Accelerated Photo‐Induced Degradation of Benzidine‐p‐Aminothiophenolate Immobilized at Light‐Enhancing TiO2 Nanotube Electrodes

Abstract Herein, the enhanced visible‐light‐induced degradation of the azo‐dye benzidine‐p‐aminothiophenolate immobilized on TiO2 nanotube electrodes is reported. Exploiting the reported photonic properties of the TiO2 support and the strong electronic absorption of the dye allowed for employing surface‐enhanced resonance Raman spectroscopy at 413 nm to simultaneously trigger the photoreaction and follow the time‐dependent decay process. Degradation rate constants of up to 25 s−1 were observed, which stand among the highest reported values for laser‐induced degradation of immobilized dyes on photonically active supports. Contrast experiments with two differently light‐enhancing TiO2 nanotube electrodes establish the direct correlation of the material's optical response, that is, electromagnetic field enhancement, on the interfacial photocatalytic reaction.


Introduction
Titanium dioxide,T iO 2 or titania, is considered one of the beststudied semiconductors to date. [1] Due to its extraordinary properties such as the high stability to chemicals and corrosion as well as the low toxicitya nd biocompatibility, thel ow-cost materialT iO 2 has becomei ndispensable for many technological applications.A sasemiconductor,T iO 2 exhibits aw ide band gap that can be resonantly excited using soft UV light.T he highly reducing and oxidizing potential of its conduction and valence band (CB and VB), respectively,h ave promoted the broad applicability of TiO 2 for hosting ar ange of light-driven reactions at its interface. [1,2] TiO 2 has been intensely investigated with respect to photocatalytic remediation of environmental pollutants. [3] Upon UV-light absorption, mobile charge carriers are generated in the TiO 2 that can reacta tt he interface to produce highly reactive molecular speciesc apable of decomposingt oxic organic molecules into nontoxic products. [3,4] Visible light-induceddegradation of pollutants at aT iO 2 interface proceeds withoutt he involvement of charge carriers from TiO 2 . [5] Here, light absorption by molecular pollutants at the interfacei sf ollowed by excitedstate electron transferi nto acceptor states in the TiO 2 ,u nlocking severalp athwaysf or the subsequent degradation of the absorbing species.
On the quest of improving the capability of TiO 2 to mediate light-induced degradations, particularly,t he ability of periodic TiO 2 nanostructures to form photonic crystals hasb een recognized as av ersatile tool to enhance photocatalysis at TiO 2 interfaces withoutt he involvement of auxiliarya toms. [6,7] TiO 2 structures with inverse-opal and nanotubularg eometry have been shown to afforde nhancedp hotocatalytic degradation of organic dyes upon visible-light irradiation. [8][9][10][11] The origin of the enhancementi st ypicallya scribed to the materials'p hotonic properties resulting from ap eriodic modification of the refractive index at the scale of the incident wavelengths. [9,12] In this way,l ight energy is "trapped" within the well-definedn anostructures forming photonic latticesand allowing for highly enhanced photon interaction (i.e.,a bsorption) with the material and associated molecules. Moreover, the formationo fn anostructures intrinsically affordsg reater active surface area. Anothera dvantage of such solid TiO 2 arrays is the ability to be readily electro-contacted and employed as an electrode to facilitate desiredi nterfacial charge-transfer reactions by applying an external bias. Furthermore, such arrays can be readily incorporated in technologically relevant photocatalytic flow cells and swiftly recycled througha nnealing, providing an edge compared with typicallye mployed nanoparticle suspensions. These qualities combinedm ake ordered nanostructured TiO 2 highly interesting for applications in photocatalytic conversion of environmental pollutants.
Recently,wer eported highly enhanced localized electromagnetic fieldsn ear the solid-liquid interface within anodizedT iO 2 nanotubular structures at 413 nm laser excitation. [13] In the presentw ork, we aim at investigating their contribution to photocatalytic degradation of organic dyes at anodized TiO 2 nanotubes (TiO 2 -NTs) interfaces.A sp hoto-active target, the azo-dye benzidine-p-aminothiophenol (BD-PATP), ak nown environmental pollutant and reported human carcinogen, was chosen as am odel dye. [14] Its degradation pathway is well understood [4,15] and resembles those of other dyes on TiO 2 [15] allowing for generalizing the findings to other environmentally relevant compounds. Exploitingt he strong electronic absorption of the dye, time-resolved surface-enhanced resonance Raman was used to concomitantly trigger and monitor the degradation of the azod ye, which has been prior immobilized on the TiO 2 interface. In contrast to typical performance-based studies, this study focuseso nt he interfacial degradation reaction and the influenceo fe nhanced electromagnetic fields on accelerating such reactions to shine light on relevant optical requirements for highly active photoactive supports.

Results and Discussion
The preparation of the dye-modified electrodes followed a three-step procedure as shown in Figure 1. Twod ifferent types of TiO 2 -NT electrodes were prepared following ar ecently published protocol. [13] Briefly,T iO 2 was obtained by anodizationo f Ti foils in fluoride-richm edia. Subsequent annealing at different temperatures afforded the different optical properties ( Figure 1, step 1). We showedp reviously that applyinga na nnealing temperature of 300 and 475 8Cr esultsi nT iO 2 -NT electrodes with lower and higher pronounced localized electromagnetic fields at the TiO 2 -NT-liquid interface, respectively. [13] In the presents tudy,s uch electrodes annealed at 300 and 475 8Cw ill be referred to as TiO 2 -NT j low and TiO 2 -NT j high, respectively.
In contrastt om any studies reporting photocatalytic degradation on TiO 2 materials, the target dye BD-PATPw as chemically attached to the TiO 2 surface. To achieve this, first am onolayer of para-aminothiophenol (PATP) was adsorbed ( Figure 1, step 2) on the TiO 2 -NT surface by incubation of the electrode in a2m m ethanolics olution.P ATP-modified electrodes are denoted as TiO 2 -NT j low j PATP and TiO 2 -NT j high j PATP.T he presence of PATP on TiO 2 -NT j high after thorough rinsing with abundant ethanol was verifiedb ys urface-enhanced Raman (SER) spectroscopy ( Figure 2). BD-PATPd ye formationw as achieved by inducing an azo-coupling reactionf ollowing reported procedures ( Figure 1, step 3). [16] BD-PATPe xhibits as trong electronic absorption due to extendedd elocalization of p electrons affording strong resonance Raman (RR) signals of the dyea t4 13 nm laser excitation within the photonic band gap of the anodized TiO 2 -NTs ( Figure 2). [13,17] Formation of BD-PATPw as therefore readily confirmed by surface-enhanced resonance Raman (SERR) spectroscopy. The resulting dye-modified electrodes are denoted as TiO 2 -NT j low j BD-PATPa nd TiO 2 -NT j high j BD-PATP. Importantly, the significantly higher signal intensity of the BD-PATPo ver PATP made it possible to perform time-dependent SERR studies.
In the employed experimentalconfiguration, the laser excitation served two purposes. On the one hand, it allowedf or sen-  sitive detection of the dye molecule on the electrode. On the other hand, the laser concomitantly induced dye degradation by resonantly exciting the molecule. [5] Upon exposing TiO 2 -NT j high j BD-PATPa nd TiO 2 -NT j low j BD-PATPt ol aser light at 413 nm in aqueousm edia, at ime-dependent decrease in SERR intensity was observed for both systems as expected for photo-induced dye degradation (Figure 3a,F igure S1, Supporting Information).
SERR analysiso ft he mediuma fter induced photodegradation by drop-casting of the surrounding water medium on a roughened Ag electrode ( Figure S2, detailsi nt he Supporting Information) showed bands that rule out the presence of the parentB D-PATP, indicating that ap hoto-induced conversion of BD-PATPh as occurred. The observed bandsm ost likely originate from degradation products( Figure S3). In this respect, a possible degradation product of BD-PATPi sb enzidine, which on TiO 2 may also have converted to higher-oxidized biphenyl species. [18] Am ere photo-induced desorption of the dye can be ruled out because in this case BD-PATPs hould have been detected in the SERR experiment on the roughened Ag electrode. Due to the immobilization of the dye onto the TiO 2 -NT surface, the overall dyec oncentrationw as too low to perform in-depth product analysis, which is also beyond the scope of this paper. The capability of TiO 2 -NT to promote dye-degradation reactions has already been demonstrated in the literature. [12] Noteworthy,n od ecreasei nS ERR intensity was observed when BD-PATPo nT iO 2 -NT was exposed to laser light that is off-resonant with the dye'se lectronic absorption, that is, at 647 nm ( Figure S4). This observation supports that the photodegradation mechanism requires excitation with light of energy sufficient to induce at ransition from the ground to the excited state of the immobilizedd ye. Importantly,b oth wavelengths are too lowi ne nergy to excite the band gap of TiO 2 . Furthermore, excitingB D-PATPi mmobilizedo nr oughened Ag electrodes with a4 13 nm laser showedn oh ints for photo-induced degradation indicating that the TiO 2 support is required to promote the reaction( Figure S5).
From the time-dependent decrease of the SERR intensity, photo-induced degradation kinetics at 413 nm excitation were derived. Note that we report the Molecular Exposure Time (MET) insteado ft he mere spectrala ccumulationt ime as the samples were moved during measurement to avoid thermal degradation (see the Experimental Section for details). Hence, the amount of time each molecule was exposed was significantly less than the actual experimental/spectral accumulation time. The intensity of the most prominent peak at 1599 cm À1 assigned to phenyl-ring stretching vibration 8a/8b mode/s (Wilson notation) [20] was normalized to the initial intensity,t hat is, t = 0a nd plotteda safunction of MET ( Figure 3b). Here, t = 0r efers to the initial intensity of the first averages pectrum (of 10 s, 10 times) obtained for the dye right at the moment it was exposed to laser.L ocated 11 cm À1 away from the nearestP ATP on TiO 2 -NT peak (1588cm À1 ), this peak can be considered characteristic for BD-PATPo nT iO 2 -NT.F itting an exponential function yielded the apparent decay constants k app (= 1 = t app )s ummarized in Ta ble 1.
It should be noted that the decreaseo ft he 1599 cm À1 band reflects the decay of the parentd ye-state over time. Thus, the kinetics we provides houldb ec onsidered as ac onvolution of all kineticp rocesses that in sum lead to irreversible decay of the parent state. Importantly,t he kinetics reflect only the relevant interfacial degradation process. On average, the decay rate constant observed for TiO 2 -NT j high j BD-PATP( 5.1 (1.4) s À1 ) was found to be approximatively 70 %f aster than for TiO 2 -NT j low j BD-PATP( 3.0 (0.6) s À1 ).
To investigate the effects of electrode polarization on the degradation kinetics, potentials were applied to the TiO 2 -NT electrodes in 0.1 m phosphate-buffered mediuma tp H7.Adistinctly different performance of the two employed materials was found. Application of À0.4 V Ag/AgCl revealed rate constants for both systems comparable to those values obtained in water at open-circuit potential( OCP) before.A lso, in this case The curved arrowi ndicatest he change in intensity of the peak at % 1599 cm À1 over time, which is used to determine the kinetics. MolecularE xposure Time denotest he time the laser was irradiating the sample. b) Normalized intensity of the % 1599 cm À1 peak derived from the SERR spectraplotteda gainst molecular exposuretime for the different TiO 2 -NT-dye systems at different appliedp otentials. An exponential fit (black and blue lines) was used to determine the decay rate constants. Data derived from SERR spectrarecorded on the respective electrodes. Conditions:1 0saccumulation time, average of 10 spectra;l aser excitation:1mW at 413 nm;sample immersedin0 .1 m phosphate buffered medium,p H7. TiO 2 -NT j high j BD-PATP( 3.6 s À1 )e xhibited higherr ates, which were averagely2 .8 times highert han those found for TiO 2 -NT j low j BD-PATP( 1.3 s À1 ). However,adesorption of the dye at À0.4 V Ag/AgCl was noted in the absence of laser irradiation. This effect was observed for both TiO 2 -NT j high j BD-PATPa nd TiO 2 -NT j low j BD-PATPs ystems and was studied in as eparate experiment on TiO 2 -NT j high j BD-PATPi nm ore detail (Supporting Information, Section2 and FigureS6). The potential-induced desorption was found to scale with the time keeping the TiO 2electrodes at negative bias, thus, pointingt oarole of the cathodic potentiali np romoting desorption. As such, k app values determineda tÀ0.4 V Ag/AgCl are probably ac onvolution of desorption and degradation kinetics.
In contrast, no desorption effects were observed at 0.4 V Ag/ AgCl butasignificant increase of the rate constant was noted compared with OCP conditions. Again, forT iO 2 -NT j high j BD-PATP (25.5 (5) s À1 )afaster rate constant by af actor of about 2.5 was observed. The presented degradation rate constants are by several orders of magnitude highert han values typicallyr eported in the literature for dye degradation mediated by TiO 2 nanostructures (usually in the order of 10 À8 to 10 À2 s À1 )( selection presented in Ta ble S1, Supporting Information). [15] This difference, however, is likely relatedt ot he fact that different systemsa re comparedc ompromising af air assessment. In fact, usually employed experimental conditions involveT iO 2 suspensions, a soluble dye, and light irradiation at various powers. In most cases, the kinetics were derived by measuring the dye concentration in solution as af unctiono ft ime. Although such configuration is closer to ar eal application,t he derived reaction kinetics will differ and also reflects low diffusionalp rocesses, for example, of the dye to or of the product away from the electrode. This being said, the TiO 2 -NT electrodes presented here surpass directly comparables ystems, that is, systemsi nw hich heterogeneous TiO 2 nanostructured supports, laser light as excitation source, and an immobilized dyew ere used. Toumazatou et al. reported degradation of methylene blue immobilized on photonically active inverse-opal TiO 2 structures under constant 514 nm laser irradiation with ar ate constant of 0.112 min À1 ,w hich is significantly lower than the reported value here even in the absence of applied potential and in aqueous conditions. [9] Latter is important because it is expected that the absence of solvent will also promote thermally induced degradations. To the best of our knowledge,n oo ther system exists that would allow for afair comparison.
The observed overall high capability of the presented TiO 2 -NT electrodes to promote photo-induced degradation can be attributed to their optical properties. As am atter of fact, improvedp hotocatalytic activity through photonic structures found for TiO 2 -NT systemsh as been demonstrated in the literature albeit not with this magnitude. [7,9] The higherperformance of TiO 2 -NT j high over TiO 2 -NT j low,u nder all measured conditions, points to this material's specifics uperior optical properties. This can be understood by considering the stronger electromagnetic fielde nhancement provided by the TiO 2 j high electrode as recently shown by us. [13] In our previousw ork, Ramans ignal enhancement of probe moleculesa tt he TiO 2 -NT interface was used to investigate the field enhancement. An approximately 7times higher Raman signal enhancement on TiO 2 j high than on TiO 2 j low was noted. This translatest oa 2.6 times highere lectromagnetic fields trength (square root of 7) and, in fact, matches nicely the factor of rate-constant increase found for TiO 2 -NT j high under all measured conditions (Table 1). Thus, the resultsindicateadirect correlation between the electromagnetic-field enhancement of as upport and its photo-induced degradation capability of nanostructures. This establishes in fact af acile and elegant wayt oa ssess am aterial's capability to promote interfaciald egradation reactions through its Raman enhancement factor.
The observed potential dependenceo ft he degradation rate constantss uggestsa ne lectronic influence. In this context, it has already been shownt hat applying an egative bias results in filling-up acceptors states in TiO 2 hindering the electron injectiona nd enhancingr ecombination processes. [21] Moreover,i t could potentially favor breaking of the TiÀSb onds and hence lead to (reductive) desorption of the dye as ac ompetings ide reaction. Such behavior has been observed for thiol-bound compounds on metal electrodes. [22] Importantly,a no verall decreasedp hotocatalytic degradation activity should be noted in this case. Both are in fact observed in our experiments if we consider the rate constants derived at À0.4 V Ag/AgCl as ar ough estimation (convolutiono fd esorption and degradation)a nd comparet hem to those obtained at 0.4 V Ag/AgCl (Table 1). Conversely, applying + +0.4 Vc ould accelerate the degradation as a result of potentially three effects.F irst, electron-injection kinetics could be enhanced due to the additional driving force as well as the availability of more unoccupied acceptor states. [21] Second, the positivep otential could assist in transporting the injected electron into the bulk away from the oxidized dye, which would enhancec harge separation efficiency.F inally,t he positive potential could afford increased adsorption of negatively charged PO 4 2À ions along with removal of K + + ions at the TiO 2 -NT interface. This corresponds to the formation of am ore negativelyc harged Helmholtz layer resulting in an interfacial electric field pointingt ot he interface. This electric field will enhance injection kinetics as well as decrease recombination rates as also discussed for cation/electric field dependence in DSSCs. [23] Conclusions We presentedt he applicationo fT iO 2 nanotube electrodes for hostingt he photocatalytic degradation of the azo-dye BD-PATP.T he extraordinary opticalp roperties of the presented TiO 2 nanotube system afforded highest degradation rates in the s À1 region under direct laser illumination in aqueous media. The degradation activity was found to be directly linked to the Ramane nhancement factor and therefore attributed to the material's ability to accommodatee nhanced electromagnetic fields within its nanostructure. The performance could be furthermore significantly increased by applying ap ositive bias. Our study demonstrates the benefit of applying surface-enhanced Raman spectroscopy to investigate interfacial dye degradation on photonic materials focusing thereby only on the important interfacial reactions. Moreover,t he approach allows for readily assessing any material's photo-induced degradationc apability at its surface through the afforded Raman enhancement factor.T he derived knowledgei so fp articular importance for rationally designing andd eveloping novel highly active photocatalysts. Further studies will focus on the activity of the TiO 2 nanotube electrodes at other laser excitation wavelengths as well as towardso ther environmentally relevant dyes.

Experimental Section
Sample preparation: The main procedure for the preparation of the sample for SERR measurement is given in Figure 1. The formation of TiO 2 by anodization is given in the literature. [24] Briefly,ap otential of 20 V( Hameg, HMP2020;w ith Ag/AgCl, 3 m KCl, leak-free as reference and Pt sheet as counter electrode) was applied on Ti foil (99.6 %, Chempur) in the presence of 0.202 m NH 4 F( 99.99 %, Sigma-Aldrich) solution in 1:1g lycerol (! 99.5 %, Sigma-Aldrich)/ water for 2h.A fter anodization, the electrode was rinsed with copious amounts of deionized water and dried under nitrogen stream. The TiO 2 -NT substrates were then annealed at 300 (TiO 2 -NT j low) or 475 8C( TiO 2 -NT j high) for 2h in aN abertherm oven to improve its crystallinity.T he ramp-up was done at 15 8Cmin À1 until the desired temperature. p-aminothiophenol, PATP (97 %, Sigma-Aldrich), was then deposited on TiO 2 -NT by immersing the electrode in 2mm ethanolic (Analytical Grade, 99.9 %, Fischer Scientific Company,G ermany) solution of the thiol and allowing self-assembled monolayers (SAMs) to form. Deionized water (resistance > 18 MW,M illipore, Eschborn, Germany) was used for washing and solution preparation.
As imple diazotization reaction adapted from Han et al. [16] was then conducted in an ice bath to yield benzidine-p-aminothiophenol, BD-PATPf rom PATP on TiO 2 -NT.B riefly,f resh 5% sodium nitrite (99.999 %, Sigma-Aldrich) solution was prepared and, together with 0.1 m hydrochloric acid solution (ACS reagent, 37 %, Sigma-Aldrich), added into the vial containing the TiO 2 -NT electrode. After 10 min, the solution was removed and aqueous benzidine (! 98 %, Sigma-Aldrich) solution (! 2mm)w as added into the TiO 2 -NT elec-trode. The reaction was stopped after 10 min by harvesting the functionalized electrode, washing it with copious amounts of water,a nd gently drying with as tream of nitrogen gas. This sample was then used for SERR measurements.
SER/SERR measurement: The functionalized TiO 2 NT electrodes were assembled in as pectro-electrochemical cell (Pt wire counter and Ag/AgCl, 3 m KCl reference electrodes) and % 4mLo fw ater or 0.1 m phosphate (from K 2 HPO 4 and KH 2 PO 4 (! 99 %, Merck)), pH 7.0, buffer solution was added. The liquid serves as ah eat sink to prevent thermal effects in the degradation procedure during laser exposure. In addition, thermal degradation was prevented by rotating the sample using an XY stage (OWIS GmbH, Germany). To prevent defocusing or shifting of position during measurement, a custom-built holder provides mechanical stability of the cell during rotation. SER/SERR measurements were performed using Kr + + laser at a4 13 nm excitation (Coherent Innova 300c) coupled to ac onfocal Raman spectrometer (Jobin-Yvon, LabRam 800 HR) with a back-illuminated CCD detector cooled by liquid N 2 .T he laser was aligned and then focused on the sample using aN ikon 20 objective (NA = 0.35, WD = 20 mm) at al aser power of % 1mW. The spectral resolution was about 1.2 cm À1 .S pectra were calibrated with respect to mercury lines and the Raman spectrum of toluene.
Time-dependent SERR measurements were performed for the degradation experiments. This entailed recording the SERR spectra in intervals of % 100 s( as an average of 10 spectra, with each spectrum accumulated for 10 s) for ad uration of at least 30 min experimental time. In potential-dependent measurements, ap otentiostat (MetrOhm) was used to apply an external potential of + +0.4 or À0.4 V( vs. Ag/AgCl, 3 m KCl).
Measurement at off-resonance was performed at 647 nm using a confocal Raman spectrometer (LabRam HR-800 Jobin Yvon) equipped with liquid N 2 -cooled CCD camera (Symphony II Horiba). Spectral analysis for background subtraction included polynomial function of the spectra. Further intensity analysis was performed using Origin.
Calculation of the Molecular Exposure Time (MET): The exposure time axis was expressed as Molecular Exposure Time (MET) to account for the the mechanical rotation of the electrode during measurements, thereby resulting to intermittent illumination of the sample at each traversed spot. Electrode rotation was employed to prevent thermal degradation as ar esult of prolonged laser exposure and to allow the molecules in the sample spot to recover thermally or prevent heating up significantly.T his was done by constantly moving the sample to yield ac ircle with a radius, r = 1mmi nthis case, that is illuminated by the laser.C onsidering that, at at ime, only af raction of the circumference was actually exposed to laser ( Figure S7, Supporting Information), and that this fraction is approximately equal to the laser spot diameter, D: [25] D ¼ð1:22 lÞ=ðN:A:Þð 1Þ in which l and N.A. refer to the laser-line excitation wavelength (i.e.,4 13 nm) and numerical aperture (i.e.,0 .35), respectively,o ne can convert the apparent experimental measurement time, t exp ,t o give the time in which each molecule was actually exposed to laser by expressing them in ratios of the distance and time of the laser path with respect to the circumference so that: Determination of apparent rate constant k app : After obtaining the SERR spectra, the intensities of the most prominent peak (1599 cm À1 )w ere divided by the initial intensity (t = 0) to express the peaks in values ! 1a nd easily see the succeeding amounts as fractions of the initial intensity.T he values were not normalized from 0t o1because the dye was still not fully degraded within the set experimental time ( % 30 min experimental time per measurement). The intensities of this peak ( % 1599 cm À1 )w ere then plotted against MET.Af it of an exponential decay function afforded the decay constant.