ARTICLE IN PRESS

0927-0248/$ - se

doi:10.1016/j.so

�CorrespondE-mail addr

Dye-sensitized solar cell using natural dyes extracted from rosella andblue pea flowers

Khwanchit Wongchareea,�, Vissanu Meeyooa, Sumaeth Chavadejb

aDepartment of Chemical Engineering, Mahanakorn University of Technology, Nong Chok, Bangkok 10530, ThailandbThe Petroleum and Petrochemical College, Chulalongkorn University, Patumwan, Bangkok 10330, Thailand

Received 6 November 2006; accepted 7 November 2006

Available online 2 January 2007

Abstract

Dye-sensitized solar cells (DSSCs) were fabricated using natural dyes extracted from rosella, blue pea and a mixture of the extracts.

The light absorption spectrum of the mixed extract contained peaks corresponding to the contributions from both rosella and blue pea

extracts. However, the mixed extract adsorbed on TiO2 does not show synergistic light absorption and photosensitization compared to

the individual extracts. Instead, the cell sensitized by the rosella extract alone showed the best sensitization, which was in agreement with

the broadest spectrum of the extract adsorbed on TiO2 film. In case that the dyes were extracted at 100 �C, using water as extracting

solvent, the energy conversion efficiency (ZÞ of the cells consisting of rosella extract alone, blue pea extract alone and mixed extract was

0.37%, 0.05% and 0.15%, respectively. The sensitization performance related to interaction between the dye and TiO2 surface is

discussed. The explanations are supported by the light absorption of the extract solution compared to extracts adsorbed on TiO2 and also

dye structures. The effects of changing extracting temperature, extracting solvent and pH of the extract solution are also reported. The

efficiency of rosella extract sensitized DSSC was improved from 0.37% to 0.70% when the aqueous dye was extracted at 50 �C instead of

100 �C and pH of the dye was adjusted from 3.2 to 1.0. Moreover, DSSC stability was also improved by the changes in conditions.

However, the efficiency of a DSSC using ethanol as extracting solvent was found to be diminished after being exposed to the simulated

sunlight for a short period.

r 2006 Elsevier B.V. All rights reserved.

Keywords: Dye-sensitized solar cells; Natural dyes; TiO2; Anthocyanin

1. Introduction

The dye-sensitized solar cell (DSSC) is a device for theconversion of visible light into electricity, based onthe sensitization of wide bandgap semiconductors [1]. Theperformance of the cell mainly depends on a dye used assensitizer. The absorption spectrum of the dye and theanchorage of the dye to the surface of TiO2 are importantparameters determining the efficiency of the cell [2].Generally, transition metal coordination compounds(ruthenium polypyridyl complexes) are used as the effectivesensitizers, due to their intense charge-transfer absorptionin the whole visible range and highly efficient metal-to-

e front matter r 2006 Elsevier B.V. All rights reserved.

lmat.2006.11.005

ing author. Tel.: +662988 3655; fax: +66 2988 4044.

ess: khwanchit9@yahoo.com (K. Wongcharee).

ligand charge transfer [3]. However, ruthenium polypyridylcomplexes contain a heavy metal, which is undesirablefrom point of view of the environmental aspects [4].Moreover, the process to synthesize the complexes iscomplicated and costly. Alternatively, natural dyes can beused for the same purpose with an acceptable efficiency[2–9]. The advantages of natural dyes include theiravailability and low cost [3].The sensitization of wide bandgap semiconductors using

natural pigments is usually ascribed to anthocyanins [2–9].The anthocyanins belong to the group of natural dyesresponsible for several colors in the red–blue range, foundin fruits, flowers and leaves of plants. Carbonyl andhydroxyl groups present in the anthocyanin moleculecan be bound to the surface of a porous TiO2 film. Thismakes electron transfer from the anthocyanin molecule to

ARTICLE IN PRESS

400 500 600 700 800Wavelength (nm)

Abs

orba

nce

(a.u

.)

(a)(b)(c)

Fig. 1. Light absorption spectra of dye solutions of: (a) rosella extract; (b)

blue pea extract; and (c) mixed rosella–blue pea extract.

the conduction band of TiO2 [3]. As reported [2–9],anthocyanins from various plants gave different sensitizingperformances. However, there is no an acceptable explana-tion behind these results, so far.

In this paper, DSSCs were prepared using natural dyesextracted from rosella (Hibiscus sabdariffa L.) and blue pea(Clitoria Ternatea) flowers as sensitizers, as these flowersare abundant in tropical countries, and rich in anthocya-nins [10,11]. The efficiency of the solar cells related to dyestructures is discussed. This would be an useful informa-tion for selecting anthocyanins and also lead to thesynthesis of dyes for DSSCs. The performance of DSSCsusing the mixed rosella–blue pea dye was also investigated.Our hypothesis was that two anthocyanins having differentabsorption characteristics would give even more synergisticeffect compared to the mixed anthocyanin–chlorophyll dyereported by Kumara et al. [8]. This is because anthocyaninshave advantages over chlorophyll as DSSC sensitizer [3].Moreover, the effects of extracting temperature, extractingsolvent and pH of the dye solution on the DSSC efficiencyand stability were also determined.

2. Experimental

2.1. Preparation of natural dye sensitizers

Fresh rosella or blue pea flower of 1 g was extracted in100ml of two different solvents at different temperaturesfor 30min. Solid residues were filtrated out to obtain cleardye solutions. A mixed dye was prepared by mixing rosellasolution to blue pea solution at a ratio of 1:1 by volume.The effect of extracting temperature was studied at 25, 50,70 and 100 �C, using water as an extracting solvent. Theeffect of extracting solvent was studied by a comparison ofdyes extracted in water and ethanol. The effect of pH ofdye solution was studied by adjusting pH from the originalpH of 3.2 using 0.1M HCl solution to four different pHs(0.5, 1.0, 2.0 and 3.0).

2.2. Preparation of TiO2 electrode (photoanode) and

counter electrode

A TiO2 film electrode (photoanode) was prepared byblending commercial TiO2 powder (Degussa, P25) of ca.0.2 g, nitric solution (0.1M) of 0.4ml, polyethylene glycol(MW10,000) of ca. 0.08 g and one drop of a nonionicsurfactant, Triton X-100. The mixture was well mixedusing an ultrasonic bath for 1 h and then the resultant pastewas spread over a conductive glass plate having 15O=cm2

which was purchased from Hartford Glass Co., Inc. Thecoated plate was then sintered at 450 �C for 2 h. The dyeswere attached to the TiO2 surface by immersing the coatedelectrodes in the aqueous solution of each dye for 24 h. Thenon-adsorbed dye was washed up with anhydrous ethanol.Pt counter electrode was prepared by deposition of Pt-catalyst T/SP paste (purchased from Solaronix SA) onanother conductive glass.

2.3. DSSC assembling

DSSCs were assembled following the procedure de-scribed in the literature [7], the catalyst-coated counterelectrode was placed on the top so that the conductive sideof the counter electrode faces the TiO2 film. The iodideelectrolyte solution (0.5M potassium iodide mixed with0.05M iodine in water-free ethylene glycol) was placed atthe edges of the plates. The liquid was drawn into the spacebetween the electrodes by capillary action. Two binder clipswere used to hold the electrodes together.

2.4. Characterization and measurement

The absorption spectra of dye solutions and dyesadsorbed on TiO2 surface were recorded using a UV–VISspectrophotometer (Shimadzu, model UV-3101). Solarenergy conversion efficiency (the photocurrent–voltage(I–V ) curve) was measured by using two computerizeddigital Keithley multimeters under simulated sunlight (AM1.5, 100mWcm�2Þ. Based on I–V curve, the fill factor (FF)is defined as

FF ¼ ðImax � VmaxÞ=ðI sc � VocÞ, (1)

where Imax and Vmax are the photocurrent and photo-voltage for maximum power output (PmaxÞ, I sc and Voc arethe short-circuit photocurrent and open-circuit photovol-tage, respectively. The overall energy conversion efficiency(ZÞ is defined as

Z ¼ ðI sc � Voc � FFÞ=Pin, (2)

where Pin is the power of incident light.

3. Results and discussion

3.1. Effect of anthocyanin sources and extracting

temperature on DSSC’s efficiency

Fig. 1 shows the UV–VIS absorption spectra of rosellaextract (red), blue pea extract (blue) and rosella–blue peaextract mixture (purple). It was found that the absorptionpeak of rosella extract is about 520 nm while those of blue

ARTICLE IN PRESS

pea extract are about 580 and 620 nm. The difference in theabsorption characteristics is due to the different type ofanthocyanins and colors of the extracts. It was reportedthat anthocyanin obtained from blue pea is ternatin whilethose from rosella are delphinidin and cyanidin complexes[10,11]. An absorption peak at high wavelength of blue peaextract is responsible for its blue color [12]. The combina-tion of rosella extract and blue pea extract did not affectthe absorption peaks, three peaks were detected, corre-sponding to the absorption nature of both extracts.

After immersion of the TiO2-coated electrode (photo-anode) in the extracts, observable colors of TiO2 filmsturned to deep purple for the rosella extract but the film

400 500 600 700 800

Wavelength (nm)

Abs

orba

nce

(a.u

.)

(a)

(b) (c)

Fig. 2. Light absorption spectra of: (a) rosella; (b) blue pea; and (c) mixed

rosella–blue pea extracts adsorbed on TiO2.

400 500 600 700 800Wavelength (nm)

Abs

orba

nce

(a.u

.)

400 500 600Waveleng

Abs

orba

nce

(a.u

.)

a b

c

Fig. 3. Light absorption spectra of dye solutions (� � � � �Þ and dyes adsorbed on

turned to light blue for the blue pea extract. Theabsorption spectra of the photoanode are shown inFig. 2. In the case of rosella extract, an absorption peakof photoanode is broader than that of the dye solution(Fig. 3(a)), with a shift to a higher wavelength (from 520 to570 nm). The difference in the absorption peak is due to thebinding of anthocyanin in the extract to the oxide surface[13]. This difference, however, was not found in the case ofblue pea extract (Fig. 3(b)), as the absorption peaks werefound at the same wavelengths for both extract andphotoanode. The structure of ternatin (Fig. 4) in blue peaextract has longer R groups compared to that of dephinidinor cyanidin complexes in rosella extract, resulted in astronger steric hindrance for anthocyanin to form bondwith oxide surface and prevents the anthocyanin moleculesfrom arraying on the TiO2 film effectively [3]. Hence, thisleads to a deficiency of electron transfer from dyemolecules to conducting band of TiO2.Fig. 5 shows the I–V (current–voltage) curve for the

sunlight-illuminated rosella extract sensitized cell. Table 1presents the performance of the DSSCs in terms of short-circuit photocurrent (I scÞ, open-circuit voltage (VocÞ, fillfactor (FF) and energy conversion efficiency (ZÞ comparedto those of other extracts. Obviously, the efficiency of cellsensitized by the rosella extract was significantly higherthan that sensitized by the blue pea extract. This is due to ahigher intensity and broader range of the light absorptionof the extract on TiO2 (Fig. 2), and the higher interactionbetween TiO2 and anthocyanin in the rosella extract leadsto a better charge transfer. Moreover, anthocyanin in the

400 500 600 700 800Wavelength (nm)

Abs

orba

nce

(a.u

.)

700 800th (nm)

TiO2 (—): (a) rosella; (b) blue pea; and (c) mixed rosella–blue pea extracts.

ARTICLE IN PRESS

0

0.5

1

1.5

2

0 100 200 300 400Voltage (mV)

Phot

ocur

rent

(m

Acm

-2)

Fig. 5. Current–voltage curve for a rosella extract sensitized solar cell.

cyanidin-3-sambubioside (R1 = OH, R2 = H, R3 = sambubiose)

dephinidin-3-sambubioside (R1 = OH, R2 = OH, R3 = sambubiose)

cyanidin-3-glucoside (R1 = OH, R2 = H, R3 = glucose)

+

Ternatin R

T-A1

T-A2

T-B1

T-B2

T-D1

T-D2

-CGCG or –CGCG

-CGCG or –CG

-CGCG or –CGC

-CGC or –CG

-CGC or CGC

-CGC or -C

where C = p-Coumaric acid and G = glucose

b

a

Fig. 4. Chemical structures of: (a) cyanidin and delphinidin in rosella dye;

(b) ternatin in blue pea dye (reprinted from [10] and [11], respectively).

Table 1

Photoelectrochemical parameters of the cells sensitized with natural

extracts

Extract sources I sc (mAcm�2) Voc (mV) FF Z (%) Ref.

Rosella 1.63 404 0.57 0.37 —

Blue pea 0.37 372 0.33 0.05 —

Mixed rosella–blue pea 0.82 382 0.47 0.15 —

Authurium flowers 2.9–3.2 0.44–0.48 — — [2]

Black rice 1.1 0.55 0.52 — [3]

Fruit of Calafate 0.96 0.52 0.56 — [5]

Skin of Jaboticaba 2.1 0.57 0.58 — [5]

rosella extract (cyanidin and delphinidin) has a shorterdistance between the dye skeleton and the point connectedto TiO2 surface compared to that of blue pea extract(ternatin) as shown in Fig. 4. This could facilitate anelectron transfer from anthocyanin in the rosella extract tothe TiO2 surface and could be accounted for a betterperformance of rosella extract sensitization [14].

The absorption spectrum of the mixed rosella–blue peaextract adsorbed on TiO2 electrodes does not showsynergistic characteristics (Figs. 2 and 3(c)) as found inabsorption spectrum of the extract solution (Fig. 1), but thespectrum was similar to that of rosella extract with aslightly narrower range and lower intensity. This indicatesa dominant influence of the rosella extract on the TiO2 film.

A DSSC sensitized by a mixed extract had an efficiencyof around the average value of those sensitized with rosellaand blue pea extracts. This result is in agreement with theabsorption spectrum of the extract coated on TiO2.However, this is rather different from the result by Kumaraet al. [8], in which a DSSC fabricated using chlorophyll andshisonin dyes showed synergistic effect of both dyes. Theresults in this work, are influenced by low interactionbetween blue pea extract and TiO2 film as mentionedabove.Under preparation and irradiation conditions, the

DSSCs prepared from a rosella extract showed a compar-able performance to the DSSCs prepared from othernatural dyes (Table 1). Therefore, the rosella extract shouldbe an alternative anthocyanin source for DSSC prepara-tion in geographical regions that rosella is widely available.Rosella extract was chosen for further study since it gave

a better sensitizing performance compared to blue-peaextract. The effect of dye extracting temperature on DSSCefficiency is shown in Table 2. Obviously, an extraction ofanthocyanin above 50 �C results in decreasing in the DSSCefficiency. This is caused by a decrease in the stability ofanthocyanin at elevated temperature [15]. At high tem-perature, the thermal degradation of anthocyanin could becaused by a loss of glycosyl moieties and a-diketoneformation [16]. On the other hand, anthocyanin extracted

ARTICLE IN PRESS

Table 2

Photoelectrochemical parameters of the cells sensitized with rosella dye

extracted at various temperature

Extracting

temperature (�C)

I sc

(mAcm�2)

Voc (mV) FF Z (%)

25 1.24 396 0.55 0.27

50 2.06 433 0.59 0.52

70 1.75 412 0.57 0.41

100 1.63 404 0.57 0.37

Table 3

Effect of extracting solvent on DSSC efficiency and stability (extracting

temperature is 50 �C)

Extracting

solvent

Irradiated

time (h)

I sc

(mAcm�2Þ

Voc

(mV)

FF Z (%)

Water a 2.06 433 0.59 0.52

Water 3 1.91 412 0.55 0.43

Ethanol a 2.51 488 0.58 0.71

Ethanol 3 0.01 212 0.25 5� 10�3

aJust finished assembling.

400 500 600 700 800Wavelength (nm)

Abs

orba

nce

(a.u

.)

(a)

(b)

(c)

(d)

Fig. 6. Light absorption spectra of rosella extract adsorbed on TiO2: (a)

water extract, just finished assembling; (b) water extract, after exposed to

the simulated sunlight for 3 h; (c) ethanol extract, just finished assembling;

and (d) ethanol extract, after exposed to the simulated sunlight for 3 h.

Table 4

Effect of pH of extract solutions on DSSC efficiency and stability

(extracting temperature is 50 �C)

pH Irradiated time (h) I sc (mAcm�2Þ Voc (mV) FF Z (%)

3.2a b 2.06 433 0.59 0.52

3.2a 3 1.91 412 0.55 0.43

3.0 b 2.18 429 0.59 0.55

2.0 b 2.32 417 0.60 0.58

1.0 b 2.72 408 0.63 0.70

1.0 3 2.68 406 0.61 0.66

0.5 b 2.26 392 0.52 0.46

aOriginal pH.bJust finished assembling.

at 25 �C gave even lower DSSC efficiency than thatextracted at 100 �C. This result can be interpreted fromthe lighter color of the extract at 25 �C compared to theothers, which is due to limitation of anthocyanin solubility.Here, the optimum extracting temperature was found at50 �C which is in between room temperature and boilingpoint of water which was performed in the literatures[3,5,6].

3.2. Effect of extracting solvent on DSSC’s efficiency and

stability

It was reported that the extracting solvent has an effecton the efficiency of DSSCs [5]. The efficiency of the DSSCswas found to increase immensely when ethanol was used forextracting anthocyanin from aged Jaboticaba skin [5]. Inthis study, similar finding was also obtained. As shown inTable 3, the just finished assembling DSSC using ethanol asa solvent shows a higher efficiency than that of using water,reported at 0.71% and 0.52%, respectively. This might oweto the fact that anthocyanin is more soluble in ethanol [17],and hence, the aggregation of dye molecules is less asexpected. A good dispersion of dye molecules on the oxidesurface could in fact improve the efficiency of the system.

Unfortunately, the efficiency of ethanol system wasfound to be diminished after being exposed to thesimulated sunlight for 3 h while there was only a slightdecrease in the efficiency for the case of using water as anextract solvent. This might be due to the photocatalyticdecomposition of anthocyanin by TiO2 in the presence ofethanol as observed by the color of the anode to becomepaler after exposed to the simulated sunlight. As shown inFig. 6, the absorption spectrum of the irradiated TiO2

photoanode adsorbed rosella ethanol extract is rather flatshowing no light absorption compares to those of otherconditions. This makes the DSSC prepared from therosella ethanol extract unable to function just after a shortoperating period. Therefore, ethanol is not suitable as ananthocyanin extracting solvent for DSSC application.

3.3. Effect of pH of extract solutions on DSSC’s efficiency

and stability

The effect of pH was also investigated in this study. Theoriginal pH of rosella extracted anthocyanin was found to

be 3.2. As shown in Table 4, the pH of extract solution hasa significant effect on the performance of DSSCs. Theefficiency was found to increase with decreasing pH andreached a maximum at the optimum pH 1.0. This is mightbe due to the fact that at pH 1.0, the photoanode madefrom the rosella extract can absorb more light, indicated bypeak intensity as shown in Fig. 7. DSSC stability was alsoimproved by adjusting the pH of the extract from 3.2 to

ARTICLE IN PRESS

400 500 600 700 800

Wavelength (nm)

Abs

orba

nce

(a.u

.)

(a)

(b)

(c)

(d)

Fig. 7. Light absorption spectra of rosella extract: (a) solution at pH 3.2;

(b) adsorbed on TiO2 at pH 3.2; (c) solution at pH 1.0; and (d) adsorbed

on TiO2 at pH 1.0.

1.0. As can be seen from Table 4, efficiency loss of DSSCsdecreased with a decreasing pH from 17.3% to 5.7%, afterbeing exposed to the simulated sunlight for 3 h. A reasonfor the better efficiency and stability is that, at pH below 2,anthocyanin existed as flavylium ion, which is stable formof anthocyanin, an increasing pH hydrated this ion toquinonoidal bases. These compounds are labile and can betransformed into the colorless carbinol pseudobase andchalcone [18]. It was evident that at low pH the formationof flavylium ion form is favorable [18]. However, the celldeterioration by acid leaching is expected as the pH goeslower (pHo1), which results in a lower efficiency [19].

4. Conclusions

In conclusion, the rosella extract has higher photosensi-tized performance as compared to the pea blue extract.This is due to the better charge transfer between the roselladye molecule and the TiO2 surface which is related to a dyestructure. Water is suitable as an anthocyanin extractingsolvent for DSSC application while ethanol gave an

adverse effect on the DSSC stability. It was also foundthat the efficiency and stability of DSSCs can be enhancedby adjusting extracting temperature and pH of the extracts.The optimum anthocyanin extracting temperature and pHof the rosella extract were found at 50 and 1:0 �C,respectively. At the optimum conditions, the rosellasensitized DSSC showed efficiency as high as 0.70%.

Acknowledgment

This work was supported by the Thailand ResearchFund (Contract Number MRG4780074).

References

[1] M. Gratzel, J. Photochem. Photobiol. C 4 (2003) 145.

[2] K. Tennakone, G.R.R.R.A. Kumara, A.R. Kumarasinghe, P.M.

Sirimanne, K.G.U. Wijayantha, J. Photochem. Photobiol. A 94

(1996) 217.

[3] S. Hao, J. Wu, Y. Huang, J. Lin, Sol. Energy 80 (2006) 209.

[4] Y. Amao, T. Komori, Biosensors Bioelectron. 19 (2004) 843.

[5] A.S. Polo, N.Y. Iha, Sol. Energy Mater. Sol. Cells 90 (2006) 1936.

[6] C.G. Garcia, A.S. Polo, N.Y. Iha, J. Photochem. Photobiol. A 160

(2003) 87.

[7] G.P. Smestad, Sol. Energy Mater. Sol. Cells 55 (1998) 157.

[8] G.R.A. Kumara, S. Kanebo, M. Okuya, B. Onwona-Agyeman, A.

Konno, K. Tennakone, Sol. Energy Mater. Sol. Cells 90 (2006) 1220.

[9] N.J. Cherepy, G.P. Smestad, M. Gratzel, J.Z. Zang, J. Phys. Chem. B

101 (1997) 9342.

[10] T. Frank, J. Clin. Pharmacol. 45 (2005) 203.

[11] N. Terahara, N. Saito, T. Honda, K. Tokis, Y. Osajima, Phytochem-

istry 29 (1990) 949.

[12] J. Oliveira, C. Santos-Buelga, A. Silva, V. Freitas, N. Mateus, Anal.

Chim. Acta 563 (2006) 2.

[13] K. Vinodgopal, X. Hua, R.L. Dalgren, A.G. Lappin, L.K. Patterson,

P.V. Kamat, J. Phys. Chem. 99 (1995) 10883.

[14] K. Sayama, S. Tsukagoshi, T. Mori, K. Hara, Y. Ohga, A. Shipou,

Y. Abe, S. Suga, H. Arakawa, Sol. Energy Mater. Sol. Cells 80 (2003)

47.

[15] M. Rubinskiene, P. Viskelis, I. Jasutiene, R. Viskeliene, C. Bobunas,

Food Res. Int. 38 (2005) 867.

[16] J.B. Adams, J. Sci. Food Agri. 24 (1973) 747.

[17] B. Lapornik, M. Prosek, A.G. Wondra, J. Food Eng. 71 (2005) 214.

[18] A. Bakowska, A.Z. Kucharska, J. Oszmianski, Food Chem. 81 (2003)

349.

[19] S. Hao, J. Wu, L. Fan, Y. Huang, J. Lin, Y. Wei, Sol. Energy 76

(2004) 745.