Chemical Physics Letters 586 (2013) 127–131

Contents lists available at ScienceDirect

Chemical Physics Letters

journal homepage: www.elsevier .com/ locate /cplet t

Riboflavin enhanced fluorescence of highly reduced graphene oxide

0009-2614/$ - see front matter � 2013 Elsevier B.V. All rights reserved.http://dx.doi.org/10.1016/j.cplett.2013.09.032

⇑ Corresponding author. Fax: +40 264 591906.E-mail address: simion.astilean@phys.ubbcluj.ro (S. Astilean).

Maria Iliut a, Ana-Maria Gabudean a, Cosmin Leordean a, Timea Simon a, Cristian-Mihail Teodorescu b,Simion Astilean a,⇑a Nanobiophotonics and Laser Microspectroscopy Center, Interdisciplinary Research Institute in Bio-Nano-Sciences and Faculty of Physics,Babes-Bolyai University, M Kogalniceanu 1, 40084, Cluj-Napoca, Romaniab National Institute of Materials Physics, Atomistilor 105b, 077125 Magurele-Ilfov, Romania

a r t i c l e i n f o a b s t r a c t

Article history:Received 5 August 2013In final form 11 September 2013Available online 18 September 2013

The improvement of graphene derivates’ fluorescence properties is a challenging topic and very few wayswere reported up to now. In this Letter we propose an easy method to enhance the fluorescence of highlyreduced graphene oxide (rGO) through non-covalent binding to a molecular fluorophore, namely theriboflavin (Rb). While the fluorescence of Rb is quenched, the Rb – decorated rGO exhibits strong bluefluorescence and significantly increased fluorescence lifetime, as compared to its pristine form. The datareported here represent a promising start towards tailoring the optical properties of rGOs, having utmostimportance in optical applications.

� 2013 Elsevier B.V. All rights reserved.

1. Introduction

Graphene, one atomically layer, attracted enormous consider-ation due to the unique electronic and optical properties whichmake it extremely promising in fields like photonics and optoelec-tronics, starting from transparent conductors, photovoltaic devicesto optical limiters and terahertz devices [1]. Recently, chemicallyderived graphene, graphene oxide (GO) and reduced grapheneoxide (rGO), has gained great interest due to the low cost, massproduction and chemical manipulation, providing the opportunityof tailoring the optoelectronic properties. While the GO is a solu-tion processable non-stoichiometric macromolecule with �2–3 nm sp2 clusters isolated by sp3 matrix, its reduction leads tothe formation of new sp2 domains that percolate between existingsp2 clusters [2]. The field which most benefits from the heteroge-nous atomic and electronic structures of chemically derived graph-ene is fluorescence spectroscopy. Moreover, due to its chemicalprocessability, the broad fluorescence band arising from GO canbe tuned by changing the sp2/sp3 ratio upon the reduction process[3–7]. Contrary, due to its gapless character, pristine graphene doesnot exhibit fluorescence [8]. The intrinsic and tunable fluorescenceof chemically derived graphene opened up exciting ways ofexploiting new applications, e.g. drug delivery, cells imaging etc.[9,10]. On the other hand, the excellent fluorescence quenchingabilities of pristine and chemically derived graphene [11,12] fillout the wide field of applications (e.g. fluorescence quenchingmicroscopy, sensing, biosensing, resonance Raman spectroscopy)[2,11].

Despite of the tunable fluorescence, the quantum yield ofgraphene derived species is relatively low [2] and sensitive to thesurrounding environment [5]. Moreover, while the reduction ofGO leads to a better conductivity, its fluorescence was reportedto suffer from the increase of the reduction time [6]. Up to now,several routes were reported for improving the fluorescence ofGO/rGO, such as the complexation with europium [13], silvernanoparticles [14], polyvinyl alcohol [15] or covalent surface pas-sivation [16] etc.

Herein we report a new route towards the improvement of thefluorescence in highly reduced graphene oxide (rGO) through non-covalent interaction with a biological molecule, namely the ribofla-vin (Rb), a fluorescent micronutrient of the utmost importance in avariety of cellular processes, with a key role in energy metabolism[17]. On the other hand, the interaction of rGO with Rb leads to thefluorescence quenching of the latter. Therefore, rGO–Rb compositecan take advantage from both: the improved optical properties ofrGO (for e.g. imaging) and the quenched fluorescence of the fluoro-phore (for e.g. bio/sensing).

2. Experimental

2.1. Chemicals

Graphite powder, 325 mesh, 99.9995% was purchased from AlfaAesar; potassium permanganate (KMnO4) 99%, sodium nitrate(NaNO3) 99%, hydrazine hydrate (NH2NH2�H2O) 24–26% in H2O(RT), Rb (C17H20N4O6, 98%) were bought from Sigma to Aldrich;hydrochloric acid (HCl) 35% was purchased from Lach-Ner; sulphu-ric acid (H2SO4) 96%, hydrogen peroxide (H2O2) 30% and ammoniasolution (NH3) 25% were obtained from local manufacturers.

128 M. Iliut et al. / Chemical Physics Letters 586 (2013) 127–131

2.2. Methods

GO was synthesized according to Hummers and Offeman’s meth-od [18]. The detailed description of the method is given in our previ-ous work [19]. Graphene oxide reduction was performed in AntonPaar Monowave 300 as follows: 50 lL of hydrazine hydrate (HH)and 20 lL of NH3 solution were added to 6 mL of GO 0.05 mg/mLsolution and kept in microwaves at 100 �C for 20 min. The measuredpH of the resulted rGO solution was �10.3. The solution was pre-pared by mixing one volume unit of Rb of a certain concentrationwith 1.7 volume units of rGO (0.05 mg/mL) to the final concentrationof 0.5� 10�4 M for Rb and 0.0315 mg/mL for rGO. The pH of the mix-ture was adjusted to the pH of the original rGO with NaOH. To avoidinner filter effect during the fluorescence measurements, the abovepreparedrGO–Rb sample was filtered several days later by using a cellulosefilter of 450 lm pore size and the resulted colorless and transparentsupernatant was collected. For the comparison measurements therGO blank solution was also filtered and the colorless solution wascollected. The samples were kept in dark to avoid any interactionwith light.

2.3. Characterization

UV–Vis absorption spectra were taken in quartz cells of 10 mmpath length using Jasco V-670 UV–Vis-NIR spectrophotometer. Thesteady-state fluorescence spectra were recorded with JascoLP-6500 spectrofluorimeter in quartz cell with 10 mm path length.The medium sensitivity was used for all measurements. FT-IRspectroscopic analyses were done in reflection configuration usinga JASCO 6200 FT-IR spectrometer with 4 cm�1 resolution, in the4000–400 cm�1 region. The measurements were performed on themixture of the sample in powder state with KBr, pressed into pellets.X-ray photoelectron spectroscopy (XPS) was performed in a Specsanalysis chamber, following the description from our previous work[19]. Raman measurements were performed with a Witec Alpha 300R confocal Raman microscope system, using a 532 nm Nd-Yag laserexcitation line, at the 0.6 mW laser power. AFM images were takenwith a Witec Alpha 300 A atomic force microscope system, operatingin non-contact mode with a standard aluminium coated silicon tip(Nanosensors). Zeta potentials were measured at 25 �C using a Mal-vern Instruments NanoZS90 Zetasizer instrument. Fluorescence life-time measurements were performed on a time-resolved confocalfluorescence microscope system (MicroTime200, from PicoQuant)equipped with an inverted microscope (IX 71, from Olympus), inthe liquid sample. The excitation was provided by picosecond diodelaser heads (LDH-D-C-375 and LDH-D-C-485, PicoQuant) operatingat 375 and 485 nm, respectively, and 40 MHz repetition rates. Thesignal collected through a UPLSAPO 60x/NA 1.2 water immersionobjective was spatially and spectrally filtered by a 50 lm pinholeand longpass emission filters before being focused on a PDM SinglePhoton Avalanche Diode (SPAD) from MPD. The detector signalswere processed by the PicoHarp 300 (PicoQuant) Time-CorrelatedSingle Photon Counting (TCSPC) data acquisition unit. Data analysisand image processing were performed with the SymPhoTimesoftware (PicoQuant). Fluorescence lifetime decay curves weretail-fitted with exponential decay curves and the quality of the fitswas evaluated by analyzing the chi-square (v2) values and the distri-bution of the residuals.

3. Results and discussions

The inset of Figure 1A shows the UV–Vis absorption spectra ofpristine Rb, rGO and rGO–Rb mixture before being filtered. Thereare four absorption bands for pristine Rb solution, in ultraviolet(UV) and visible (Vis) regions, assigned to the p–p⁄ transitions

(223 and 267 nm) and to the combination of n–p⁄, p–p⁄, and p–p⁄ transitions (374 and 445 nm), respectively. The last of the tran-sitions corresponds to the HOMO to LUMO transition [20,21]. TherGO presents a characteristic peak at 269 nm due to the C–C p–p⁄ transition discussed in the characterization section from supple-mentary information (ESI). In rGO–Rb mixture, the Rb bands arered shifted from 267 and 445 nm to 271 and 455 nm, respectively,while the band at 223 nm disappears. These shifts could be due tothe changing in surrounding environment of Rb. After the rGO–Rbmixture was filtered (Figure 1, rGO–Rb), it is observed from thespectrum that most of the Rb was removed, since the absorptionbands of Rb disappeared. This is probably due to the interactionof Rb with rGO removed by filtering, as well as intercalation of un-bounded Rb between rGO sheets. The resulted solution exhibits aweak broad absorption band in UV region and a continuousabsorption over the whole visible range, characteristic to rGO.These features are similar to the blank rGO (Figure 1, rGO), withspecification that in the former case the overall absorption is in-creased and the UV band of rGO is almost vanished. A possibleexplanation could be the p–p stacking interaction of rGO with Rbwhich, on one side affects the p–p⁄ inter-band transition in rGO,while on the other side disturbs the intramolecular electron tran-sitions in Rb. Figure 1B shows the photoluminescence (PL) spectraof pristine rGO and rGO–Rb solutions. The rGO–Rb aqueous sus-pension exhibits strong PL under UV irradiation (exc. 375 nm)which can be seen by naked eye (see inset), with emission maximaat 440 nm and a shoulder centered at �490 nm. Unlike rGO–Rb,the pristine rGO shows no visible fluorescence when excited atthe same excitation wavelength (see optical photograph from in-set) and has weak fluorescence spectrum with the maximum at�450 nm and a shoulder at �500 nm. Additionally, the observedblue – shift (from 450 to 440 nm) of the blue band in rGO–Rb rel-ative to rGO can be an evidence of rGO–Rb interaction. It wasestablished [3,5] that GO possesses two emission bands, a blueband in 400–500 nm range and a second broad one in the blue tored/NIR region. After reduction, the latter band disappears [3,6],remaining only the blue band and some residues of the broad bandas a shoulder, in agreement with our measurements. This is consis-tent with the increase in number of small sized sp2 domains uponthe reduction [3,6]. It is also noticed that no emission from Rb isdetected (pristine Rb has emission at 530 nm, data is not shown),supporting the binding of Rb to the rGO sheets.

It was showed in most of graphene – related fluorescence stud-ies [4,5,7] that the spectral position of the fluorescence emission ofchemically derived graphene (kem) is dependent on the excitationwavelength (kexc), due to different possible transitions from thebottom of CB (conduction band) and nearby localized states tothe wide-range VB (valence band) [4]. In order to see whetherthe rGO–Rb composite preserves these features, the emission spec-tra of rGO–Rb were recorded from 375 to 470 nm by varying thekexc with 5 nm (Figure 2B). The blue band decreases in intensityand red-shifts with increasing the kexc (from 440 to 460 nm), asin the case of pristine rGO (Figure 2A). The same red-shift (from490 to 530 nm) and intensity decrease is observed for the bandin 500–700 nm region with increasing the excitation wavelength.Such as, at kexc of 470 nm, the emission spectrum only features abroad band in 530 nm region, which might correspond to the blueband and its shoulder. While there is only a slight shift in the peakposition of the PL of rGO–Rb relative to rGO, the excitation spectraof rGO–Rb taken at the three different emission wavelengths (Fig-ure 2A and B) are very different in shape and position compared torGO. While the pristine rGO exhibits broad excitation bands cen-tered at �318 nm, for all emission wavelengths, the rGO–Rb haswell defined sharp excitation spectra, at 358 and 362 nm foremission at 440 and 460 nm, respectively, becoming broad withno defined maxima at kem of 530 nm.

Figure 3. Fluorescence lifetime measurements of rGO, Rb and rGO–Rb solutions, atthe excitation wavelengths 375 and 485 nm.

Figure 1. UV–Vis absorption spectra of rGO and rGO–Rb filtered solutions (A). Inset shows the UV–Vis absorption spectra of rGO, Rb and rGO–Rb solutions before filtering;fluorescence spectra of rGO (dotted line) and rGO–Rb (solid line) (B). Inset shows the optical photographs of cells containing rGO (bottom) and rGO–Rb (top) solutions at the375 nm excitation.

Figure 2. Fluorescence excitation (at kem 440, 460 and 530 nm) and emission spectra of rGO (A) and rGO–Rb (B) recorded at different kexc, from 375 to 470 nm with avariation of 5 nm between two subsequent excitations. The sharp peak in rGO spectra is due to the Raman scattering.

M. Iliut et al. / Chemical Physics Letters 586 (2013) 127–131 129

The time resolved fluorescence (TRF) measurements are furtherperformed as the most definitive method, by using two excitationwavelengths: 375 nm and 485 nm, for analyzing the fluorescencein the regions 400–700 nm and 500–700 nm, respectively. Forcomparison, rGO and Rb pristine solutions were measured withexcitation at 375 nm. The fluorescence decay curves are shown inFigure 3 and the corresponding derived values are listed in Table 1.While Rb shows only one lifetime component of �4.76 ns, the pris-tine rGO presents three lifetime components: s1 = 5 ns, s2 = 1.2 nsand s3 = 0.2 ns, with the shorter component having the higher con-tribution (�84%), in accordance with other findings [3,5]. However,for the simplicity reason, the lifetime decay of rGO–Rb was fittedby two components. For the kexc = 375 nm, the rGO–Rb givess1 = 4.7 ns and s2 = 1.4 ns, with a higher contribution from thelonger lifetime component (�58.6%). The components shorter than�1.4 ns could be included in the s2 component. At the kexc = 485nm the contribution of the short lifetime component greatlydecreases, however still having important contribution to theoverall lifetime (�20%). The above measurements lead us toconclude that the short lifetime component has the main contribu-tion in the blue emission, whereas its contribution decreases in500–600 nm emission range. On the other hand, the long lifetimecomponent has the higher contribution in the 500–600 nm

emission range. The measured significant increase of the fluores-cence lifetime of rGO–Rb (�3.5 ns) compared to rGO (0.6 ns),

Table 1Fluorescence lifetime values and corresponding relative amplitudes for pristine rGO,Rb and rGO–Rb.

Samples kexc-375 nm kexc-485 nm

Relativeamplitude(%)/overallamplitude(kCnts)

Fluorescencelifetime (s) (ns)/average lifetimehsi/v2

Relativeamplitude(%)/overallamplitude(kCnts)

Fluorescencelifetime (s) (ns)/average lifetimehsi/v2

rGO A1 = 5.79 s1 = 4.99 ± 0.13A2 = 10.73 s2 = 1.218 ± 0.051A3 = 83.48 s3 = 0.22 ± 0.002

hsi = 0.6v2 = 1.046

Rb A1=100 s1 = 4.755 ± 0.004v2 = 1.017

rGO–Rb A1=58.6 s1 = 4.725 ± 0.000 A1 = 79.2 s1 = 4.737 ± 0.006A2=41.4 s2 = 1.362 ± 0.004 A2 = 20.8 s2 = 1.356 ± 0.000

hsi = 3.3 hsi = 4v2 = 1.070 v2 = 0.999

Note: The average lifetime values hsi were determined according to Ref. [15].

130 M. Iliut et al. / Chemical Physics Letters 586 (2013) 127–131

indicates on rGO–Rb interaction and the formation of a more stableexcited state.

In order to explain the observed increase of the PL intensity ofrGO in the presence of Rb and the formation of a stable exited statecomposite, it is important to first identify the potential factorsaffecting the PL of rGO in its pristine form. The weak fluorescenceobserved in the case of pristine rGO can be considered aconsequence of several factors. First, the hydrazine induced exten-sive GO reduction (see XPS and FT-IR data from ESI) increases theinterconnectivity between sp2 clusters which facilitates the hop-ping of excitons to nonradiative recombination centers [6]. Second,the remained after reduction oxygen containing groups (carboxyl,hydroxyl) are highly deprotonated at our experimental pH (�10)(zeta potential is ��40 mV) [22] and, together with epoxy groups,

Figure 4. AFM image of rGO (A) and rGO–Rb (B); Raman spectra of Rb powder (red line) aintegration of Rb peaks (732–837 cm�1) (D) and rGO G band (1563–1600 cm�1) (E). (For inthe web version of this article.)

will induce fluorescence quenching in rGO by the nonradiativeelectron transfer to the holes of sp2 carbon rings [15,23]. Finally,the collision with solvent molecules will contribute as well to theoverall PL quenching [15,16].

As our results demonstrate, the coupling of Rb with rGO canovercome the above described problems and lead to the enhance-ment of the PL of rGO. In this context, in order to establish the ori-gin of the interaction between rGO and Rb, several structuralaspects connecting the two components are considered. Firstly,the reduction of GO into rGO (see rGO characterization, ESI) in-creases the availability of sp2 domains due to the removal of oxy-gen groups. Secondly, similar to other fluorophores, Rb has anamphiphilic character due to the hydrophilic ribityl side chainand hydrophobic isoalloxazine ring [20,24]. Therefore, the p-likestructure of the isoalloxazine ring promotes the p–p type interac-tion between Rb and rGO [25]. This interaction and the resultedloss of rGO planarity (see further AFM) may induce changes in pelectronic conjunction, determining a lesser conjugated systemwhere the nonradiative hopping of the excitons will be limited. An-other possible interaction of Rb with rGO is via hydrogen bonds be-tween Rb and the remained oxygen species of rGO [21]. However,considering the experimental pH of the solutions (pH = 10), itshould be noted that at this pH Rb exists in solution in neutralRFloxH and anionic RFlox � forms [24]. Unlike anionic form, theneutral form contains a hydrogen atom at the N3 position (seeRef. [24]). On the other hand, the remained oxygen groups ofrGO are highly deprotonated at given pH, in accordance with zetapotential measurements (�40 mV) and Ref. [22]. Accordingly, thehydrogen bonds will be only favored at NH position of the neutralRb and will occur between the NH groups, deprotonated (COO�,O�) and other unreduced oxygen species (epoxy, carbonyl) ofrGO. Therefore, the hydrogen bond formation will inhibit the elec-tron donation capabilities of the oxygen groups to the holes of sp2

domains, resulting in a more stable excited state. Additionally, therGO ‘decoration’ with Rb prevents the extensive interaction and

nd rGO–Rb (black line) dried on glass substrate (C); Raman maps obtained from theterpretation of the references to colour in this figure legend, the reader is referred to

M. Iliut et al. / Chemical Physics Letters 586 (2013) 127–131 131

the quenching of excitons with the solvent molecules. Altogether,the above itemized factors can be considered as contributing tothe overall fluorescence enhancement in rGO.

In order to prove experimentally the assumed interactions, wefirst employed AFM to get a visual image of surface morphologyand sheet thickness of rGO upon interaction with Rb molecules.Different from flat and �0, 7 nm thick pristine rGO (Figure 4A),the rGO–Rb (Figure 4B) presents an increased surface roughnesswhich can be clearly seen from the height profile, with the thick-ness varying between �2 nm and �5 nm. The observed morphol-ogy is a consequence of Rb’s stacking on both sides of rGO andits probable affinity to form nanoaggregates on the rGO surface.On the other hand, the Rb attachment determines the loss of theflakes planarity and the apparition of the wrinkleless, which couldexplain the different thicknesses observed in the height profile. Inorder to confirm whether the entities observed at the surface ofrGO are Rb molecules, Raman spectroscopy measurements wereperformed on the composite sample. Figure 4C shows the Ramanspectrum of rGO–Rb sample (black line) deposited and dried ontothe substrate. For comparison, the Raman spectrum of Rb powder(red line) was recorded, following the peaks assignment found inRef. [26]. It can be seen that the Raman spectrum of rGO–Rb con-tains the peaks from both rGO and Rb, in 1250–1600 cm�1 (seerGO characterization, ESI) and 570–870, 1100–1400 cm�1 rangerespectively (see Ref. [26]). Moreover, the Raman maps obtainedby integrating the rGOs G band (1563–1600 cm�1) and Rb peaks(732–837 cm�1) and displayed in Figure 4E and D, respectively,show an excellent match, confirming the bonding between Rb mol-ecules and rGO. Interestingly, while the Rb’s Raman spectrum pre-sents a fluorescent background, in rGO–Rb the fluorescence iscompletely suppressed, in accordance with above observed (seePL spectra, Figures 1B and 2B) rGO induced fluorescence quenchingof Rb. Because of the rGO with Rb ground state interaction andfavorable positioning of HOMO–LUMO levels of the Rb (Eg = 2.8 eV)relative to the work function of rGO (EF = �4.5 eV) [27], the chargetransfer from Rb to rGO will be favored, rationalizing the observedquenching of Rb fluorescence induced by rGO.

4. Conclusions

In this study we proved that the non-covalent p–p and hydro-gen bond interaction between Rb and rGO leads, on the one hand,to the fluorescence quenching of Rb and, on the other hand, to thefluorescence enhancement and increase in average fluorescencelifetime of rGO. Moreover, the rGO–Rb composite shows the sameexcitation – dependent emission characteristic to graphene-likematerials. This study can be a good start towards designing of

graphene-based hybrids with tunable optical properties, of impor-tance in optical applications.

Acknowledgements

This work was supported by CNCS – UEFISCDI Romania, underthe project number PNII-ID-PCCE-0069/2011 and Sectoral Opera-tional Programme for Human Resources Development 2007–2013, co-financed by the European Social Fund, under the projectnumber POSDRU/107/1.5/S/76841 with the title ‘‘Modern DoctoralStudies: internationalization and Inter-disciplinarity’’. The authorsthank Mircea Dan Puia FT-IR measurements.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.cplett.2013.09.032.

References

[1] F. Bonaccorso, Z. Sun, T. Hasan, A.C. Ferrari, Nat. Photonics 4 (2010) 611.[2] K.P. Loh, Q. Bao, G. Eda, M. Chhowalla, Nat. Chem. 2 (2010) 1015.[3] C-T. Chien et al., Angew. Chem. Int. Ed. 51 (2012) 6662.[4] J. Shang, L. Ma, J. Li, W. Ai, T. Yu, G.G. Gurzadyan, Sci. Rep. 2 (2012) 792.[5] X.-F. Zhang, X. Shao, S. Liu, J. Phys. Chem. A 116 (2012) 7308.[6] G. Eda et al., Adv. Mater. 21 (2009) 1.[7] C. Galande et al., Sci. Rep. 1 (2011) 85.[8] K.F. Mak, L. Ju, F. Wang, T.F. Heinz, Solid State Commun. 152 (2012) 1341.[9] X. Sun, Z. Liu, K. Welsher, J.T. Robinson, A. Goodwin, S. Zaric, H. Dai, Nano Res. 1

(2008) 203.[10] Z. Liu, J.T. Robinson, X. Sun, H. Dai, J. Am. Chem. Soc. 130 (2008) 10876.[11] J. Kim, L.J. Cote, F. Kim, J. Huang, J. Am. Chem. Soc. 132 (1) (2010) 260.[12] R.S. Swathi, K.L. Sebastian, J. Chem. Phys. 129 (2008) 054703.[13] B.K. Gupta et al., Nano Lett. 11 (12) (2011) 5227.[14] Y. Wang et al., Nanoscale 5 (2013) 1687.[15] A. Kundu, R.K. Layek, A. Kuila, A.K. Nandi, Appl. Mater. Interfaces 4 (2012)

5576.[16] Q. Mei, K. Zhang, G. Guan, B. Liu, S. Wang, Z. Zhang, Chem. Commun. 46 (2010)

7319.[17] B.F.M. George, Riboflavin in Vitamins in Foods, Analysis, Bioavailability, and

Stability, Taylor and Francis Group, New York, 2006. 168–175.[18] W.S. Hummers, R.E. Offeman, J. Am. Chem. Soc. 80 (1958) 1339.[19] M. Iliut, C. Leordean, V. Canpean, C-M. Teodorescu, S. Astilean, J. Mater. Chem.

C 1 (2013) 4094.[20] M.M.N. Wolf, C. Schumann, R. Gross, T. Domratcheva, R. Diller, J. Phys. Chem. B

112 (2008) 13424.[21] P. Routh, R.K. Layek, A.K. Nandi, Carbon 50 (2012) 3422.[22] B. Konkena, S. Vasudevan, J. Phys. Chem. Lett. 3 (2012) 867.[23] A. Kundu, R.K. Layek, A.K. Nandi, J. Mater. Chem. 22 (2012) 8139.[24] P. Drossler, W. Holzer, A. Penzkofer, P. Hegemann, Chem. Phys. 282 (2002) 429.[25] V. Georgakilas et al., Chem. Rev. 112 (2012) 6156.[26] F. Liu, H. Gu, Y. Lin, Y. Qi, X. Dong, J. Gao, T. Cai, Spectr. Acta Part A 85 (2012)

111.[27] H.-X. Wang, Q. Wang, K.-G. Zhou, H.-L. Zhang, Small 9 (2013) 1266.