Cost‐Effective Dyes Based on Low‐Cost Donors and Pd‐Free Synthesis for Dye‐Sensitized Solar Cells

The structural design of photosensitizers for dye‐sensitized solar cells (DSSCs) usually involves modification of the donor and π‐spacer moieties using expensive Pd crosscoupling reactions. Herein, the concept of more sustainable and cost‐effective dyes to realize large‐scale production is presented. Two dyes, coded BzC and PTZC, with low‐cost donors, phenyl and phenothiazine, respectively, and 4H‐cyclopenta[2,1‐b:3,4‐b’]dithiophene π‐spacer, are synthesized by Pd‐free reactions, including Horner–Wadsworth–Emmons, with high overall yield (68% for BzC and 55% for PTZC). The best photovoltaic performances exhibit 3.00% and 5.92% together with relatively low ideality factor of 1.03 and 1.12 for BzC and PTZC, respectively, indicating efficient retardation of electron recombination from surface trap states. Finally, the synthesis costs for BzC and PTZC are estimated to be about 7.5–9 times lower than that of LEG‐4 (as a commercial reference dye). The cost performances of BzC and PTZC are estimated to be around 5 times better than that of LEG‐4.

These modifications, however, increase the costs associated with the already high synthesis cost of the organic dyes.
In this work, we designed and synthesized two new dyes, BzC and PTZC (Figure 1), which do not involve Pd cross-coupling reaction. The reaction used to connect the donor and π-spacer was based on the Horner-Wadsworth-Emmons (HWE) reaction. Moreover, low-cost PTZ and phenyl moieties were used as the donor to further lower the synthesis cost. PTZ also possesses electron-donor ability [18] in which we expect it can substitute the triarylamine-based moiety leading to lower synthesis cost. The use of phenyl ring as the donor in BzC is for a control dye structure so as to compare the effect of the donor group within this dye design. For π-spacer, commonly used CPDT was chosen due to its facile modifications, such as introduction of long alkyl chains, and large conjugation. To demonstrate the low synthesis cost, estimation was made not only for the synthesis costs but also for the cost performance per cm 2 per %PCE of these two dyes in comparison with highly efficient commercial dye LEG-4.

Synthesis
BzC was synthesized according to Scheme 1. CPDT was first alkylated with hexyl chains to give compound 1 which subsequently underwent the Vilsmeier-Haack reaction to yield compound 2. To achieve a conjugated structure suitable for a dye, CPDT unit was joint to the phenyl ring by a double bond using the HWE reaction. Benzyl bromide was changed to compound 3 by the Michaelis-Arbuzov reaction using triethyl phosphite (P(OEt) 3 ). After that, compound 3 and compound 2 underwent the HWE reaction to give compound 4. The Vilsmeier-Haack reaction was performed again with compound 4 to synthesize compound 5 which was subsequently transformed into BzC by the Knoevenagel condensation with cyanoacetic acid catalyzed by piperidine, yielding a pink solid BzC with the overall yield of 68.5%.
Scheme 2 shows the synthesis procedure of PTZC which started with the N-alkylation of PTZ with hexyl bromide giving compound 6. The Vilsmeier-Haack reaction was used to formylate compound 6 to give compound 7. The formyl group in www.advancedsciencenews.com www.solar-rrl.com compound 7 was reduced by NaBH 4 to yield compound 8. We first tried the Wittig reaction to connect alkylated PTZ with compound 2 via double bond, as shown in Scheme S1, Supporting Information. This method has been used to connect PTZ and the π-spacer with a double bond in previous literature [21][22][23]25,26] for DSSC applications. However, it leads to a mixture of (E)-and (Z )isomers in which the (E)-isomer is the major product (around 80%) as seen from 1 H NMR spectrum of compound 10 in Figure S1, Supporting Information. To convert the (Z )-isomer into (E)-isomer, isomerization of the mixture of isomers with a catalytic amount of iodine is often carried out. However, this treatment leads to more synthetic steps and gives relatively low yield. [23,25] The isomerization reaction can be avoided using the HWE reaction instead of the Wittig reaction. To perform the HWE reaction, a conversion of compound 8 to 9 was necessary which can be done by the Michaelis-Arbuzov reaction via PTZ-CH 2 Br. The attempts to convert compound 8 to PTZ-CH 2 Br as shown in Scheme S2, Supporting Information, by either reacting with PBr 3 or the Appel reaction (with CBr 4 and triphenylphosphine) were unsuccessful. This may be due to the instability of the intermediate or PTZ-CH 2 Br. To solve the problem of the unsuccessful attempts to prepare PTZ-CH 2 Br, a direct conversion [27] from compound 8 to compound 9 using ZnI 2 and P(OEt) 3 was successfully performed, as shown in Scheme 2. After obtaining compound 9, it underwent the HWE reaction with compound 2 to yield compound 10, as shown in Scheme 2. The Vilsmeier-Haack reaction was performed on compound 10 giving compound 11. The final dye PTZC was synthesized by the Knoevenagel condensation reaction of compound 11 with cyanoacetic acid and piperidine as the catalyst, giving rise to a purple solid of PTZC with the overall yield of 55.6%. To the best of our knowledge, compound 9 in Scheme 2 is novel and can be prepared easily by the direct conversion from its alcohol counterpart. This can open up a new way to synthesize bulky PTZ-bearing phosphonate reagents which may be widely used for the HWE reaction, avoiding the formation of isomers.

Photophysical and Electrochemical Properties
The UV-vis absorption and normalized emission spectra of BzC and PTZC in DCM are presented in Figure 2A. Their corresponding results are shown in Table 1. BzC and PTZC exhibit a prominent peak in the visible region. BzC shows the maximum absorption (λ max ) and extinction coefficient of 523 nm and 58 500 M À1 cm À1 , respectively, while PTZC exhibits λ max at 547 nm, which has a bathochromic shift of 838 cm À1 compared with BzC, with extinction coefficient of 63 000 M À1 cm À1 . These prominent peaks (λ max ) are ascribed to an ICT process. The bathochromic shift of PTZC reflects the larger conjugation of the PTZ unit compared with the phenyl ring, leading to a smaller optical gap. It is worth mentioning that their extinction coefficient values are astonishingly high with such simple dye structures. The high extinction coefficients may be due to the fact that their structure is highly conjugated because of the planar structure due to the double bond as well as the CPDT unit which facilitates ICT process. For the photoluminescent results of BzC and PTZC in DCM (Figure 2A, S2, Supporting Information and Table 1), BzC shows the emission peak maximum (λ em ) at 589 nm with Stokes shift of 2142 cm À1 , whereas λ em of PTZC is observed at 624 nm with Stokes shift of 2256 cm À1 . Moreover, the emission peak of PTZC exhibits a small shoulder around 650 nm, which is not seen for BzC. The larger Stokes shift and small shoulder of PTZC may imply that PTZC experiences different processes of structural reorganization during the electronic transition. [28] As shown in Figure 2A, the intersection point of the absorption and emission spectra can be used to estimate the optical gaps (E 0-0 ) of BzC and PTZC, which are 2.23 and 2.10 eV for BzC and PTZC, respectively.
The spectral response of BzC and PTZC adsorbed on the 4 μmtransparent TiO 2 film was determined by UV-vis absorption and the results are depicted in Figure 2B. When compared with the corresponding absorption spectra in DCM (Figure 2A), their spectral profiles on film become broader due to H-aggregates. [29] The broader spectral profile can benefit the light-harvesting efficiency. As shown in Figure 2B, the absorption from around 350 to 450 nm from both dyes improves significantly when compared with the corresponding absorption in DCM.
Cyclic voltammetry (CV) and square-wave voltammetry (SWV) were used to characterize the electrochemical properties of BzC and PTZC. Their CVs in positive potential range ( Figure S3, Supporting Information) reveal that both BzC and PTZC show both forward and return waves, representing their chemical and electrochemical reversibility upon oxidation reaction. It is worth mentioning that although BzC shows incomplete chemical reversibility at slow scan rate ( Figure S3, Supporting Information), this improves with increasing scan rate, confirming the chemical stability of BzC in the context of a DSSC photosensitizer, given that dye regeneration is usually in a microsecond timescale. According to Figure 3A and Table 1, SWVs of BzC and PTZC show that the oxidation process of BzC occurs at 0.57 V against Fc/Fc þ or 1.20 V against normal hydrogen electrode (NHE), which is more positive than the first oxidation potential of PTZC (0.22 V against Fc/Fc þ or 0.85 V against NHE). Moreover, PTZC shows two oxidation processes within the potential range (from À1.70 to 1.35 V). The difference in oxidation potential between BzC and PTZC can be explained in terms of the difference in the donor group. The less-positive oxidation potential of PTZC is attributed to more extended conjugation of the PTZ unit than phenyl ring in BzC. In contrast to oxidation processes, the reduction processes of BzC and PTZC occur at similar potential (around À2.0 V against Fc/Fc þ or À1.4 V against NHE, Figure 3A and Table 1, respectively). The similar reduction potentials imply that the reduction process is attributed to cyanoacrylic acid.
The evaluation of the optical and electrochemical properties is crucial as they provide information about the driving force for dye regeneration, charge injection, and light-harvesting capability. Figure 3B illustrates a visual diagram of energy-level alignment of TiO 2 /dye/electrolyte (I À =I À 3 ). It is clearly seen that the oxidation potentials of BzC and PTZC are well below the redox potential of I À =I À 3 (0.35 V vs NHE), [30] which indicates favorable driving force for dye regeneration. When compared with the conduction band edge energy level (E CB ) of TiO 2 , the excited oxidation potential (E Ã ox ) values of the dye molecules are higher than E CB of TiO 2 (À0.5 V vs NHE), [30] confirming sufficient driving force for electron injection to the conduction band of TiO 2 .

Computational Study
The frontier molecular orbitals and electronic energy level of BzC and PTZC in DCM solvation model were investigated by DFT calculations at B3LYP [31] level of theory with 6-31G(d) basis set. To ensure that all optimization calculations were at local minima, frequency calculations were also performed together with optimization calculations. As shown in Figure 4, the HOMO   The optical gap (E 0-0 ) was determined by the intersection point of normalized absorbance and emission in DCM. b) Potentials were measured in DCM with 0.3 M TBAPF 6 as a supporting electrolyte versus Fc/Fc þ and were converted to be against NHE by addition of 0.63 V. [59] c) The HOMO and LUMO energies were estimated by: E HOMO(LUMO) ¼ À4.8À(E ox ( red ) vs Fc/Fc þ ). [ www.advancedsciencenews.com www.solar-rrl.com of BzC is distributed across almost the entire structure. On the other hand, LUMO is localized on cyanoacrylic acid with a small contribution on the phenyl ring. Similarly, for PTZC, the HOMO is mostly located on PTZ with some extent on CPDT. The LUMO is shifted toward cyanoacrylic acid with negligible contribution from PTZ. By comparing the effect of substituents (phenyl and PTZ) on the HOMO-LUMO gap, the trend in which the HOMO energy is raised (less negative in vacuum scale) with more extended conjugation of PTZ corresponds to the experimental values (electrochemical results), whereas the LUMO energies are unaffected. As a result, the HOMO-LUMO gap is reduced upon more conjugation present in the structure.
To shed light on the spectroscopic properties of BzC and PTZC, the optimized geometry of BzC and PTZC was subjected to time-dependent DFT (TDDFT) calculations at CAM-B3LYP [32] level of theory with 6-31G(d) basis set in DCM solvation model. The experimental and calculated absorption spectra are shown in Figure S4, Supporting Information, and the corresponding transitions are summarized in Table 2. According to Figure S4, Supporting Information, the calculated spectra exhibit λ max at 491 and 517 nm for BzC and PTZC, respectively, which are slightly underestimated, compared with experimental results. However, the trend of phenyl and PTZ units on absorption and extinction coefficient is in line with the experimental values, where PTZC exhibits bathochromic shift and higher extinction coefficient with respect to BzC. Table 2 also highlights the difference in the molecular orbital (MO) contributions to electronic transition at calculated λ max of BzC and PTZC as well as oscillator strength ( f ). For BzC, the major %MO contribution to the transition at 491 nm (523 nm from experiment) originates from HOMO to LUMO, accounting for 93%. On the contrary, the major MO contributions to the transition at 517 nm (547 nm from experiment) for PTZC are attributed to a mixture of HOMO to LUMO and HOMO-1 to LUMO in which the former has a larger contribution of 74%. The oscillator strength is related to the probability of the transition from ground state to excited state [33] and is proportional to extinction coefficient. [34] As shown in Table 2, the oscillator strength of PTZC is larger than that of BzC, which indicates that the probability of the transition in PTZC is higher, and that contributes to higher extinction coefficient.

Photovoltaic Performance
Solar cells based on BzC and PTZC were fabricated with doublelayer TiO 2 (4 μm transparent layer and 4 μm scattering layer) and I À =I À 3 electrolyte. The photovoltaic performances were measured under AM 1.5G simulated solar light (100 mW cm À2 ). The corresponding results are shown in Table 3, and their champion current-voltage ( J-V ) curves are shown in Figure 5A. The devices with LEG-4 were also fabricated for cost analysis discussion only and the results are tabulated in Table 3. From the results of BzC and PTZC, the best device with BzC exhibited short-circuit current density ( J sc ), open-circuit voltage (V oc ), and fill factor (FF) of 7.88 mA cm À2 , 0.52 V, and 0.73, respectively, giving PCE of 3.00%, whereas the best device with PTZC achieved J sc , V oc , and FF of 12.92 mA cm À2 , 0.61 V, and 0.75, respectively, leading to PCE of 5.92%. It is clearly seen that the higher PCE obtained from PTZC is due to higher J sc and V oc .
To gain more insight into the photovoltaic performances, the incident photon-to-electron conversion efficiency (IPCE) measurements were carried out ( Figure 5B). IPCE can be expressed in terms of light harvesting efficiency LHE(λ), injection efficiency (φ inj ), charge collection efficiency (η col ), and dye regeneration efficiency (φ reg ), as shown in Equation (1) [17] As mentioned earlier, E ox of BzC and PTZC are more positive than the redox potential of I À =I À 3 , which can ensure sufficient driving force for dye regeneration. When compared with the CB level of TiO 2 , their E Ã ox are more negative, and the difference between E Ã ox and CB is greater than 0.2 V, which indicates that fast electron injection rate is possible. [35] Moreover, the E Ã ox of PTZC is more negative than that of BzC, reflecting larger driving force for electron injection. In terms of LHE, it can be expressed in relation with absorbance of the film (A f ) as shown in Equation (2). [36] LHE λ ð Þ ¼ 1 À 10 ÀA f (2)  Figure 5B, PTZC exhibits broad IPCE spectrum ranging from 400 to 600 nm with the maximum IPCE around 75%. The onset of the IPCE of PTZC appears around 700 nm, which is consistent with the absorption spectrum. On the other hand, BzC shows much narrower IPCE spectrum with the maximum around 70% at 450 nm. This difference in IPCE profile between BzC and PTZC originates from their absorption properties in which PTZC exhibits higher extinction coefficient and broader absorption spectrum on transparent TiO 2 ( Figure 2B). Apart from the absorption profiles, the amount of adsorbed dye on the TiO 2 surface also determines the LHE. The analysis of dye uptake by dye desorption reveals that the dye loading of PTZC is higher than that of BzC (Table 3). The combination between higher extinction coefficient, boarder absorption spectrum, and greater amount of adsorbed dye from PTZC leads to superior LHE. In terms of η col , it can be determined by Equation (3). [7] Figure 4. Calculated frontier molecular orbitals in DCM solvation model with experimental energy levels ( Table 1) of BzC and PTZC.  LEG-4 results given in brackets are averaged from three individual cells www.advancedsciencenews.com www.solar-rrl.com where R t and R rec are transport and recombination resistances, which can be obtained from electrochemical impedance spectroscopy (EIS) measurement. More details on the determination of these parameters can be found in the EIS section. The results show that BzC exhibits η col around 95% which is higher than that of PTZC (around 88%). Nonetheless, the effect of higher η col of BzC on IPCE is less pronounced than the better LHE of PTZC.
As a result, the overall parameters determining IPCE of PTZC are superior to those of BzC.

EIS Measurement
EIS is a powerful technique that can be used to analyze the processes governing the performance of DSSCs, especially V oc . The EIS measurements were performed in the frequency range from 1 MHz to 0.1 Hz under white LED illumination at various forward biases. The results were fit using a transmission line model [37] and the equivalent circuit is depicted in Figure S5, Supporting Information. In general, the EIS spectra ( Figure 6) show a typical DSSC feature with two semicircles where the first semicircle at high-frequency range represents Pt/electrolyte  www.advancedsciencenews.com www.solar-rrl.com interface and the second semicircle at intermediate-frequency range represents TiO 2 /dye/electrolyte interface. [38] As shown in Figure 6B, the EIS spectrum of BzC also shows a minor feature of the third semicircle at low-frequency range which represents electrolyte diffusion. [38] However, due to an unclear semicircle, parameters related to electrolyte diffusion are omitted in the equivalent circuit. To represent the EIS spectra close to their respective V oc , Figure S6, Supporting Information, shows the Nyquist plots at À0.48 V for BzC and at À0.56 V for PTZC. As shown in J-V curve results, the difference in the dye structure also plays a role in V oc determination in the full device. V oc of the device is determined by the difference between quasi-Fermi level of TiO 2 (E Fn ) and the redox potential of electrolyte (E F,redox ) as depicted in Figure 7 and described by Equation (4) [39] where q is elementary charge. As the two devices were made with the same electrolyte (I À =I À 3 ), E F,redox is considered unchanged. Therefore, the difference in V oc between the two devices stems from the variation of E Fn . This alteration of E Fn can be expressed by Equation (5). [39] where E CB is the conduction band edge of TiO 2 , k B is Boltzmann constant, T is temperature, n c is the free carrier density, and N c is the density of states in CB of TiO 2 taken as 7 Â 10 20 cm À3 . [40] Equation (5) implies that raising E CB leads to an increase in E Fn , hence, higher V oc . The parameters mentioned above are influenced by the dye properties. For example, some dyes can cause a large upward shift in E CB which might lead to higher V oc . Moreover, changing dye may influence the trap states [41] which, subsequently, impacts the performance of the devices. EIS measurement allows us to analyze these variables affected by dye through fitted EIS parameters, such as transport resistance (R t ), chemical capacitance (C μ ), and recombination resistance (R rec ). The following discussion will focus on each fitted parameter associated with the variable affected by dye properties.
The transport resistance (R t ) is one of the useful parameters extracted from EIS fitting with the equivalence circuit shown in Figure S5, Supporting Information. It provides information about E CB and its relationship with voltage can be displayed in Equation (6) [42]    where A denotes film area, L is the film thickness (8 μm), p is porosity taken as 0.6, [43] and μ 0 is electron mobility of TiO 2 taken as 4 cm 2 V À1 s À1 . [40] Fitting R t as a function of voltage ( Figure 8A) shows an exponential decrease in R t with increasing voltage. Figure 8A indicates that PTZC shows a positive shift in voltage around 100 mV with respect to BzC. Assuming that A, L, μ 0, and N c are identical across the devices, [37,42] this positive shift implies the upward displacement of E CB in the case of PTZC. To give quantitative estimation of E CB of BzC and PTZC, which is explained in terms of E CB ÀE F,redox (see Figure 7 for clarity), the intercepts from ln(R t ) versus V plot ( Figure 8A and Equation (6)) were used, and the results are shown in Table 4. The trend in R t is consistent with the results from the J-V curve measurement in which PTZC shows higher V oc than BzC.
The chemical capacitance (C μ ) at the TiO 2 /dye/electrolyte interface is also a useful parameter which can also give information about E CB , a trap distribution parameter denoted as α (the exponential shape of TiO 2 in Figure 7) and the number of trap states denoted as N t . Equation (7) [44] expresses the exponential relationship between C μ and voltage together with the dependency of C μ on parameters listed above.
As seen from the equivalent circuit in Figure S5, Supporting Information, constant phase element (CPE) was used for fitting instead of capacitor. CPE was converted to the equivalent capacitance, hereafter C μ , using Equation (8). [38] where R is the resistor parallel to the corresponding CPE, Q is the CPE prefactor, and w is the CPE index. The linear fitting to Equation (7) allows α to be extracted, giving 0.42 and 0.36 for BzC and PTZC, respectively. Figure 8B exhibits that BzC has lower C μ than PTZC at the same voltage. From Equation (7), it suggests that the lower C μ in BzC may originate from lower N t at the same voltage with similar α. To show the trend in C μ quantitatively from BzC and PTZC, N t was estimated by Equation (7) together with the corresponding E CB ÀE F,redox (from R t results in Table 4) and α (from linear fitting to Equation (7)) through the intercepts of ln(C μ ) versus V ( Figure 8B). The results of estimated N t are tabulated in Table 4. It is clearly seen that PTZC device possesses %9 times larger N t than BzC. As suggested in Equation (7) that C μ is proportional to N t , the consequence of lower N t results in lower C μ in BzC at the same voltage. Thus, from the results of C μ and R t , they suggest that BzC and PTZC modify the trap states differently, and PTZC shows more upward displacement in E CB . EIS results also allow the recombination resistance (R rec ) at the TiO 2 /dye/electrolyte interface to be determined. The dependency of R rec on voltage can be expressed in Equation (9) [44] where λ is reorganization energy of acceptor species in the electrolyte, c ox is the concentration of acceptor species in the electrolyte, k r denotes the rate constant for recombination kinetics, N s is the total number of surface states contributing to interfacial  www.advancedsciencenews.com www.solar-rrl.com recombination, and m denotes ideality factor explaining the coefficient determining nonlinear charge transfer. Fitting R rec against voltage exhibits an exponential decrease in R rec with increasing voltage, as shown Figure 9A. The results show that PTZC exhibits larger R rec than BzC, which implies more efficient charge recombination retardation from PTZC. Furthermore, this fitting enables m to be determined and the results are shown in Table 4. Normally, the value of m for DSSCs is larger than 1 due to nonlinear charge transfer through surface trap states. [40] Therefore, the relatively low values of m (close to 1) from BzC and PTZC may suggest efficient suppression of electron recombination from surface trap states with higher degree of the suppression from BzC. Recently, Zhang et al. reported a low value of m (1.08) obtained by dye MS5. [45] As shown in Table 4, PTZC shows m value of 1.12 which is close to the above value reported. Moreover, BzC achieved even lower m (1.03), which is the lowest value reported.
To gain further insight into recombination kinetics between BzC and PTZC, electron lifetime (τ n ) was calculated by Equation (10). [46] τ n ¼ C μ R rec (10) As highlighted by Barnes et al., charge concentration can be used to decouple the contribution of voltage to differences in electron recombination observed between devices. [41] Therefore, total electron density (n) should be used as a basis to compare the electron recombination instead of voltage. The total electron density (n) can be determined by Equation (11) [41] n ¼ C μ k B T q 2 L α (11) Figure 9B illustrates τ n as a function of n in which both dyes show exponential decrease in τ n with increasing n. It is clearly seen that PTZC has much longer electron lifetime than BzC at the same electron density. This indicates that the recombination kinetics of electron with acceptor species in electrolyte is much slower for PTZC than BzC. Thus, the longer electron lifetime for PTZC enables higher V oc , which is consistent with J-V curve results (Table 3).
Overall, the EIS results reveal the different influence of each dye to the energetics and kinetics of the device. Changing from BzC to PTZC gives rise to both an upward shift in E CB , and also slower electron recombination kinetics under the same total electron density, which prolongs electron lifetime. These points together explain the higher V oc for the device sensitized with PTZC than that sensitized with BzC. Moreover, analysis of R rec results reveal that both dyes effectively suppress interfacial charge recombination from surface states as seen by relatively low m values.
To rationalize these effects in terms of dye properties, the effect of dye structure on E CB is discussed below. The energetics of TiO 2 can be explained in terms of dipole moment of dyes and dye coverage, as suggested in Equation (12) [47] ΔE CB ¼ À qμ normal γ ε 0 ε where ΔE CB is the shift in TiO 2 CB, γ is the dye surface concentration,μ normal is the dipole moment component of individual molecule perpendicular to TiO 2 surface, and ε 0 and ε are the dielectric constant of the monolayer and the permittivity of the vacuum, respectively. Equation (12) suggests that higher dye surface concentration and largerμ normal cause the negative shift (upward shift) in E CB . The total dipole moment (μ total ) is used to compare the effect of dipole moment on E CB shift by assuming that the orientation of the dye is perpendicular to TiO 2 surface. From DFT calculations results, the dipole moments of BzC and PTZC are 11.08 and 12.07 D, respectively ( Figure 10). Although μ total of BzC and PTZC are similar, the angle of μ total with respect toμ normal for BzC is larger than that of PTZC. It suggests that μ total of PTZC induces more upward shift in E CB , compared with BzC, as the direction of μ total of PTZC is more perpendicular to TiO 2 surface or closer tõ μ normal . [48] Moreover, the dye uptake results (Table 3) show that PTZC exhibits higher dye-loading amount than BzC, implying better dye coverage (relating to γ) from PTZC. Thus, the combined results between the dipole moment and dye coverage indicate the trend in more upward shift in E CB from PTZC. Apart from the shift in E CB , PTZC structure is composed of PTZ unit featuring the hexyl chain. It is well known that the long alkyl chain can retard electron recombination with acceptor species in the electrolyte. [49,50] This may also contribute to slower recombination kinetics for PTZC.

Synthesis Cost Analysis
Most reported DSSC dyes were synthesized using versatile Pd cross-coupling reactions. In such studies, the amount of Pd catalyst may not be optimized and the typical %loading range may fall around 5-10%. This may be more than the reaction needs, leading to higher synthesis costs. Although efforts to optimize the %loading of Pd catalysts could make some contribution to lower costs, using alternative Pd-free reactions circumvents the issue entirely. Hence, in our case, we have used the HWE reaction to connect the donor and π-spacer. To demonstrate the concept of alternative reactions for Pd cross-coupling reactions and low-cost donor moiety, the synthesis costs of BzC and PTZC are calculated. This estimation is only based on the preparation cost in which the work-up and purification costs are not included as they depend on the www.advancedsciencenews.com www.solar-rrl.com scale of the synthesis and the purification techniques may be substituted by other appropriate techniques for the larger-scale synthesis. The estimated cost analysis was carried out following Maciejczyk et al., [51] which is based on 1 gram of product. The synthesis yield at each step was taken into account in the cost estimation (Scheme 1 and 2). Table S1 and S2, Supporting Information, depict the estimated synthesis cost for the CPDT unit from twostep reactions. The overall cost for the CPDT unit for the next steps is $37.43 g À1 . The synthesis of BzC comprises four steps and the corresponding cost analysis for each step is shown in Table S3-S6, Supporting Information. The total estimated synthesis cost for BzC is $50.06 g À1 . In the case of PTZC, there are 7 steps involved in the synthesis and the corresponding cost analysis for each step is depicted in Table S7-S13, Supporting Information. The total estimated synthesis cost for PTZC is $61.94 g À1 . From the estimation, one can see that the large share of the synthesis cost is the CPDT unit, accounting for %63% of the total synthesis cost.
To prove the concept of the cost-effective synthesis, LEG-4, one of the efficient commercial dyes, was chosen to compare with BzC and PTZC (see the synthetic method and structure in Scheme S1, Supporting Information). Unless stated otherwise, the synthesis methods and yields were taken from Gabrielsson et al. [11] and only material costs are considered so as to allow the direct comparison with BzC and PTZC. Table S14-S22, Supporting Information, show 10 steps of LEG-4 synthesis, including relevant intermediates, and the estimated synthesis cost of LEG-4. The total synthesis cost of LEG-4 is estimated to be $467.63 g À1 . It is clearly seen that approximately 65% ($310) of the total synthesis cost falls in the synthesis of the donor part as it involves an expensive boronic acid derivative. Moreover, the prices of the Pd catalysts for the Suziki-Miyaura crosscoupling used to synthesize LEG-4 are more than or equal to the total synthesis price of BzC and PTZC, respectively. When compared with BzC and PTZC, the total synthesis cost of LEG-4 is around 7.5-9 times higher than those for BzC and PTZC. This indicates the possibility of alternative Pd-free reactions for dye synthesis. For example, in our case, we show that the HWE reaction to link the donor with π-spacer is endowed with the potential for cost-effective synthesis which may be able to replace Pd cross-coupling reaction. It is also worth highlighting the use of hazardous pyrophoric reagents, such as n-BuLi, when considering large-scale synthesis. Although the synthesis of CPDT may need n-BuLi as one of the reagents, we have avoided the use of pyrophoric reagents in subsequent steps (Scheme 2), and this can also serve as a guide toward more sustainable dye design and synthesis in the future.
Finally, for a complete picture of the cost analysis, the photovoltaic performance needs to be taken into account to determine the cost performance. Note that the associated cost of glass, TiO 2 , Pt, and electrolyte is not taken into account. The DSSC devices with LEG-4 were also fabricated with I À =I À 3 electrolyte to reference the cost performance. The photovoltaic results show that the best device (from three individual cells) with LEG-4 attained PCE of 7.41% (Table 3), which compares favorably well with literature. [52] The cost performance calculation was determined by cost per unit area per %PCE following Tanaka et al. [53] The calculated results are shown in Table 5. Although the synthesis cost of BzC is lower than that of PTZC, the cost performance of PTZC is only 3.12% higher than that of BzC when considering the photovoltaic performance. This is due to higher photovoltaic performance obtained from PTZC and comparable synthesis cost of PTZC with BzC. When compared with LEG-4, it is clearly seen that the cost performance of BzC and PTZC are around 5 times cheaper than that of LEG-4. This is due to the much lower synthesis cost and broadly comparable PCE (5.92% for PTZC and 7.41% for LEG-4). The cost performances of PTZC could be even lower if meticulous device optimizations were carried out.

Conclusion
Two new dyes (BzC and PTZC) based on low-cost donor moiety and simple structures were successfully synthesized with Pd-free reactions. The aims of Pd-free synthesis are not only to reduce the synthesis cost but also to highlight environmental concerns, especially when the Stille crosscoupling is used in the synthesis. For optical properties, both dyes exhibit high extinction coefficient with PTZC showing slightly higher extinction coefficient compared with BzC. Due to large conjugation in PTZC, it shows the first oxidation potential of about 0.35 V less than that of BzC. The study on photovoltaic performance reveals that PTZC shows better efficiency with around 6% obtained from the best device. The higher J sc for PTZC is attributed to better light harvesting efficiency and more dye loading on the TiO 2 film. The EIS measurements reveal that BzC and PTZC exhibit relatively low ideality factor (m) with the values of 1.03 and 1.12, respectively, indicating efficient suppression of electron recombination from surface trap states. The device with PTZC shows slower recombination kinetics and upward shift of TiO 2 conduction band, leading to higher V oc than BzC. We also demonstrated the cost analysis for the synthesis of BzC and PTZC, which are estimated to be around $50 g À1 and $60 g À1 , respectively. When compared to LEG-4 used as the reference dye, the synthesis costs of BzC and PTZC are around 7.5-9 times lower. Moreover, the overall yields of BzC and PTZC are very high (around 55-70%) which can be scalable. The comparison of the BzC and PTZC cost performances with LEG-4 reveals that they are around 20% of the cost of LEG-4 per cm 2 per %PCE. Indeed, the efficiency does matter but we also believe that when it comes to practical use, that is, large-scale production, the associated price should be taken into account.

Experimental Section
Dye Synthesis: All reagents were purchased from Sigma-Aldrich (Merck), Alfa-Aesar, Fisher Scientific, Acros Organics, and Fluorochem unless stated otherwise and were used without further purification. CPDT was purchased from Shanghai Qinghang Chemical Co. Ltd., The values were estimated from dye desorption (Table 3).
www.advancedsciencenews.com www.solar-rrl.com China, and was used as received. The synthetic methodology of BzC and PTZC is described as follows. (1): CPDT (1.50 g, 8.42 mmol), 1-bromohexane (3.6 mL, 25.7 mmol), and KI (0.14 g, 0.84 mmol) were dissolved in 60 mL of DMSO and bubbled with N 2 for 10 min. Ground KOH (1.92 g, 34.2 mmol) was added into the mixture, and the resulting mixture was stirred overnight at room temperature in the dark. After adding distilled water, the mixture was extracted with hexane. The organic phase was collected, dried over anh. Na 2 SO 4 , filtered, and evaporated to remove hexane. The residue was purified by column chromatography on SiO 2 gel column eluted with hexane to obtain light yellow oil (2.67 g, 91.6%) 1 Compound 1 (0.26 g, 0.75 mmol) was dissolved in 5 mL anh. 1,2-dichloroethane. The solution was cooled to 0°C with an ice bath. To the solution, 0.1 mL (1.29 mmol) of anh. DMF was added and the mixture was stirred at the same temperature under N 2 . Then, 0.11 mL (1.18 mmol) of POCl 3 was added to the cooled mixture dropwise, and the resulting mixture was stirred at the same temperature for 4 h (TLC monitor). To the completed reaction was added saturated aqueous solution of NaOAc resulting in a separation into two phases. The mixture was stirred at 50°C for about 1.5 to 2 h. After that, the mixture was extracted with EtOAc, dried over MgSO 4 , filtered, and evaporated to remove EtOAc. The pure product was purified by silica plug eluted with 1:1 v/v of hexane: DCM to obtain yellow oil (0.26 g, 94.7%). 1  Synthesis of (E)-4,4-Dihexyl-2-Styryl-4 H-Cyclopenta[2,1-b:3,4-b 0 ]Dithio-Phene (4): Compound 3 (0.28 g, 1.23 mmol) and compound 2 (0.29 g, 0.78 mmol) were added to a reaction flask and dissolved in 5 mL of anh. THF. The mixture was cooled down to 0°C and stirred under N 2 . In another container, potassium tert-butoxide (KO t Bu) (0.29 g, 2.58 mmol) was dissolved in 10 mL of anh. THF. The KO t Bu solution was added to the cooled mixture drop wise, stirred at 0°C, and gradually warmed up to 25°C before being left to stir overnight under N 2 . After which time, water was added to the mixture before it was extracted with EtOAc and dried over MgSO 4 , filtered, and evaporated to remove EtOAc. The crude product was purified by column chromatography SiO 2 eluted with gradient elution of a mixture of hexane and DCM (20% increment of DCM to 80%DCM) to obtain yellow oil (0.32 g, 91.4%). 1   Synthesis of (E)-4,4-Dihexyl-6-Styryl-4 H-Cyclopenta[2,1-b:3,4-b 0 ]dithiophene-2-Carbaldehyde (5): Compound 4 (0.31 g, 0.69 mmol) was dissolved in 5 mL anh. 1,2-dichloroethane, and the solution was cooled down to 0°C with an ice bath. To the solution, 0.17 mL (2.19 mmol) of anh. DMF was added and the mixture was stirred at the same temperature under N 2 . Then, 0.10 mL (1.07 mmol) of POCl 3 was added to the cooled mixture drop wise, and the resulting mixture was stirred at the same temperature for 4 h (TLC monitor). To the completed reaction was added saturated aqueous solution of NaOAc resulting in a separation into two phases. The mixture was stirred at 50°C for about 1.5 to 2 h. After that, the mixture was extracted with EtOAc, dried over MgSO 4 , filtered, and evaporated to remove EtOAc. The pure product was purified by silica plug eluted with a mixture of hexane and DCM with 20% increment of DCM to 80%DCM to obtain orange wax (0.32 g, 95.8%). 1  Synthesis of (E)-2-Cyano-3-(4,4-Dihexyl-6-((E)-Styryl)-4 H-Cyclopenta-[2,1-b:3,4-b 0 ]-Dithiophen-2-yl)acrylic Acid (BzC): 0.18 Â g (2.12 mmol) of cyanoacetic acid was added in a reaction flask equipped with a condenser. Compound 5 (0.25 g, 0.53 mmol) was dissolved in 10 mL of anh. CHCl 3 and transferred to the reaction flask. The mixture was stirred at 75°C for about 2 min before adding 0.42 mL (4.24 mmol) of piperidine. The resulting mixture was refluxed at 75°C for 6 h (TLC monitor). After that, the mixture was diluted with DCM and extracted with 1 M HCl and the DCM phase was collected, dried over Na 2 SO 4 , filtered, and evaporated to remove DCM. The crude product was purified by dry column vacuum chromatography (DCVC) technique with gradient elution (hexane to DCM with 10% increment of DCM, pure DCM for 10 fractions, and DCM to 20% MeOH with 2% increment of MeOH). The pure product was diluted with DCM, extracted with 1 M HCl, dried over MgSO 4 , filtered, and evaporated to obtain pink solid (0.26 g, 90.2%). 1  Synthesis of 10-Hexyl-10 H-Phenothiazine (6): Phenothiazine (PTZ) (3.00 g, 15.05 mmol), 1-bromohexane (3.2 mL, 22.87 mmol), and KI (0.4 g, 2.41 mmol) were dissolved in 50 mL of DMSO and bubbled with N 2 for 10 min. Ground KOH (2.60 g, 46.3 mmol) was added into the mixture, and the resulting mixture was stirred overnight at room temperature in the dark. After adding distilled water, the mixture was extracted with ether. The organic phase was collected, dried over anh. MgSO 4 , filtered, and evaporated to remove ether. The residue was purified by column chromatography on SiO 2 gel eluted with 2% EtOAc in hexane to obtain pale yellow oil (4.05 g, 95.0%). 1  Synthesis of 10-Hexyl-10 H-Phenothiazine-3-Carbaldehyde (7): Compound 6 (0.46 g, 1.62 mmol) was diluted in 10 mL of anh. 1,2-dichloroethane in a reaction flask equipped with a condenser and the solution was cooled down to 0°C. In another container, 0.45 mL (4.83 mmol) of POCl 3 was added dropwise to a cooled solution of anh. DMF (0.63 mL, 8.14 mmol) in 5 mL of anh. 1,2-dichloroethane. The mixture of POCl 3 and DMF was added to the solution of compound 6 dropwise while being stirred at 0°C. After that, the mixture was refluxed at 90°C for 4 h. To the mixture was added saturated aqueous solution of NaOAc and stirred at 50°C for 1 h. Then, the mixture was extracted with EtOAc, dried over MgSO 4 , filtered,and evaporated to remove EtOAc. The crude product was purified by column chromatography on SiO 2 and eluted with 15% Synthesis of (10-Hexyl-10 H-Phenothiazin-3-yl)methanol (8): Compound 7 (0.40 g, 1.28 mmol) was dissolved in 12 mL of THF:MeOH (5:1 v v À1 ) mixture. After that, NaBH 4 (0.1 g, 2.64 mmol) was added to the solution portion wise, and bubbles formed upon adding NaBH 4 . The mixture was stirred at room temperature for 1.5 h during which time, the colour of solution changed from bright yellow to pale yellow. After that, the mixture was evaporated and redissolved in EtOAc, followed by extraction with brine, dried over MgSO 4 , filtered, and evaporation. The crude product was purified by silica plug with 70%hexane:30%EtOAc elution to obtain pale yellow oil (0.40 g, quant.). 1 (9): In an oven-dried flask equipped with a condenser, ZnI 2 (0.32 g, 1.00 mmol) was added, followed by evacuation, and refilled with N 2 for three times. The solid ZnI 2 was suspended in 5 mL anh. THF before adding 0.22 mL (1.28 mmol) of triethylphosphite. In another container, compound 8 (0.20 g, 0.65 mmol) was dissolved in 10 mL of anh. THF. After that, the solution of compound 8 was added in the above mixture of ZnI 2 and triethylphosphite, followed by reflux at 75°C for 16 h. The mixture was diluted with ether and extracted with 2 M NaOH. The ether phase was collected, dried over MgSO 4 , filtered, and evaporated. The crude product was distilled under vacuum to remove excess triethylphosphite to obtain light yellow oil (0.25 g, 90.0%). 1  Synthesis of (E)-3-(2-(4,4-Dihexyl-4 H-Cyclopenta[2,1-b:3,4-b 0 ]dithiophen-2-yl)vinyl)-10-Hexyl-10 H-Phenothiazine (10): Compound 9 (0.58 g, 1.34 mmol) and compound 2 (0.42 g, 1.12 mmol) were dissolved in 10 mL of anh. THF and the solution was cooled down to 0°C. In another container, KO t Bu (0.31 g, 2.75 mmol) was dissolved in 10 mL of anh. THF. The KO t Bu solution was added to the cooled mixture dropwise, stirred at 0°C, and gradually warmed up to 25 0°C before being left to stir overnight under N 2 . After this, water was added to the mixture before it was extracted with DCM and dried over MgSO 4 , filtered, and evaporated to remove DCM. The crude product was purified by column chromatography SiO 2 with 20%DCM:80%hexane elution to obtain yellow wax (0.70 g, 95.6%). 1 carried out by O. McCullough at London Metropolitan University using a Carlo Erba CE1108 Elemental Analyzer.

Synthesis of 4,4-Dihexyl-4 H-Cyclopenta
Optical Characterization: UVÀvis absorption spectra were recorded on Jasco V-670 UV/vis/NIR spectrophotometer controlled by the SpectraManager software. All solutions were prepared in DCM. All extinction coefficients were determined by the Beer-Lambert plot with various concentrations in the range of 10-35μM of corresponding dye. The dye desorption was carried out by diluting 1 M of tetrabutylammonium hydroxide (TBAOH) in methanol to 5 mM TBAOH with DCM. The sensitised films were soaked in 5 mM TBAOH for 30 min before making up the volume with DCM. The absorption spectra were measured in a similar manner to UV-vis absorption characterization.
Electrochemical Characterization: All voltammetry measurements were performed in anhydrous DCM with 0.3 M TBAPF 6 as supporting electrolyte in a three-electrode system at room temperature. Each solution was purged with N 2 prior to measurement. The working electrode was a Pt disk while the counter electrode and reference electrode were a Pt rod and Ag/AgCl in 2 M LiCl in ethanol, respectively. All measurements were carried out using μAUTOLAB Type III potentiostat driven by the electrochemical software GPES. The scan rates used in CV were 25, 50, 100, 200, and 400 mV s À1 . SWV were conducted at a step potential of 2.1 mV, amplitude of 250 mV, and frequency of 25 Hz, which gave a scan rate of 52.5 mV s À1 . All measurements were referenced to Ferrocene/ Ferrocenium (Fc/Fc þ ) as an internal standard.
Theoretical Calculations: The molecular dye structures were built in Avogadro [54] and the drawn molecule was adjusted using basic optimization in Avogadro. The drawn molecules were optimized by using Gaussian 09 [55] at B3LYP [31] level of theory with 6-31G(d) basis set under vacuum. After that, the optimized structures were subject to re-optimization in DCM (PCM solvation model). [56] The TDDFT calculations used the optimized structures in solvation model and were conducted by using Gaussian 09 at CAM-B3LYP [32] level of theory with 6-31G(d) basis set and in DCM (PCM solvation model). The 50 states of singlet electronic transitions obtained from TDDFT calculations were processed with the GaussSum software package. [57] For the dyes' dipole moment, the DFT optimized dyes in DCM solvent model were subjected to GaussView 6 [58] for visualization. Solar Cell Fabrication: Fluorine-doped tin oxide (FTO)-coated glass (Merck, 7 Ω sq À1 ) was cleaned using 2% Hellmanex solution with sonication in deionized water for 30 min, followed by sonication in ethanol for 30 min. The cleaned glass was treated with UV-O 3 treatment for 20 min. After that, the cleaned and treated glass was pretreated with 40 mM aqueous solution TiCl 4 at 70°C for 30 min, which was subsequently rinsed with deionized water and ethanol, respectively. The TiCl 4 -treated glass was sintered at 500°C for 30 min. Upon cooling down, the sintered glass was screen printed with commercial transparent TiO 2 paste (Ti-Nanoxide T/ SP, Solaronix), followed by drying at 120°C for 10 min. The scattering TiO 2 paste (Ti-Nanoxide T/SP, Solaronix) was screen printed on printed glass, followed by drying at 120°C for 10 min. The films were annealed at 500°C for 15 min using programmable hotplate. The resulting TiO 2 film thickness was 8 μm with the area of 0.28 cm 2 . The sintered films were allowed to cool down and were treated with 40 mM aqueous solution TiCl 4 at 70°C for 30 min, rinsed with deionized water and ethanol, and were sintered again at 500°C for 30 min. When the temperature dropped to about 90°C, the films were soaked in 0.5 mM dye bath containing 2.5 mM CDCA for BzC and PTZC and containing 5 mM CDCA for LEG-4 in tert-butyl alcohol:acetonitrile 1:1 v/v for 20 h. The sensitized working electrodes were removed from dye bath and washed with acetonitrile to remove unadsorbed dye molecules. The counterelectrodes were predrilled on FTO glass, washed by sonicating in 0.1 M HCl in ethanol for 15 min, and followed by sonicating in ethanol for 15 min and deionized water for 15 min, respectively. The cleaned predrilled glass was doctor bladed with Platisol (Solaronix) and sintered at 450°C for 15 min. The DSSCs were assembled using hot-melt sealing film (Solaronix). The composition of electrolyte was 0.1 M LiI, 0.05 M I 2 , 0.6 M DMPII, and 0.5 M 4-tert-butyl pyridine (4-tBP) in acetronitrile. The electrolyte was injected into the assembled cells using Vac'n'Fill Syringe (Solaronix). The hole was covered by cover glass with hot-melt sealing film (Solaronix).

Solar Cell Characterization:
The photocurrent-voltage ( J-V ) measurements were carried out on an Autolab potentiostat (Metrohm), driven by electrochemical software GPES, with class AAA SLB300A solar simulator (Sciencetech) as the light source at University of Edinburgh. The light intensity was calibrated to AM 1.5G (100 mW cm À2 ) using a silicon reference cell. A black metal mask with a circular aperture of 0.0707 cm 2 was applied when measuring J-V curves. The EIS spectra of I À =I À 3 electrolytebased devices setup was similar to the J-V measurements, except for using white LED as the light source and Frequency Response Analyser (FRA) software. The EIS spectra were recorded in the frequency range between 1 MHz and 0.1 Hz at various forward biases set to the corresponding voltage produced by white LED illumination. The obtained spectra were fit with a transmission line model [37] using Zview (Scribner Associates) software. IPCE measurements were conducted at University of Strathclyde. The IPCE of the solar cells were measured with a photospectrometer setup (Bentham PVE300) by illuminating the solar cell with modulated monochromatic light (Xenon and quartz halogen lamps) through 1.85 mm slit. The intensities of the lamps were calibrated with a silicon photodiode. The photospectrometer was operated in DC mode and the spectral resolution was set to 5 nm.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.