Novel nanographene/porphyrin hybrids – preparation, characterization, and application in solar energy conversion schemes

Four novel nanographene/porphyrin hybrids were prepared, characterized, and probed in solar energy conversion schemes. Exfoliation of graphite by means of immobilizing four different porphyrins onto the basal plane of graphene is accompanied by distinct electronic interactions in both the ground and the excited states. In the ground state, a strong loss in oscillator strength goes hand-in-hand with a notable broadening of the porphyrin transitions and, as such, attests to the shift of electron density from the electron donating porphyrins to nanographene. In the excited state, a nearly quantitative quenching of the porphyrin fluorescence is indicative of full charge transfer. The latter is corroborated by femtosecond transient absorption measurements, which reveal the generation of the one-electron oxidized radical cation of the porphyrins with absorption maxima at 490 and 625 nm in the visible region and conduction band electrons in nanographene with features at 890 and 1025 nm in the near infrared region. We have demonstrated the applicability of the new nanographene/porphyrin hybrids in, for example, solar cells. In this regard, the presence of flakes is crucial in terms of influencing the injection processes, preventing aggregation, and reducing recombination losses, which are commonly encountered in porphyrin-based DSSCs.


Introduction
Nature chose carbon to provide the basis for life on earth. As a matter of fact, carbon is the key to many technological applications, which over the years have become indispensable in 25 our daily life and have influenced the world's civilization for centuries. Notably, the structural diversity of organic building blocks results in sheer endless chemical and physical properties. Altering the periodic binding motifs in sp 3 -, sp 2 -, and sphybridized networks represents the toolbox for constructing a 30 wide palette of carbon allotropes. To this end, the past two decades have served as a test-bed for probing the physicochemical properties of nanocarbons in reduced dimensions starting with the advent of 0D fullerenes, followed by 1D carbon nanotubes (CNTs) and by 2D graphene. 1-5 35 Turning to 2D graphene, single layers of graphene were first prepared successfully in 2004 by mechanical exfoliation of graphite using Scotch Tape. 6 Alternative fabrication strategiesepitaxial growth 7 and solubilization from bulk graphite 8 -have recently been demonstrated and they are, indeed, paving the way 40 to systematic experiments and technological applications. 9 At room temperature, charge transport measurements show that a flat single layer graphene is a zero-gap semiconductor and exhibits a remarkably high electron mobility with values exceeding values from 2000 (CVD 10 ) to 200000 (suspended 11 ) 45 cm 2 / V s. The symmetry of the experimentally measured conductance indicates high mobility not only for electrons but also for holes. In addition, an ideal monolayer has an optical transmittance of 97.7%. 12 The aforementioned calls for its implementation into transparent conducting electrodes as a viable 50 alternative to indium tin oxide. 13 Graphite is the most desired starting material en-route towards fabricating graphene. Going from graphite to graphene requires, however, an activation step owing to the thermodynamic stabilization of the earlier. The activation might come in the form 55 of oxidizing graphite to graphitic acid evolved providing valuable insights into the structure and constitution of graphite. 14 At basic pH, exfoliation results in monolayers of graphene oxide (GO). 14 Nevertheless, the final step, namely reduction of GO monolayers, fails to completely remove the introduced oxygen atoms and 60 reduction leads irreversibly to partially amorphous carbon. 15 Not surprising, the correspondingly reduced GO (rGO) features properties different than those predicted for real graphene. Another form of activation involves the intercalation of graphite with alkali metals. 16 Dissolving the resulting negatively charged 65 graphene sheets in solution, enables solutions of bi-and trilayer graphene. 17 To avoid any of the aforementioned pre-treatments, but still to activate graphite requires ultrasound as a chemical free treatment in liquids. 8,18 A common denominator is a suitable amphiphilic intercalator and / or solvent that guarantees the 70 efficient exfoliation. 19 As such, the versatility lies in the use of water and / or organic solvents to exfoliate graphite and / or to directly functionalize nanographene sheets by a variety of noncovalent means. 4 Nevertheless, recent results have documented that charge transfer emerges as a necessity to harvest the full potential of single layer graphene by tailoring the electronic properties for new complex 5 electronics. 20,21 To this day, only a few scattered reports are known that come, however, short to provide a coherent picture. 20,[22][23][24] Particularly relevant are advances on self-ordering / selfassembling [25][26][27][28] , on one hand, and π-stacking 29,30 as well as charge transfer interactions 30,31 , on the other hand, to yield high- 10 quality single layer flakes of nanographene that are stable and dispersable. With the help of recent studies a number of challenges in studying nanographene have been unveiled. Clearly, the lack of spectroscopic signatures, which render spectroscopic assignments 15 quite ambiguous -similar to early contributions to the field of CNTs -stands out among these challenges. In preliminary investigations, we have realized high-quality flakes of nanographene, which were, however, few layer graphene rather than single layer, by means of solution processing graphite with 20 tightly interacting porphycenes, phthalocyanines and porphyrins. 29,[32][33][34] In terms of spectroscopy, we were able to establish spectroscopic as well as kinetic evidences not only for ground state interactions but also for interactions in the excited state. In the excited state, radical ion pair states are formed 25 including the radical cation of the electron donating phthalocyanines and / or porphyrins and electrons that are delocalized within the basal plane of graphene. As a matter of fact, planar, aromatic macrocycles are highly versatile to exfoliate graphite and to afford functional nanographene based charge 30 transfer hybrids. 4,29,35 In the current work, we report on the formation and characterization of novel nanographene / porphyrin hybrids and their implementation into solar cells. Particular emphasis was placed on a top-down preparation of stable dispersions -starting 35 from natural graphite rather than graphene oxide -whilst preserving the intrinsic properties of graphene. To the best of our knowledge, the benefits of introducing such hybrid materials in DSSCs have rarely been established to date. 40

Materials
Natural graphite was purchased from Graphit Kropfmühl AG. Spectroscopic quality solvents tetrahydrofuran (THF) and toluene were ordered from Sigma-Aldrich and were used without further 45 purification. The 5,10,15,20-tetrabutylporphyrin and its copper (CuTBP) complex were synthesized according to literature procedures. 36 The other reagents were obtained commercially. Phosphorous oxychloride (4.0 mL, 43 mmol) was added dropwise into N-methylformanilide (6.9 mL, 56 mmol) and the resulting orange solution was stirred at room temperature for 10 minutes until it solidified. A solution of CuTBP 1 (0.344 g, 0.58 mmol) in chloroform (50 mL) was added and the resulting red mixture 55 was stirred at 60 °C for 1.5 h. The mixture was poured into cold water (200 mL), the organic layer separated, washed with water then diluted ammonia, dried over magnesium sulphate, and evaporated to dryness under reduced pressure at 50 °C . The remaining crude product was dissolved in a minimal amount of 60 chloroform and filtered through a pad of silica. The solvent was removed under reduced pressure at 50 °C and the resulting dark oil treated with methanol to give 2 (0. 24  base porphyrin aldehyde 3 (0.73 g, 1.30 mmol) and 18-crown-6 (52 mg, 0.2 mmol) were dissolved in benzene (80 mL), brought to reflux then (1,3-dioxolan-2-ylmethyl)triphenylphosphonium bromide (0.56 g, 1.30 mmol) with anhydrous, finely-powdered, potassium carbonate (0.18 g, 1.30 mmol) (3x) was added every 2 95 h for 6 h. Afterwards the reaction mixture was cooled down and filtered through a pad of silica using chloroform as eluent. The solvents were removed under reduced pressure at 50 °C and the remaining solid was dissolved in a mixture of tetrahydrofuran (50 mL) and chloroform (13 mL). Concentrated hydrochloric acid 100 (6.5 mL) was added and the resulting green mixture was vigorously stirred for 45 minutes. The acid was neutralized with concentrated ammonia and the organic phase after dilution with dichloromethane (100 mL) was separated, dried over magnesium sulphate, and evaporated to dryness under reduced pressure at 50 105 °C . The crude product was dissolved in a minimal amount of dichloromethane, filtered through a pad of silica, and evaporated to dryness under reduced pressure at 50 °C . The product was recrystallized from dichloromethane-methanol to give the allyl aldehyde 4 (0.61 g, 80 %) as a dark purple powder. 1 30 ammonium acetate (0.11 g, 1.38 mmol) were dissolved in a mixture of tetrahydrofuran (7 mL) and glacial acetic acid (7 mL). The resulting mixture was stirred at 65 °C for 1.5 h then quenched by cold water (30 mL). The dark powder was filtered off, washed several times with water and dried under vacuum at 60 °C to give 35 the zinc acid 6a (0.15 g, 88 %) as a dark greenish powder. 1   23 mmol), malonic acid (0.14 g, 1.38 mmol), and ammonium acetate (0.11 g, 1.38 mmol) were dissolved in a mixture of tetrahydrofuran (7 mL) and glacial acetic acid (7 mL). The 50 resulting mixture was stirred at 65 °C for 1.5 h and quenched by cold water (30 mL). The dark powder was filtered off, washed several times with water, and finally dried under vacuum at 60 °C to give the zinc diacid 6b (0.13 g, 77 %) as a dark green powder. 1

Preparation of nanographene hybrids
Natural graphite (1 mg) was added to a 2 x 10 -5 M stock solution (3 mL) of 6a/b and 7a/b in THF or THF/water mixtures (1:1, 1:10, or 1:100 v/v) and subsequently ultrasonicated for 10 minutes with a Bransonic 52 from BRANSON at room 10 temperature with 37 kHz and an effective power of 330 W. The considered solvents were THF and THF/water mixture. Next, graphite (1 mg) was added again and the dispersion was kept under ultrasonic treatment for an additional 10 minutes. Afterwards, the heterogeneous dispersion was centrifuged for 15 15 min at 500 rpm with a Fresco 21 centrifuge from Thermo Scientific at room temperature. The steps were repeated two times with the resulting supernatant. The final step included centrifugation at 5000 rpm for 15 min to remove any larger particles of graphite. 20

Physico-chemical characterization
Steady-state absorption spectra were recorded with a Perkin-Elmer Lambda 35. Steady-state emission spectra were recorded with a Fluoromax-P-spectrometer from HORIBA Jobin Yvon. Time-correlated single photon counting (TCSPC) spectra were 25 taken with a Fluorolog system (HORIBA Jobin Yvon). Signal acquisition was gathered by a Hamamatsu MCP photomultiplier (type R3809U-50). The time profiles were recorded at the emission maxima. All samples were measured in a quartz glass cuvette with a width of 10 mm. Femtosecond transient 30 absorption spectra were obtained with a Ti:sapphire laser system CPA-2101 (Clark-MXR Inc.) in combination with a Helios TAPPS detection unit from Ultrafast Inc. The initial laser excitation wavelength was 775 nm with a pulse width of 150 fs. The excitation wavelength employed was 387 nm, which was 35 generated with a SHG crystal. For the generation of the white light, a sapphire crystal of adequate thickness was used. The chirp-effect between 420 and 770 nm was approximately 350 fs.
The detection was carried out with two CCD cameras, each for a specific measuring range. The spectral windows were therefore 40 415 to 770 nm and 770 to 1600 nm. The delay line allowed spectral acquisition up to time delays of 8000 ps. All samples were measured in a fused quartz glass cuvette with a width of 2 mm. Data acquisition was done with the software HELIOS Visible/nIR (Newport / Ultrafast Systems). 45 Raman measurements were carried out with a LabRAM ARAMIS Raman-spectrometer from HORIBA Jobin Yvon that was equipped with a confocal microscope and an automated XYZ-

Results and Discussion
110 Scheme 1 displays the structures of porphyrins 6a/b and 7a/b, which were used in this work to realize the multifunctional nanographene hybrids. Porphyrins 6a/b and 7a/b have several key features. Firstly, the hydrophobic nature of the porphyrin core is of utmost importance for the successful exfoliation of graphite to yield stable suspensions of nanographene hybrids. Secondly, the conjugated β-pyrrolic side chain bears one or two carboxylic groups that facilitate grafting of TiO 2 nanoparticles 5 and the integration of the hybrids into the TiO 2 electrodes. Additionally, the cyano group or corresponding carboxylic acid group is electron withdrawing and assists in electron transfer to the TiO 2 .
The general features of the syntheses of porphyrins 6a/b and 7a/b 10 have been previously described for the synthesis of 7a. 38 As indicated in Scheme 1, the porphyrins were synthesized in an analogous fashion to that which we have previously used for tetraarylporphyrin dyes. 37, 39 The Vilsmeier formylation of tetraalkylporphyrins has been reported previously albeit in very 15 low yields. 40,41 We also found that using the DMF-POCl 3 formylation complex at 80-90 °C produces mostly decomposition products. Shortening the reaction time to 50 min gave 2 in low yield (<20 %) with recovery of some unreacted 1, whilst lowering the reaction temperature below 70 °C did not give appreciable 20 product. However, replacing DMF with the more active Nmethylformanilide allowed us to decrease the reaction temperature to 60 °C yet still provide 2 in satisfactory yield (67 %) with no starting material remaining. It is notable that a much larger excess (x100) of the Vilsmeier reagent is required than 25 normally used for porphyrin formylation. Porphyrin demetallation reactions usually require strong acids such as conc. sulfuric acid. 37 Usage of such harsh conditions decreases the yield of the free base tetraalkylporphyrin. 41 As a result, we have utilized here the milder procedure described by 30 Ponomarev et al. 42 A cooled solution of porphyrin 2 dissolved in a small amount of POCl 3 is treated with water (10 % by volume). After a few minutes a vigorous reaction takes place and a large quantity of hydrogen chloride is evolved. A high yield of the resulting free base porphyrin 3 is thus obtained. The reaction is 35 so vigorous that using more than 300 mg of the starting materials is not advised. In addition, using larger amounts of porphyrin only results in partial demetallation. The allylaldehyde derivative 4 was synthesized by a modified procedure of Ishkov et al. 43 Porphyrin aldehyde was reacted with 40 the ylide generated in situ from (1,3-dioxolan-2ylmethyl)triphenylphosphonium bromide, using crown ether as a phase transfer catalyst and potassium carbonate as a base. Since we found that the ylide was not stable under the reaction conditions, the addition of an excess of the phosphonium salt and 45 the base had to be carried out every 2 hours for 6 hours. Interestingly, the attempts to replace benzene with the less toxic toluene did not give satisfactory results. The reaction in boiling toluene gave lower yields due to the apparent decomposition of the starting materials. Lowering the temperature of the reaction 50 in toluene to 90 °C did result in some product formation but the reaction could not be driven to completion. The product of this condensation is a mixture of E and Z isomers of the protected aldehyde 4. It was previously reported that attempts to purify a similar mixture on silica gel resulted in 55 partial deprotection of the porphyrins. 43 As a result we carried out deprotection and isomerisation on the crude Wittig reaction mixture, which gave, after purification, the porphyrin 4 in high yield (80 %). The zinc insertion to give 5 was achieved by a standard procedure 60 with almost quantitative yield. 37 The resulting zinc complex 5 and free base 4 were then subjected to Knoevenagel condensation with malonic and cyanoacetic acids to give the corresponding porphyrin acids 6a/b and 7a/b in excellent yields. We found that the reactivity of the tetraalkyl derivatives after formylation did 65 not differ significantly from their tetraaryl equivalents. 37 To determine the interactions of the porphyrins with graphene, we decided to firstly elucidate the ground and excited state features of 6a/b and 7a/b based on steady-state absorption and fluorescence spectroscopy as well as their electrochemical  The electrochemical features of 6a/b and 7a/b were measured in dichloromethane with 0.1 M tetrabutylammonium hexafluorophosphate (TBAPF 6 ) as electrolyte. A silver wire, which was calibrated versus the Fc/Fc + reference redox couple, was used as a quasi-reference electrode. As typically observed in 95 porphyrins, two reductions -the first reversible, the second quasi-reversible -were noted for 7a and 7b at -1.6 and -1.8 V followed by a quasi-reversible oxidation at +0.4 V. In contrast, 6a and 6b reveal two oxidations -the first reversible, the second quasi-reversible -at +0.2 and +0.6 V. The quasi-reversible 100 reduction was detected for 6b at -1.5 V, whereas 6a did not show any appreciable reduction within the electrochemical window of up -2.0 V. In general, when comparing zinc porphyrins 6a and 6b with free base porphyrins 7a and 7b the earlier feature lower oxidation, that is, +0.2 versus +0.4 V, while the reduction potentials are -1.5 and -1.6 V. The excited state characteristics of 6a/b and 7a/b were 5 determined by fluorescence spectroscopy in THF. As Figure 2 illustrates, upon exciting 7a and 7b at 410 nm broad fluorescence bands centered at 662 / 720 nm and 660 / 718 nm evolved, respectively. Notably, the fluorescence quantum yields are 0.035 for 7a and 0.061 for 7b. These facts are in line with the  30 Having established the key features of 6a/b and 7a/b, we turned to the preparation of the respective nanographene / porphyrin hybrids G6a/b and G7a/b following a previously reported procedure that was, however, slightly modified. 38 The procedure consists of several consecutive steps, namely: i) ultrasonication of 35 solutions of the porphyrins with natural graphite, ii) centrifugation of the suspension to eliminate the remaining graphite particles, iii) ultrasonication of the supernatant with newly added natural graphite, and iv) repetition of the steps -see Experimental section for more details. Such a treatment resulted 40 in the direct exfoliation of graphite and the concomitant immobilization of the porphyrins onto the basal plane of graphene. For 7a and 7b, steps i), ii), and iv) were sufficient to yield stable nanographene/porphyrin hybrid suspensions.
However, in the case of 6a and 6b, despite the optimization of 45 different parameters such as force, time, and temperature during the ultrasonication, only the addition of step iii) resulted in acceptable results. An overall enrichment of exfoliated graphite required the addition of new natural graphite to decrease the amount of free porphyrins. Interestingly, the stability of the 50 nanographene / porphyrin dispersions varies as a function of relative THF/water ratios -see Figure S3. In fact, the overall dispersion stability benefits from the presence of water as it enforces nanographene / porphyrin interactions. Initial insights into the graphite exfoliation came from Raman 55 spectroscopy performed with G6a/b and G7a/b. Laser excitation at, for example, 532 nm is in resonance with 6a/b and 7a/b and with nanographene. Owing to the quantitative quenching of the porphyrin fluorescence in G6a/b and G7a/b -vide infra -the Raman peaks of 6a/b and 7a/b and nanographene evolve -see 60 Figure S4. In particular, at 1096 (δ(C β -H)), 1122, 1194, 1220 (ν(C α -N)), 1463 (ν(C α -C β ) + δ(C β -H)), 1481 and 1509 cm -1 ((ν(C α -C m )) porphyrin-centered Raman peaks are observed. 48 Interesting is the fact that G7a and G7b gave rise to much better results in terms of exfoliating graphite into few layer graphene 65 than G6a and G6b. This conclusion came from Raman spectra that revealed 2D/G ratios between 0.45 and 1 as well as FWHM values that ranged from 46 to 94 cm -1 . Responsible for the aforementioned trend is the fact that porphyrin metallation to afford 6a and 6b leads to an increase in steric demand and, in 70 turn, to a weaker interaction with the graphene surface. 49 Despite the successful graphite exfoliation and porphyrin immobilization, the 2D/G ratios suggest a limited extent of doping, which is in line with a recent report on porphyrin interactions with the basal plane of graphene. 50 75 The Raman spectrum of G7b confirms the presence of turbostratic and electronically decoupled single layer graphene -  Additional experiments focused on screening different solvent mixtures with respect to graphite exfoliation starting with pure THF, to which increasing amounts of water were subsequently added -vide supra. The best results in terms of graphite exfoliation and nanographene functionalization were obtained in pure THF. Raman spectra gave 2D/G ratios in the range from 0.5 5 to 0.6 and a FWHM of 94 and 82 cm -1 . Nevertheless, greater dispersion stabilities and greater graphene enrichments resulted in THF/water (1:1 v/v). The presence of even more water does not lead to any significant difference in terms of exfoliation efficiency, not taking dispersion stabilities into account. In fact, 10 the Raman assays suggest that the porphyrin aggregates are difficult to dissemble. AFM images shed light on the nanographene / porphyrin flake height and size distribution. The graphene flakes were about 2 -6 nm in height -note that a calibrated single layer is with this 15 apparatus in the range of 1 nm -owing to the presence of strongly folded and intertwined sheets. Figure 4 illustrates that the lateral sizes are up to 500 nm. Self-aggregation, reaggregation, and the rolling up of graphene sheets should be considered in the interpretation of the AFM height profiles. Along the same lines, the TEM images reveal folded as well as regularly stacked sheets that reach sizes of up to 1.2 µm² with around 10 % of them exceeding 2.0 µm² - Figures 5 and S6. It is noteworthy that the graphene flakes of sizes below 1 µm² are 30 more likely to be thinner than the larger graphene flakes -a finding that goes hand-in-hand with the Raman experiments. The use of a contrast aperture enabled us to detect even thinner flakes, which usually render difficult to be seen in TEM investigations. 51 With stable and concentrated dispersions of nanographene / 35 porphyrin hybrids G6a/b and G7a/b in hand, their photophysics were probed. Overall, the absorption spectra of G6a/b and G7a/b lack the distinct and well-resolved features of 6a/b and 7a/b. In particular, the absorption spectra of G7a and G7b in THF reveal weak maxima at 255/421 nm and 255/417 nm, 40 respectively. The latter feature corresponds to the Soret band of 7a and 7b - Figure 6. The amount of dispersed graphene was calculated using the absorption coefficient a = 36.0 mL/mg 52 cm at 660 nm, resulting in concentrations of approximately 0.006 mg/mL for G7a and 0.004 mg/mL for G7b. 18 In contrast, the Noteworthy, the initial concentrations of 6a/b and 7a/b were in all of the experiments on the order of 2 x 10 -5 M. Concentration losses during the work up procedure cannot be ruled out, especially during the removal of the larger graphite particles after centrifugation. Nevertheless, 55 we correlate the loss in oscillator strength with the strong electronic interaction between 6a/b and 7a/b and the basal plane of graphene. 53 wavelength analyses of these newly developed charge transfer states reveal their metastable character with lifetimes in THF of 250 ± 20 ps for G7a, 260 ± 20 ps for G7b, 390 ± 50 ps for G6a, and 420 ± 50 ps for G6b. The shorter lifetime in the earlier two when compared to the latter two is ascribed to the weaker 40 interactions of the more hydrophilic zinc porphyrins as well as the fact that the more planar free base porphyrins 7a/b better immobilize onto the basal plane of the graphene. The observation that tighter nanographene / porphyrin interactions lead to shorter lifetimes is reinforced in THF/water (1:1 v/v), in 45 which the lifetime values are 130 ± 20 ps for G7a and G7b and 190 ± 20 ps for G6a and G6b -Figures S10, and S11. To further exploit the properties of the porphyrins, the carboxylic groups of 6a/b and 7a/b were probed as an anchor for nanoparticles. In this context, three batches of G7a with 50 increasing loads of TiO 2 nanoparticles were prepared by dispersing them with TiO 2 nanoparticles in water. The resulting composites were deposited and dried onto copper grids for TEM imaging - Figure S12. The batch with the highest concentration of TiO 2 nanoparticles yielded large TiO 2 agglomerates that are 55 adhered to the functionalized nanographene surfaces. A decrease of TiO 2 nanoparticle concentrations, on the other hand, results in a more homogenous coverage of the graphene flakes. Most importantly, the presence of TiO 2 nanoparticles is restricted exclusively to areas covered with G7a. 60 Intrigued by the features described for the new nanographene / porphyrin hybrid materials, we decided to investigate their impact on the performance of dye-sensitized solar cell (DSSC) devices. Details regarding device fabrication and characterization are summarized in the experimental section. soaked with G7a in THF and THF/water (1:1 v/v) for 120 hours. 70 We initially focused our attention on elucidating which type of TiO 2 electrode -transparent (TiO 2 -T) or light-scattering (TiO 2 -S) -and which sort of suspension -THF/water (1:1 v/v) or THF solvents -were optimal for fabricating DSSCs with the nanographene / porphyrin hybrids G6a/b and G7a/b. To this 75 end, we immersed the previously prepared electrodes coated with thin films of each type of TiO 2 into the nanographene / porphyrin dispersions in each solvent for a period of 120 hours to ensure sufficient coating of the TiO 2 film by the graphene composite. By naked eye, we observed that the adsorption of the 80 nanographene hybrids and porphyrins from THF/water (1:1 v/v) was poor in comparison with the colorful electrodes obtained from THF suspensions - Figure 8. when dealing with such hybrid materials. For instance, as is seen in Figure 8, the G7a coverage is better in the case of the TiO 2 -S electrode than that of the TiO 2 -T electrode. This is presumably due to the larger particle size of the TiO 2 -S creating a more porous electrode into which the graphene composite can 95 intercalate. Although we performed cross-sectional SEM with the TiO 2 electrodes ( Figure S13) Additionally, probing the modified TiO 2 electrodes at high magnification reveals that, to a notable extent, re-aggregation of the graphene flakes takes place during the integration of G7a.
Despite the presence of aggregates, their chemical and physical nature is clearly graphene-like rather than graphite-like due to the presence of the porphyrin. This indicates that the nature of the nanographene / porphyrin hybrids is preserved when attaching to the TiO 2 interface. To corroborate this notion, we also recorded 15 SEM images for nanographene / porphyrin hybrids G6a/b and G7b leading essentially to the same results -Figures 9 and S14. Loading of G6a/b and G7a/b for 120 hours onto both TiO 2 -T and TiO 2 -S electrodes, produced active devices albeit with low photocurrents - Figure 10. Not surprisingly, the larger the 20 amount of nanographene composite on the photoanode, the higher the photocurrent with TiO 2 -S films featuring higher values than TiO 2 -T films. In order to optimize the adsorption process of the nanographene / porphyrin hybrids on the TiO 2 -S electrode, we focused our investigations on the impact of different soaking 25 times, that is, 2, 4, 8, and 16 h, exert on the device performance using both THF and THF/water (1:1 v/v) suspensions. Overall, the best performances were observed for devices using photoanodes prepared from THF composite suspensions (Tables  S1 and S2). Therefore, we will focus on the trends obtained from 30 these DSSCs. The photovoltaic characteristic trends are given in Figure 11 and the best DSSC performance values given in Table  1. Overall, the device parameter versus soaking time dependence is quite similar for all hybrid materials. While open-circuit voltages (V oc ) remain nearly constant, short-circuit currents (I sc ) increase 40 and reach maximum values at around 8 h for most of the nanographene / porphyrin composites. The impact of the latter is seen when inspecting the efficiencies (η) - Figure 11. We note a sizeable enhancement of I sc and η in G6a/b and G7a/b devices when compared to those of 6a/b and 7a/b under comparable 45 conditions -Tables S1 and S2. Such a phenomenon is likely to relate to the benefits from the presence of graphene flakes in terms of influencing the injection processes, the recombination processes, and the adsorption behavior on TiO 2 . To get insights into the penetration of the nanographene / 50 porphyrin hybrids into the TiO 2 network, we reduced the TiO 2 film thickness from 8 to 4 µm. Please note that cross sectional imaging was unsuccessful -vide supra. Important is the fact that the trends on the thinner device performance are virtually identical to those previously mentioned - Figure S15. This leads 55 us to conclude the depth of penetration of the nanographene / porphyrin hybrids is not a limiting factor. To elucidate the injection mechanism that is operative in nanographene / porphyrin based DSSCs, incident photon-tocurrent efficiency spectra (IPCE) were recorded. Importantly, the 60 IPCE spectrum of Figure 12 matches the absorption spectrum of Figure 6 and, as such, confirms that the porphyrin generates the photocurrent. The overall broadening suggests that the electronic coupling between the porphyrin and graphene is even on TiO 2 still intact. For comparison, IPCE spectra were taken of devices 65 with graphene and porphyrin only photoanodes. The earlier is featureless with a maximum of around 0.2 % indicating no contribution of this moiety while the latter feature a maximum at around 425 nm similar to the nanographene / porphyrin hybrid devices but, in line with the photogenerated currents, a much lower value was noted ( Figure 12). In addition, SEM images show that the adsorption of graphene flakes at these electrodes is 5 rather poor ( Figure S16) and, in turn, the photocurrent is very low (3 µm/cm 2 ). In light of the aforementioned, it is safe to assume that graphene on its own plays no significant role in the overall photosensitization process. Our hypothesis implies that the photoexcited porphyrins inject electrons either directly into the conduction band of TiO 2 or 25 indirectly via graphene as a conducting mediator. Both processes are driven by a sizeable thermodynamic driving force as has already been shown. [54][55][56] The competition between the direct and indirect route is seen when comparing G6a/b and G7a/b, in which the earlier give rise 30 to weaker interactions with the basal plane of graphene -vide supra -and, in turn, favor the direct route. Likewise, comparing G6a/7a with G6b/7b, the presence of two carboxylic groups in the latter shifts the competition to the direct route. Nevertheless, graphene is also known to act as a Schottky barrier, 57 which 35 prevents the recombination process that is a typical drawback observed in porphyrin-based DSSCs. 54,56 In other words, the presence of the graphene seems to be crucial in the overall enhancement of device efficiency by means of affecting the injection process and decreasing the recombination 40 process. Finally, the fact that the chemical and also the physical nature of the nanographene / porphyrin hybrids is clearly graphene-like rather than graphite-like, as revealed by the SEM images (Figures 8 and 9), cannot be neglected. Thus, its contribution to minimize porphyrin aggregation and thereby to 45 enhance porphyrin regeneration should not be underestimated.
To corroborate the latter, we are currently directing our attention to perform femtosecond pump probe experiments.

Conclusions
In summary, two major milestones in the preparation and use of 50 graphene in solution are provided in this work. Firstly, we have demonstrated the versatility of free-base porphyrins 7a/b to realize nanographene hybrids, while their metalated analogous 6a/b give rise to a much poorer performance in terms of exfoliating graphite. The success has been corroborated by 55 means of Raman, AFM, TEM, and time-resolved pump probe experiments. Importantly, the latter provides solid evidence for shifting electron density from the porphyrins to nanographene upon excitation. In line with our previous assays, we conclude that the synergy of π-π and electronic interactions is crucial to produce highly stable graphene sheets in solution. Secondly, TEM images of nanographene hybrid, to which TiO 2 nanoparticles were grafted, and SEM images of TiO 2 electrodes have undoubtedly shown that, even when the porphyrin interacts 5 with the basal plane of graphene, its hydrophilic chain is operative to implement this novel hybrid into solar cells. In this context, the device data prompt to the benefits when the nanographene hybrids are present. Quite likely, aggregation is prevented, which leads to benefits in the regeneration process of the porphyrins, on one hand, and implementation of a Schottky barrier reducing the recombination rate process, on the other hand. These aspects lead to a significant boost of the overall device performance. Specifically, in the optimum photosensitizer uptake for nanographene / porphyrin, the enhancement amounts 15 to a 50 % increase of the device efficiency compared to DSSCs constructed only the respective porphyrins were noted. Currently, we are directing our efforts to further unveil the underlying mechanism of DSSCs that are based on nanographene hybrids. However, it should be noted that the use of 20 nanographene / porphyrin hybrids opens the way for the introduction of other nanoparticles into solar cells.