A greener procedure for the synthesis of [Bu₄N]₂-cis-[Ru(4-carboxy-4′-carboxylate-2,2′-bipyridine)₂(NCS)₂] (N719), a benchmark dye for DSSC applications

Abstract

A previously reported protocol for the synthesis of the commercial dye [Bu₄N]₂[Ru(4-carboxy-4′-carboxylate-2,2′-bipyridine)₂(NCS)₂] (N719) was thoroughly optimized in terms of the amount of input materials needed, reaction times and temperatures achieving significant reductions in all the three synthetic steps. This optimization allowed a 81% reduction of the Sheldon's E factor of the overall process.

---

Introduction

The commercial dye N719 ([Bu₄N]₂-cis-[Ru(4-carboxy-4′-carboxylate-2,2′-bipyridine)₂(NCS)₂], Fig. 1) still remains an important benchmark for dye-synthesized solar cell (DSSC) applications.^2–8

Fig. 1 Structure of dye N719.

According to the original procedure reported by Grätzel and co-workers⁹ the synthesis of N719 requires that RuCl₃ is reacted with 4,4′-dicarboxylic acid 2,2′-bipyridine (H₂dcbpy) at reflux in N,N-dimethylformamide (DMF) to obtain the cis-[Ru(H₂dcbpy)₂Cl₂] complex, which is refluxed again in DMF with excess KNCS to obtain the corresponding cis-[Ru(H₂dcbpy)₂(NCS)₂] complex. N719 ([Bu₄N]₂-cis-[Ru(Hdcbpy)₂(NCS)₂]) is then obtained by treatment with excess Bu₄NOH followed by a slow titration to pH 3.8 with 0.1 M HNO₃ and precipitation.

Later reports¹⁰ outlined that the material obtained by this procedure contains some impurities, the most important among them being isomeric complexes containing S-bound thiocyanate groups^11,12 (“S-isomer”) and that repeated chromatographic purifications over Sephadex-LH20 are necessary to obtain pure N719.^12–14

In order to avoid the tedious and expensive repeated purifications over Sephadex-LH20, Rawling et al.¹ reported an alternative procedure for the synthesis of N719, which involves a protection/deprotection strategy that allows the use of conventional silica gel in the single chromatographic purification step needed (Scheme 1).

Scheme 1 N719 synthesis as reported in ref. 1.

Although a protection/deprotection strategy such as that adopted by Rawling et al.¹ is generally not considered good green-chemistry practice,¹⁵ in this case it represents an extremely good trade-off, since it allows to avoid the repeated tedious chromatographic separations over Sephadex-LH20 and also because no additional deprotection step is actually needed. In fact, the saponification of the isobutyl esters occurs during the final treatment with tetrabutylammonium hydroxide, which is needed anyway for the synthesis of N719. So the only additional step introduced in the synthesis is the initial protection of the 2,2′-bipyridine-4,4′-dicarboxylic acid.

However, in view of possible scale-up application, this modified synthesis still presents some drawbacks in terms of the greenness: the use of three different solvents in the three steps depicted in Scheme 1,¹⁶ the use of large amounts of solvents in some steps, the use of a problematic solvent such as acetonitrile in the last step of the synthesis.

In this paper we present the results of a thorough optimization we performed on this modified synthesis of N719, which led to an 81% reduction of the Sheldon's E factor of the process.¹⁷

Results and discussion

The literature synthesis of the complex N719 consists in a three-step procedure (Scheme 1).¹ Step one is the protection of the carboxylic acid of the ligand by Fisher esterification. In the second step, the protected bipyridine is coordinated to the metal centre and the “dithiocyanate-complex” is formed.

The synthetic crude needs to be purified by chromatographic column before the third step that leads to the formation of the final product (Fig. 1) via saponification of esters and precipitation with acid. Each synthetic step was optimized with respect to the literature procedure. The most significant data for each step are discussed in detail in the following section.

Step 1 – esterification of H₂dcbpy

The optimization of the first step of the process concerned (i) the reduction of the solvent volume, (ii) the reduction of the sulphuric acid amount, and (iii) the nature of the catalyst.

Table 1 highlights the results we obtained after the optimization of the solvent volume employed for the reaction and of the sulphuric acid amount. In particular, we found that the volume of isobutanol employed can be reduced down to 10 mL g⁻¹ of bipyridine (Table 1, entry 3) instead of the 40 mL g⁻¹ used in the original work. Also, the molar ratio between sulphuric acid and the starting H₂dcbpy can be decreased from 3 to 0.75 (Table 1, entry 3). Both these modifications did not affect the yields of the reaction.

Table 1 Optimization results for the esterification step

Entry	H2dcbpy (g)	iBuOH/H2dcbpy (mL g^−1)	H2SO4/H2dcbpy (mmol mmol^−1)	Yield^a (%)
a Isolated yield.
Ref. 1	1.5	40	3.00	92
1	2.5	40	3.00	96
2	2	10	3.00	96
3	1.8	10	0.75	92

---

These optimizations also show positive effects on the workup procedure, which in view of possible industrial application are extremely desirable. Less solvent means a faster and less energy demanding final evaporation. Less acid means less base needed for neutralization and less aqueous waste produced. In our hands, even when run under the same reaction conditions described in the literature, this reaction was complete in more than the 4 h reported in the literature. Careful optimization of the reaction times allowed us to find that this esterification is reproducibly complete after 16 h of reflux.

The nature of the acid was also explored and alternative homogeneous (H₃PO₄) and heterogeneous acids (Dowex® 50Wx8, Zeolite HY, Nafion-H and silica) were tested in this step. In particular, heterogeneous acids were considered in the perspective of a possible flow application of this step. However, the results in terms of yields and reaction times were all rather unsatisfactory and sulphuric acid remains the best acid among those tested.

Step 2 – formation and purification of the [Ru(i-Bu₂dcbpy)₂(NCS)₂] complex

This step is a “one-pot” procedure that must be performed under inert atmosphere and with degassed solvent (Scheme 2). In step 2.1, i-Bu₂dcbpy is reacted at reflux with RuCl₃ and when the formation of the dichloro complex is complete, excess ammonium thiocyanate is added to the mixture and the formation of the desired product occurs (Scheme 2, step 2.2).

Scheme 2 Step 2 of the synthesis of N719.

When we initially reproduced the original procedure (Scheme 1, step 2), we found that the desired complex cis-[Ru(i-Bu₂bpy)₂]₂(NCS)₂] was always accompanied by two slower-eluting impurities we named as impurity B and impurity C. In particular, impurity B elutes very close to the desired tetra-i-Bu ester and makes the silica gel chromatographic purification very difficult. Even using large amounts of silica and solvents, a complete recovery of the desired tetra-i-butyl ester complex was not possible, as part of it was always lost in mixed fractions along with the impurity B. Therefore, in order to avoid these difficulties in the purification step we focused our efforts on identifying these two impurities with the aim to avoid their formation and simplify the chromatographic purification.

Three fractions were obtained after the chromatographic purification: the first fraction is composed only of product A, which was clearly identified by NMR and ESI-MS as the desired [Ru(i-Bu₂dcbpy)₂(NCS)₂] complex; the third fraction contains only impurity C, which was clearly identified by NMR and ESI-MS as the undesired “S-isomer” (Fig. 4C). The second fraction instead could never be obtained as a pure compound and was collected as a mixed fraction, where the major component [Ru(i-Bu₂dcbpy)₂(NCS)₂] complex was always accompanied by variable amounts of impurity B as the minor component. Comparison of HPLC chromatograms of the second and third fraction were made with that of the crude product of step 2 (Fig. 2). HPLC analysis of the second fraction (Fig. 2, middle) clearly shows that impurity B is actually a mixture of two different compounds.

Fig. 2 HPLC chromatograms of step 2 reaction crude (top), second fraction after purification containing A and impurity B (middle), and third fraction containing impurity C (bottom).

The ¹H NMR spectrum of this mixture (Fig. 3) shows the presence of aromatic signals almost coincident with those of A. Two minor sets of signals are clearly distinguishable in the aliphatic portion of the spectrum, apart from the presence of the signals due the isobutyl ester groups of the major A component (Fig. 3, insets) and ascribable to ethyl ester groups. The ESI MS analysis of this mixed fraction (A + B) exhibits the ion peak of the species A (m/z = 930) and, an additional peak with a m/z = 902.

Fig. 3 ¹H NMR spectrum of the fraction A + B obtained after step 2.

All these experimental evidences suggest that impurity B may be the mixture of the two isomeric different mono-transesterifications species (Fig. 4B). The formation of these molecules probably occurs via a random transesterification of one of the isobutyl ester group with ethanol, which is used as the solvent in this step.

Fig. 4 Structure of the components obtained after step 2 of the process.

Species B were never isolated, so, in order to confirm our hypothesis, fraction A + B was tested as a starting material for step 3, and indeed, the final saponification procedure was found to be effective also on species B. In fact, under this reaction conditions its conversion occurs with no detectable difference in both kinetic and yield with those performed on the compound A alone.

Therefore, having identified species B led us to the conclusion that species A and B can be collected together, greatly facilitating the chromatographic purification performed at the end of step 2 of the process.

The identification of the species B was of great importance for the purposes of the optimization of the process, indeed, we decided to take this finding one step forward and investigated the possibility to avoid altogether the formation of B, by carrying out step 2 in i-BuOH instead of ethanol.

Nature of the solvent and the way to add ruthenium chloride. Initially, we performed step 2 under the reaction conditions reported described in the literature, using ethanol as the reaction solvent, and we found this reaction to be very robust and reproducible. By careful monitoring the progress of the reaction by HPLC, we found that the reaction times for both the substeps 2.1 and 2.2 are much shorter than what reported in the literature (Table 3, entries 1–3).²¹

The first modification adopted with respect to the literature procedure concerned the nature of the solvent, in particular, by changing EtOH with i-BuOH, we wanted both to avoid the formation of the trans-ester (Fig. 4B) and to simplify the process by unifying the solvent of the first and second step.

When we initially ran this step in ethanol as the solvent, we found that the addition of the ruthenium salt to the reaction mixture was done most conveniently as an ethanolic solution. When we switched to isobutanol as the reaction solvent, we tried to maintain the same mode of ruthenium addition via an isobutanolic solution. However, we observed that the much lower solubility at room temperature of the ruthenium trichloride in this solvent gave lots of difficulties in the preparation of ruthenium trichloride solutions in isobutyl alcohol and in the addition itself. Therefore, in order to avoid this solubility problem, the metal precursor was introduced directly to the reaction flask as a solid powder, which is the same mode of addition used in the original procedure.¹

The progress of the reaction was monitored by HPLC and we were pleased to find that both the change of the reaction solvent and of the ruthenium mode of addition did not affect the result. Moreover, the use of the higher boiling i-BuOH, in place of ethanol, allowed a further reduction of the reaction times involved in step 2 from 60 and 90 min to 30 and 60 min respectively (Table 2, entry 3 and 4).

Table 2 Results of the optimization of step 2

Entry^a	Reaction scale^b	Solvent	Solvent/RuCl3 (mL mmol^−1)	NH4SCN/RuCl3 (mmol mmol^−1)	Time (min)		Yield^c (%)
					Step 2.1	Step 2.2
a All reactions were run with a i-Bu2dcbpy/RuCl3 molar ratio = 2 at reflux in the corresponding solvent.b Expressed as the molar ratio between the RuCl3 used and the RuCl3 used in ref. 1.c Isolated yield after silica gel chromatography.d Product not isolated, yield estimated.
Ref. 1	1	EtOH	215	40	210	150	49
1	3.7	EtOH	42	40	210	150	40
2	4.7	EtOH	42	40	60	90	65
3	2	i-BuOH	42	40	30	60	[51]^d
4	1.9	i-BuOH	42	20	30	60	[53]^d
5	1.9	i-BuOH	42	5	30	60	[55]^d
6	1.9	i-BuOH	42	4	30	60	54
7	2.6	i-BuOH	42	4	30	60	54
8	5.2	i-BuOH	42	4	30	60	60

---

Step 2.1 was considered complete when the peak corresponding to the starting iBu₂dcbpy ligand disappeared in the HPLC analysis. HPLC monitoring of step 2.2 (Scheme 2) showed that the intermediate dichloro complex disappears after a very short time. However, initially the reaction forms large amounts of the kinetic “S-isomer” (Fig. 4C) which then equilibrates over the time to the thermodynamic desired complex (Fig. 4A) up to a point where, after about 60 min, the ratio between the HPLC area of the peak of A and that of C approximately reaches the value of 3.3. Leaving the reaction at reflux for several additional hours, does not change this ratio.²¹

As previously mentioned, ruthenium trichloride is poorly soluble in isobutanol, therefore this reaction was first conducted in heterogeneous phase. In order to run the reaction under constantly homogeneous conditions we tried to use the more lipophylic ruthenium p-cymene dimer as the ruthenium source for this reaction. This complex has less problems of hygroscopicity compared to the ruthenium trichloride and is far more soluble in i-BuOH.

However, when using ruthenium p-cymene dimer it was noticed that the reaction rate is much lower with respect to reactions in which ruthenium trichloride was used. To speed up the rate of the reaction we also tried to remove cymene co-distilling it out of the reaction mixture with part of i-BuOH, which was constantly replaced with freshly degassed i-BuOH, added via a dropping funnel. The presence of free cymene in the distilled-off solvent was observed by GC analysis. HPLC analysis of the reaction mixture showed additional peaks not attributable neither to the desired product (A) nor to the “S-isomer” (C). Finally, ¹H NMR spectrum of the crude displays several aromatic signals probably resulting from the formation of additional ruthenium complexes obtained with p-cymene and bipyridine ligand. We therefore concluded that ruthenium p-cymene dimer is not a viable ruthenium source for this process and that ruthenium trichloride a better precursor.

It is also relevant to note that the solvent amount needed to run this step can be drastically reduced down to a fifth with respect to volumes reported in the literature (Table 2, entry 1), without affecting the yields.

Optimization of the amount of ammonium thiocyanate. The ammonium thiocyanate is used exclusively in step 2.2 to replace the chloride atoms in the dichloro complex cis-[Ru(i-Bu₂dcbpy)₂(Cl)₂] (Scheme 2). All the published procedures report the use of very large excesses of the thiocyanate inorganic source (potassium and ammonium being the most common), ranging from 10 to 20 times that stoichiometrically needed. From an environmental and industrial point of view, use of a large excess of thiocyanates represents a cost for both its disposal and its toxicity. In addition to this, excess thiocyanate produces a lot of insoluble inorganic material in the reaction crude, which hampers the following isolation and purification of the complex. We therefore wanted to verify carefully if such a large excess of NH₄SCN was really needed.

We then started to progressively reduce to molar ratio amount between the thiocyanate and the ruthenium used in the reaction and we found this ratio could be reduced from the reported value of 40 to 4 without affecting yields and reaction times (Table 2, entries 3–6).

The final optimized conditions for step 2 (Table 2, entry 6) use a fifth of the solvent, a tenth of NH₄SCN and is kept at reflux for a quarter of the time. To test the robustness of these optimized conditions several reactions were carried out without any significant variation of yield and reaction time. Finally, the reaction was tested also with a moderate increase of scale, up to five times the scale reported in the literature (Table 2, entries 7 and 8) obtaining good results.

Step 3 – formation of N719

Nature and amount of solvent. In the original literature the saponification of cis-[Ru(iBu₂dcbpy)₂(NCS)₂] occurs in acetonitrile, by addition of a solution of tetrabutylammonium hydroxide. After evaporation of the acetonitrile solvent, the residue was dispersed in water and precipitated in almost quantitative yield by the addition of acid to reach pH 3, 8. Finally, the compound was recovered by filtration (Scheme 3).

Scheme 3 Step 3 of the synthesis of N719.

One of the main objectives of this work was the replacement of acetonitrile used in this step with an alternative friendlier solvent such as an alcohol. MeOH, EtOH, i-PrOH and i-BuOH were all tested and we observed that, as expected, the solubility of the starting tetraester increases on going from MeOH to i-BuOH while the reaction proceeds in good yields with all of these solvents. Isobutanol was finally chosen as the preferred solvent of this step with the aim to unify the solvent throughout the entire process. Optimization of the solvent volume allowed us to run this step with a quarter of the solvent volume reported in the literature procedure, without affecting yield and quality of the final material. Finally, we did not put any effort to reduce the reaction time because half an hour was considered already short enough.

E factor assessment of the process

The optimized process was compared to the previous reference by means of Sheldon's E factor.¹⁷ The comparison between the two processes was carried out considering the charge of all input materials for each step.^18,19 The results of this comparison are summarized in Table 3.

Table 3 Comparison of the E factor for the two processes²¹

	Ref. 1 ^a		Present work^a		% Diff
a All values are expressed as mass amount normalized to the same mass unit of desired N719.
H2dcbpy		0.98		0.76	−22
i-BuOH		31.44		6.10	−81
H2SO4		1.20		0.23	−81
Waste of step 1		32.31		6.07	−81
RuCl3·3H2O		0.49		0.37	−24
Solvent	EtOH	312.16	i-BuOH	48.74	−84
NH4NCS		5.64		0.44	−92
Waste of step 2		318.76		49.78	−84
Solvent	CH3CN	152.38	i-BuOH	35.32	−77
Bu4NOH		9.05		7.98	−12
Waste of step 3		161.27		43.10	−73
E factor		512		99	−81

---

The table reports the E factor for both processes expressed as the mass ratio of waste to desired N719 product.^17,21 To have a clearer picture, the overall E factor is broken down to the single contribution of any single input material used and to the waste contribution for each of the three steps.

A quick comparison of the data for the two processes shows that in the present optimized process, the E factor was significantly reduced in all of the three steps. The reduction calculated over the entire process is 81%. The use of any single input material was reduced in the optimized process. One the most significant reductions was achieved in the solvents' usage. In fact, over the three steps, the original process employs three different solvents amounting to a total of 496 g of solvents per gram of N719 produced. The optimized process, over the three steps, uses i-BuOH as the only solvent amounting to only 90 g of per gram of N719 produced.

Since the solvents are always removed by evaporation, this important reduction of the solvents' amounts needed not only translates into a significant reduction of the waste but also translates into a significant reduction of time and energy required for these evaporations. Clearly, this reduction, accompanied to the fact that only one solvent is used over the entire process, greatly facilitates the recovery and reuse of the process solvent.

As a final consideration, it should be outlined that the overall E factor of the entire optimized process is lower than the contribution of the problematic acetonitrile solvent alone used in the third step of the literature process.

Characterization and performance assessment of the final product

Having introduced so many modifications on the original procedure it was of course crucial to fully characterize the material obtained with our optimized procedure and to test its performance as a DSSC²⁰ dye comparing it with commercially available N719. The final product was therefore characterized by ¹H-NMR, ESI-MS, HPLC, UV-Vis analyses and in term of thermal properties.

¹H-NMR spectroscopy is extremely useful to test N719 for the presence of the unwanted “S-isomer” which, when present, is clearly visible in the aromatic region and to estimate how many tetrabutylammonium groups are present in the sample.

Fig. 5 compares the aromatic region of the ¹H NMR spectra of the dye prepared with our optimized procedure (Fig. 5, bottom) with an N719 commercial sample. Both spectra show the characteristic six aromatic signals of the pyridine rings coordinated to the metal centre. Also the four aliphatic signals belonging to resonate, specifically, the ratio between the first aromatic proton and the last aliphatic one should be 1/12. As it can be observed from the spectrum, the dye is not affected by the presence of impurities and the value between the integrals is very close to the theoretical value. This first evidence confirms that the used procedure led to the desired product and NMR characteristics of our dye are entirely analogous to those of the commercial product.

Fig. 5 Comparison of ¹H NMR spectrum of commercial N719 (top) and of N719 obtained by the present optimized procedure (bottom).

The material was also analysed by HPLC (Fig. 6). The analysis of a commercial sample shows peaks eluting at retention times shorter than 6 minutes, followed by the peak corresponding to the “S-isomer” and finally, at 10.31 min, the peak corresponding to N719. In order to obtain quantitative information, our sample was quantitated against the commercial sample, which was used as a relative standard. Most of the samples obtained through this optimized procedure showed purity very close or even higher to that of the commercial sample.

Fig. 6 Comparison of HPLC chromatogram for commercial N719 (top) and of N719 obtained by this optimized procedure (bottom).²¹

Also the comparison of the two UV-Vis absorption spectra (Fig. 7) shows that the product obtained by our optimized procedure has a pattern quite similar to that of the commercial dye.

Fig. 7 UV-Vis spectra obtained by solution of dye 0.3 mM in ethanol for N719 obtained by the present optimized procedure (continuous line) and commercial N719 (dashed line).

The thermal stability of the two molecules is almost the same up to about 200 °C while above this temperature the commercial sample appears to be a bit more stable. Commercial N719 present a smaller peak at 400 °C and a shoulder around 300 °C not present in synthetic N719 (Fig. 8).

Fig. 8 TG/DSC analysis of dyes powder in air for N719 obtained by the present optimized procedure and commercial N719.

Finally, dye performance was assessed in DSSC test devices and was compared to that of a commercial dye. Thus, three sets of DSSC test devices were built:²¹ (i) a reference set of 4 DSSC cells containing commercial N719 dye (commercial); (ii) a set of 20 DSSC cells containing N719 obtained with several runs of our optimized procedure (homemade); (iii) a set of 4 DSSC cells containing N719 obtained with our optimized procedure in which chromatographic purification after step 2 was omitted (homemade not purified).

The characterization of the photovoltaic devices was accomplished using a Solar Simulator (Abet Technologies Sun 2000-11018) with AM 1.5G filter as the light source unit. I–V curves were obtained with a potentiostat unit (Autolab PGSTAT100).^22,23

To analyse the stability of the synthetized materials we performed an accelerated aging introducing the cells in a climatic chamber at 85 °C and a relative humidity of 15%. The photovoltaic efficiency was checked at 0, 50, 500 and 1000 hours of aging, respectively.

The average data for the three sets of DSSC test devices (Fig. 9) show how N719 prepared according to our improved procedure and maintaining the chromatographic purification step (set homemade) gives similar efficiency and stability over accelerated aging compared to the commercial dye. We also observed that the removal of the chromatographic purification that follows step 2 (set homemade not purified) causes a significant efficiency reduction but doesn't affect stability over time.

Fig. 9 Efficiency comparison of commercial N719 dye (dashed line) and N719 dyes obtained by the present optimized procedure, with (continuous line) and without purification process (dotted line). The reported values are average data of 4, 20 and 4 devices respectively.

Experimental section

The ¹H NMR spectra were recorded at 300, 400 or 600 MHz. The chemical shifts (δ) are given in ppm relative to residual signals of the indicated solvent. Coupling constants are given in Hz. following abbreviations are used to indicate the multiplicity: s, singlet; d, doublet; t, triplet; q, quartet; m, multiplet; bs, broad signal. ESI-MS were collected on a Waters Micromass ZQ 4000 probe with capillary potential 3.5 kvolts, source cone 20–25 V, source temperature 70 °C, and direct infusion of 20 μL min⁻¹. Column chromatography purifications were performed on Merck silica gel (Geduran, Si60, 63–200 μm particle size). TLC analyses were performed over Merck precoated TLC plates (Silica gel 60 GF₂₅₄ 0.25 mm). Organic solutions were concentrated under reduced pressure on a Büchi rotary evaporator. HPLC analyses were performed on an Agilent HP1100 instrument equipped with a quaternary pump with inline degasser, a column thermostating compartment (25 °C), a variable UV detector, a manual Rheodyne injector with a 20 μL loop, and an Alltech Alltima C18 5 μm column (250 × 4.6 mm) using the mobile phase composition and flow rates indicated below.

Commercial samples of N719 were purchased from Sigma-Aldrich. 2,2′-Bipyridine-4,4′-dicarboxylic acid and ruthenium salts were purchased from Alfa Aesar and used as received, while all other materials were purchased from Sigma-Aldrich and used as received.

Preparation of i-Bu₂dcbpy

2,2′-Bipyridine-4,4′-dicarboxylic acid (1.80 g, 7.37 mmol) was suspended in a mixture of isobutyl alcohol (18 mL) and conc. sulfuric acid (0.30 mL). The mixture was heated at reflux for 16 h, during which time the solution became clear.

The solution was cooled and the remaining isobutyl alcohol was removed using a rotary evaporator. The residue was taken up in water (25 mL) and saturated aqueous Na₂CO₃ was added until a pH of 8–9 was obtained. The precipitate was collected by filtration, washed with water and dried under vacuum to yield 2.41 g (92%) of product as an off-white solid, which was used in the next step without further purification. ¹H NMR δ_H (300 MHz, CDCl₃) 8.95 (dd, J 0.9, 1.5 Hz, 2H), 8.87 (dd, J 0.9, 5.0 Hz, 2H), 7.90 (dd, J 1.5, 4.8 Hz, 2H), 4.19 (d, J 6.9 Hz, 4H), 2.15 (sep, J 6.9 Hz, 2H), 1.05 (d, J 6.9 Hz, 12H).

Preparation of cis-[Ru(i-Bu₂dcbpy)₂(NCS)₂]

i-Bu₂dcbpy (2.05 g, 5.75 mmol) and RuCl₃·3H₂O (0.752 g, 2.87 mmol) were dissolved in deoxygenated isobutyl alcohol (122 mL) and the reaction vessel shielded from light. The resulting solution was refluxed under a nitrogen atmosphere for 0.5 h. Ammonium thiocyanate (0.89 g, 11.48 mmol) was then added to the reaction mixture and heating at reflux continued for a further 1 h. The solvent was then removed by rotary evaporation and the residue suspended in water (140 mL). The solid was collected by filtration and was purified by silica gel column chromatography eluting with acetone/dichloromethane (2 : 98). After removal of the solvent, 1.61 g (60%) of the desired compound were obtained as a dark red solid. ¹H NMR δ_H (600 MHz, DMSO-d₆) 9.47 (d, J 5.7 Hz, 2H), 9.16 (d, J 1.2 Hz, 2H), 8.99 (d, J 1.5 Hz, 2H), 8.45 (dd, J 1.8, 6.0 Hz, 2H), 7.82 (d, J 6.0 Hz, 2H), 7.62 (dd, J 1.6, 6.0 Hz, 2H), 4.26 (d, J 6.6 Hz, 4H), 4.10 (d, J 6.6 Hz, 4H), 2.16 (sep, J 6.6 Hz, 2H), 2.02 (sep, J 6.6 Hz, 2H), 1.07 (d, J 6.6 Hz, 12H), 0.96 (d, J 6.9 Hz, 12H). HPLC: mobile phase, A = acetonitrile (60%) + water (40%), B = acetonitrile; analysis conditions: isocratic A (77%), B (23%); flow rate; 2.1 mL min⁻¹; UV = 254 nm. ESI-MS (m/z) 930 (M⁺, 100%).

Preparation of N719

A solution of 40% wt tetra-n-butylammonium hydroxide in water (9.95 mL, 10.75 mmol) was added to a solution of [Ru(i-Bu₂dcbpy)₂(NCS)₂] (1.00 g, 1.07 mmol) in 56 mL of i-BuOH. The resulting mixture was stirred at room temperature for 30 min. The solvent was removed by rotary evaporation and the residue dissolved in water (40 mL). The pH of the solution was adjusted to 3.8 using 0.1 M nitric acid, then precipitation occurred. The suspension was left in a refrigerator at −3 °C overnight, then the solid was collected by filtration, yielding 1.26 g (98%) of the desired compound as a very dark red solid. ¹H NMR δ_H (600 MHz, MeOD) 9.56 (d, J 5.4 Hz, 2H), 9.04 (d, J 1.2 Hz, 2H), 8.88 (d, J 1.2 Hz, 2H), 8.28 (dd, J 5.7 Hz, 2H), 7.74 (d, J 6.0 Hz, 2H), 7.62 (dd, J 6.0 Hz, 2H), 3.23 (t, J 8.2 Hz, 16H), 1.67 (quin, J 8.1 Hz, 16H), 1.40 (sep, J 7.3 Hz, 16H), 1.01 (t, J 7.3 Hz, 24H). HPLC: isocratic conditions, mobile phase methanol/water 45 : 55 + 0.1% v/v of tetra-n-butylammonium hydroxide 40% solution, flow rate = 0.5 mL min⁻¹; UV = 310 nm. UV λ_max/nm (ε/10⁴ M⁻¹ cm⁻¹) 534 (1.08), 390 (1.02), 312 (3.63). ESI(−)-MS (m/z) 1187 (M − 1)⁻, 946 (M − Bu₄N)⁻, 593 (M − 2)⁻².

Experimental fabrication and characterization of DSSC test devices

Transparent conductive fluoride doped tin oxide coated glasses (Pilkington-TEC15, 4 mm, 15 Ω per square, 2 cm × 5 cm) were washed with neutral Carlo Erba Ausilab 101 cleaner, sonicated in an ultrasonic cleaning machine (CEIA USCM1G-1400) and flushed with technical ethanol, then the glasses were dried at 60 °C for 30 minutes. Both glasses were perimetrally screen printed with a lead-free glass frit paste²⁴ with low softening point.

After a drying step the photoanode, consisting in a 1.95 cm² layer of commercial titania dioxide paste (Dyesol 18NR-T), was screen printed on the front glass. The cathode was screen printed on the back glass using a commercial nanoplatinum paste (3D-Nano). Two tiny holes were previously drilled on the back glass for the final injection of the organic material.

The two glasses were sintered in a convection oven (Nabertherm N120/65HAC) up to 450–500 °C with a slope of 2 °C min⁻¹; the thickness of the TiO₂ anode was measured with a contact profilometer (KLA Tencor P-10) giving about 6–7 μm after sintering. After that, the anode and the cathode glasses were overlapped (Fig. 10) and sealed together using a custom thermo-press by melting the glass frit sealant up to 500 °C. The cell gap spacing after sealing between the photo anode and the counter electrode is about 20–30 μm.

Fig. 10 Schematic drawing of test device assembly.

After cooling, a dye solution was fluxed through the holes over 90 min at 60 °C using a syringe pump: 0.30 mM N719 solution was prepared in a dried bottle by dissolving the dye in a 1 : 1 mixture of ACN/t-BuOH adding the magnetic stir. The mixture was sonicated for 10 minutes and stirred at room temperature overnight.²⁵ After the coloring process, a ionic liquid-based electrolyte was injected with a syringe. Finally, the holes were sealed with an UV-curable resin (Fig. 11).

Fig. 11 Back side of a DSSC test cell: in it possible recognize the glass frit frame, the dye-functionalized titania, the electrolyte and the sealed holes.

Conclusions

The main purpose of this work is the optimization of the dye N719 synthesis. We moved positively towards this goal by identifying, for each step of synthesis and purification, new procedures and measures aimed at simplifying the process, preparing the synthesis for a larger scale and cutting off its environmental and economic impact. Table 4 summarizes the most important modifications introduced to the previously reported synthesis of N719.

Table 4 Summary of the optimizations introduced in the reported N719 synthesis

Parameter	Ref. 1	This work
Step 1
Solvent amount (mL g^−1 of H2dcbpy)	40	10
Conc. H2SO4 (mmol mmol^−1 of H2dcbpy)	3	0.75
Step 2
Solvent	EtOH	i-BuOH
Solvent amount (mL mmol^−1 RuCl3)	215	42
NH4SCN (mmol mmol^−1 RuCl3)	40	4
Step 2.1 reflux time (min)	210	30
Step 2.2 reflux time (min)	150	60
Step 3
Solvent	MeCN	i-BuOH
Solvent amount (mL mmol^−1 [Ru(iBu2dcbpy)2(NCS)2])	214	52
E factor	512	99

---

This new process can be entirely conducted in isobutanol as a single solvent.

The possibility to reduce the volume of the solvent and the amount of the catalyst was verified on the first synthetic step. We also clarified which other impurities were involved in the original process by characterising the trans-esterification species which form when the second step of the process is run in ethanol according to the original procedure.

We optimized reaction times for each phase of the second step, both in ethanol and isobutanol; in the latter, we also verified that even by largely reducing the excess amount of thiocyanate the reaction time remain unchanged.

We also verified that the single chromatographic purification stage performed after step two of the process cannot be eliminated; in fact, N719 dye obtained omitting this purification has a significant lower photo-generation performance in DSSC devices.

Acetonitrile used as the solvent for the third step in the original literature procedure was one of our major environmental concerns. Different alcohols were tested as replacement for acetonitrile and isobutanol was chosen. It was also possible to reduce the amount of solvent to one quarter of the initial value.

In terms of yield, the first step stands at 92%, the second reaches 60%, while the third step is semi-quantitative.

The final product was characterized in detail by HPLC, ¹H NMR, UV-Vis, DSC and TGA. All the performed analyses confirmed a comparable quality of the synthesis product with respect to a commercial N719 reference.

NMR spectroscopy and elemental analysis are mutually complementary in the characterization of the final product confirming a number of TBA⁺ and water molecules of hydration approximately equal to 2 per unit of complex.

The thermogravimetric analyses and DSC provided useful information about the thermal stability of the product and confirmed a behaviour of the synthetic product similar to the commercial one.

Finally, the efficiency and aging stability of N719 dye obtained with our improved procedure was tested in DSSC devices. The measurements showed that N719 prepared according to our procedure has similar performance than the commercial dye and comparable stability.

Acknowledgements

The authors thank F. Martina for fruitful discussions; C. Martelli and L. Armiento for experimental support; P. Ziosi for laboratory support and help. The work presented in this paper was made possible by the Italian Ministry of Economic Development (MiSE) under funded project “MOSAIC” (D. M. 07/07/2009 – B01/0740/00/X16).

Notes and references

1. T. Rawling, F. Buchholz and A. M. McDonagh, Aust. J. Chem., 2008, 61, 405.
2. B. O'Regan and M. Grätzel, Nature, 1991, 353, 737.
3. M. Grätzel, Nature, 2001, 414, 338.
4. M. Grätzel, J. Photochem. Photobiol., C, 2003, 4, 145.
5. A. Hagfeldt and M. Grätzel, Chem. Rev., 1995, 95, 49.
6. A. Hagfeldt, G. Boschloo, L. Sun, L. Kloo and H. Pettersson, Chem. Rev., 2010, 110, 6595.
7. M. K. Nazeeruddin, R. Splivallo, P. Liska, P. Comte and M. Grätzel, Chem. Commun., 2003, 1456.
8. M. Grätzel, Inorg. Chem., 2005, 44, 6841.
9. (a) M. K. Nazeeruddin, S. M. Zakeeruddin, R. Humphry-Baker, M. Jirousek, P. Liska, N. Vlachopoulos, V. Shklover, C. H. Fischer and M. Grätzel, Inorg. Chem., 1999, 38, 6298; (b) Md. K. Nazeeruddin, A. Kay, I. Rodicio, R. Humpbry-Baker, E. Mueller, P. Liska, N. Vlachopoulos and M. Grätzel, J. Am. Chem. Soc., 1993, 115, 6390.
10. Md. K. Nazeeruddin, F. De Angelis, S. Fantacci, A. Selloni, G. Viscardi, P. Liska, S. Ito, B. Takeru and M. Grätzel, J. Am. Chem. Soc., 2005, 127, 16847.
11. T. P. Brewster, W. Ding, N. D. Schley, N. Hazary, V. S. Batista and R. H. Crabtree, Inorg. Chem., 2011, 50, 11938–11946.
12. O. Kohle, S. Ruile and M. Grätzel, Inorg. Chem., 1996, 38, 4787.
13. G. Smestad, C. Bignozzi and R. Argazzi, Sol. Energy Mater. Sol. Cells, 1994, 32, 259.
14. S. M. K. Rendon, D. Mavrynsky, A. Meierjohann, A. Tiihonen, K. Miettunen, I. Asghar, J. Halme, L. Kronberg and R. Leino, Rapid Commun. Mass Spectrom., 2015, 29, 2245.
15. P. T. Anastas and J. C. Warner, Green Chemistry: Theory and Practice, Oxford University Press, New York, 1998, p. 30.
16. For a remarkable example of process solvent simplification, see: (a) D. J. Dale, P. J. Dunn, C. Golightly, M. L. Hughes, P. C. Levett, A. K. Pearce, P. M. Searle, G. Ward and A. S. Wood, Org. Process Res. Dev., 2000, 4, 17; (b) P. J. Dunn, S. Galvin and K. Hettenbach, Green Chem., 2004, 6, 43.
17. (a) R. A. Sheldon, Chem. Ind., 1992, 903; (b) R. A. Sheldon, CHEMTECH, 1994, 38R. A. Sheldon, Chem. Ind., 1997, 12; (c) R. A. Sheldon, Pure Appl. Chem., 2000, 72, 1233; (d) R. A. Sheldon, Green Chem., 2007, 9, 1273; (e) R. A. Sheldon, Chem. Commun., 2008, 3352.
18. To achieve meaningful comparisons of processes, water is generally excluded from the calculation of the E factor. See: (a) M. Lancaster, Green Chemistry: An Introductory Text, Royal Society of Chemistry, Cambridge, 2002; (b) ref. 17d.
19. It was not possible to include in this comparison the materials used for work-ups, since amounts of those materials are never reported in journal articles. Anyway, for both processes the work up most impacting the waste is of course the chromatographic separation performed after step 2. We do not expect a big difference between the two processes in this regard. Actually, we believe that the change of solvent we introduced in step 2-from EtOH to isobutanol, which eliminates transesterification products – eases a little bit this chromatographic purification.
20. T. Kitamura, K. Okada, H. Matsui and N. Tanabe, J. Sol. Energy Eng., 2010, 132(2), 21105.
21. See ESI† for details.
22. (a) F. Fabregat-Santiago, J. Bisquert, E. Palomares, L. Otero, D. Kuang, S. M. Zakeeruddin and M. Grätzel, J. Phys. Chem., 2007, 111, 6550; (b) J. Halme, P. Vahermaa, K. Miettunen and P. Lund, Adv. Energy Mater., 2010, 22, E210.
23. P. Wen, M. Xue, Y. Ishikawa, H. Itoh and Q. Feng, ACS Appl. Mater. Interfaces, 2012, 4, 1928.
24. Daunia Solar Cell SRL, US Pat., 8 927 445, 2010.
25. Y. Jo, C. Jung, J. Lim, B. H. Kim, C. Han, J. Kim, S. Kim, D. Kim and Y. Jun, Electrochim. Acta, 2012, 66, 121.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra13550a

---