Rhodium-doped titania photocatalysts with two-step bandgap excitation by visible light — in ﬂ uence of the dopant concentration on photosensitization e ﬃ ciency †

Development of photocatalysts active under visible light has been achieved through the modi ﬁ cation of titania with rhodium (Rh) ions. Depending on the concentration of the precursor of Rh species, various crystalline phases and modi ﬁ cation modes have been obtained. Lower Rh concentrations (below 0.5%) resulted in doped rutile particles, while higher (above 1%) gave mainly surface modi ﬁ ed anatase particles. High photocatalytic activities were obtained for surface enriched doped rutile-titania with an extremely low amount of trivalent rhodium (Rh 3+ ) ions (about 0.0005 mol%). Almost four orders of magnitude higher Rh concentration (2%) also provided highly active materials, however, they were composed of mainly anatase particles with photoactive surface RhO x species. Extremely low Rh concentration allowed to limit the amount of intrinsically present tetravalent rhodium (Rh 4+ ) species, which may decrease charge mobility in the titania lattice and promote the recombination route in Rh-doped materials. Photoactive Rh species created in low concentrated samples were able to drive reversible redox processes controlled by the range of light energies. Photocatalytic activity tests, action spectra and photoelectrochemical measurements allowed to propose a unique photosensitization mechanism, involving an Rh 3+ /Rh 4+ redox couple acting as built-in redox mediator, in which both electrons in the conduction band (CB) and positive holes in the valence band (VB) might be generated upon irradiation with visible light. Such a mechanism was not valid for materials containing higher amounts of Rh composed of mainly anatase. Detailed studies on the properties of the materials containing various amounts of Rh species (0.0001 – 3 mol%) showed the dependency of e ﬃ ciency of photoinduced processes on Rh concentration.

Rhodium-doped titania photocatalysts with twostep bandgap excitation by visible light-influence of the dopant concentration on photosensitization efficiency † Introduction Development of visible-light active inorganic materials able to drive simultaneously redox processes that require high oxidation and reduction potentials, such as oxidation of water and reduction of oxygen, is still facing problems concerning charge-carrier generation efficiency and stability.There are only a few bare semiconductors that possess high oxidation as well as reduction power in the excited state, such as titanium(IV) oxide (titania), strontium titanate, cadmium sulde, zinc sulde or zirconium oxide, moreover, most of them are unstable or photocatalytically inactive under visible light. 1 Photosensitization of suitable wide band-gap semiconductors usually does not solve the problem.Commonly used methods for the activation of semiconductors toward visible range of light, such as dye sensitization or doping with metal or non-metal ions, [2][3][4][5][6] seem to provide materials maintaining only one of the intrinsic oxidative or reductive properties.In such materials it is impossible to induce one-step band-gap excitation only with visible-light irradiation.To overcome this problem and provide simultaneous generation of highly active holes and electrons, a complex z-scheme-resembling systems, in which reduction and oxidation processes have been targeted to two different semiconductors, have been proposed. 7Activity of such systems is based on the electron-transfer reaction between particles of two materials with anodic and cathodic characters, mediated with an external redox pair or carried out without any redox mediators. 8Nevertheless, z-scheme photocatalysis has many drawbacks arisen from complexity of the systems and limited efficiency of interfacial electron transfer between particles of different origins.Thus, another concept for developing efficient visible-light active systems might evolve from the introduction of the idea of z-scheme mechanism into a single photocatalyst system.It should be possible if a suitable inner redox mediator would be introduced to (or combined with) semiconductor particles of proper redox potentials of conduction and valenceband edges.
Such idea should be feasible in doped materials, where new electronic states are introduced into the band gap of base material.Titania, due to its redox abilities, seems to be a good candidate for development of such single photocatalyst system.There are many transition metal ions, e.g.Mo, Co, Cr, Fe 6,9,10 that may efficiently reduce the bandgap of titania leading to the red shi of the optical absorption edge, however, only a few are able to induce appearance of intermediate bands which divide the energy gap into sub-gaps. 9][13] Rh-doped titania has been already suggested as an efficient photocatalyst for O 2 evolution under visible light. 145][16][17] The electronic structure of doped materials depends strongly on the valency of introduced ions.Due to the charge compensation insertion of Rh 3+ induces Rh 4+ species and oxygen vacancies generation.Theoretical studies, focused on the materials containing about 1.4% of Rh, show that in such conditions the donor states just at the top of VB associated with Rh 3+ ions and interband-states arising from the presence of Rh 4+ , which may act as acceptors of electrons from VB, should be created. 13owever, in Rh-doped titania, as in the case of titania modied with Rh 3+ species located only at the surface, [18][19][20] visible-lightinduced activity is assigned mainly to the excitation of electron from Rh 3+ to the CB of semiconductor, which is followed by the chemical reduction of generated Rh 4+ .Same mechanism was presented by Niishiro et al. for 1% Rh-doped SrTiO 3 . 15,16lthough the authors shown that part of intrinsically present Rh 4+ species may be photoreduced with electrons from the VB in the initial step of photocatalytic process, it is believed that consequent irradiation of the photocatalytic system induces mainly excitation of Rh 3+ .Photo-regenerated in the initial step Rh 3+ ions are regarded as "unstable" compared with intrinsically present Rh 3+ species stabilized with oxygen vacancies. 15n the case of both, titania and SrTiO 3 based Rh-doped materials, the possibility of photogeneration of Rh 4+ from intrinsically present Rh 3+ and the role of photoreduction of Rh 4+ in over whole photocatalytic process have not been investigated.Moreover, Rh 4+ ions formed in the doping process are believed to play a role of recombination centers, what was a main reason of photocatalytic inactivity of as prepared Rh-doped SrTiO 3 , 15,21 and Rh-doped titania containing 1 mol% of Rh. 14 Thus, up to now, to restore photoactive Rh 3+ species in Rh 4+ reach-materials or to prevent the creation of Rh 4+ species two strategies have been used: reduction with hydrogen ow, and, more successful, codoping with ions of different valency (e.g.Sb 5+ ) allowing charge compensation within the host lattice. 14,15,21However, total reduction of Rh 4+ ions within the lattice and stabilization of Rh 3+ species provided by second strategy results in the disappearance of available interband states, 13,14 what was calculated and indirectly observed as bleaching of absorption band associated with Rh 4+ in codoped samples, and put the possibility of visible-light-induced excitation of valence band electrons into question.
Our approach is to decrease the number of recombination centres while maintaining the pot of "unstable" Rh 3+ species by decreasing the concentration of the dopant to the extremely low value.As it was studied by Bloh et al. effective concentration of dopant is strongly dependent on particles size, and may reach very small values. 22,23If the amount of Rh 3+ ions is low enough and they are enriching the surface of the semiconductor particles the charge compensation by Rh 4+ generation should be partly suppressed.Such conditions should allowed creation of Rh 3+ ions which are able to drive reversible redox processes.In these processes photogeneration of stable Rh 4+ ions as well as their photoreduction should be possible.2][23][24] Recently our studies on such "depleted" Rh-doped titania have shown that photogenerated Rh 4+ species play important role in overall photocatalytic activity. 25We developed materials presenting unique photosensitisation mechanism assuring two-step bandgap excitation involving Rh 3+ /Rh 4+ coupled redox pair.However, still the inuence of the intrinsic Rh 3+ and Rh 4+ concentration on the efficiency of photoinduced processes was not well understood.
In the present work we investigated the inuence of Rh concentration on properties of the materials and photosensitization mechanism by detailed studies on the morphology, crystalline structure, photoabsorption, photocatalytic and photoelectrochemical properties of materials containing broad range of the Rh concentration: from 0.00001 to 3 mol%.We propose the best conditions for the highest efficiency of twostep bandgap excitation, and discuss future perspectives of improvement of the performance of visible-light-active materials acting by such mechanism.

Results and discussion
The Rh-TiO 2 materials have been obtained by impregnation of titania (anatase) with RhCl 3 and calcination at 700 C for 3 h.The calcination conditions were sufficient for RhCl 3 decomposition.The concentration of Rh signicantly inuenced the crystalline structure of host material.XRD analysis revealed that materials containing up to 0.1% of Rh had a single phase of rutile.Additionally, no changes in the lattice constants could be observed for less concentrated samples.It may indicate not efficient incorporation of Rh ions into the host lattice or, more likely, too small changes in the lattice parameters to be detected due to low Rh concentration.Further increase of Rh content caused enlargement of rutile lattice constants, what suggests that Rh ions are introduced to rutile crystalline lattice in more concentrated samples (Fig. 1S in ESI †).Though, it is hard to conclude if there are any changes in anatase phase parameters.Together with an increasing Rh concentration retardation of anatase-rutile transformation could be also observed (Fig. 1), and the samples containing 2% were composed mainly of anatase particles.Moreover, for all samples there is no evidence of any other crystalline phases (e.g.Rh 2 O 3 ) occurring in studied materials.
Particle size of obtained materials depended on their crystalline structure.Samples containing mainly rutile phase were characterized of particles with an average diameter included within the range 70-120 nm.With an increase of anatase-phase content a slight decrease in particles size has been observed.The changes in particle size estimated on the basis of SEM and TEM images (Fig. 2 and 3) were in agreement with the changes in surface area which was estimated to 11.8 m 2 g À1 for bare titania and 10.2, 10.0, 17.3, 24.2 for 0.001%, 0.05%, 1%, and 3% Rh-TiO 2 , respectively.TEM images (Fig. 3) and STEM/EDX analysis (Fig. 4) revelled also that in the case of low concentrated photocatalysts (up to 0.5% of Rh) composed of mainly rutile phase the materials are homogenous and Rh ions are uniformly located within all of the particles.No surface signicant accumulation or localized aggregation of Rh species could be observed.The materials containing higher Rh concentration and anatase content are characterized of inhomogeneous Rh distribution.On the basis of the size and resistance of the observed particles to laser ablation during the measurement, it is suggested that Rh species accumulated at the anatase particles.
The valency of Ti and Rh species located at the surface and sub-surface area of studied materials has been investigated by XPS measurements.Oxidation states were determined by the binding energies of Rh 3d 5/2 and Ti 2p 3/2 .
Fig. 1 XRD patterns of Rh-TiO 2 and a reference sample without rhodium.
Fig. 2 SEM images of Rh-TiO 2 materials containing 0% (a), 0.001% (b), 0.01% (c), 0.1% (d), 0.5% (e), 1% (f), 2% (g) and 3% (h) Rh.Samples with concentration of Rh up to 0.5% contained overlapped signals about 309.0 eV and 309.8 eV, assigned to Rh 3+ and Rh 4+ , respectively (Fig. 5 le). 26,27The Rh 4+ /Rh 3+ ratio was increasing with Rh concentration reaching the highest value for 1% Rh-TiO 2 .Together with that change a slight shi of the binding energy-value of Rh 3d 5/2 signal, especially Rh 4+ component (from 309.87 eV to 309.68 eV), toward lower energies has been observed.It may indicate enrichment of the surface with doped Rh 4+ ions that should change their surroundings.Decomposition of RhCl 3 in air results in rhodium oxide generation and Rh 2 O 3 is more in the creation of oxides agglomerates characterized mainly by Rh 3+ signal.Thus, the presence of Rh 4+ species may indicate introduction of Rh ions into the titania lattice, at least close to the surface.Above 1% of Rh the ratio Rh 4+ /Rh 3+ decreased and signal of Rh 3+ became predominant, what indicates competition between incorporation and agglomeration processes at the titania surface in such a conditions.What is more, the broadening of the signal recorded for highly concentrated samples could be observed.Deconvolution of the spectra revealed that new component appeared at ar. 308.2 eV (Fig. 6) what may be associated with Rh 3+ species with different surroundings, such as in Rh 2 O 3 agglomerates. 27,28The concentration of such species located at the surface of titania increased gradually with amount of Rh.Insignicant change in the ratio of Rh 4+ to Rh 3+ primary signals in samples containing more than 1% of Rh clearly shows that mainly new species were generated above that concentration.The observed change correlated with an increase of anatase structure contribution with Rh content.The analysis of Rh valency in materials containing less than 0.5% of Rh was insufficiently reliable due to the detection limits of used technique, which was around 0.1%.
Titanium present at or near to the surface of bare titania gave peak of the energy around 458.45 eV assigned to 2p 3/2 of Ti 4+ (Fig. 5 right). 17Increase of Rh concentration in material caused progressive shi of the binding energy of Ti 2p 3/2 signal up to 457.96 AE 0.03 eV, which was observed for 1% and higher concentrations of Rh.It may indicate changes in Ti ions surrounding and Fermi level shis caused by doping process. 17or materials containing more than 2% of Rh new component of Ti 2p 3/2 signal at binding energies around 456.9 eV was observed, which might suggest possible contribution of Ti 3+ ions.
Ratio of Rh/Ti calculated from XPS data is higher than the nominal one, what indicates that the surface is enriched with Rh ions.
Obtained samples kept in the dark were characterized of pale yellow to orange colour which was reversibly turning to sage tinge upon day light irradiation.Samples containing more than 1% of Rh became more brownish and did not change the colour under the light.Intensity of the colours was gradually increasing with Rh concentration.
DRS absorption spectra of the samples containing 0.0001-0.001% of Rh comprise characteristic for titania band-gapabsorption accompanied with a weak absorption with an offset at 550-590 nm assigned to Rh 3+ strongly interacting with TiO 2 (Fig. 7 inside). 14,17For samples with more than 0.005% of Rh two bands within visible range of light can be distinguished.More intensive one at 445 nm and broad weak band (resembling tailoring for less concentrated samples) at ar. 620 nm associated with Rh 3+ and Rh 4+ species strongly interacting with titania lattice, respectively.The former one results from the electron transition from Rh 3+ ion to the CB of titania, and the latter one originates from the electron transition from VB of titania to Rh 4+ ion. 14,17The observed absorption does not resemble the absorption spectra of bare Rh 2 O 3 and a 0.01% mixture of Rh 2 O 3 with titania (Fig. 2S in ESI †), what indicates, that does not originate from the rhodium oxide microcrystalline species.The changes observed for the latter material aer calcination at 700 C may suggest partial introduction of Rh ions into the crystalline lattice of titania in a such prepared sample.
Samples containing more than 0.5% of Rh are characterised of strong absorption within whole range of visible light increasing with the Rh content and, in result, covering characteristic band at 620 nm in the case of 2% and 3% Rh-TiO 2 .It suggests creation of new types of Rh species at the surface of doped titania particles.Observed increased absorption above 750 nm may be assigned to d-d charge transfer transition between Rh 3+ and Rh 4+ species, as it was suggested for Rhdoped SrTiO 3 . 29,30ght induced photoabsorption changes Materials with lower Rh concentration show unusual spectroscopic behavior under exposure to light.Similarly to the changes observed for the previously studied 0.01% Rh-TiO 2 material (calcined at 650 C) 25 irradiation with light from the range overlapping the band assigned to Rh 3+ (440-550 nm) caused decrease of the intensity of that band and increase of the absorption intensity of band at.620 nm assigned to Rh 4+ (Fig. 8 le).The process was reversed through the irradiation with light of the range 590-730 nm, covering the range of the Rh 4+ band.It is clear that former irradiation causes electron transition from Rh 3+ to the CB of titania.The latter one should induce photo reduction of generated Rh 4+ by the electron from the VB. 14,17he photoreduction of Rh 4+ has been observed previously in Rh-doped SrTiO 3 , however only in the induction period of photocatalytic reaction in which just intrinsically present Rh 4+ ions participated. 15,16Due to that in such materials two kinds of Rh 3+ ions have been distinguished: Rh 3+ stabilized by oxygen vacancies created during the doping process, and Rh 3+ "unstable", which may be created from intrinsically present Rh 4+ by the photoreduction process at the beginning of photocatalytic reaction.However, in published papers there was little discussion on fate of Rh 3+ ions participating in photoexcitation process, and no information about photoreduction process of Rh 4+ ions created through the photooxidation of Rh 3+ species have been provided.Results of our studies indicate that concentration of "unstable" Rh 3+ ions capable of reversible photoredox processes in the studied materials is much higher than concentration of Rh 3+ ions just created upon reduction of intrinsically present Rh 4+ (included into calculations).As it was shown before, 25 only a very little fraction of Rh 3+ species participating in reversible redox processes, is oxidized to Rh 4+ in air in the dark.Most of such Rh 3+ ions are stable in the presence of O 2 and the only oxidizing factor is light (Fig. 8

le).
Such a light induced photoabsorption changes were observed for the samples containing from 0.001% up to 1% of Rh.The analysis of the absorption changes recorded for the samples containing less than 0.001% of Rh was difficult due to the very small ratio of the observed differences in the KM function values to the noise.
The concentration of reversibly oxidized Rh 3+ ions increased with Rh concentration reaching the maximum for samples with 0.5% of Rh species (Fig. 8 right). 25For samples containing 1% of Rh the amount of photogenerated Rh 4+ slightly decreased.Increase of Rh concentration caused also extension of the time necessary for total Rh 3+ recovery in air, which became signicant for samples containing more than 0.01%.Moreover, the range of the red light efficiently inducing recovery of Rh 3+ was slightly shied toward lower energies for more concentrated  samples.The materials prepared by calcination of the mixture of Rh 2 O 3 and TiO 2 did not show any signicant light-induced absorption changes (Fig. 3S in ESI †), as well as visible lightphotocatalytic activity, what indicates importance of the strong interaction of Rh ions with titania lattice in these processes.
Materials with higher Rh concentration (2 and 3%) did not present such spectroscopic behaviour.In those samples UV-vis light as well as red light caused only very slight decrease in the absorption intensity within whole range of visible light.The reason of observed changes is still under considerations.

Photocatalytic and photoelectrochemical activity
Mechanistic considerations.Applied modication improved visible-light-induced photocatalytic activity of titania studied in the reaction of acetaldehyde (AcH) oxidative decomposition in air.Dependency of the activity on Rh concentration shows two maxima (Fig. 9).One is within the range of extremely low content of Rh, second can be noticed for much higher concentrations (several rows of difference).In the low range of Rh concentration the most active materials contain 0.0005-0.001% of Rh.From 0.01% of Rh increase of Rh content caused progressive decrease of the intrinsic titania activity, which itself showed noticeable photocatalytic activity probably due to some surface states weakly absorbing within visible range of light.In the case of samples containing more than 1% of Rh sharp increase of the photocatalytic activity was observed reaching the highest value for 2% Rh-TiO 2 .Turnover number calculated for bare and modied titania from the molar ratio of the amount of generated CO 2 to the total amount of TiO 2 did not exceed unity (in the case of 0.0005% Rh-TiO 2 it was 0.002 aer 6 h of irradiation).However, the turnover number calculated for modied materials vs. amount of Rh present in the system was much higher (in the case of 0.0005% Rh-TiO 2 it was ca.5000 aer 6 h of irradiation), what seems to be better approximation for the studied systems since during irradiation with visible light Rh species have a crucial role in the yield of photoinduced processes.Moreover, turnover number estimated for highly concentrated samples vs. amount of Rh was much lower compared with "depleted" doped materials (in the case of 2% Rh-TiO 2 it was 0.18 aer 6 h of irradiation).Thus, materials with extremely low Rh concentration showing slightly lower visible light-induced photocatalytic activity compared with those with 4 Â 10 3 times higher concentration of Rh, are more protable with respect to the future applications.
Since in the samples with lower Rh concentration two types of dependent on each other photoinduced redox processes can be distinguished, the photocatalytic activity should result from the efficiency of both processes.
As it was shown in our previous paper, transient Rh 3+ /Rh 4+ ratio in irradiated samples in air depends strongly on the wavelength range of incident light. 25Red shi of the used range of visible light causes increase of the ratio (Fig. 4Sa in ESI †).Similar dependency was observed in a course of photocatalytic reaction, however in the presence of organic electron donor (AcH) in the system photoreduction of Rh 4+ was enhanced which was explained by improved consumption of holes generated within VB of titania (Fig. 4Sb in ESI †). 25 Nevertheless, if the range of light was limited to 440-550 nm signicant decrease of Rh 3+ /Rh 4+ was observed.Such changes indicated that photoreduction of Rh 4+ is one of the main photoinduced processes occurring not only in the initial step but also in a coarse of the photocatalytic process conducted under whole range of visible light.It was shown also, that decrease of Rh 4+ concentration during the photocatalytic reaction may be achieved by utilization of light characterized of proper wavelength ranges.Such procedure should improve the efficiency of over whole photocatalytic process through the elimination of recombination centers together with the creation of VB holes.
The evidence of the importance of Rh 4+ photoreduction process has been also shown in double beam photocatalytic activity measurements. 25Simultaneous irradiation of the 0.01% Rh-TiO 2 (650 C) material with two light ranges: 440-550 nminducing the electron transfer from Rh 3+ to titania CB, and 590-730 nminitiating the excitation of the electrons from the VB to Rh 4+ , gave synergetic effect which was not observed for bare titania (Fig. 5S in ESI †).This experiment indicated also that induction of the electron transfer from Rh 3+ to CB of titania is essential for the occurrence of the process of Rh 4+ photoreduction.Such coupling of these two processes is conrmed by  the results of action spectra and photoelectrochemical measurements performed for the applied potentials from the range 0.5-1.0V (vs.Ag/AgCl).The experiments showed that monochromatic light induces only the photoexcitation of Rh 3+ (Fig. 10, Fig. 11 le).Similar results of action spectra measurements were obtained for 1% Rh-doped SrTiO 3 . 16It was suggested that obtained dependency of the apparent quantum yield of photocatalytic reaction on the energy of monochromatic light proved the photosensitization mechanism based mainly of the Rh 3+ -CB transition.Considering occurrence of both dependent on each other photoinduced processes in a coarse of photocatalytic reaction the analysis of results of the experiments performed with monochromatic light should be careful.
According to obtained results it might be expected that materials with higher content of Rh containing higher concentration of Rh 3+ able to drive reversible redox processes will show higher photocatalytic activity under visible light.However, increase of Rh concentration caused progressive deactivation of the materials.It might be explained by two phenomena: hindrance of the absorption of titania within its band edge caused by overlaid Rh species located at the surface or negative impact of doped ions to charges mobility in titania CB and enhanced recombination. 22,23The results of photoelectrochemical measurements clearly showed that the titaniaoriginated photoactivity induced by light from the range of bandgap absorption decreased with Rh concentration, even in the case of extremely small amount of Rh (Fig. 10 le), what indicates validity of the latter assumption.What is more, the photoelectrochemical activity within the range of visible light resulting only from the efficiency of Rh 3+ excitation increased with Rh content.Similarly, the photocatalytic activity tests performed under the light from the range limited to 550-730 nm, under which bare titania shows negligible activity, showed improvement of photocatalytic activity of materials with an increase of Rh-concentration (Fig. 12).Both improvements were observed up to 0.1% and 0.01% of Rh, respectively.For higher concentrations, up to 1%, both titania-and Rh-originated activity decreased, as it might be seen in the Fig. 10 (le).
Materials containing more than 1% of Rh revealed different features.The action spectra measurements performed for 2% Rh-TiO 2 have shown that maximal efficiency of the photocatalytic process was obtained upon irradiation with 515 nm light (Fig. 11, right).Taking into account also a lack of the noticeable amount of Rh 3+ ions able to carry on reversible redox reactions and the results of XPS measurements, improvement of the visible light-induced photocatalytic activity of more concentrated materials should result from the activity of different type of surface Rh species compared with low concentrated samples.The signicance of surface character of the modication is proved by the high value of photocurrents in the spectral range of titania absorption resulting from low efficiency of recombination processes (Fig. 10 right), what suggests negligible concentration of Rh ions within the lattice of anatase.Thus, the formation of surface species seems to be favorable for particles with anatase crystalline structure, which is dominant in more concentrated samples.It is in agreement with previous studies showing that Rh-doping process is more preferential in the case of rutile than anatase. 12What is more, rutile phase seems to be essential for the creation of Rh 3+ ions able for reversible redox reactions.Low visible light-induced electrochemical activity observed for 2% Rh-TiO 2 compared with strongly improved photocatalytic activity in that light range suggests that created surface species may also partially cause an improvement of separation of photogenerated charges.However, the complicity of the composition of highly concentrated samples signicantly impedes mechanistic studies of observed processes and the exact mechanism of visible light-induced activity of highly concentrated materials was not fully examined.

Location of new electronic states on the scale of potentialsspectroelectrochemical measurements
Spectroelectrochemical measurements performed in a broad range of the applied potentials allowed to study the redox potential of Rh 3+ species in less concentrated samples.During the experiments samples were electrochemically oxidized and changes in the reectance at 625 nm, reecting the presence of Rh 4+ species, have been recorded.The potential of CB edge of studied materials have been studied by following the changes in the reectance at 780 nm during negative potential application.
The consecutive oxidation of the samples containing from 0.01 to 0.5% of Rh resulted in the increase of the absorption at 625 nm (Fig. 13, le).In the case of all samples potential ar.0.9-1.1 V vs. Ag/AgNO 3 electrode (depending on the amount of Rh),  The reference experiments in which spectra of 0.5% Rh-TiO 2 and bare titania have been recorded at 0.5, 1.0, and 1.5 V of applied potential clearly showed that appearing absorption originated from Rh 4+ species (Fig. 14 le).It resembles absorption which appeared during the irradiation of the Rh-modied samples with white light (Fig. 8

le).
The results of studies on the redox properties of titania (both modied and bare) indicate that the potential of conduction band edge is around À1.4 V vs. Ag/AgNO 3 electrode regardless of the type of the sample (Fig. 13, right).Comparison of the potentials of the conduction band edge, additional states created upon modication with Rh and optically determined band gap energies (ca.3.0 eV for all samples) draw to conclusion that acceptor electronic states are located within the band gap of titania (Fig. 15).

Location of photoactive species in low-concentrated samples
The location of Rh ions strongly depends on the Rhconcentration.The characterisation of materials containing extremely low amount of Rh was very hard due to limitations of used methods, thus the nature of the modication was deduced from the tendency observed in the case of more concentrated samples.
Used preparation method should have assured uniform distribution of the Rh species, even in the case of very low Rh contents.The lack of aggregation of Rh species during the preparation was indicated by the linear increase of the absorption band within visible range of light with Rh concentration observed for impregnated but not calcined samples with extremely low Rh concentration (Fig. 6S in ESI †).The STEM/ EDX analysis (Fig. 4) also proved homogenous distribution of Rh ions within the particles for more concentrated materials.What is more, the results of XRD measurements performed for materials with higher Rh concentrations indicated that the used method allowed to introduce Rh ions into the rutile lattice.Moreover, XPS analysis indicated enrichment of the surface area with Rh ions.Thus, it is concluded that Rh-species in lowconcentrated samples should strongly interact with titania lattice, however doping process may occur mainly in the surface area of the titania particles.It is expected that the depth of penetration of Rh-ions increases with Rh-content, thus for extremely low concentrated samples the ratio of Rh doped species located at the surface to the species located in the bulk should be the highest.Doping process was also indicated by the light induced absorption changes observed for studied materials.Such reversible redox processes controlled by light occurred only in the samples in which photogenerated Rh 4+ ions were stable, and were not observed in the case of anatasebased materials, where doping process is less favorable, nor in the case of materials prepared by mixing titania with rhodium(III) oxide.What is more, such changes were also not reported for titania graed with Rh 3+ ions, in which rhodium(III) oxide species are located at the surface of semiconductor. 20n the case of materials with extremely low Rh-concentration such surface located Rh-species appeared to be the main photoactive species responsible for the improvement of the   photocatalytic activity under visible light irradiation.Photoabsorption and photocatalytic activity measurements performed for these materials kept in water at 90 C, rinsed four times with water and dried at room temperature in air showed that such intensive rinsing caused slight depletion of the absorption bands in the visible range of light accompanied with large decrease of photocatalytic activity within same range of radiation (Fig. 7S in ESI †).The moderate changes in the intensity of the Rh-originated absorption band upon rinsing may indicate that other, less active, Rh species are doped more deeply in the bulk of titania.The inuence of the type of Rh species on titania properties is still under investigations.

Conclusions
The obtained Rh-modied titania showed visible-light activity towards oxidative decomposition of AcH.The most active materials appeared at 0.0005% and 2% of Rh concentrations.The obtained results indicate that lower Rh concentrations (below 0.5%) resulted in doped rutile particles, while higher (above 0.5%) gave mainly surface modied anatase particles.Due to the complicity of the composition of more concentrated samples and great advantage from economical point of view of extremely low doping ratio, the mechanistic studies were focused mainly on the low concentrated materials.
Low Rh concentration allowed creation of Rh species able to drive reversible photo-induced redox processes controlled by the range of light energies.Such species play crucial role in the unique photosensitization mechanism revealed for studied materials.It is shown that photosensitization mechanism suggested for materials with extremely low Rh content is based on two-step bandgap excitation, in which excitation of the electrons from Rh 3+ to the CB is followed by the light induced-electron transfer from the VB to photogenerated Rh 4+ .What is more, the mechanism assures not only generation of both electrons and holes within titania bands but also recovery of photoactive Rh 3+ ions during the photocatalytic process.
It was found that essential condition for the formation of such a Rh species that can induce photoinduced reversible redox reactions is rutile structure of host material.In the case of more concentrated doped materials (2%, 3% Rh-TiO 2 ) anataserutile transformation was retarded, which caused the formation of materials with different Rh species and photosensitization mechanism.
In the case of rutile-based materials increase of the amount of photoactive Rh species capable of reversible redox reactions could be achieved by an increase of nominal Rh content, which was conrmed by the photoabsorption, photocatalytic activity and photoelectrochemical measurements.The efficiency of the visible-light-induced Rh-originated activity increased with Rh content reaching the maximum at about 0.01-0.1% of Rh.However, increase of Rh concentration simultaneously caused enhancement of recombination and/or deterioration of charges mobility within titania bands.Introduction of extremely low amount of Rh allowed limit the amount of intrinsically present tetravalent rhodium (Rh 4+ ) species, which is believed to decrease charge mobility in titania lattice and promote recombination route in Rh-doped materials.It probably contributed also to the limitation of the doping area mainly to the surface of the semiconductor.
To obtain materials with higher photocatalytic activity induced by visible light it is necessary to increase the amount of photoactive Rh species with simultaneous inhibition of recombination processes.The preparation method which can lead to growth of concentration of active and homogenously spread Rh 3+ species able for reversible redox processes incorporated mainly into the surface of titania particles should cause improvement of the activity resulting from two-step bandgap excitation.Moreover, bringing doping area to the surface and subsurface of host material may reduce undesired consequence of bulk doping, such as accelerated recombination.Created Rh species, due to the creation of new electronic states within the bandgap resulting from the strong interaction with TiO 2 lattice could be called built-in redox mediators.
Synthetic procedures.Rh-TiO 2 materials were obtained by the impregnation of commercial TiO 2 with proper amount of RhCl 3 followed by calcination process.6 g of titania were suspended in 55 ml of Mili-Q water and proper amounts of RhCl 3 solution were added.Aer mixing components in the dark at room temperature for 1 h, the suspension was heated up (75-80 C) and mixed until almost complete evaporation of water.
Materials were dried at 120 C overnight and calcined in muffle furnace at 700 C for 3 h.Obtained materials contained form 0.0001 to 3% (molar percentage) of Rh.As a reference material bare TiO 2 (0% Rh/TiO 2 ) was prepared according to the same procedure, without the addition of RhCl 3 solution.

Characterization of the materials
Absorption properties.The diffuse reectance spectra were measured with UV-vis Jasco V-670 spectrophotometer.Samples were prepared in a form of pellets, and barium sulphate was used as a reference material.The obtained reectance spectra were converted to Kubelka-Munk function values, F(R N ), dened as F(R N ) ¼ (1 À R N ) 2 /2R N ; where R N stands for reectance.For determination of photoinduced absorption changes materials in a form of thin layers placed in spectroscopic cell were illuminated in quartz vessel with light of proper wavelength range.Aer a set time of irradiation the reectance spectra were recorded.In order to verify the stability of the changes the spectra were recorded several times in differed time gaps aer the irradiation.
Crystalline structure and XPS spectra.Crystal structure analysis was performed with Rigaku X-ray diffractometer equipped with PDXL soware for data analysis.XPS spectra were obtained using JEOL JPS-9010MC instrument and Spec-Surf soware.Carbon 1s signal was taken as a reference for the correction of the position of other signals.The deconvolution of the signals have been performed using SpecSurf and OriginPro soware.
BET measurements.Specic surface area of the materials has been determined by BET adsorption method using Quantachrome Autosorb-6.
SEM, TEM, and STEM/EDX measurements.SEM and TEM measurements were performed with JEOL JSM-7400F scanning and JEM-2100F transmittance electron microscopes, respectively.STEM observations were performed using high resolution analytical transmission electron microscope (FEI Tecnai Osiris) equipped with a X-FEG Schottky eld emitter (200 kV) and Super-X EDX (Energy Dispersive X-ray) windowless detector system with 4-sector silicon dri detector (SDD).The Z-contrast images were acquired using a High Angle Annular Dark Field (HAADF) detector in the scanning mode.The STEM images coupled with EDX elemental mapping were acquired with applied sample dri correction using Bruker Esprit soware in order to investigate the spatial distribution of the constituent elements within the sample.
Photocatalytic activity tests.Photocatalytic activity of obtained materials was studied in the reaction of oxidative decomposition of acetaldehyde (AcH) in gas phase.The 0.05 g of the material homogenously distributed in glass cell (dimensions of sample layer: 10 Â 10 Â 0.2 mm) was inserted in leak-proof reactor, and closed tight.The sample was preirradiated for 1 hour with UV-vis light using Xe lamp (150 W) as a source of light (the light intensity was ca.193 mW per 1 cm 2 ).Aerwards the gas phase inside the reactor was exchanged with the compressed air (50% of H 2 O; ow: 100 ml min À1 ; 15 min), and 0.08 ml of pure AcH was injected.The reactor was kept in the dark for ap.110 min for obtaining the equilibrium in adsorption process.The amount of AcH and CO 2 was measured before and during the irradiation with micro GC 3000A (Agilent Technologies).Sample was irradiated with visible light using the Xe lamp equipped with cut off and cut on lters (440, 470, 500, 570, and 590 nm), and IR lter (CF1600) (the light intensity in case of 440 nm cut off lter was ca.43 mW per 1 cm 2 ).The spectra of used light source equipped with various optical lters recorded with Photonic Multichannel Spectral Analyzer PMA-11 (Type No. C7473-36) consisting of thermoelectric-cooling type BT-CCD are shown in Fig. 8S in the ESI.† The values of light intensities estimated for each wavelength range are also provided.
Action spectra measurements.Action spectra were determined for the reaction of oxidative decomposition of AcH in air.Monochromatic light, with full-width at half-maximum intensity (FWHM) around 15 nm, was obtained using diffraction grating-type illuminator (JASCO, CRM-FD) equipped with 300 W Xe-lamp (Hamamatsu Photonics, C2578-02).The light intensity at each wavelength was measured with power meter (HIOKI 3664).Reaction was conducted in tight cylindrical quartz cuvettes with total volume around 16 ml.Samples were prepared by spreading 10 mg of studied material on glass plate and drying at 120 C for 1 h.Before injection of 10 ml of AcH, cuvette was purged with synthetic air (50% of H 2 O) for 1 min.Aer the injection cuvettes were kept in the dark for 1.5 h to reach adsorption equilibrium.The molar amount of CO 2 generated in photocatalytic process was measured by gas chromatograph (Shimadzu, GC-14 B) equipped with methanizer (Shimadzu, MTN-1) and FID.
Photoelectrochemical measurements.Photoelectrochemical measurements were performed in three-electrode set-up using 0.1 mol dm À3 KNO 3 water solution as an electrolyte.Working electrode was prepared by spreading an aqueous suspension of studied material on ITO-covered (indium-tin oxide) transparent foil dried aerwards in the stream of warm air (ca.40-50 C).Ag/AgCl and platinum were used as reference and auxiliary electrodes, respectively.The measurements were done using the electrochemical analyzer Autolab PGSTAT (NLAB).A 150 W Xe lamp (Instytut Fotonowy) equipped with automatically controlled monochromator with a shutter (Instytut Fotonowy) was used as a source of monochromatic light in the whole range of UV and visible light.Working electrodes were irradiated from the backside.Photocurrents are dened as a difference between currents recorded aer and before shutter opening.Spectroelectrochemical measurements.Spectroelectrochemical measurements were performed using UV-vis diffuse reectance spectroscopy combined with setup for electrochemical measurements. 31The electrochemical setup consisted of three-electrode cell, with platinum wire and Ag/AgNO 3 as a counter and reference electrodes, respectively.Potential of the silver electrode vs. SHE estimated on the basis of the measurements of ferrocene redox potential in 0.1 M LiClO 4 solution in acetonitrile amounted to 0.56 V. Working electrode consisted of platinum foil covered with studied materials (covered area: 1 cm Â 1 cm).The potential control was provided by the electrochemical analyzer Bio-Logic SP-150.The measurements were performed in Teon cuvette with a quartz window in 0.1 mol dm À3 LiClO 4 solution in acetonitrile.The oxygen was removed from the system by purging with argon 30 min before and during the experiments.During the measurement the changes in reectance at various wavelengths upon the applied changing potential (scan rate 0.5 mV s À1 ) have been recorded by Perkin Elmer UV-vis Lambda 12 spectrometer equipped with a 5 cm dia.integrating sphere.The UV-vis spectra of the materials have been recorded aer 20 min of application of desirable potential.

Fig. 7
Fig. 7 Diffuse reflectance spectra of Rh-TiO 2 with various Rhconcentration and a reference sample without Rh species.

Fig. 8
Fig. 8 Diffuse reflectance spectra and differential spectra (inside) in the representation of Kubelka Munk function (F(R)) of 0.01% Rh-TiO 2 material kept in the dark and irradiated with the light in the ranges: 440-730 nm, and 590-730 nm (left).Changes in the value of F(R) observed at 620 nm for Rh-TiO 2 materials with various Rh-concentration induced by the irradiation of the samples with light in the range 440-730 nm (right).The changes were calculated in respect to the F(R) values recorded for samples irradiated with low-energy light (590-730 nm).

Fig. 9
Fig. 9 Rates of CO 2 liberation in first two hours of the process of acetaldehyde oxidative decomposition in the presence of Rh-TiO 2 with various Rh concentrations irradiated with visible light (l > 440 nm).A horizontal line shows CO 2 -generation rate in the presence of 0% Rh-TiO 2 .

Fig. 10
Fig. 10 Photocurrent as a function of wavelength of incident light recorded in aerated 0.1 M KNO 3 for Rh-TiO 2 materials with various Rh-concentration (left and right).Applied potential: 0.95 V vs. Ag/AgCl.

Fig. 11
Fig. 11 Action spectra together with absorption spectra recorded for bare TiO 2 , and 0.001% Rh-TiO 2 (left), and 2% Rh-TiO 2 (right) in the reaction of oxidative decomposition of AcH in air.

Fig. 12
Fig. 12 Amounts of CO 2 and AcH monitored during process of AcH oxidative decomposition in the presence of Rh-TiO 2 with various Rhconcentration under irradiation with light of the range 550-730 nm.

Fig. 13
Fig. 13 Changes of KM function at 625 nm (left) and 780 nm (right) with the applied potential recorded for 0.5% Rh-TiO 2 (left) and 0.01% Rh-TiO 2 and bare TiO 2 (right) deposited at the surface of platinum plate.The measurements were carried out in deoxygenated 0.1 mol dm À3 LiClO 4 solution in acetonitrile.

Fig. 14
Fig. 14 Reflectance spectra in the representation of KM function recorded for 0.5% Rh-TiO 2 (left) and bare titania (right) deposited at the surface of platinum plate at the various applied potentials (E vs. Ag/ AgNO 3 ).The measurements were carried out in deoxygenated 0.1 mol dm À3 LiClO 4 solution in acetonitrile.

Fig. 15
Fig. 15 Scheme of the bands structure on the scale of potentials and possible light-induced electron transfer processes presented for Rh-TiO 2 .