Combined experimental and DFT–TDDFT study of photo-active constituents of Canarium odontophyllum for DSSC application

Chemical Physics Letters 585 (2013) 121–127

Contents lists available at ScienceDirect

Chemical Physics Letters

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

Combined experimental and DFT–TDDFT study of photo-activeconstituents of Canarium odontophyllum for DSSC application

0009-2614/$ - see front matter � 2013 Elsevier B.V. All rights reserved.http://dx.doi.org/10.1016/j.cplett.2013.08.094

⇑ Corresponding author at: Applied Physics Group, Faculty of Science, UniversitiBrunei Darussalam, Jalan Tungku Link, Gadong BE1410, Brunei Darussalam. Fax:+673 2461502.

E-mail addresses: [email protected] (P. Ekanayake), [email protected] (M.R.R. Kooh), [email protected] (N.T.R.N. Kumara), [email protected] (A. Lim), [email protected] (M.I. Petra),[email protected] (N.Y. Voo), [email protected] (C.M. Lim).

Piyasiri Ekanayake a,b,⇑, Muhammad Raziq Rahimi Kooh a, N.T.R.N. Kumara b, Andery Lim c,Mohammad Iskandar Petra d, Voo Nyuk Yoong a,b, Lim Chee Ming a,b

a Energy Research Group, Faculty of Science, Universiti Brunei Darussalam, Jalan Tungku Link, Gadong BE1410, Brunei Darussalamb Applied Physics Group, Faculty of Science, Universiti Brunei Darussalam, Jalan Tungku Link, Gadong BE1410, Brunei Darussalamc Biology Group, Faculty of Science, Universiti Brunei Darussalam, Jalan Tungku Link, Gadong BE1410, Brunei Darussalamd Faculty of Integrated Technology, Universiti Brunei Darussalam, Jalan Tungku Link, Gadong BE1410, Brunei Darussalam

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

Article history:Received 2 July 2013In final form 22 August 2013Available online 30 August 2013

The active constituents of Canarium odontophyllum (COP) were investigated experimentally and theoret-ically for dye sensitized solar cell (DSSC) application. Three main flavonoid pigments (cyanidin, pelargon-idin and maritimein) were detected in COP showing photo-energy conversion efficiencies of 1.43%, 0.87%and 0.60%, respectively. The molecular geometries, electronic structures, optical absorption spectra andproton affinity of these molecules were investigated with DFT/TDDFT. All three molecules displayedp?p� transition dominant in HOMO?LUMO transition. The anchoring groups onto TiO2 surface werededuced from combined experimental and calculated data. All the constituents of COP are potential sen-sitizers for DSSC.

� 2013 Elsevier B.V. All rights reserved.

1. Introduction

The main motivations for solar cell research are to discover safe,low cost, high-efficiency, reliable and easy to process materials.One such system is the dye-sensitized solar cell (DSSC). It is re-garded as a third generation solar cell, capable of providing lowercost and ease of device preparation [1]. The DSSC system was firstreported by O’regan and Grätzel in 1991, and this had led to thediscoveries of various dyes, morphology of materials, and cell de-signs that show photo-energy conversion efficiency comparableto low-end semiconductor based solar cells. The record forphoto-energy conversion efficiency of DSSC is currently at 12%.This is achieved with porphyrin-sensitizers and cobalt(II)/(III) re-dox couple [2]. In recent developments, researchers have beenlooking into the use of natural dyes which are easy to extractand are stable. This Letter reports on the use of natural dyes forDSSC, namely, Canarium odontophyllum (COP) which contains theactive photo-energy conversion molecules of cyanidin, pelargoni-din and maritimein.

Carotenoids, anthocyanins, betalains and chlorophylls are themajor classes of plant pigments. Anthocyanins are responsible forthe orange-to-blue colours, while both carotenoids, betalains areyellow-to-red colourants, and chlorophylls are green [3].

Anthocyanins is a class of flavonoids, that can be found in manyflowers, leaves and fruits, and are the glucosides of anthocyanidins.Anthocyanidins form stable flavylium structure at pH below 3 andare redder than anthocyanins [3]. Aurones are rare class of flavo-noids and give flower bright yellow colours [4].

In this study, dyes were extracted from the skin of COP. COP isone of the exotic fruits on the island of Borneo. The skin of the fruitis black, and ethanol extraction yield a purple dye, and the compo-nents of this purple dye were extracted using column chromatog-raphy, and they were identified by their UV–vis absorptionproperties. The optical active components found in the purpledye of COP consist of cyanidin, pelargonidin and maritimein. Bothcyanidin and pelargonidin are anthocyanidins, while maritimein isan aurone.

DSSC are produced using these dyes as sensitizers. The solar en-ergy conversion efficiency of DSSC is investigated, and UV–vis andcyclic voltammetry data were used to estimate the HOMO (highestoccupied molecular orbital) and LUMO (lowest unoccupied molec-ular orbital) energy levels. This report shows the performance dataof COP constituents as sensitizers, their estimated energy HOMO–LUMO levels, with comparison to DFT–TDDFT results and lastly,the anchoring groups onto TiO2 surface were deduced from com-bined experimental and calculated data.

122 P. Ekanayake et al. / Chemical Physics Letters 585 (2013) 121–127

2. Computational and experimental methods

2.1. Computational methods

The molecular structures of the pigments were computed usingSPARTAN’10 software package to obtain the molecular geometry coor-dinates [5]. Both DFT and TDDFT calculation were performed usingGAUSSIAN’09W software package [6] and calculated using the B3LYPhybrid functional and 6-31g(d) basis set [7,8]. The geometric opti-mization and the solvation effect were included in the calculations.The electronic structures of HOMOs and LUMOs of the dyes wereplotted using GaussView version 5.0.9 [9]. The deprotonation ener-gies of COP constituents (proton affinity) were calculated in B3LYP/6-31g(d) level under vacuum condition [10].

2.2. Experimental methods

The dye was extracted from COP fruit skin in 70% ethanol andthe pigments separation was done by column chromatography,whereby the glass column was filled with silica gel (type 60) usingslurry method and mixtures of acetonitrile and chloroform as thedeveloping solvent. Methanol was used for final flushing. Furtheranalysis were carried out using thin layer chromatography (TLC),UV–vis absorption spectroscopy and cyclic voltammetry.

The photoelectrodes were prepared by the Doctor Blade meth-od, where the TiO2 paste Solaronix (nanoxide-T, colloidal anataseparticles size: �13 nm, �120 m2 g�1 (BET)) were coated on pre-cleaned fluorine-doped conducting tin oxide (FTO) glasses (Nipponsheet glass 10–12 X q�1). The films were sintered at 450 �C for30 min, and the resulted thickness of the TiO2 electrode was�9 lm (Dektak profilometer; Veeco, Dektack 3) [11]. The sinteredTiO2 were stored in vacuum oven at temperature of 40 �C. The TiO2

electrodes were dipped in the COP constitutes for 14 h at roomtemperature and dried under a flow of nitrogen. The active solarcell area was measured for each cells (�0.25 cm2). The cells wereassembled using Dyesol’s Test Cell Assembly Machine, the Surlyn(50 lm, Dyesol) sealant and the redox electrolyte. The redox elec-trolyte consisted of tetrabutylammonium iodide (TBAI; 0.5 M)/I2

(0.05 M) in a mixture of acetonitrile and ethylene carbonate (6:4,v/v). The redox electrolyte was introduced through pre-drilled holein platinum counter electrode [11]. DSSC cells were tested withDyesol Solar Simulator LP-156B at one sun level of irradiation.N719 used as our reference standard, and it was measured at5.8% photo-energy conversion efficiency, open circuit voltage(Voc) at 0.564 V, short circuit current density (Jsc) at 22.91 mA cm�2

and fill factors (ff) at 0.603.UV–vis absorption spectroscopy was carried out using Shima-

dzu UV-1800 spectrophotometer. The cyclic voltammetry (CV)measurements were carried out using eDAQ potentiostat equippedwith e-corder 410. The CV system consisted of a glassy carbonworking electrode, platinum wire as counter electrode and 4 MKCl saturated Ag/AgCl reference electrode. 20 lL of the dye solu-tions were separately placed on the glassy carbon working elec-trode and dried at room temperature before immersing it in0.1 M KNO3 as the supporting electrolyte.

3. Results and discussion

3.1. Experimental data

During the chromatographic separation process, three differentbands were observed; yellow band was being firstly recovered, fol-lowed by reddish-blue band and eventually dark blue band. TLCwas run with various combination of solvents such as ethanol:hex-ane, chloroform:hexane, methanol:chloroform and methanol:ethyl

acetate of different ratios, to check the purity of the recovereddyes, whereby one spot was observed in all the chromatograms.

Identification of the flavonoids was done by comparing UV–visdata to specific dye standards from the literature [12,13]. COP dyehas three groups of flavonoids, namely aurone (maritimein) andanthocyanidin (pelargonidin and cyanidin). A small peak at the re-gion of 500–550 nm was observed for maritimein despite one spotobserved in TLC.

The best performance in photo-energy conversion among theCOP constituents, at 1.43%, was obtained from the DSSC sensitizedwith cyanidin. HOMO and LUMO energy levels of the dyes wereestimated using UV–vis and CV data as proposed by Tauc et al.[14]. Details of the estimation can be found in our previous work[15]. All the HOMO levels are below the redox potential whileLUMO levels are above the conduction band of TiO2 (see Figure 2)indicating COP constituents meet basic requirements to sensitizeDSSC. The photo-energy conversion efficiencies of pelargonidinand maritimein were 0.87% and 0.60%, respectively. The Voc ofcyanidin, pelargonidin and maritime measured at 0.350, 0.357and 0.388 V, respectively, while the Jsc at 9.74, 6.57 and3.81 mA cm�2, and the ff at 0.546, 0.484 and 0.529. These experi-mental data are only for the comparison purposes with DFT/TDDFTcalculations.

3.2. Computational calculations

3.2.1. DFT calculationDFT calculations were performed on COP constituent. The calcu-

lations were repeated for geometrically optimized structures andthe results are summarized in Table 1.

It was observed that the data, especially the band gap values,obtained after geometrical optimization, agree well with theexperimental data (see Table 1 and Figure 2). Cyanidin and pelarg-onidin have several resonances in addition to the fundamentalstructure with band gaps of 2.81 eV and 2.90 eV, respectively[16]. A few of these resonances are shown in Figure 1. The resultsof our calculations revealed that the resonances do not producesignificant deviations in the data from the fundamental structure.The resonances were not calculated for maritimein molecule sincethere is no flavylium ion group in the structure.

The resonances of pelargonidin reproduced the same data of thefundamental structure while difference of 0.05 eV was producedby the resonances of cyanidin.

3.2.2. Solvation effectBy considering the solvation effect in the DFT calculations for

HOMO and LUMO lead to the reduction in the differences betweenthe data of experiment and calculation. HOMO and LUMO of cyani-din under vacuum condition are �9.19 eV and �6.42 eV, respec-tively, while these values become closer to the experimental datavalues of �6.28 eV and �3.48 eV, respectively when solvation ef-fect of ethanol is considered. The calculations involving the solva-tion effect does not produce significant changes to the band gapbut only shows shifts in the absolute values of the HOMO andLUMO (see Table 1). This indicates the calculations in vacuumoverestimate the energy levels but has no significant effect to theband gap. Similar results were obtained for pelargonidin and mari-timein. In addition, the energy levels show no significant differencewhen different solvation effect of polar solvents were introducedinto the calculations.

3.2.3. HOMO–LUMO energy levelsThe lowest band gap, 2.81 eV, calculated using DFT, was ob-

served from cyanidin, which correlates with the highest conversionefficiency exhibited by this molecule as a sensitizer in DSSC. TheHOMO and LUMO energy levels of COP constituents, as calculated

Figure 2. Energy levels diagram showing various energy levels of COP constituents, redox potential and conduction band of TiO2. Both calculated and experimentallyestimated data are shown. Computational results were obtained using B3LYP/6-31g(d) level and with geometric optimization.

Table 1Summary of DFT computational calculation of COP constituents in vacuum and ethanol, as well as, with and without geometric optimization.

Solvent Effect No geometric optimization With geometric optimization Experimental

HOMO LUMO Energy gap HOMO LUMO Energy gap Energy gap(eV) (eV) (eV) (eV) (eV) (eV) (eV)

Cyanidin In vacuum �9.87 �6.93 2.94 �9.19 –6.48 2.70Water �6.83 �3.72 3.11 �6.2 �3.39 2.81Ethanol �6.91 �3.82 3.1 �6.28 �3.48 2.81 2.11Methanol �6.88 �3.78 3.1 �6.25 �3.45 2.81Acetone �6.94 �3.85 3.1 �6.31 �3.51 2.80

Pelargonidin In vacuum �9.28 �6.73 2.55 �9.13 �6.29 2.84Water �6.52 �3.55 2.97 �6.19 �3.29 2.90Ethanol �6.59 �3.64 2.96 �6.27 �3.38 2.90 3.28Methanol �6.57 �3.61 2.96 �6.24 �3.35 2.90Acetone �6.62 �3.67 2.95 �6.3 �3.4 2.89

Maritimein In vacuum ��5.86 �2.28 3.58 �5.55 �2.13 3.42Water �5.97 �2.44 3.53 �5.61 �2.29 3.32Ethanol �5.97 �2.44 3.53 �5.61 �2.28 3.32 3.15Methanol �5.97 �2.44 3.53 �5.61 �2.29 3.32Acetone �5.96 �2.43 3.53 �5.6 �2.28 3.32

Figure 1. Molecular structures and selected resonances of COP constituents. The sites of deprotonation (hydroxyl groups) were labelled as H1, H2, H3, H4, H5, H6 and H7.

P. Ekanayake et al. / Chemical Physics Letters 585 (2013) 121–127 123

Table 2Computed excitation energies in (eV), (nm) and the oscillator strength (f) of the COP constituents obtained by TDDFT calculations at B3LYP/6-31g(d) level with the inclusion ofgeometric optimization under vacuum condition and the solvation effect of ethanol.

Optimized, in vacuum Optimized, in ethanol

Calculated energy Oscillator strength MO configuration(coefficient)*

Calculated Energy Oscillator strength MO configuration(coefficient)*(eV) (nm) (f) (eV) (nm) (f)

Cyanidin 2.53 490 0.435 H?L = 0.650 2.51 494 0.563 H-1?L = 0.167H?L = 0.680

2.81 440 0.092 H-1?L = 0.650H?L = 0.243

2.87 433 0.079 H-1?L = 0.675

3.16 393 0.149 H?L = 0.149 3.21 386 0.144 H-2?L = 0.694

Pelargonidin 2.68 463 0.448 H-1?L = 0.300H?L = 0.639

2.59 478 0.539 H-1?L = 0.224H?L = 0.670

3.08 402 0.290 H-1?L = 0.624 3.07 404 0.328 H-1?L = 0.6603.47 357 0.031 H-2?L = 0.695 3.65 340 0.018 H-2?L = 0.605

Maritimein 3.19 388 0.614 H?L = 0.681 3.03 409 0.727 H?L = 0.6933.31 374 0.0002 H-3?L = 0.699 3.43 362 0.008 H-1?L = 0.6443.55 349 0.015 H-1?L = 0.563 H?L = 0.126 3.46 358 0.000 –

* Molecular orbital with configuration coefficient <0.1 are not shown.

0.0

0.5

1.0

1.5

2.0

2.5

250 350 450 550

ε(M

-1cm

-1)

x 10

000

Wavelength (nm)

(A) ethanolVacuum

0.0

0.5

1.0

1.5

2.0

2.5

3.0

200 300 400 500 600 700

ε(M

-1cm

-1)

x 10

000

Wavelength (nm)

(B) EthanolVacuum

0.0

0.5

1.0

1.5

2.0

2.5

300 400 500 600 700 800

ε(M

-1cm

-1)

x 10

000

Wavelength (nm)

(C) Ethanol

Vacuum

Figure 3. Simulated absorption spectra of (A) maritimein (B) pelargonidin (C) cyanidin, plotted to show the effect of the vacuum versus solvation effect in ethanol.

124 P. Ekanayake et al. / Chemical Physics Letters 585 (2013) 121–127

from DFT and from experiments are plotted with respect to vac-uum level, in the Figure 2 and the superimposed on the plots arethe energy levels corresponding to the conduction band of TiO2

and the redox couple [17]. The results in Figure 2 show that theCOP constituents satisfy the conditions for photo-energy genera-tion. These conditions in brief are as follow. The electron injectionfrom the excited dye molecule to the conduction band of TiO2 ismore efficient if the LUMO level is higher than conduction bandedge of TiO2. For efficient regeneration of the oxidized dye mole-cule to its original state by the hole conductor, the energy differ-ence between HOMO and energy level of redox couples must besufficiently high. LUMO must be above the conduction band ofTiO2 while HOMO must be below the energy level of redox couples[18].

3.2.4. TDDFT and optical propertiesTDDFT is used to calculate the optical properties of the dyes and

to simulate the UV–vis absorption spectra. TDDFT also provides themean to determine the oscillator strength (f), an indicative prop-erty for effective sensitizers [19].

Table 2 shows the singlet–singlet transitions in the absorptionbands of dyes cyanidin, pelargonidin and maritimein. The simu-lated data shows the initial and final states of electronic transitionsof the first absorption bands of the dyes in solution to lie mostly inHOMO and LUMO, respectively. The oscillator strengths in cyani-din, pelargonidin and maritimein are 0.563, 0.538 and 0.727,respectively. The molecular orbital configuration coefficient corre-

sponding to these transitions in cyanidin, pelargonidin and mari-timein are 0.69, 0.67, and 0.69, respectively.

The next prominent transitions are from HOMO�2 to LUMO incyanidin, and HOMO�1 to LUMO in pelargonidin. However, inmaritimein the electronic transitions are significant only fromHOMO to LUMO. There are little contributions from HOMO�1 toLUMO in cyanidin and from HOMO�2 to LUMO in pelargonidin.

Figure 3 depicts the calculated UV–vis absorption spectra ofCOP constituents. The inclusion of the solvation effect of ethanolleads to a red shift in the absorption spectra. The wavelengths ofmaximum absorption, kmax,, in the latter case are 494 nm,478 nm and 409 nm for cyanidin, pelargonidin and maritimein,respectively and their corresponding red shifts are 4 nm, 15 nmand 21 nm. We have calculated the solvation effect of water, meth-anol and acetone, however, we found no significant deviation fromthe solvent effect of ethanol. The absorption peaks in the visible re-gion are further red shifted, as shown in Figure 4A. Figure 4B com-pares the simulated absorption spectra of these dye molecules,with the inclusion of solvation effect of ethanol.

The experimental UV–vis absorption spectra revealed the kmax

values of cyanidin, pelargonidin and maritimein are 534 nm,534 nm and 335 nm, respectively. Cyanidin and pelargonidin spec-tra are red-shifted from the calculated absorption data in solventby 40 nm and 56 nm, respectively, while maritimein is blue-shiftedby 74 nm.

The colouration of these dyes is believed to be contributed bythe first (strongest) excited state [19]. However, our results alsosuggest a prominent role played by the higher excited states for

0.0

0.5

1.0

1.5

2.0

2.5

3.0

200 300 400 500 600 700 800

ε(M

-1cm

-1)

x 10

000

Wavelength (nm)

Mari�meinPelargonidinCyanidin

(B)(A)

Figure 4. (A) Absorption spectra obtained experimentally where the kmax of maritimein, pelargonidin and cyanidin at visible range were 335 nm, 530 nm and 532 nm,respectively; (B) simulated absorption spectra of maritimein, pelargonidin and cyanidin in solvation effect of ethanol and under geometric optimization, where the kmax are409 nm, 478 nm and 494 nm, respectively.

P. Ekanayake et al. / Chemical Physics Letters 585 (2013) 121–127 125

cyanidin and pelargonidin. Similar results of TDDFT calculations ofanthocyanidins were reported by Woodford [16]. From the threeexcitations predicted by TDDFT, except for maritimein, two havenon-negligible oscillator strengths (f > 0.1).

High molar extinction coefficient is one of the importantrequirements in DSSC to ensure efficient light harvesting efficiency[20]. Absorbance, A, (in arbitrary unit) is related to molar extinc-tion coefficient, e, by the Beer–Lambert law [21]. The Beer–Lambertlaw is expressed as:

A ¼ ebc ð1Þ

where b is the path length, in cm and is a constant, while c is themolar concentration.

TDDFT estimates the molar extinction coefficient of the dyemolecule. However, they are linearly related according to Eq. (1).The molar extinction coefficients at the strongest absorption bandof cyanidin, pelargonidin and maritimein in ethanol are 24500,25000 and 29800 M�1 cm�1, respectively. The oscillator strengthf reflects the strength of the particular transition; in other wordsthe light harvesting efficiency (LHE). The values of f determinethe usefulness the sensitizer in DSSC [1,19]. Relationship of LHEto f is expressed as:

LHE ¼ 1� 10�f ð2Þ

The f corresponding to transition from HOMO to LUMO of cyani-din, pelargonidin and maritimein are 0.56, 0.54 and 0.72, respec-tively. TDDFT results showed the existence of HOMO�1?LUMOtransitions, within the same excited state, for all these dyes. How-ever, the molecular orbital configuration coefficient of these transi-tions is very low (�0.20). It can be observed that molecular orbitalconfiguration coefficient involved in the dominant transition(HOMO to LUMO) of cyanidin, pelargonidin and maritimein are0.68, 0.67 and 0.69, respectively. The TDDFT results show thatmaritimein have the highest capability to capture photons, of thethree dyes within the first absorption band (first excited state).However, maritimein has only this absorption band while theTDDFT results show cyanidin and pelargonidin have the capabilityto absorb photons also in the second absorption band. The molarextinction coefficients of cyanidin and pelargonidin, in the latterband, are 8460 and 18900 M�1 cm�1, respectively. When the molarextinction coefficients and oscillator strengths of each excitedstates are considered, we would expect pelargonidin to exhibitthe highest photo-energy conversion efficiency while cyanidin tobe the next highest. However, our experiments indicated thephoto-energy conversion efficiencies of cyanidin, pelargonidinand maritimein are 1.43%, 0.87% and 0.60%, respectively. Although

the calculations resulted in higher molar absorption coefficients forpelargonidin and maritimein, in the experiment, their absorptionin the visible region is relatively low compared to that of cyanidin(see Figure 4a). The band gaps of these transitions are presented inTable 1. The dye with the smallest band gap is cyanidin, and it hasbeen shown to have the highest photo-energy conversion effi-ciency. The results demonstrate the high degree of correlation be-tween UV–vis absorption, band gaps and the photo-energyconversion efficiency.

The HOMO–LUMO energy gap of some of these dyes wereinvestigated using photo-electron spectroscopy and the resultsdo not show any significant difference from the band gap esti-mated from UV–vis and cyclic voltammetry data. It is suggestedthat further calculations, such as electronic structure and protonaffinity, are needed to provide better understanding of the sensitiz-ing capabilities of these dye molecules.

3.2.5. Electronic structureThe absorption capacity of the dyes in the visible spectrum is

crucial for photo-energy conversion efficiency of DSSC, and thiscan be determined by TDDFT. The results of TDDFT calculationsare summarized in Table 2. The optimized molecular structure ofCOP constituents in ethanol and the molecular orbital surface ofHOMO�2, HOMO�1, HOMO and LUMO are shown in Figure 5.

All the ground state energy levels (HOMO�2, HOMO�1, HOMO)of cyanidin are p type. At the HOMO�2 level the molecular orbitalis mostly localized in the benzene-diol unit. The molecular orbitalin the chromenylium unit is localized in the benzene ring, whereasthe molecular orbitals of the HOMO�1 and HOMO are delocalizedover the entire molecule of cyanidin. The LUMO of cyanidin is p⁄

type and the molecular orbital is delocalized over the entire mole-cule. The molecular orbital is observed to be denser in chromeny-lium unit than the benzene-diol unit. Since cyanidin generatescurrent by absorbing light in DSSC, there must be electron injectionfrom the excited molecule to the conduction band of TiO2. Thiselectron injection is most probable from the chromenylium unitas it can accumulate denser molecular orbital in the p⁄ state.Therefore, the anchoring groups of cyanidin molecule that mayefficiently inject electrons into the TiO2 are deduced to be the hy-droxyl groups H2, H4 and H5, please refer to Figures 1 and 5.

The HOMO�1 and HOMO�2 of pelargonidin have the similarelectronic pattern as that of HOMO�2 of cyanidin. The molecularorbital patterns of HOMO and LUMO of pelargonidin are similarto cyanidin. The anchoring groups of pelargonidin are deduced tobe hydroxyl group H1, H3 and H4 as the molecular orbitals are ob-served to be localized in these hydroxyl groups at the LUMO level.

Figure 5. Diagram showing the calculated optimized molecular structure of cyanidin, pelargonidin and maritimein in ethanol and the molecular orbital surface of HOMO+2,HOMO+1, HOMO and LUMO with isovalue of contour = 0.03.

Table 3Single deprotonation energies (kcal mol�1) of the COP constituents.

Sensitizers Deprotonation energies(kcal mol�1)

�H1+ �H2+ �H3+ �H4+ �H5+ �H6+ �H7+

Cyanidin 245.9 243.9 236.9 235.8 240.1Pelargonidin 244.3 241.8 241.8 241.9Maritimein 331.7 319.3 314.4 344.5 348.6 340.7 348.1

126 P. Ekanayake et al. / Chemical Physics Letters 585 (2013) 121–127

In maritimein molecule the HOMO�2, HOMO�1, HOMO are allp type and the molecular orbital is localized only to benzyl-furanand the benzylidene units. Dense molecular orbital localization isseen in the benzene ring of benzyl-furan, in the energy levels ofHOMO�2 and HOMO�1. The LUMO of maritimein is also p⁄ typeand the molecular orbital is localized only to the benzyl-furanand benzylidene units. The glucose unit of maritimein is observedto have no molecular orbital localization in both the HOMO andLUMO states. At the LUMO level, the only hydroxyl group withmolecular orbital localization is the hydroxyl group H2, thus thishydroxyl group is deduced as the most likely anchoring group thatcan provide efficient electron injection to the TiO2.

The ability to generate current in DSSC arrangement, by the COPconstituents, is photo-induction whereby electron transfer processtakes place, and the excitation generates the charge separatedstates. In the pelargonidin molecule the degree of extension ofthe LUMO molecular orbital towards the hydroxyl group that linksthe dye to the TiO2 surface is lower compared to that of cyanidin.This indicates a relatively weak electronic coupling between theLUMO of pelargonidin and the acceptor states of TiO2. This is ingood agreement with the experimental results which shows higherphoto-energy conversion efficiency of cyanidin compared to that ofpelargonidin.

3.2.6. Proton affinityDeprotonation energy is obtained by calculating the energy dif-

ference between the optimized protonated and deprotonated dyemolecules in B3LYP/6-31g(d) level of theory under vacuum condi-tion [10]. Our computational calculation of the deprotonation or-der did not include the solvation effect because the solvatedmodel artificially restricts molecules to a solvation sphere, and thisdoes not provide a realistic model [10]. The details on the effect ofsolvated models and isolated models on the deprotonation ener-gies can be found in a study by Trout and Kubicki [22].

The lowest values of the deprotonation energy indicate themost probable group that anchors to the TiO2 [19]. The commonanchoring group of natural dyes are the hydroxyl (–OH) groupand the carboxyl group (–COOH) [19].

Single deprotonation energies of the constituents of COP areshown in Table 3. Cyanidin molecules contain five hydroxyl groupsand their proton affinities vary only slightly. H2 and H3 have thesmallest deprotonation energies, whereas H1 has the highestdeprotonation energy. The ascending order of the energyvalues of the proton affinity (deprotonation order) isH4 � H3 < H5 < H2 < H1 and the energy difference between H4and H1 is 10.1 kcal mol�1. Therefore, the most probable anchoringgroups of cyanidin were deduced to be H4 and H3. We have alsocalculated double deprotonation energies choosing hydroxylgroups that have lowest single deprotonation energy. The doubledeprotonation was calculated on the two hydroxyl groups withleast single deprotonation energies, H3 and H4. The calculateddouble deprotonation energy (H3,4) is 552.6 kcal mol�1, which ismuch larger than single deprotonation. This shows that singledeprotonation is more probable, indicating the lower possibilityto anchor onto TiO2 by two hydroxyl groups, simultaneously.When the two most probable anchoring groups H4 and H3 arecompared, molecular orbital localization only occurs on H4 butnot on H3, deducing H4 as the anchoring group for cyanidin.Although H3 may still anchor to the TiO2, this may lead to ineffec-tive injection of electron and this can lead to reduction in photo-energy conversion efficiency. This information is valuable as thedye can be subjected to modification to prevent the undesirableanchoring group from attaching to the TiO2.

Pelargonidin consists of four hydroxyl groups. The ascendingorder of deprotonation energies are H2 = H3 � H4 < H1. All four

P. Ekanayake et al. / Chemical Physics Letters 585 (2013) 121–127 127

hydroxyl groups are equally favourable as the anchoring group dueto negligible difference in deprotonation energies. The doubledeprotonation calculation was carried out (on H2,3) and the corre-sponding energy is 559.0 kcal mol�1, which is much larger thansingle deprotonation and thus double deprotonation is less likelyto occur. The molecular orbital localization only occur at H1, H3and H4 in LUMO level. Therefore, we can deduce that pelargonidinanchors to TiO2 through H1, H3 or H4 hydroxyl groups.

Maritimein consists of seven hydroxyl groups, in which fourhydroxyl groups are located at the glucose component. Theascending order of the deprotonation energies are H3 < H2 <H1 < H6 < H4 < H7 < H8. Double deprotonation energy of (H2,3)was found to be 684.4 kcal mol�1 which is much larger than singledeprotonation energy. It was initially deduced, from the electronicstructure, that the hydroxyl groups in the glucose unit of the mari-timein are not desirable anchoring groups since there is no molec-ular orbital localized at the region. The deprotonation energies ofall the hydroxyl groups in glucose unit are much higher than thatof H1, H2 and H3, hence they are less likely to be the anchoringgroup.

As H3 has the lowest single deprotonation energy, it is mostlikely the anchoring group in maritimein. The differences of depro-tonation energies between the single deprotonation of H3 and H2is small (4.9 kcal mol�1), which suggests that H2 would be the nextfavourable anchoring group. However, at the LUMO level, the onlyhydroxyl group with the localization of molecular orbital is H2.Thus H2 group is considered as the most profitable anchoringgroup of maritimein. In practice both H2 and H3 may compete toanchor on TiO2 resulting in lower performance as a sensitizer inDSSC.

4. Conclusions

The combined experimental and computational studies of C.odontophyllum (cyanidin, pelargonidin and maritimein) showphoto-energy conversion efficiency of cyanidin, pelargonidin andmaritimein to be 1.43%, 0.87% and 0.60%, respectively. The compu-tational calculations revealed the energy levels, especially bandgaps, are closer to the experimental data when the structures wereoptimized with the inclusion of solvation effect. TDDFT resultsshowed that the dominant energy transition of all three natural

dyes is from HOMO to LUMO. The molecular orbital plots diagramshowed all three dyes displayed a p?p� transition from HOMO toLUMO. The deprotonation analysis of cyanidin suggests that hydro-xyl groups H4 and H3 are the most probable anchoring group toTiO2. All the four hydroxyl groups in pelargonidin are favourablefor anchoring, whereas in maritimein the hydroxyl groups H2and H3 are more favourable. When deprotonation order, electronicstructure and photo-energy conversion efficiencies are considered,the desirable anchoring groups that promote efficient electroninjection are H1, H3 and H4 for pelargonidin, H4 for cyanidin andH2 for maritimein.

Acknowledgments

Universiti Brunei Darussalam research grant UBD/PNC2/2/RG/1(176) and Brunei Research Council (BRC) Science and Technologygrant S&T 17 are acknowledged for the financial support.

References

[1] J. Wang, S. Cong, S. Wen, L.K. Yan, Z.M. Su, J. Phys. Chem. C 117 (2013) 2245.[2] A. Yella et al., Science 334 (2011) 629.[3] Y. Tanaka, N. Sasaki, A. Ohmiya, Plant J. 54 (2008) 733.[4] A. Mortensen, Pure Appl. Chem. 78 (2006) 1477.[5] B. Deppmeier et al., Spartan’10, Wavefunction Inc., Irvine, CA, 2011.[6] M.J. Frisch et al., Gaussian 09, Gaussian, Inc., Wallingford CT, 2010.[7] A.D. Becke, J. Chem. Phys. 98 (1993) 1372.[8] C. Lee, W. Yang, R.G. Parr, Phys. Rev. B 37 (1988) 785.[9] R. Dennington, T. Keith, J. Millam, GaussView, Semichem Inc., Shawnee

Mission, KS, 2009.[10] C. Qin, A.E. Clark, Chem. Phys. Lett. 438 (2007) 26.[11] G.K.R. Senadeera, T. Kitamura, Y. Wada, S. Yanagida, Sol. Energy Mater. Sol.

Cells 88 (2005) 315.[12] J.B. Harborne, T.A. Geissman, J. Am. Chem. Soc. 78 (1956) 829.[13] J. Harborne, Biochem. J. 70 (1958) 22.[14] J. Tauc, R. Grigorovici, A. Vancu, Phys. Status Solidi B 15 (1966) 627.[15] N.T.R.N. Kumara, P. Ekanayake, A. Lim, M. Iskandar, L.C. Ming, J. Sol. Energy

Eng. 135 (2013) 031014.[16] J.N. Woodford, Chem. Phys. Lett. 410 (2005) 182.[17] H.C. Chu et al., Dyes Pigm. 93 (2012) 1488.[18] G. Calogero et al., Photochem. Photobiol. Sci. 12 (2013) 883.[19] T.R. Heera, L. Cindrella, J. Mol. Model. 16 (2010) 523.[20] P. Wang, C. Klein, R. Humphry-Baker, S.M. Zakeeruddin, M. Grätzel, J. Am.

Chem. Soc. 127 (2005) 808.[21] D. Calloway, J. Chem. Educ. 74 (1997) 744.[22] C.C. Trout, J.D. Kubicki, Geochim. Cosmochim. Acta 70 (2006) 44.