Push-Pull N,N-Diphenylhydrazones Bearing Bithiophene or Thienothiophene Spacers as Nonlinear Optical Second Harmonic Generators and as Photosensitizers for Nanocrystalline TiO2 Dye-Sensitized Solar Cells.

A series of push-pull heterocyclic N,N-diphenylhydrazones were prepared to study the effect of structural modifications (different π-spacers and electron-withdrawing groups) on the optical (linear and nonlinear) and electronic properties of the molecules. The photovoltaic response of dye-sensitized solar cells assembled using nanocrystalline titania photosensitized with the synthesized dyes was also studied. These heterocyclic push-pull conjugated dyes involve N,N-diphenylhydrazones as electron donors linked to bithiophene or thieno[3,2-b]thiophene spacers and were functionalized with carboxylic acid, cyanoacetic acid, or dicyanovinyl acceptor groups. A combination of Suzuki-Miyaura cross-coupling, Vilsmeier formylation, and condensation reactions was used to synthesize the intermediates and final products. Density functional theory (DFT) and time dependent-DFT calculations were used to obtain information on conformation, electronic structure, and electron distribution, both for the free dyes and those adsorbed on TiO2. The results of this multidisciplinary study indicate that dyes 5b and 6b have the strongest second-order nonlinear optical response with hyperpolarizability values in the range of β = 2330 × 10-30 to 2750 × 10-30 esu, whereas photovoltaic power conversion efficiencies reach values in the range of 0.7-3.0% for dyes 5a-b and 7c and were enhanced by coadsorbing deoxycholic acid (0.8-5.1%).

Introduction

Push–pull heteroaromatic π-conjugated systems are established compounds in materials science, which combine easy synthesis with tunability of structures to produce systems with predictable and unique optoelectronic properties. The molecular arrangement of these D−π–A systems leads to terminal electron donor and acceptor group communication through a π-conjugated bridge. This intramolecular charge transfer (ICT) involves a new low-energy molecular orbital, accessible through visible light excitation, which is responsible for the polarization of the molecule.[1,2]

Optoelectronic properties of these push–pull systems may have applications in field-effect transistors,[3] light-emitting diodes,[4] nonlinear optics [second harmonic generation (SHG) and two-photon absorption],[5−7] photovoltaics,[8−13] and near-infrared absorbing dyes.[14]

These push–pull systems may be tailored toward specific applications by changing electron donor or acceptor moieties, spacer (conjugation length, planarity, and electronic nature), and the overall chromophore arrangement. An efficient method of tuning the electronic properties of these systems involves incorporation of heterocyclic units, which provide high polarizability, thermal and chemical robustness, and possibilities for further structural changes. In addition, they can behave as both efficient spacers and auxiliary electron donor/acceptor groups. A variety of π-bridges and donor and acceptor groups have been used in the design of heterocyclic push–pull heterocyclic dyes.[1,2]

Aromatic hydrazone-based compounds have been studied in several areas of materials chemistry, such as organic nonlinear optical (NLO)[15−18] and hole-transport materials, and so forth.[19,20] These compounds have been of particular interest because of the easy synthetic access through reaction of carbonyl compounds (particularly aldehydes) with relatively low-cost starting materials, avoiding expensive catalysts. They are also tolerant to the presence of water and/or oxygen. Other interesting characteristics are the extended conjugation when compared to the corresponding amines, good thermal and chemical stability, and enhancement of charge mobility through delocalization of the terminal nitrogen atom lone pair into the π-conjugated system.[16−20]

Although application of compounds with a hydrazone moiety as efficient electron donors in bulk-heterojunction solar cells has previously been reported,[21−23] few have been studied as sensitizing dyes for nanocrystalline TiO2 dye-sensitized solar cells (DSSCs).[24−28] The excellent electronic properties of the thiophene moiety have led to its widespread application in the design of ICT chromophores.[29,30]

On the basis of the interesting results reported by us and other groups[1,2,15−18,24−28] on push–pull heterocyclic π-conjugated systems for NLO[31−41] and photovoltaic applications,[42−45] we have synthesized five organic dyes bearing electron-rich heteroaromatic groups (N,N-diphenylhydrazone and thiophene) as the donor moiety or/π-spacer and carboxylic, cyanoacetic acid, or dicyanovinyl moieties as the strong electron-withdrawing/-anchoring groups. Their optical and redox properties have been evaluated, and they have been tested in DSSC photovoltaic devices.

Results and Discussion

Synthesis and Characterization

A series of push–pull heterocyclic chromophores were prepared to study the effect of different spacers (bithiophene and thieno[3,2-b]thiophene) and electron-withdrawing moieties (carboxylic acid, cyanoacetic acid, and dicyanovinyl) on their linear and NLO properties, redox behavior, and potential as sensitizers in DSSCs based on nanocrystalline titania. All of the compounds bear a N,N-diphenylhydrazone donor unit that was chosen because of its high charge mobility properties as an electron donor.[25] The thiophene units in the π-spacer show excellent charge-transfer properties, efficient π-conjugation, and low geometrical relaxation upon oxidation.[29,31−47]

The aldehyde precursor 2c was prepared in good yield (53%) by Suzuki–Miyaura cross-coupling of 5-bromothiophene-2-carboxylic acid and 5-formyl-2-thiopheneboronic acid. Intermediates 3a–b and push–pull chromophore 7c were synthesized in moderate to good yields (55–72%) through condensation of the respective aldehydes 2a–c with N,N-diphenylhydrazone 1 in ethanol. The intermediate aldehydes 4a–b were obtained (45–58% yields) by Vilsmeier–Haack formylation. Knoevenagel condensation (piperidine catalyst) in refluxing ethanol of the aldehydes 4a–b with 2-cyanoacetic acid gave the push–pull chromophores 5a–b in fair to good yields (27–58%), whereas dyes 6a–b were prepared in fair yields (22–29%) by condensation of the same aldehydes with malononitrile (Scheme ).

Scheme 1

Synthesis of N,N-Diphenylhydrazone Derivatives 3–7: (i) EtOH, rt; (ii) DMF, POCl3; (iii) 2-Cyanoacetic Acid, EtOH, Piperidine, Reflux; and (iv) Malononitrile, EtOH, Piperidine, Reflux

Recently, the synthesis of precursor 3a was reported by Roncali and collaborators through condensing N,N-diphenylhydrazine with aldehyde 2a, in 63% yield, using different experimental conditions: MeOH/tetrahydrofuran in the presence of sodium acetate. Aldehyde 4b was also reported, in 37% yield, using a one-step methodology by condensing N,N-diphenylhydrazine with 5,5′-diformylbithiophene. The same investigators also published the synthesis of dyes 6a–b using different reaction conditions (NET3 as the base and CHCl3 as the solvent).[23]

The structures of these compounds were confirmed by standard procedures (detailed characterization and Table S1 with experimental data in the Supporting Information).

Electrochemical Study

To obtain energies of highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO) redox properties of the compounds, 5a–b and 7c were studied using cyclic voltammetry (CV). In DSSCs, dye N719 is most commonly used; its band gap energy is appropriate for absorption in the visible region of the solar spectra; energies of LUMO and HOMO fit well for electron injection into TiO2 and regeneration of the dye. Data on the electrochemical behavior of N719 are presented as a “reference’’. The energies of HOMO and LUMO of 5a–b and 7c were calculated. The energy level of ferrocene (4.39 eV) was used as the reference. Table summarizes the data. For compounds 5 and 7, the redox behavior is reversible; whatever scan rate of potential is applied, the potentials of cathodic and anodic peaks of current remain constant. Potentials for oxidation of compounds 5a, 5b, and 7c are 0.73, 0.67, and 0.64 V, respectively. These potentials are higher than the potential of the 3I–/I3– couple (0.42 V), which in the case of application in solar cells, reduces back transfer of electrons to the electrolyte.[48,49] The oxidation potential clearly depends on the electronic nature of the heteroaromatic spacer unit: the stronger the auxiliary electron-donating ability, the easier it is to oxidize the molecule. The bithiophene derivative 5b displays the lowest oxidation potential (highest energy of HOMO), suggesting a slightly stronger electron-donating ability when compared to compound 5a (this is in accordance with the UV–vis spectral data). The LUMO level of the dyes indicates the thermodynamic possibility of the electron injection into the conduction band of TiO2 such that the LUMO level of the dye must be higher than that of the conduction band. Compound 5b shows the lower LUMO level of all studied dyes (−2.65 eV), indicating the lowest thermodynamic driving force for electron injection. Compound 7c exhibits the highest band gap among the compounds studied (2.92 eV), indicating that its photoexcitation is less efficient at long wavelengths—excitation requires higher energies. Even though the HOMO level is similar to that of compound 5b, the LUMO level is much higher, which could be ascribed to the lower ability for electron capture by the carboxylic group present in 7c when compared to the cyanoacetic moiety in 5b.

Table 1

Electrochemical Data for the Dyes 5a–b, 7c, and N719

Cpds	reductiona/V	oxidationa/V	EHOMOb/eV	ELUMOb/eV	band gapc/eV
5a	–1.94	0.73	–5.12	–2.45	2.67
5b	–1.74	0.67	–5.06	–2.65	2.41
7c	–2.29	0.64	–5.02	–2.10	2.92
N719	–2.04	0.46	–4.85	–2.35	2.50

Measured in 1.0 mM solution of the dye in DMF with the addition of 0.1 M tetrabutylammonium tetrafluoroborate. Glassy carbon was used as a working electrode. Scan rate of potential: 0.01 V s–1. Potentials (E) are indicated with respect to Fc+/Fc. Epa and Epc stand for the potentials of anodic and cathodic peaks, respectively.

ELUMO = −(Ered + 4.39) (eV) and EHOMO = −(4.39 + Eox) (eV).

Calculated as the difference between the onset potentials of oxidation and reduction.

Optical Studies

Linear Optical Properties

The absorption and emission spectra of heterocyclic compounds 3–7 were studied in ethanol, at room temperature. Table summarizes the data, whereas Figure displays the spectra of chromophores 4–7. All of the push–pull chromophores studied exhibit at least one strong broad absorption band between 416 and 456 nm that is assigned to an ICT between the electron donor and acceptor groups. The peak of these absorption bands varies according to the spacer and electron-withdrawing moieties employed. The substitution of thieno[3,2-b]thiophene for the bithiophene moiety as the spacer induces a bathochromic shift in the longest wavelength absorption (Δλ = 5 nm for compounds 3, Δλ = 30 nm for compounds 4, Δλ = 21 nm for compounds 5, and Δλ = 25 nm for compounds 6) that is explained by the slight increase of the π-conjugation path and the enhanced auxiliary donor effect of bithiophene. Shifts of the absorption maxima are also observed when comparing the acceptor groups; generally, the higher the conjugation length and electron-withdrawing ability, the longer is the wavelength of maximum of absorption. For compounds 3 and 4, the introduction of a formyl group leads to bathochromic shifts in the range of 42–67 nm. Comparing compounds 4 and 5, the substitution of the formyl group for the 2-cyanoacetic acid moiety induces another bathochromic shift of 16–25 nm because of the stronger acceptor characteristics of the cyanoacetic group as well as the increase of conjugation. Comparison of compounds 4a–b and 6a–b shows that the dicyanovinyl group causes bathochromic shifts around 8–13 nm. Moreover, a hypsochromic shift of 24 nm is observed on comparing absorption spectra of compounds 4b and 7c because of the substitution of the electron-withdrawing formyl group by a carboxylic acid group. This result was not expected, bearing in mind the lower conjugation of the aldehyde compared to the carboxylic acid moiety. The novel cyanoacetic chromophores 5a–b have high molar extinction coefficients (20 141–30 740 M–1 cm–1) compared to the dicyanovinyl derivatives 6a–b (19 221–26 766 M–1 cm–1). Exciting compounds 3–7 at the maximum absorption wavelength, their emission spectra were recorded under room-temperature conditions (Figure , Table ). The fluorescence was significantly influenced by the nature of the spacer and acceptor groups; generally, an increase in the extent of the π-conjugated system or functionalization with stronger electron donor or acceptor groups shifts the spectra to longer wavelengths. With the exception of compound 3b (ΦF = 0.20), all heterocyclic systems showed very weak emissive properties, with relative fluorescence quantum yields ranging from 0.01 to 0.03. When a nitrogen heteroatom is involved in the π-system, the n → π* transition may be the lowest lying transition and is characterized by a radiative lifetime at least 100 times longer than that of low lying π → π* transitions. Under these conditions, nonradiative processes are dominant, leading to low fluorescence quantum yields of many azo compounds as well as some compounds containing carbonyl groups and nitrogen heterocycles. As expected, a red shift of the emission of the compounds is observed upon increasing the strength of the acceptor group (aldehyde < carboxylic acid < cyanoacetic acid < dicyanovinyl). Thienothiophenes 4a, 5a, and 6a show emissions at 480, 519, and 564 nm, respectively, whereas bithiophenes 4b, 5b, 6b, and 7c exhibit emissions at 536, 574, 591, and 501 nm, respectively. A bathochromic shift is also observed upon increasing the π-conjugation path length though the bithiophene spacer, comparing compound 4a with 4b (7 nm), compound 5a with 5b (52 nm), and compound 6a with 6b (18 nm). All of the studied push–pull systems show large Stokes’ shifts (3557–5910 cm–1), indicating that significant structure relaxation occurs in the excited state upon absorption. Moreover, the compounds functionalized with stronger electron-withdrawing cyanoacetic acid and dicyanovinyl moieties 5–6 exhibit larger Stokes’ shifts than the corresponding aldehyde precursors 4. This is consistent with the possibility of increased ICT upon absorption, particularly in the case of the dicyanovinyl moieties for which the Stokes shift increase is greater than 50%.

Table 2

UV–Visible Absorption and Fluorescence Data for N,N-Diphenylhydrazone Derivatives 3–7 in Ethanol Solution at Room Temperature

	UV–vis	fluorescence
Cpds	λmax/nm	ε/M–1 cm–1	λem/nm	ΦF	Stokes shift/cm–1
3a	368	24 867	428	0.03	3809
3b	373	26 060	454	0.20	4783
4a	410	29 750	480	0.01	3557
4b	440	23 588	536	0.01	4071
5a	435	30 740	519	0.01	3721
5b	456	20 141	574	0.03	4508
6a	423	26 766	564	0.01	5910
6b	448	19 221	591	0.01	5401
7c	416	12 562	501	0.01	4078

Figure 1

Normalized absorption (full solid lines) and emission (dashed lines) spectra for compounds 4–7 in ethanol at room temperature.

NLO Properties

Hyper-Rayleigh scattering[50,51] (HRS) was used to characterize the molecular first hyperpolarizabilities β of push–pull N,N-diphenylhydrazone derivatives 5–7. The incident laser beam had a fundamental wavelength of 1064 nm, and the chromophores were dissolved in dioxane. The mean hyperpolarizability values, β, were determined by comparison with a p-nitroaniline () reference solution.[52,53] The extent that the second harmonic signal might be contaminated by multiphoton-induced fluorescence was taken into account by measuring the HRS signal over different spectral bandwidths (see Supporting Information for more details). The static hyperpolarizabilities β0 were estimated via a simple two-level model neglecting damping.[54−56] Both an increase in the π-conjugation of the spacer and an intensification of the electronic acceptor ability of the withdrawing group clearly influence the nonlinearities of compounds 4–7 as quantified in Table . First, the measured β values show that the bithiophene spacer leads to higher molecular hyperpolarizability values than the thieno[3,2-b]thiophene moiety. Therefore, aldehyde 4b has a higher β value (β = 290 × 10–30 esu) than 4a (β = 205 × 10–30 esu), as does cyanoacetic acid 5b (β = 2330 × 10–30 esu) when compared to 5a (β = 930 × 10–30 esu) and dicyanovinyl derivative 6b (β = 2750 × 10–30 esu) in comparison with 6a (β = 265 × 10–30 esu). The static hyperpolarizability β0 values of the compounds follow the same trend. The more extensive π-conjugation and stronger auxiliary electron-donating ability of the bithiophene moiety are probably the main reasons for these trends. We also notice a general increase of the hyperpolarizability β as the electron-accepting ability is increased and as the electronic conjugation of the acceptor end moiety becomes larger. In the case of compounds bearing thieno[3,2-b]thiophene (4a, 5a, and 6a), the SHG response is enhanced by an increase of the acceptor group strength upon changing the aldehyde group (β = 205 × 10–30 esu for 4a) to a dicyanovinyl group (β = 265 × 10–30 esu for 6a) and then for a cyanoacetic acid moiety (β = 930 × 10–30 esu for 5a). A similar tendency is observed in compounds bearing bithiophene as the spacer (4b, 5b, 6b, and 7c). Derivative 4b shows the lowest hyperpolarizability β, followed by the carboxylic acid 7c. Chromophores 5b and 6b exhibit the highest values of β (2330 × 10–30 esu for 5b and 2750 × 10–30 esu for 6b); however, in this case, the dicyanovinyl derivative 6b has higher SHG response than the cyanoacetic derivative 5b. The lower hyperpolarizability β value of 5b might be a result of the distorted conformation of the π-system due not only to the bithiophene spacer but also the cyanoacetic acid moiety when compared to the dicyanovinyl group. This could result in a suppression of electron transfer, even though the dominant conformer is planar between the spacer and acceptor moieties.[43] This result is in accordance with the electrochemical study of dyes 6, previously reported by Roncali et al.,[23] where we noticed that the LUMO levels in dyes 6 do not follow the same trend as in dyes 5: the introduction of the bithiophene spacer in dye 6b leads to a destabilization of the LUMO (−3.45 eV for 6a and −3.39 eV for 6b), which is expected because of the stronger donating ability of the auxiliary electron, whereas in dye 5b, the opposite occurs. The static hyperpolarisability β0 values for the compounds show the same trend.

Table 3

UV–Vis Absorption Data in 1,4-Dioxane and β and β0 Values for Chromophores 4–6 and 7c

	UV–visa
Cpds	λmax/nm	ε/M–1 cm–1	βb × 10–30 esu	β0c × 10–30 esu
4a	408	30 871	205	72
4b	438	28 010	290	77
5a	464	31 534	930	180
5b	486	20 125	2330	300
6a	423	27 032	265	81
6b	447	20 022	2750	910
7c	422	12 444	370	114
pNA	352		40.1

Wavelength of the absorption maximum of the respective compound when dissolved in dioxane.

Average hyperpolarizability values, β, reported in the T-convention.

Estimate of the static hyperpolarizability using the two-level model, β0 = β[1 – (λmax/1064 nm)2][1 – (λmax/532 nm)2], and neglecting damping factors.

Performance in DSSCs

Recently, several groups have reported that photovoltaic efficiencies ranged from 3.30 to 7.74% for organic hydrazine dyes bearing different π-spacers/electron donor groups (thiophene, pyrrole, furan, triphenylamino, and tetrahydroquinoline) and functionalized with various anchoring moieties (cyanoacetic acid and rhodanine acetic acid).[24−28] Therefore, we decided to expand these earlier studies as well as our work concerning organic sensitizers bearing bithiophene and thienothiophene spacers[42−45] to the synthesis of the novel bithiophene and thienothiophene dyes 5a–b and 7c to be evaluated as sensitizers for DSSCs.

Performance metrics of the DSSCs (efficiency—η, filling factor—FF, max. power point—MPP, voltage at open circuit—VOC, current density at short-circuit—JSC) sensitized with dyes 5a–b and 7c and N719 reference dye, as well as the results from the coadsorption with deoxycholic acid (DCA) are presented in Table . The current–voltage characteristics of the prepared DSSCs are presented in Figure S1 (Supporting Information) and Figure and show shapes typical for photodiodes for all of the cells. Incident photon-to-current conversion efficiency (IPCE) spectra are presented in Figure .

Table 4

Photovoltaic Parameters of the Cells Sensitized with Dyes 5a–b and 7c and N719 Reference Dye, as Well as Coadsorbed with DCA

sensitizer	VOC/V	JSC/mA cm–2	MPP	FF	η/%
5a	0.64	6.41	2.67	0.67	2.98
5a + DCA	0.62	6.82	2.98	0.70	3.22
5b	0.61	6.77	2.78	0.67	3.01
5b + DCA	0.64	10.54	4.71	0.70	5.10
7c	0.54	2.01	0.67	0.62	0.72
7c + DCA	0.55	1.85	0.73	0.72	0.79
N719	0.74	16.11	8.51	0.72	9.25

Figure 2

Current–voltage characteristics recorded under illumination and in the dark for the DSSCs assembled using synthesized dyes 5a–b and 7c (dashed lines) and with the DCA coadsorbent molecule.

Figure 3

Spectra of IPCE for the cells assembled using N719, 5a–b, and 7c dyes. The sensitization was in 0.5 mM ethanol solutions of the corresponding dye (dashed lines) and with 50 mM DCA as an additive for coadsorption (solid lines).

The DSSC prepared using N,N-diphenylhydrazone dye 7c exhibits the lowest efficiency (0.72%). The very low JSC (2.01 mA cm–2) could be related to the low extinction coefficient (12562 M–1 cm–1), narrow absorbance spectrum, and high electronic band gap (2.92 eV) exhibited by this dye. The IPCE spectra of the DSSC with 7c dye supports these assumptions; only the higher energy photons from the spectral region 350–450 nm are able to turn the dye to its photoexcited state and contribute to photocurrent. However, this could turn to the advantage of this dye for “greenhouse” concept, where high transparency of the integrated photovoltaic device is eventually important.

DSSCs prepared using dyes 5a–b show very similar photovoltaic efficiencies: 2.98% for 5a and 3.01% for 5b. The slightly higher efficiency of the DSSC sensitized with 5b dye (about 33% of the efficiency of the DSSC with N719 dye), bearing a bithiophene moiety as the π-spacer/auxiliary electron donor, results from the enhanced JSC (6.77 mA cm–2) that could be due to the longer π-conjugation of the sensitizing dye and lower electronic band gap when compared to dye 5a.

To enhance the efficiency of the DSSCs sensitized with dyes 5a–b and 7c, we used coadsorption with DCA. Bile acids such as DCA and chenodeoxycholic acid are able to improve the efficiency of the cells because of the reduction of dye aggregation on the surface of TiO2, suppressing dark current and improving electron lifetime.[25,57]

The photovoltaic response of the cells with dyes 5a and 7c was increased slightly by the addition of DCA as the coadsorbent (from 2.98 to 3.22% for dye 5a and from 0.72 to 0.79% for dye 7c), whereas the efficiency of the cell with dye 5b improved significantly—from 3.01 to 5.10%. It can be seen from Table and Figure that the addition of DCA does not affect VOC of the cells based on dyes 5a and 7c—the variations are within the statistical deviation. The slight enhancement of the performance of these cells is due to the improvement of the FF via suppressing the dark current. At the same time, for dye 5b, coadsorption of DCA leads to noticeable increase of VOC.

It is noteworthy that the addition of DCA to the sensitization solutions causes perceptible bathochromic shifts in the transmittance and absorption spectra for dyes 5a–b (Figure ).

Figure 4

Transmittance (a) and absorbance (b) spectra for the ethanol solutions (0.5 mM) of the synthesized dyes 5a–b and 7c (dashed lines) and with the addition of 50 mM of DCA (solid lines).

This bathochromic shift is accompanied by a slight increase of the quantum yield above 500 nm for the cell based on dye 5a. Noteworthy, a pronounced improvement of quantum yield above 550 nm for the cell based on dye 5b was observed (Figure ). The overall performance of the DSSC with dye 5b coadsorbed with DCA was found to be 5.01%, which is promising for metal-free organic dyes.

Computational Studies

The metal-free organic sensitizers 5a–b and 7c considered in our study have many possible conformations. Different conformers can have very different degrees of conjugation; therefore, it is important to determine the lowest energy conformers responsible for the observed absorption spectra of the dyes. For dyes 5a and 5b, 12 conformers were considered in each case, differing in the relative arrangements of the carboxylic acid, cyano, bithiophene (or thienothiophene), and N,N-diphenylhydrazone groups. For dye 7c, we investigated the structures and energies of four different conformers. For dyes 5b and 7c, a simplification was introduced by considering only the thiophene–thiophene trans conformers. This is justified in view of previous studies,[43,58] which showed that the cis conformation usually accounts for ca. 10% of the population of conformers in this type of structures. Figure S2 (Supporting Information) shows the most stable conformers of these dyes and their relative energies calculated at the density functional theory (DFT) level. The energy differences found for these forms indicate a population of about 58% for conformer 5a-I, 77% for conformer 5b-I, and 53% for conformer 7c-I. Because these are the dominant conformers, our subsequent analysis will be carried out for these forms, and for simplicity, we will refer to conformers 5a-I, 5b-I, and 7c-I as dyes 5a, 5b, and 7c, respectively.

Figure presents the structures of the dominant conformers. In the three dyes, the cyano and the carboxylic groups are coplanar with the thienothiophene (in 5a) or bithiophene (in 5b and 7c) π-bridges, showing strong conjugation between the π-orbitals of these groups. This is convenient because it will improve the efficiency of electron transfer from the bridge to the electron acceptor. Dye 7c is slightly distorted in the region between the donor and the π-bridge.

Figure 5

DFT B3LYP/6-31G(d,p) optimized geometries of conformers (a) 5a-I (5a), (b) 5b-I (5b), and (c) 7c-I (7c).

Figure shows the DFT-calculated energies for the frontier molecular orbitals of the three dyes. The larger extension of the conjugation in the chain in 5b leads to a smaller band gap for this dye. Dye 7c has the smallest conjugated chain, and for this reason, it has the highest band gap of the three dyes considered. This leads to a red shift of the first band in the absorption spectra of 5b and a blue shift of the first band in the spectra of 7c, compared with 5a. The LUMO levels are significantly lower for 5a and 5b, compared with 7c, in agreement with the cyanoacetic group being a better electron acceptor than the carboxylic one. The LUMO levels are above Ec of titania (−4.0 eV vs vacuum),[59] and the levels of HOMO are below the redox potential of 3I–/I3– (−4.6 eV vs vacuum).[60] For utilizing in DSSCs, this means that the necessary driving force is provided for electron injection into the semiconductor and the regeneration of the dyes by the electrolyte. The data acquired from CV are provided in Figure (green solid lines). One can see that HOMO and LUMO energies match very well the experimental CV values.

Figure 6

Energy of the frontier molecular orbitals calculated at the DFT B3LYP/6-31G(d,p) level of theory for the dyes 5a, 5b, and 7c. The experimental values extracted from CV are marked with the green solid lines. The magenta dashed line points to the energy of the conduction band of TiO2 (−4.0 eV for anatase), and the gray dashed line indicates the 3I–/I3– redox potential (−4.6 eV). LUMO and HOMO are indicated as “L” and “H”, respectively.

The UV–vis absorption spectra (vertical electronic excitations) were calculated for the free dyes at the time-dependent-DFT (TD-DFT)/CAM-B3LYP level, and the main contributions to the transitions with maximum absorption involve the HOMO – 1, HOMO, LUMO, and LUMO + 1 orbitals (Figure and Table ). For 5a, the maximum absorption transition has a computed wavelength of 446 nm (experimental value 435 nm in ethanol) and corresponds predominantly to a HOMO → LUMO excitation (85%) with small contributions from HOMO – 1 → LUMO (8%) and HOMO → LUMO + 1 (3%) configurations. The maximum absorption of 5b (calculated value is 470 nm and experimental value in ethanol is 456 nm) is red-shifted relatively to 5a, as discussed above, and the dominant contributions to this transition are the HOMO → LUMO (79%), HOMO – 1 → LUMO (14%), and HOMO → LUMO + 1 (5%) excitations. Dye 7c has its maximum absorption blue-shifted relative to 5a and 5b. The calculated maximum appears at 399 nm (in ethanol solvent, the experimental value is 416 nm), and as with dyes 5a and 5b, the main contribution is a HOMO → LUMO excitation (86%). From Figure , one can see the type of the orbitals involved in these transitions. The HOMOs are π orbitals, mainly localized on the donor and the thienothiophene (in 5a) or bithiophene (in 5b and 7c) π-bridges, whereas the LUMOs are π* orbitals essentially on the acceptor group and on the π-bridges. The HOMO – 1 → LUMO and HOMO → LUMO + 1 excitations tend to decrease the charge separation between the ground and excited states in these dyes; however, their contributions to the transition are very small. Therefore, there is efficient charge transfer between the donor and the acceptor upon excitation.

Figure 7

B3LYP/6-31G(d,p) frontier molecular orbitals of the free dyes 5a, 5b, and 7c.

Table 5

TD-DFT/CAM-B3LYP Calculated Vertical Excitation Energies (Eex)a, Absorption Maxima (λmax), Main Contributions to the Excited State, and Oscillator Strengths (f) for Dyes 5a, 5b, and 7c

Eexa/eV	λmax/nm	λmaxb/nm	main contributionsc/%	f
			Dye 5a
2.78	446	435	H → L (85%), H – 1 → L (8%), H → L + 1 (3%)	1.747
			Dye 5a Adsorbed to (TiO2)9
2.78	446		H → L (37%), H → L + 3 (36%), H → L + 1 (4%), H – 1 → L (4%), H → L + 4 (4%), H → L + 6 (4%), H – 1 → L + 3 (3%)	1.859
			Dye 5b
2.64	470	456	H → L (79%), H – 1 → L (14%), H → L + 1 (5%)	1.546
Dye 7c
3.10	399	416	H → L (86%), H – 1 → L (7%), H → L + 1 (5%)	1.246

Results from TD-DFT calculations at the CAM-B3LYP/6-311G(d,p) level of theory.

Experimental data from UV–vis spectra of the ethanolic solution.

Only the contributions ≥3% are indicated.

Adsorption of the dye on TiO2 causes a reorganization of the electronic states of the semiconductor.[61−63] This can affect the electronic absorption spectra of the dyes and also the efficiency of injection of electrons into TiO2. It is, therefore, important to evaluate the effect of adsorption of the dyes on their structural and electronic properties at the interface. Because of computational limitations, we only carried out this analysis for the dye 5a adsorbed on TiO2. Two types of approaches have been used to model the dye···TiO2 interface. One is based on the simulation of the dye adsorbed to an infinitely large particle of TiO2, in one or multiple dimensions, employing a periodic condition and using plane-wave basis sets. The second approach considers the dye adsorbed to an isolated cluster of (TiO2) and the calculations use a localized basis set.[61−68] Both methods present advantages, and in our study, we have used the second one. The size of the cluster, (TiO2)9, and the bidentate mode of adsorption in which the dye is covalently linked to a Ti atom of the cluster through the two oxygen atoms of the anchoring carboxylic group were also chosen based on the literature.[62,64,69,70] To keep electroneutrality of the system, the hydrogen atom coming from deprotonation[71] of the anchoring carboxylic group was linked to a cluster oxygen atom.[62,66] The geometry of the 5a/TiO2 system was optimized at the DFT level; the effect of dye adsorption on the energy levels of the orbitals is presented in Figure . The LUMO and the next virtual orbitals in the combined 5a/TiO2 system are stabilized relative to the same orbitals of the free dye. This is a consequence of the introduction of additional electronic states by the semiconductor. The HOMO and the next occupied orbitals are also slightly stabilized. Interestingly, however, there is no shift in the absorption maximum of the dye in the combined system, as we can see from Table . For the dye 5a adsorbed to (TiO2)9, TD-DFT calculations estimate an absorption maximum at 446 nm, the same value as for the free dye. The maximum absorption transition in the combined 5a/TiO2 system has now significantly different contributions. Two dominant contributions come from the HOMO → LUMO (37%) and HOMO → LUMO + 3 (36%) excitations, together with minor contributions from HOMO → LUMO + 1 (4%), HOMO – 1 → LUMO (4%), HOMO → LUMO + 4 (4%), HOMO → LUMO + 6 (4%), and HOMO – 1 → LUMO + 3 (3%).

Figure 8

Energy of selected molecular orbitals calculated at the DFT B3LYP/6-31G(d,p) level of theory for the (TiO2)9 cluster and the dye 5a adsorbed on (TiO2)9. The dashed line (magenta) points to the level of the conduction band edge of TiO2 (−4.0 eV for anatase), and the gray dashed line indicates the 3I–/I3– potential (4.6 eV). “L” and “H” stand for LUMO and HOMO, respectively.

It should be noted that these energy gaps are different in the two systems. Smaller contributions come also from excitations with larger energy gaps, such as HOMO → LUMO + 25, HOMO → LUMO + 26, HOMO → LUMO + 27, and so forth, which are absent in the case of the free dye. The sum of all of these contributions in the case of 5a/TiO2 gives a resulting vertical excitation energy which is similar to one of the free dye.

Although their energy suffers slight changes, the HOMO and HOMO – 1 in the combined 5a/TiO2 system (Figure ) are essentially equivalent to these orbitals in the free dye. In contrast, the LUMO and the remaining virtual orbitals involved in the maximum absorption transition are localized predominantly on the cluster. The LUMO + 3, LUMO + 4, and LUMO + 6 also extend to the acceptor and π-bridge. The distribution of electron density in these orbitals shows that there will be efficient electron injection from the excited dye into TiO2 in a DSSC.

Figure 9

Optimized geometry and selected molecular orbitals of dye 5a adsorbed on (TiO2)9 calculated at the B3LYP/6-31G(d,p) level.

To discuss further the relation between the performance of the chromophores as sensitizers for TiO2 DSSCs and their second-order NLO response, we focus on a comparison between compounds 5a and 5b. These chromophores differ only in their π-electron bridge and should have similar coupling efficiencies for electron transfer to the TiO2 substrate. Given the tendency of many D−π–A molecules to form π-stacked aggregates on the surface of TiO2 clusters, which will encourage the back transfer of electrons and, consequently, decrease DSSC efficiency, we further restrict our attention to the measurements using the coabsorber DCA, known for its ability to dissociate π-aggregates. The ratio of the efficiency of the 5b + DCA DSSC to that of the 5a–DCA DSSC is 5.10%:3.22% ≈ 1.6, roughly equal to the ratio of the extrapolated hyperpolarizabilities, 300:180 ≈ 1.7. Interestingly, the DFT-estimated oscillator strengths of the first singlet transition listed in Table indicate that chromophore 5a has a stronger dipole optical transition moment than chromophore 5b. This suggests that the asymmetry in the optical field-induced polarizability of chromophore 5b is greater, which might facilitate the transfer of photoexcited electrons from the donor to the substrate. We note that the DFT study carried out on chromophore 5a coupled to a TiO2 cluster indicated that because of the interaction between the chromophore and the substrate, the character of the first singlet transition changed; a substantial mixing of the LUMO + 3 state was accompanied by a strong reduction in the participation of the LUMO state (see Table ). As the electron density of LUMO + 3 extends far into the π-bridge, this could result in less efficient electron collection by the substrate. The existence of a stronger charge-transfer asymmetry in chromophore 5b might counteract this delocalization of the electron density from the TiO2 cluster, promoting higher DSSC efficiencies.

Conclusions

N,N-Diphenylhydrazone derivatives 3–7 were obtained in fair to good yields, using commercially available precursors and simple, convenient synthetic and purification procedures. Condensation of aldehydes 2a–c with N,N-diphenylhydrazine gave intermediates 3a–b and 7c. Further functionalization of precursors 3a–b through Vilsmeier formylation allowed the preparation of formyl derivatives 4a–b that, through Knoevenagel condensation, gave the push–pull compounds 5–6.

Experimental and theoretical characterization of the absorption, fluorescence, and NLO and electrochemical properties of the systems were carried out as a means to evaluate their potential as photosensitizers for nanocrystalline TiO2-based DSSCs. The effect of different π-spacers and acceptor groups was evaluated. These comparative studies demonstrate that the incorporation the bithiophene π-bridge enhances the electron donor effect when compared with thienothiophene and that the dicyanovinyl acceptor group, although less conjugated than cyanoacetic acid and a weaker electron-withdrawing moiety, induces less distortion to the heterocyclic system, which can be advantageous.

Chromophores 5b and 6b exhibit the highest hyperpolarizability β (2330 × 10–30 esu for 5b, 2750 × 10–30 esu for 6b). The lower hyperpolarizability β value of 5b, which is the more strongly conjugated system, could be attributed to the distorted conformation of the π-system due not only to the bithiophene spacer but also to the cyanoacetic acid moiety.

Compound 5b, bearing a bithiophene π-spacer, exhibits the best conversion efficiency (3.01%) when used as a sensitizer for nanocrystalline titania in DSSCs because of the lower electronic band gap and longer π-conjugation of the sensitizing dye. Studies of coadsorption were performed by adding DCA to the sensitizing mixtures, revealing slight improvements for the cells based on dyes 5a and 7c, whereas the efficiency of the DSSCs with dye 5b was significantly enhanced from 3.01 to 5.10%. This improvement originated from the bathochromic shift of the adsorption spectra of the dye 5b in the presence of DCA and consequential widening of the spectral response of the photocurrent quantum yield.

Further optimization of the molecular structure of all chromophores should lead to enhanced hyperpolarizabilities and photovoltaic efficiencies.

Experimental Section

Materials and Methods

N,N-Diphenylhydrazine hydrochloride, 5-bromothiophene-2-carboxylic acid, thieno[3,2-b]thiophene-2-carbaldehyde, 2,2′-bithiophene-5-carboxaldehyde, phosphorous oxychloride, malononitrile, and 2-cyanoacetic acid were obtained from Aldrich, whereas 5-formylthiopheneboronic acid was obtained from Acros Organics. All commercial reagents and solvents were used without further purification. The progress of the reaction was checked by means of thin-layer chromatography on 0.25 mm thick precoated silica plates (Merck Fertigplatten Kieselgel 60 F254); the spots were visualized using UV light. Silica gel column chromatography (Merck Kieselgel, 230–400 mesh) was used in the purification of the compounds. NMR spectra were performed on a Bruker Avance II 400 (working frequency of 400 MHz for 1H and 100.6 MHz for 13C), and the solvent peak was used as the internal reference. The solvents are specified in parenthesis before the chemical shifts values (δ relative to tetramethylsilane). Peak assignments were obtained by comparison of chemical shifts, peak multiplicities, and J values and were sustained by spin decoupling-double resonance and bidimensional heteronuclear multiple bond correlation and heteronuclear multiple quantum coherence techniques. Infrared spectra were obtained on a Bomem MB 104 spectrophotometer. UV–vis absorption spectra were recorded with a Shimadzu UV/2501PC spectrophotometer. Fluorescence spectra were obtained with a FluoroMax-4 spectrofluorometer, and relative fluorescence quantum yields were determined using the reference standard fluorescein in 0.1 M aqueous solution of NaOH (ΦF = 0.79)[72] or 9,10-diphenylanthracene in ethanol.[73] Melting points were determined on a Gallenkamp machine. Mass spectra analysis was performed at the C.A.C.T.I.—Unidad de Espectrometria de Masas of the University of Vigo, Spain. The experimental procedures for the synthesis of precursor 2, N,N-diphenylhidrazone derivatives 3a–b, 7c, aldehydes 4a–b, the final push–pull cyanoacetic acid 5a–b, and dicyanovinyl derivatives 6a–b are described in the Supporting Information.

Cyclic Voltammetry

Autolab PGSTAT302N was used for electrochemical measurements. The measurements were performed in a three-electrode cell. Glassy carbon served as the working electrode, Pt was as the counter-electrode, and the reference was Ag/AgCl. Solutions were prepared using dry dimethylformamide (DMF). Concentrations of the dyes were 1 mM. [NBu4][BF4] (0.1 M) was added to achieve sufficient conductivity. All solutions were deaerated by bubbling of N2. As an internal redox reference, Fc+/Fc was used.[42,74]

NLO Measurements

The orientationally averaged first hyperpolarizability β of the push–pull chromophores 4–7 was characterized by measuring the intensity of HRS provoked by incident q-switched laser pulses with a pulse duration of approximately 12 ns and an energy of 2 mJ. The experimental setup has been previously described[31,38] and is based on the pioneering work of Clays and Persoons,[50] (see Supporting Information for further details).

Preparation and Characterization of DSSCs

DSSCss were prepared as described in the Supporting Information. The photovoltaic performance was obtained with an Oriel Class solar simulator and a ZENNIUM workstation, using a metal mask with an aperture area of 0.25 cm2, as previously reported.[42] (See Supporting Information for further details).

Computational Studies

DFT calculations were used to obtain the geometries and optoelectronic properties of the dyes 5a, 5b, and 7c. The geometries of the free dyes were optimized with the B3LYP exchange correlation functional (combination of the hybrid exchange functional of Becke[75] together with the correlation functional LYP[76]), using the all-electron 6-31G(d,p) double-ζ plus polarization basis sets for all atoms. No symmetry constrains were imposed and the polarizable continuum model[77,78] was used to account for the effects of chloroform as the bulk solvent. The molecular structures of the most stable conformers of the dyes were also optimized with the CAM-B3LYP[79] functional. The harmonic vibrational frequencies were calculated to assess the nature of the optimized stationary points found in the potential energy surface and all were confirmed to be true minima (i.e., with no imaginary frequencies). The properties of the dye 5a adsorbed to a (TiO2)9 cluster were also calculated. The initial geometry for the (TiO2)9 cluster was obtained from previous theoretical studies.[80,81] This structure was then optimized using the B3LYP functional and the Los Alamos effective core potential plus double-ζ basis set (LANL2DZ)[82−84] on titanium and the 6-31G(d,p) basis set for oxygen. The same computational procedure was used to optimize the geometry of the dye/TiO2 combined system. TD-DFT calculations were carried out to analyze the absorption spectral properties of the free dyes and of the dye/TiO2 system and also to calculate the vertical excitation energies for the lowest singlet excited states of the systems. The CAM-B3LYP[79] functional and the 6-311G(d,p) triple-ζ plus polarization basis sets were used in these calculations. The bulk solvent effects of chloroform were also considered in the TD-DFT calculations. All calculations were carried out using the GAMESS-US code.[85]