as a fuel feedstock by taking advantage of the readily available solar energy. This article reviews the basics of CO2 photoreduction mechanisms, limiting steps, possible strategies to enhance photoreduction efficiency, and the state-of-the-art photocatalytic systems for CO2 reduction. In particular, a comparison between different catalytic systems, including biological (plants and algae), inorganics (semiconductors), organics (molecular complexes), and hybrid (enzyme/semiconductors) systems is provided.


INTRODUCTION
Fossil fuels, such as coal, petroleum, and natural gas, are the major conventional energy sources in the world due to their availability, stability, and high energy density (Roy et al., 2010;Biswas et al., 2011).The continuous reliance on burning fossil fuels has raised atmospheric carbon dioxide (CO 2 ) levels by approximately 100 parts per million (ppm) by volume over the last century (Metz et al., 2005).CO 2 emissions are widely considered as the major cause of global warming.Thus there is an increasing need to mitigate CO 2 emissions.In June 2013, the U.S. President Barack Obama announced a plan to address climate change brought upon by CO 2 emissions.A proposal to regulate anthropogenic CO 2 emissions from existing and future electricity generating power plants as a part of a greater initiative to reduce and displace CO 2 emissions globally was suggested.Several strategies to accomplish this are being considered.Among them, carbon capture and sequestration (CCS) technologies are generally considered as an efficient and viable method to mitigate CO 2 emissions.CO 2 mitigation by these CCS technologies can be realized by pre-, post-, and oxy-fuel combustion technologies, which can be followed by compression and geological sequestration.However, these technologies require high energy inputs and are generally high cost, making it economically disincentive for widespread use at its current stage of development.
As fossil fuels will continue to be used globally in the near future, it is imperative to consider efficient CO 2 mitigation methodologies.While CCS sequestration in geological formations will be essential due to the necessary large scale of CO 2 emission reductions; it is also important to evaluate methodologies for conversion of the captured CO 2 to useful products and promoting use of carbon neutral methodologies.Carbon dioxide reduction can be realized by several different ways, such as biological reduction by plants (Blankenship et al., 2011), and thermal (Chueh and Haile, 2009), electrochemical (Abe et al., 1996;Jitaru et al., 1997), or photocatalytic reduction (Inoue et al., 1979) using synthetic systems.Among these options, CO 2 photoreduction is gaining increasing attention since it can potentially consume alternative forms of energy by harnessing solar energy, which is abundant, cheap, and ecologically clean and safe.In a typical semiconductor-based photocatalyst system, the catalyst needs to absorb light energy, generate electron-hole pairs, spatially separate them and transfer them to redox active species across the interface (Fujishima et al., 2008).In spite of its economic and environmental benefits, the photocatalytic pathways, however, are very complex and always suffer from low efficiency due to several limiting factors, such as fast electron-hole (e --h + ) recombination rates, complicated backward reactions, narrow light absorption wavelength range (in particular to wide bandgap semiconductors such as titania (TiO 2 ) and zinc oxide (ZnO)), and low CO 2 affinity of the photocatalysts.
There are several reviews on CO 2 photoreduction (Sutin et al., 1997;Song, 2006;Usubharatana et al., 2006;Kitano et al., 2007;Indrakanti et al., 2009;Morris et al., 2009;Dorner et al., 2010;Roy et al., 2010;Biswas et al., 2011;Windle and Perutz, 2012;Hong et al., 2013), covering both fundamental CO 2 photoreduction mechanisms and practical photocatalyst development.This article intends to provide a state-of-the-art overview of the basics of CO 2 photoreduction pathways, and a comparison of different CO 2 photoreduction systems as listed below: (1) Biological systems, including mainly algae; (2) Inorganic photocatalysts, mostly transition metal oxides (or semiconductors), in particular TiO 2 -based catalysts; (3) Organic photocatalysts, including mainly metalorganic complexes; and (4) inorganic and organic/biological hybrid, or the so-called biomimetic systems, consisting of enzyme-activated or dye-sensitized semiconductors.A summary and outlook of CO 2 photoreduction have been made for future development of efficient photocatalytic systems.

Typical Mechanisms
The mechanism of CO 2 photoreduction occurs through a set of complicated steps.Taking semiconductor photocatalysts as an example, to reduce CO 2 , the semiconductor should be activated by light with the energy (E hv ) equal to or higher than its bandgap (E g ) (e.g., E hv ≥ 3.2 eV for anatase TiO 2 ).The bandgap excitation (photoexcitation) leads to the formation of electrons in conduction band (CB) and holes in valence band (VB), which serve as the sites for photoreduction and photooxidation, respectively.The catalyst is also required to have its electrons with higher energy compared to the corresponding CO 2 reduction potential.In an ideal system, this means that the CB potential (or the so-called band edge) of the catalyst should be well above (more negative than) the reduction potential for a certain product at a predetermined pH.Meanwhile, the holes should be able to oxidize water to O 2 and produce protons (H + ), i.e., the VB edge should be well below (more positive than) the water oxidation level for an efficient photolysis.Fig. 1 shows the energy levels of several semiconductors and the corresponding redox potentials in the CO 2 photoreduction system (Inoue et al., 1979).
The possible reactions that can occur in the reduction of CO 2 in aqueous medium are shown in Scheme 1 (Inoue et al., 1979;Sutin et al., 1997;Fujita 1999;Usubharatana et al., 2006;Indrakanti et al., 2009).It is apparent from Scheme 1 that CO 2 photoreduction involves multiple steps.As explained above, the initial step in the photocatalytic reduction of CO 2 is the generation of e --h + pairs (Eq.( 1)).After photoexcitation, the e --h + pairs should be separated spatially, and be transferred to redox active species across the interface and minimize electron hole recombination.Fig. 2 shows the possible charge transfer pathways of the TiO 2 photocatalyst (Fujishima et al., 2008), which involves several critical charge transfer steps.Lifetime of the e --h + pairs is only a few nanoseconds, but this is adequate for promoting redox reactions.However, the time scale of e --h + recombination (Eq.( 2)) is two or three orders of magnitude faster than other electron transfer processes (Indrakanti et al., 2009).This is thus considered as one of the major limiting steps in photocatalysis.Therefore, any process which inhibits e --h + recombination would greatly increase the efficiency and improve the rates of CO 2 photoreduction.
Ideally, after photoexcitation, water (or other hole scavengers if not in aqueous systems) is oxidized by holes to form oxygen and protons (H + ) (Eq. ( 3)), which are essential  for photoreduction.However, as indicated in Scheme 1, in addition to the desired reactions, there are also several competing reactions, such as the formation of H 2 and H 2 O 2 , which consume H + and e -, respectively (Eqs.( 4) and ( 5)).Research on minimizing the competing reactions is still going on.For example, some researchers have attempted to replace water with other reductants, such as methanol and ethanol (Liu et al., 1998).This provides a high reaction yield and high selectivity to desired products by changing the mechanism.However, water in practice remains the primary reactant since it is readily available and inexpensive.
As for photoreduction reaction pathways, some researchers proposed that CO 2 activation involves the formation of a negatively charged CO 2 •-species, which are 23 electron radical anions (Yahaya et al., 2004) and can be detected via infrared (IR) spectroscopy (Hartman and Hisatsune 1966) and electron paramagnetic resonance (EPR) spectroscopy (Perissinotti et al., 2001).However, the single electron transfer to CO 2 occurs at potentials as negative as -1.9 V (Eq.( 6)), which is highly endothermic because of the negative adiabatic electron affinity of CO 2 (Lehn and Ziessel, 1982;Freund and Roberts 1996;Indrakanti et al., 2009).The activation of CO 2 molecules in both gas and liquid phases and the formation of CO 2 •-are still open questions.Many researchers now agree that this pathway is a multi-electron transfer (MET) process rather than a single electron transfer process; for the later, the electrochemical potential of -1.90 V precludes the likelihood of a single electron transport mechanism (Lehn and Ziessel, 1982;Yahaya et al., 2004).
As shown in Scheme 1, various products could be formed based on MET photoreduction reactions (Eqs.( 7)-( 12)).Carbon monoxide (CO) is the most common product, since the reaction needs only two protons and two electrons.Formic acid (HCOOH) is also a common product found in the CO 2 photoreduction system, which consumes the same amount of protons and electrons, but requires slightly higher reduction potential than that for CO formation.The other products are generally difficult to form in gas-solid systems since they need more electrons and protons even though the reactions are thermodynamically favorable.It should be noted that the MET reactions listed in Scheme 1 are just the simplest examples.Some products may be formed through multiple reactions.For example, methane (CH 4 ) could be formed through Eq. ( 11) or through the following reactions as well (Varghese et al., 2009;Roy et al., 2010).
In addition to these desired redox reactions, some backward reactions are also possible, making the CO 2 photoreduction system even more complex.For example, the strong oxidation power of the photoexcited holes, protons, OH radicals, or O 2 could oxidize the intermediates and products to form CO 2 .

Limiting Steps and Strategies for Enhancement
The kinetics of CO 2 photoreduction are dependent on many factors, such as reaction temperature, reactant adsorption, product desorption, CO 2 activation, incident light intensity, fraction of the incident light absorbed by the photocatalyst, the specific area of the photocatalyst absorbing the light, and surface as well as crystalline properties of the photocatalysts.For example, CO 2 adsorption on the photocatalyst surface is always considered as a very critical step for CO 2 photoreduction, which could be a pseudo-first-order reaction since the CO 2 photoreduction rate was found to increase with initial CO 2 concentration (Lo et al., 2007).The activation of CO 2 or the electron transfer from the activated photocatalyst to CO 2 is also a major limiting step, where the mechanisms are rather complex and still debatable (Indrakanti et al., 2009).In addition, the e --h + recombination is also widely considered as a major limiting step, which draws mounting attention.Slowing down the e --h + recombination rates has become a key area of research to develop efficient photocatalyst systems.To address the above issues, in particular, to slow down the e --h + recombination rates, several strategies have been developed to enhance the photoreduction efficiency, which are provided as follows.
The first is choosing semiconductors with appropriate bandgap energies/edges.The proper structure of the photocatalyst can improve the product selectivity and yield and increase the rate of reaction.Therefore, photocatalyst preparation is one of the most important steps that can enhance the capacity of a photoreduction system.Wide bandgap semiconductors are the most suitable photocatalysts for CO 2 photoreduction, because they provide sufficient negative and positive redox potentials in conduction bands and valence bands, respectively.The disadvantage of using wide bandgap semiconductors is the requirement for high energy input.Semiconductors with narrow bandgaps, like cadmium sulfide (CdS, 2.4 eV) work in the visible range, but it is not sufficiently positive to act as an acceptor.This causes the photocatalyst to decompose with the hole formation.Table 1 lists several semiconductors with detailed information on their bandgap and band edges.
The second common method is surface modifications of the photocatalysts.For example, one way is to coat metal particles/islands on the semiconductor surface (Fig. 3, left).These metal islands serve as electron sinks to separate e --h + pairs and increase their lifetime (Linsebigler et al., 1995).Formation of composites of two or more semiconductors is also an interesting way.As shown in Fig. 3 (right), in a coupled CdS-TiO 2 system, due to the different band edges of two components, electrons generated from CdS conduction band can transfer to TiO 2 conduction band easily, which increases the charge separation and efficiency of the photocatalytic process (Linsebigler et al., 1995).Doping with non-metals, such as nitrogen is also a promising way to extend the absorption range of the wideband semiconductors from UV to visible range (Sathish et al., 2005).More recently, dye-sensitized and/or enzyme-modified semiconductor systems have also been developed to mimic nature and enhance the photoreduction efficiency.Details of these modifications are provided in the following sections.

Biological Systems
In this section, we will start by presenting a comparison of synthetic systems and their potential for CO 2 conversion; the potential of algal systems for photoconversion is discussed first.Many algae species have been thoroughly investigated for their use in production of biofuels.This motivation stems from two premises: firstly, algae require CO 2 to grow, thus removing CO 2 from the atmosphere; secondly, while the algae grow, many species produce oils and other refinable byproducts that can be used as potential biofuel sources (Kumar et al., 2010;Ho et al., 2011).Therefore, by researching ways in which to maximize algae growth using anthropomorphic CO 2 sources, effective enhancement of a natural biological system that uses solar energy to convert CO 2 into potential fuel sources can be achieved.Fig. 4 shows how photosynthesis occurs in algae.In algae, the fixation of CO 2 occurs as a result of two separate reaction steps.In the light-dependent set of reactions, photoexcited electrons are used to reduce the coenzyme NADP + to NADPH as well as create the highenergy molecule ATP.In the second set of reactions, called the light-independent reactions, these reduced molecules are used to convert CO 2 to organic compounds that can then Table 1.List of semiconductors with detailed bandgap information (Koffyberg and Benko, 1982;Matsumoto, 1996;de Jongh et al., 1999;Xu and Schoonen, 2000;Carlson et al., 2008;Li et al., 2010a;Chikata et al., 2013)  be used by algae as a source of energy.The oils produced by algae which are used for biofuel production contain the carbon that was fixed in the Calvin cycle.It can be seen that this biological system is similar in function to synthetic systems, with the added step of electron transfer to ADP and NADP + to form high energy molecules for subsequent reduction to occur in the Calvin cycle (Zeng et al., 2011).In a synthetic system, light energy is used to excite electrons in the photocatalyst which are then used to reduce CO 2 .Thus, light energy is stored in the bonds of the reduced CO 2 products.It is apparent that the steps carried out in these synthetic systems very much resemble what is accomplished with algae.Tables 2 and 3 summarize different species of algae and synthetic systems with their CO 2 conversion rates, respectively.In order to make a fair comparison between these systems it is important to use one consistent unit of measurement for CO 2 conversion, such as µmol CO 2 converted per gram catalyst per hour (µmol/g/h).Out of the many ways CO conversion is reported in the literature for synthetic systems, this unit seems to the best to demonstrate the performance of the system, as it truly reflects the potential of conversion of the system without bias of amount of catalyst used or amount of time used to run the experiment.For algae systems, a commonly reported unit in the literature is mg of CO 2 consumed per liter of culture fluid per day (mg/L/d).For   CO/H 2 15.6 TON (Hori et al., 2002) a (TON = turnover number) results for organic complex systems are reported in turnover numbers because of lack of information for conversion to the metric µmol/g/h.The turnover number is the maximum number of substrate molecules that the molecular catalyst can convert to product, per site available on the catalyst (Morris et al., 2009).instance, Ho et al. (2011) reported the CO 2 consumption rate of the algae species in Table 2 in the unit mg/L/d.We converted this unit to µmol/g/h for comparison with synthetic systems.This unit conversion was accomplished by assuming 1 gram of dry algae in the culture fluid, which is a typical range of dry algae reported for 1 liter of culture fluid.
As we can see from the two tables, in general, synthetic systems conversion rates fall below the algal systems.However, it should be noted that in order to produce well performing algae systems a lot of resources must be consumed.The tanks for the algae must be cleaned frequently in order to prevent such as water and nutrients buildup that blocks light penetration.As mentioned earlier, for just one gram of algae up to an entire liter of water is required.Therefore, a lot of space and water resources are required to get the performance as outlined in the table above.The CO 2 uptake occurs in the algal biomass that requires cumbersome separation steps for the extraction of the oil product.Furthermore, algae systems are in general less robust in stability compared to synthetic systems.Algae are living organisms that require a narrow range of living conditions, thus small changes in their environment can impact biomass growth as well as the amount of byproducts produced, for which a lot of time and energy is required in process control to guarantee optimal performance.Very few, if any, synthesis have demonstrated uptake of CO 2 from concentrated systems exceeding a few percent.Synthetic systems however, do not have these growth associated problems of a biological system.While it is true that current synthetic systems have decreased activity after a certain amount of time leading to a decrease in conversion rates, there is a lot of room to improve and modify these systems; whereas most of the problems associated with algae systems are due to algae being a living organism, which makes the problem for improvement comparatively more complicated.Thus it is important to not only improve the conversion rates of synthetic systems but also to improve the lifetime of these systems for practical and scalable use in the future.

Semiconductor Systems
There are many types of synthetic systems covered in the literature.The most common and thoroughly researched synthetic systems are semiconducting materials, such as TiO 2 .

TiO 2 -Related Photocatalysts
TiO 2 is a commonly used wide bandgap semiconductor for CO 2 photoreduction because of its relatively low cost, high availability, resistance to photo-induced corrosion and low toxicity (Indrakanti et al., 2009).
TiO 2 Nanomaterials.The first CO 2 photoreduction system involved the use of semiconductor powders such as TiO 2 suspended in a beaker of water (Inoue et al., 1979).Improvements have been made to these systems with the advent of semiconductor nanoparticles (NPs) which led to an increase in the surface area of a semiconductor catalyst.By using TiO 2 NPs, the catalytically active surface area is increased, leading to an increase in the frequency between CO 2 and TiO 2 , a crucial factor for the electron transfer in reductive reactions in Scheme 1, as well as to leading to a decreased amount of e --h + recombination that occurs on the TiO 2 surface.P25 is a commercially available source of anatase-rutile mixed phase TiO 2 NPs that is commonly purchased and employed in photocatalytic research.Its popularity stems from its high catalytic activity that arises from the repressed recombination rates that is a result of the ratios of mixed anatase and rutile phases employed in the P25 formulation (Hurum et al., 2003).
TiO 2 crystal structure and morphology studies are commonly explored in the literature as methods for increasing TiO 2 photocatalytic activity.TiO 2 nanosheets with 95% exposed {100} facet are a good example to demonstrate the crystal structure effect.Marked photocatalytic activity was achieved since the {100} facet was considered to be the active facet (Xu et al., 2013).By changing the morphology of TiO 2 , the electron transport through the TiO 2 could be controlled.The most studied morphology is the porous arrays or spheres.For instance, porous microspheres of MgO-patched TiO 2 were synthesized by Liu et al. (2013).The microspheres exhibited 10 times higher activity toward CO production when the reaction temperature increased from 50 to 150°C.However, this morphology also suffers from having multiple grain boundaries that cause erratic electron transport through the network leading to undesired electron recombination.By changing to a patterned film structure or nanowire/rod configuration, the electron transport can be oriented to be more direct in one path so that electron recombination between two individual TiO 2 structures is minimized.For example, the synthesis and characterization of anatase phase nanorods modified with rutile phase NPs (TiO 2 -RMA) for the photoreduction of CO 2 were reported (Wang et al., 2012a).
The rutile phase NPs served as an electron sink in place of more commonly used and more expensive Pt NPs.Wang et al. (2012a) showed that this TiO 2 -RMA structure showed greater conversion rates than just anatase phase nanorods.While the transition from semiconductor powders to semiconductor NPs has effectively increased the surface area of TiO 2 catalytic systems, additional modifications to TiO 2 are required to enhance activity that is a result of its wide band-gap, which restricts excitation to the UV range, which constitutes only 3-5% of the solar spectrum.Furthermore, modifications to reduce e --h + recombination rates and increase CO 2 availability at the catalyst surface must be made to see the upper-level efficiencies possible for TiO 2 systems.
Modifications of TiO 2 Systems.Common modifications made to TiO 2 include doping of the TiO 2 lattice with metal and nonmetals, leading to reduced e --h + recombination rates and visible sensitivity, respectively.The formation of heterojunctions by incorporating two or more components (metals or semiconductors) with different energy levels is another effective approach for this purpose.Catalytically inactive supports such as SiO 2 are used to physically separate TiO 2 NPs while effectively increasing surface area.Crystal phase modification is also an interesting topic to explore.
By doping metals such as Cu, Pt, Fe in the TiO 2 lattice or coating the metals/metal oxides onto the TiO 2 surface, it has been found that the conversion rates of the photocatalytic reduction for these systems are higher than their non-modified counterparts.In the case of doping or coating metals, this can be explained by considering the metal dopants/islands as sinks for the excited TiO 2 electrons (Usubharatana et al., 2006).When the excited electron is injected into the conduction band of the TiO 2 and starts to travel across the TiO 2 crystal lattice, it becomes trapped by the metal dopants because of the relative lower energy state that it can obtain if it goes to the metal dopant.Thus the metal dopant/island serves as a means in which to separate the excited electrons from the holes that are in the TiO 2 lattice, thereby reducing e --h + recombination.For heterojunctions with mixed semiconductors, the difference between bandedges is the major driving force for the enhancement of charge transfer and subsequently the photocatalytic performance.As from Fig. 5, upon deposition of Cu species (Cu 2 O in this case) on the TiO 2 surface not only did the amount of CO 2 reduced and thus product yielded increase, but also formed a new product, CH 4 .The formation of pn junction between Cu 2 O and TiO 2 is considered as the major reason for the improvement.
Electron sinks that are noble metals, such as Pt and Rh, also have implications in multi-electron transfer reactions that are crucial for high efficiency photocatalytic reduction of CO 2 to take place (Indrakanti et al., 2009).It has also been found that the size of the noble metal NP used to serve as the electron sink is a crucial variable in the conversion rates that can be achieved.Wang et al. (2012b) reported a facile synthesis for Pt-TiO 2 nanostructured films using an aerosol chemical vapor deposition route (ACVD) to deposit the TiO 2 film and a tilted-target sputtering (TTS) technique to deposit Pt NPs.It was observed that if the Pt NP is too small, then quantum confinement effects cause the energy band separation of the Pt NP to be too high for an excited electron to be transferred to the NP.Conversely, if the Pt NP is too big, the NP merely serves as a recombination center capturing both excited electrons and holes, vastly reducing the efficiency of the system (Wang et al., 2012b).Thus precise control over the size of the metal dopant plays a role in the CO 2 conversion rates for the system.Fig. 6 provides a graphic description of the photocatalytic steps in this system along with the change in energy levels of the Pt NPs as a function of their size.Doping of Au or Ag can also extend the sensitivity of the photocatalyst to the visible spectrum due to their localized surface plasmonic resonance properties (Li et al., 2007).
Nonmetal doping has also been used for TiO 2 photocatalytic reduction systems.However the function desired from nonmetal doping is much different from that of metal doping.The doping of nonmetals such as nitrogen or carbon is motivated by trying to extend the electromagnetic region in which the photocatalyst is able to use to generate excited electrons by narrowing the bandgap of the photocatalyst system (Hamadanian et al., 2009).The nonmetal dopants effectively redshift the TiO 2 photocatalyst  system thus allowing for a greater portion of the incoming light's spectrum to be used for generating e --h + pairs (Yu et al., 2002).Sathish et al. (2005) also reported higher photocatalytic activity for a N-TiO 2 system.The photocatalytic activity of the N-TiO 2 system exposed to visible light was reported to be higher than pure TiO 2 and P25.The shift from UV only to visible sensitization is shown clearly in Fig. 7.
TiO 2 Crystal Structure.TiO 2 exists in three distinct crystal forms called rutile, anatase and brookite.Of these crystal phases, rutile and anatase are the best understood and studied.
There is much research being conducted on exploring the effect of TiO 2 crystal structure on photocatalytic activity.Pan et al. (2011) reported a synthesis technique for faceted anatase rods.These faceted rods showed higher photocatalytic conversion of CO 2 to CH 4 when loaded with Pt NPs than P25 loaded with Pt did, thus showing how significant a simple change to crystal structure can greatly affect and improve a photocatalytic system.
Out of the three phases, brookite is the least studied and the optimum composition for an anatase-brookite photocatalyst for the reduction of CO 2 has only been explored once in the literature to date (Liu et al., 2012;Zhao et al., 2013), while rutile-anatase systems are more common and can be purchased in the form of P25.Zhao et al. (2013) has reported an optimal composition of anatase-brookite systems as 75% anatase and 25% brookite.This optimized system showed higher photocatalytic activity than P25, which suggests that the interaction leading to greater photocatalytic activity for the rutile/anatase combination for P25 compared to anatase and rutile alone is not as effective as the interaction between anatase and brookite.It is interesting to note that the composition, 75% anatase and 25% brookite that was found to be optimal is the same fractions used for many commercially available anatase-rutile P25, further supporting the idea that the anatase-brookite interaction is more favorable than the anatase-rutile interaction for enhanced photocatalytic activity.
TiO 2 in Mesoporous Supports.By incorporating SiO 2 as a mesoporous support into the TiO 2 lattice it has been found that the photocatalytic activity of the system increases (Anpo et al., 1986(Anpo et al., , 1998)).While the SiO 2 in the system is not photocatalytically active, it is understood that the SiO 2 mesoporous network serves to not only spatially separate the TiO 2 NPs but also serves to increase the surface area of the system for CO 2 to bind and be reduced.Furthermore, the dispersion of TiO 2 NPs into the SiO 2 network reduces contact between the TiO 2 NPs leading to less e --h + recombination occurring between individual particles (Wang et al., 2011c).Wang et al. (2011c) reported a low cost, rapid synthesis technique using a furnace aerosol reactor (FuAR) for the synthesis of Cu-TiO 2 -SiO 2 photocatalyst composite particles formed by evaporation driven self-assembly.The size, composition, and porosity of the composite particle were controlled by adjusting the temperature of the reactor, the precursor concentrations and the ratio of each constituent present in the feed to the FuAR.Fig. 8 shows a schematic of the experimental setup for catalysis synthesis and analysis.It was reported that a relatively high conversation rate of 20 µmol/g/h of CO was obtained, showing the high potential for aerosol pathways for the synthesis of photocatalysts compared to more time consuming wet chemical based techniques.The synergistic effect of deposition with Cu 2 O and dispersing the TiO 2 NPs in a silica mesoporous matrix has been reported in the literature and it was found that the photocatalytic conversion rates and selectivity for CH 4 was increased (Fig. 5, Li et al., 2010b).

Other Semiconductors
Current photocatalytic systems for the reduction of CO 2 are not limited to just TiO 2 .There are many other semiconductor-based systems reported in the literature.Some of these semiconductors include CdS, ZnO, GaP, SiC, WO 3 , and various other metal oxides.One of the possible advantages of using non TiO 2 semiconductors is a narrower bandgap that results in a wider range of wavelengths to induce excitation of electrons (Navalon et al., 2013).This has implications in visible light range sensitivity without the use of sensitizers that are subject to photo-degradation, thus allowing for a possible more stable system under visible light conditions.Praus et al. (2011) reported the synthesis of CdS NPs deposited on montmorillonite (MMT) and found that the photocatalytic activity of the resulting CdS-MMT nanocomposite was 4-8 times greater than P25.Cao et al. (2011) reported the use of carbon NPs covalenty functionalized with gold and platinum to concentrate the photogenerated electrons for the ultimate reduction of CO 2 .This functionalization gives carbon NPs strong absorption in the visible range and possible near-IR range as well.It was reported that when tested for its photocatalytic activity, formic acid was found to be the main product with a quantum yield of ~0.3% which was said to be an order of magnitude higher than what was obtained for TiO 2 using similar experimental conditions except with exclusion of the visible range.It can be seen that research surrounding these alternative semiconductor systems that naturally absorb greater in the visible region is relative, as their photocatalytic capabilities in comparison to more traditional TiO 2 systems are promising.Metal sulfide systems have also been paired with titaniabased systems to sensitize for visible light.Li et al. (2012b) reported the use of TiO 2 nanotubes (TNTs) with Bi 2 S 3 and CdS modification that led to both increased visible light absorption and catalytic activity of the TNTs.As from Table 3, these values are rather high showing promise for systems that incorporate non titania based semiconductors.

Metal-Organic Complex Systems
CO 2 reduction systems employing large molecular compounds can be divided into two types.The first type of system is characterized by the electron excitation and catalytic reduction of CO 2 taking place at two different sites, while the second type of system involves the excitation and catalytic reduction taking place on the same large molecular complex (Morris et al., 2009).The first system can be thought of as a photo-sensitized system, which commonly uses Rudyes for the sensitization.These types of systems have been discussed previously in this study, however were not specifically classified until now as the use of dye sensitizers in semiconductor systems to extend sensitization of the semiconductor catalyst into the visible range is so ubiquitous in the literature.For this review, we will focus mostly on type two systems in which the sensitizer and catalyst of the system are the same large molecule and type one systems in which the catalyst is not a metal semiconductor, but rather another molecular complex covalently bonded to the sensitizer.Bruckmeier et al. (2012) reported the photocatalytic activities of type 2 rhenium(I) based catalysts, which stated that the transfer of the second electron for CO 2 reduction to CO originates from the long lived one-electron reduced species (OER), dominates if the proximity of the coordination centers is adjusted according to the lifetime of the OER.In the case of Re-NCS type complexes with a naturally long lived OER, the bimetallic mechanism will occur by simply increasing the concentration of the catalyst, which changes the proximity of the coordination centers, thus it has often been the case in the literature that Re-NCS photocatalysts are reported as the best rhenium(I) based photocatalysts.However, Bruckmeier et al. (2012) reported that in fact Re-Br photocatalysts showed better photocatalytic activity if the centers are brought in closer proximity through covalent linkage, which would be expected considering the ease with which the bromo-ligand leaves thus providing a new coordination site.Ettedgui et al. (2011) reported the synthesis of a metalorganic polyoxometalate hybrid complex, Re I (L)(CO 3 )CH 3 CN-MHPW 12 O 40 (L = 15-crown-5phenanthroline and M = Na + , H 3 O + ), for the photoreduction of CO 2 in visible light.This system showed high conversion rates and the structure of this complex is provided in Fig. 9. Chauvin et al. (2011) reported the use of type 2 osmium complexes for the photoreduction of CO 2 in visible light with the use of an amine sacrificial agent.It was reported that conversion rates were comparable to rhenium based systems thus making osmium complexes a possible alternative.These systems have the advantage of not requiring the use of expensive co-catalyst metals such as Pt, Au.Furthermore, the amount of modifications and changes that can be made to the structure of the molecules involved in each system provides a lot of variables for experimentation and improvement of this type of photocatalytic system.However, a problem worth noting is that many of these systems suffer from photo-degradation that ruins the catalytic behavior of the dye shortly after exposure to a light source.Furthermore, many of these systems require a sacrificial agent for restoration of the depleted sensitizer, which is not desired in the long term for CO 2 photoreduction systems.There is much literature regarding the synthesis of novel molecular complexes for the photocatalytic reduction of CO 2 .However, it is interesting to compare these types of systems with their parallel in biology.In biology, large molecular complexes called proteins are responsible for both electron transport, that results from excitation upon incidental visible light, as well as for the conversion of CO 2 in plants.We will now provide a discussion of some of these bio-inspired systems that use enzymes for the catalyzed CO 2 conversion step on a semiconducting material.

Hybrid Systems
A more trendy area of research for CO 2 photoreduction includes incorporating enzymes to carry out the catalytic step.The goal behind using enzymes is to set a benchmark of what synthetic systems should be capable, based on the fact that enzymes have evolved to perform the catalytic step with high efficiency (Woolerton et al., 2012).Enzymes are highly substrate specific, show high rates of conversion, and have evolved to be highly efficient at carrying out their catalytic functions.Thus they are an attractive option for proof-of-concept CO 2 photoreduction systems.
As early as 1992, Inoue et al. (1992) used malic enzymes isolated from chicken liver to carry out the catalytic reduction of CO 2 .The enzymes were paired with both TiO 2 microcrystals and CdS particle suspension systems and showed conversion of pyruvic acid to malic acid.Woolerton et al. (2012) conducted similar experiments using a carbon monoxide dehydrogenase (CODH) enzyme.Their basic system involved the CODH enzyme and a Ru-based dye adsorbed onto the surface of TiO 2 NPs.The dye served as a visible light sensitizer and upon excitation injects its electrons into the conduction band of the TiO 2 nanoparticle which then serves to transport the electrons to the enzyme.The enzyme then carries out the catalyzed reduction step.This system showed high selectivity for CO and the conversion rate was reported as 250 µmol/g/h.Below in Fig. 10 is visual representation for this system.Chaudhary et al. (2012) reported co-attching CODH enzyme to both CdS quantum dots (QDs) and CdS nanorod (NR) morphologies.It was reported that while the QD-CdS-CODH system showed a higher conversion rate at 60 µmol/g/h.The NR-CdS-CODH showed a much higher turnover rate at 1.23 s -1 compared to 0.25 s -1 for the QD-CdS-CODH system.Yadav et al. (2012) reported the use of a novel graphene-based visible light active photocatlyst that incorporates a covalently bonded chromophore, such as multianthraquinone substituted porphyrin to the graphene oxide (GO) as a photocatalyst coupled with the activity of the enzyme, formate dehydrogenase.In this system, the chromophore is excited and transports its electrons into the GO.The electrons are then transferred to a rhodium complex in solution which becomes reduced.The reduced rhodium complex then reduces the co-enzyme NAD + to NADH which is consumed by the formate dehydrogenase enzyme to carry out the reduction of CO 2 to formic acid, thus completing the catalytic cycle (Yadav et al., 2012).The conversion rate for this system was high, showing not only how enzyme incorporation into a photocatlytic system can provide high efficiency, but also shows the future for systems that incorporate graphene-based materials in CO 2 photoreduction systems.

SUMMARY AND OUTLOOK
Photoreduction of CO 2 is a promising approach to reduce carbon emission while simultaneously recycles it as a fuel feedstock by harnessing readily available solar energy.The reduction mechanisms, however, are complex, involving various limiting steps.Overcoming the limiting steps is the major task for developing efficient photocatalyst systems.Several strategies have been reviewed for this purpose, including selection of semiconductors with suitable band edges, coating electron sinks, forming heterogeneous junctions, designing novel morphologies, and developing hybrid structures.A state-of-the-art comparison between different photocatalytic systems for CO 2 reduction was made aiming to demonstrate the advances in this area and provide an overview of the research trend for future development of photocatalysts for CO 2 photoreduction in a large scale.

Fig. 1 .
Fig. 1.Schematic diagram of the energy correlation between semiconductor catalysts and redox couples in water at pH = 5. (Reprinted with permission from Inoue et al., 1979; Copyright 1979 Nature Publishing Group).

Fig. 6 .
Fig. 6.Schematic of CO 2 photoreduction on Pt modified TiO 2 thin films with highly directed electron flow.On the right is provided a bandgap description of the effect of Pt size on CO 2 photoreduction (Reprinted with permission from Wang et al., 2012b; Copyright 2012 American Chemical Society).

Fig. 8 .
Fig. 8. Experimental setup for aerosol processing of catalysts and CO 2 photoreduction analysis (Reprinted with permission from Wang et al., 2011c; Copyright 2011 Royal Society of Chemistry).

Table 2 .
Summary of algae systems for CO 2 fixation and conversion.

Table 3 .
Literature summary of CO 2 photoreduction by semiconductors, metal-organic, and hybrid systems.