Straightforward and Relatively Safe Process for the Fluoride Exchange of Trivalent and Tetravalent Group 13 and 14 Phthalocyanines.

To avoid the use of hydrofluoric acid, a series of fluorinated trivalent and tetravalent metal-containing phthalocyanines (MPcs) were synthesized using a straightforward one-step halide substitution process using cesium fluoride (CsF) as the fluoride source and by reflux in N,N-dimethylformamide for less than an hour. The resulting fluoro MPcs were characterized and compared to the parent chloro MPcs. In some cases, very little change in properties was observed between the fluoro MPcs and the chloro MPcs. In other cases, such as fluoro aluminum phthalocyanine, a blue shift in the absorbance characteristics and an increase in oxidation and reduction potential of as much as 0.22 V was observed compared to the chloro derivative. Thermo gravimetric analysis was performed on all halo-MPcs, indicating that the choice of halo substitution on the axial position can have an effect on the decomposition or sublimation temperature of the final compound. After initial establishment and characterization of the fluoro MPcs, the halide substitution reaction of difluoro silicon phthalocyanine (F2-SiPc) was further explored by scaling the reaction up to a gram scale as well as considering tetrabutylammonium fluoride (TBAF) as an additional safe fluoride source. The scaled-up reactions producing F2-SiPc using CsF and TBAF as fluoride exchange sources were successfully reproducible, resulting in reaction yields of 100 and 73%, respectively. Both processes led to pure final products but results indicate that CsF, as the fluoride exchange reagent, appears to be the superior reaction process as it has a much higher yield.

Introduction

Phthalocyanines (Pcs) are a family of aromatic macrocycles which can coordinate a variety of divalent, trivalent, or even tetravalent metals.[1,2] Metal-containing Pcs (MPcs) have been utilized in a plethora of applications ranging from pigments/dyes,[3,4] to active materials in organic electronics.[5−8] Whereas the majority of organic electronic studies utilize the divalent MPcs (M = Zn or Cu), a growing interest in trivalent and tetravalent MPcs has emerged. The use of trivalent or tetravalent metals (group 13 and group 14, respectively) leads to one or two additional bonding opportunities to moieties such as halide, hydroxy, phenoxy, and so forth, resulting in a series of derivatives with ranging properties. Those synthetic modifications allow fine-tuning of the chemical, optical, and electrophysical properties of the MPcs without changing the aromatic structure of the Pc macrocycle;[9−11] even the crystal packing can be engineered.[11] Some recent examples utilizing such approaches include chloro aluminum Pc (Cl-AlPc) and chloro gallium Pc (Cl-GaPc). These examples have been paired with C60 for the fabrication of a bilayer and even blended organic solar cells.[12−17] Cl-AlPc and Cl-GaPc have also been used as interlayers between PEDOT:PSS and a blended solution of poly(3,3″-didodecylquaterthiophene) (PQT-12) and [6,6]-phenyl-C61-butyric acid methyl ester to form bilayer bulk heterojunction solar cells.[18,19]

Our group has recently found that incorporation of fluorine groups on boron subphthalocyanine (a class of Pcs) resulted in a change in solid-state arrangement, a decrease in sublimation temperature, and a change in electrochemical properties.[9,20] Our interest has since then expanded to the substitution of the chloride of group 13 and group 14 MPcs with fluoride to enable the physical study of these materials.[16,21] Through anecdotal observations, researchers have identified that fluoro gallium Pc (F-GaPc) and fluoro aluminum Pc (F-AlPc) have unique linear stacking motifs in the solid state where the macrocycles overlap to form a pseudo polymer-like arrangement, which is not the case for the respective chlorides.[1,22−24] This unique solid-state stacking motif influences the optical/absorption and electronic properties of the base AlPc and GaPc chromophore. However, short of these two, there are very few examples of the synthesis of axially fluorinated MPcs that can be found in the literature. The synthesis of F-GaPc and F-AlPc (Scheme ) was reported by a two-step process where the chloro group III MPc was converted into the hydroxy derivative by treating with ammonium hydroxide and pyridine followed by the substitution of the hydroxy group with a fluorine group using hydrofluoric acid.[1,25]

Scheme 1

Synthesis of Fluoro and Difluoro Group 13 and Group 14 MPcs, Respectively

Hydrogen fluoride (HF) gas and hydrofluoric acid are the primary industrial sources of fluorine or fluoride for a wide range of applications, including the synthesis of organofluoro materials and fluorinated Pcs.[25−29] Although HF is readily available, with over 1 million tons produced annually, HF is extremely dangerous. Both forms of HF are highly corrosive and highly caustic to human tissue, causing both acute and chronic toxicity. HF attacks the human bone structure through the formation of calcium fluoride, causing severely corrosive damage known as skeletal fluorosis.[30] HF has an extremely low short-term exposure limit of 2 ppm according to the occupational exposure limits for the Province of Ontario (Canada) workplaces.[31] For this reason, safer fluorinating reagents/fluoride sources need to be considered for relevant chemical processes.

Because of these serious safety concerns, our group has explored the use of fluoride salts such as cesium fluoride (CsF), sodium fluoride (NaF), potassium fluoride (KF), or tetrabutylammonium fluoride (TBAF) as alternatives with minimized safety concerns.[16,21] We report herein that a simple halide substitution can be performed by heating the respective MPc chlorides within a polar aprotic organic solvent in the presence of the fluoride salt, surpassing the use of hydrofluoric acid,[16,21] yielding the desired fluorides of group 13 and 14 MPcs. The resulting MPc fluorides were then characterized beyond what is currently available in the literature. Thereafter, with a focal point of difluoro silicon phthalocyanine (F2-SiPc), the halide substitution reaction was further explored. On the basis of detailed evaluation of safety, environmental sustainability, cost, and performance, both CsF and TBAF emerged as the most desirable fluoride sources to produce MPc fluorides.

Experimental Section

Materials

Dimethylsulfoxide (DMSO, 99.9%) was purchased from ACP, CsF (>99.0%) was purchased from Tokyo Chemical Industry, TBAF (98%) was purchased from Oakwood Chemical, and dicyclohexano-18-crown-6 was (98.0%) purchased from Aldrich, whereas all other solvents were purchased from Caledon Laboratories Ltd. All chemicals were used as received unless otherwise specified. Dichlorosiliconphthalocyanine (Cl2-SiPc, 1)[3,32,33] dichlorogermaniumphthalocyanine (Cl2-GePc, 2),[5,34] chloroaluminumphthalocyanine (Cl-AlPc, 3),[16,18] and chlorogaliumphthalocyanine (Cl-GaPc, 1)[35] were synthesized according to the literature.

Synthesis of Difluoro Silicon Phthalocyanine (F2-SiPc, 2)[10]

Method (a): In a 50 mL three-neck round-bottom flask with a reflux condenser and nitrogen inlet, Cl2-SiPc (0.1 g, 0.163 mmol) and CsF (0.06 g, 0.395 mmol) were dissolved in DMF (1 mL). The mixture was stirred and heated at 150 °C under nitrogen for 30 min. The crude product was allowed to cool to 130 °C and was precipitated into 150 mL of isopropanol. The final product was gravity-filtered, resulting in a fine dark indigo powder. Yield 0.068 g (71%). UV–vis (DMSO): λmax 688 nm; HRMS [M+]: calcd, 578.1231; found, 578.1235; method (b.1). CsF procedure: in a 250 mL three-neck round-bottom flask with a reflux condenser, Cl2-SiPc (2.052 g, 3.355 mmol) and CsF (1.285 g, 8.460 mmol) were dissolved in DMF (20 mL). The mixture was stirred and heated at 150 °C for 60 min. The crude product was allowed to cool to 130 °C and was precipitated into 300 mL of isopropanol. The final product was gravity-filtered, washed with methanol and ethanol, and resulted in a fine dark indigo powder. Yield 1.942 g (100%).

Method (b.2). (b) TBAF procedure: similarly, in a 250 mL three-neck round-bottom flask with a reflux condenser, Cl2-SiPc (2.067 g, 3.380 mmol) and tetrabutylammonium fluoride (2.222 g, 8.500 mmol) were dissolved in DMSO (30 mL). The mixture was stirred and heated at 150 °C for 60 min. The crude product was allowed to cool to 130 °C and was precipitated into 300 mL of isopropanol. The final product was vacuum-filtered, washed with water and acetone, and resulted in a fine dark indigo powder. Yield 1.419 g (73%).

Synthesis of Fluoro Aluminum Phthalocyanine (F-AlPc, 5)[16]

The synthesis of F-APc was performed under similar conditions to that of F2-SiPc (method a), except for using Cl-AlPc as the starting material. Yield 0.049 g (49%). UV–vis (DMSO): λmax 671 nm; HRMS [M+]: calcd, 558.1324; found, 554.1341.

Synthesis of Fluoro Gallium Phthalocyanine (F-GaPc, 6)

The synthesis of F-GaPc was performed under similar conditions to that of F2-SiPc (method a), except for using Cl-GaPc as the starting material, as well as the addition of a crown ether. For example, Cl-GaPc (0.1 g, 0.16 mmol), CsF (0.03 g, 0.2 mmol), and dicyclohexano-18-crown-6 (0.074 g, 0.2 mmol) were dissolved in DMF (1 mL) and reacted at 150 °C under nitrogen for 30 min. The product was purified in a similar fashion to that of F2-SiPc, resulting in a yield of 0.042 g (42%). UV–vis (DMSO): λmax 676 nm; HRMS [M+]: calcd, 600.0747; found, 600.0738.

Synthesis of Difluoro Germanium Phthalocyanine (F2-GePc, 3)

The synthesis of F2-GePc was performed under similar conditions to that of F-GaPc, except for using Cl2-GePc as the starting material and using 18-crown-6 instead of dicyclohexano-18-crown-6. Yield 0.914 g (96%). UV–vis (DMSO): λmax 680 nm; HRMS [M+]: calcd, 624.0678; found, 624.0690.

Characterization

Ultraviolet–visible (UV–vis) spectroscopy was performed using PerkinElmer Lambda 1050 with a 10 mm quartz cuvette. Cyclic voltammetry was performed using a three-electrode cell assembly at room temperature in a 0.1 M dichloromethane and tetrabutylammonium perchlorate (TBAP) electrolyte solution. The working electrode was a glassy carbon disk electrode, the counter electrode was a polished platinum wire, and the reference electrode was Ag/AgCl. An internal standard of decamethyl ferrocene (bis(pentamethylcyclopentadienyl)iron, Sigma-Aldrich) as well as a scan rate of 100 mV/s was used for all measurements. The samples were bubbled using nitrogen until no dissolved oxygen was present (30–60 min prior to each run). Photoemission measurements were conducted on a PHI 5500 Multitechnique system using a monochromated Al Kα photon source (hν = 1486.7 eV) for X-ray photoelectron spectroscopy (XPS) and a non-monochromated He Iα photon source (hν = 21.22 eV) for ultraviolet photoelectron spectroscopy (UPS). Work function and valence-band measurements were carried out using UPS with the sample tilted to a take-off angle of 89° and under an applied bias of −15 V. The analysis chamber base pressure was ≈10–10 torr.

Results and Discussion

Initial Process Development

The synthesis of fluoro aluminum phthalocyanine (F-AlPc), fluoro gallium phthalocyanine (F-GaPc), difluoro silicon phthalocyanine (F2-SiPc, method (a)), and difluoro germanium phthalocyanine (F2-GePc) was achieved by substitution of the chloride atom within the respective precursor with a fluoride atom by heating in a polar aprotic solvent in the presence of a fluoride salt (Scheme ); a method to avoid the use of hydrofluoric acid.[16,21] In order to consider fluoride sources and appropriate aprotic solvents, a literature survey on the solubility of various fluoride salts in multiple aprotic solvents was conducted to ensure maximum concentration and, therefore, reactivity of the fluorides in solution. Many publications detail the solubilities of halide salts in various solvents such as formamide, N-methylformamide, acetamide, N-methylacetamide, acetonitrile, DMF, N,N-dimethylacetamide, and DMSO; the data are presented in Tables S1.1–S1.5.[2−4,6,25,26,35−39] After such a survey, DMF was identified as the solvent with the highest fluoride salt solubility varying from 1 mg/kg for NaF to 96 mg/kg for CsF.

We had previously developed a simple thermodynamic model to examine and predict the spontaneity of the phenoxylation reaction of multiple chloro Pc derivatives.[40] Here, we developed a similar thermodynamic model for the halide exchange reaction for the chloro/dichloro MPcs of interest, using different cesium halide salts (i.e., CsF, CsBr, and CsI, eq ). The model enables the consideration of a thermodynamic driving force for the halide exchange alongside solubility considerations. The premise of the model is the assumption that enthalpy of reaction (ΔHR) is solely dependent on the bond dissociation/formation energies of the bonds involved in the reaction with a negligible change in entropy. At constant entropy, the sign and magnitude of ΔHR can indicate at least qualitatively the spontaneity of the reaction (large negative values being spontaneous). ΔHR were calculated using bond dissociation (formation) energies.[41] ΔHA–B is the bond dissociation (formation) energy between atom A and atom B.

On the basis of the model, thermodynamically the exchange of chloride with fluoride is highly favorable. Fluorination of any of the chloro derivatives using CsF was then expected to be spontaneous because of the largely negative calculated values of ΔHR (Figure ). Experimental results support this conclusion. The synthesis of F-AlPc, F-GaPc, F2-SiPc, and F2-GePc was easily accomplished in a short time by substitution of chloride with fluoride by heating the respective chloride precursors in DMF in the presence of CsF (Scheme ).[16,21] Whereas the yields of producing F-AlPc, F2-SiPc, and F2-GePc using this chemistry were high (≈70–80%), the yield of producing F-GaPc under the same conditions was very small (<5%). Because ΔHR of the fluoride exchange of Cl-GaPc is negative and large in magnitude (although much lower than that of Cl-AlPc) the reaction was still expected to be thermodynamically favorable and therefore must have been kinetically limited. One of the methods employed to overcome this kinetic limitation was to change the solvent properties to encourage the dissolution of CsCl and increase the solubility of CsF, which would increase the likelihood of reaction and reduce the probability for the reverse reaction. The addition of crown ethers has been known to favorably increase the fluoride salt’s solubility compared to the chloride salt and was therefore explored for the efficient synthesis of F-GaPc.[42,43] We determined that the addition of 1 equiv (to the CsF) of dicyclohexano-18-crown-6, under consistent reaction conditions, resulted in the complete synthesis of F-GaPc at high yield (>70%).

Figure 1

Estimated enthalpy of halogen exchange reactions (ΔHR) for different chloro/dichloro MPcs.

ΔHR was also calculated for the fluoride exchange reaction using HF, the reactant we are aiming to avoid. The thermodynamic force for HF is clearly lower than that of the reaction using CsF. Although the reactions of Cl-GaPc and Cl2-GePc was seen as thermodynamically unfavorable, the small magnitude of ΔHR (<20 kJ/mol) is well within the uncertainties of many of the used bond dissociation (formation) energies. Another interesting observation resulting from the model is whereas the bromination of the chloro derivatives is expected to be unfavorable as bromides are stronger leaving groups than chlorides, the bromination of Cl2-GePc seems slightly favorable. Again, the magnitude of ΔHR is small, causing uncertainty over the thermodynamic favorability of this reaction.

These synthetic and model results overall indicate a straightforward and versatile method for obtaining analytically pure fluoro and difluoro group III and group IV MPcs at high yields, without the need to use a hazardous chemical such as hydrofluoric acid. CAUTION: Should one consider variations in this process, be cautious as the use of acids in this process, even during workup and isolation of product, increases the risk of producing HF as a by-/side-product. This can/could result in releasing HF to the surrounding environment and increase the chance of human exposure without proper waste management or other process considerations.[44,45]

Optical Characterization

After synthesis and purification, UV–vis spectroscopy was performed on F-AlPc, F-GaPc, F2-SiPc, and F2-GePc, along with their respective chloride starting materials as a point of comparison. The characterization of the AlPc derivatives has been previously reported and therefore was reproduced for ease of comparison.[16,21] All compounds were analyzed in DMSO, chloroform, and toluene because of ranging solubilities and for comparison. The optical band gap and max absorbance (λmax) were determined from the absorbance spectra and are compiled in Table S2. Figure shows the characteristic normalized absorption spectra of the axial fluorinated versus chlorinated MPcs (M = Al, Ga, Si, and Ge) in DMSO solution. The maximum absorbance of Cl2-SiPc is almost identical to that of F2-SiPc (Table S2, Figure b). Unlike for F2-SiPc, the peak absorption of F-AlPc is slightly blue shifted compared to the chlorinated material (Figure a);[16] this small blue shift in absorbance corresponds to an increase in optical bang gap from Eg,opt = 1.78 eV to Eg,opt = 1.81 eV, for Cl-AlPc to F-AlPc, respectively (Table S2).[16] As a point of comparison, these results are consistent with the literature values for halo boron subphthalocyanines (BsubPcs), where little change was observed in the maximum absorbance of Cl-BsubPc and F-BsubPc, at 565 and 562 nm respectively.[9] No significant change in the maximum absorbance peaks was found for the Ga- and Ge-based MPcs comparing the fluoride to chloride derivatives (Figure c,d). Figure illustrates the effect of solvent on the absorbance of F2-SiPc and F-AlPc, each dissolved in DMSO, chloroform, and toluene. Of the three solvents used, DMSO was found to be the most effective in consistently solubilizing each material, whereas some compounds were slightly soluble in chloroform and most were only sparingly soluble in toluene. The broad absorption spectrum is indicative of the insolubility and formation of a dispersion in toluene. When examining Figure a, it is apparent that a small change in the maximum absorbance for F2-SiPc is observed between that of DMSO, with λmax = 699 nm, and chloroform, with λmax = 694 nm. Similarly, F-AlPc experienced a slight red shift with chloroform, with λmax = 688 nm, compared to DMSO with λmax = 671 nm. These results illustrate that the fluoride exchange of trivalent and tetravalent MPcs translates to only a small red shift of up to 9 nm. Second, the use of various solvents for the analysis on the absorbance can change the solubility of the product, resulting in a fluctuation in λmax of up to 19 nm.

Figure 2

Normalized absorbance spectra of (A) F-AlPc and Cl-AlPc, (B) F2-SiPc and Cl2-SiPc, (C) F-GaPc and Cl-GaPc, and (D) F2-GePc and Cl2-GePc in DMSO.

Figure 3

Normalized absorbance spectra of (a) F2-SiPc and (b) F-AlPc in DMSO, chloroform, and toluene.

Electrochemical Characterization

The fluoro/difluoro MPcs and, as a comparison, the corresponding chloro-/dichloro MPcs were then characterized by cyclic voltammetry. An overview of the electrochemical results can be found in Table S3. Because of limited solubility, in general electrochemical signals were weak but for each result a peak oxidation or reduction potential could be identified. However, the weak signals prevented the identification of half-wave oxidation or reduction potentials. In all cases, the oxidation peak (EOX,peak) is observed at a reduced potential of up to ≈0.2 V for the fluoride MPc derivative compared to the respective chloride MPc derivative. For example, F-AlPc exhibited an EOX,peak = 1.18 V and ERed,peak = −0.81 V where the Cl-AlPc exhibited an EOX,peak = 1.40 V and ERed,peak = −0.71 V (Table S3, Figure A,B). Some compounds experienced less significant changes in electrochemical characteristics, for example F2-SiPc exhibited an EOX,peak = 1.23 V and ERed,peak = −0.63 V, where the Cl2-SiPc exhibited an EOX,peak = 1.27 V and ERed,peak = −0.74 V (Table S3, Figure C,D). We previously reported a similar decrease in oxidation potential of 0.05 V between chloro boron subphthalocyanine (Cl-BsubPc) and fluoro boron subphthalocyanine (F-BsubPc).[9]

Figure 4

Characteristic electrochemical spectra for the (A) oxidative scanning and (B) reductive scanning of Cl-AlPc and F-AlPc as well as (C) oxidative scanning and (D) reductive scanning of Cl2-SiPc and F2-SiPc. Electrochemistry was performed at a 100 mV/s scan rate (three cycles) using a reference electrode of Ag/AgCl and internal standard of decamethyl ferrocene.

UPS Characterization

To verify the abovementioned trends, UPS was employed to firmly determine the highest occupied molecular orbital (HOMO) energy levels of the chloro and fluoro MPcs as thin solid films. The pure compounds were deposited by sublimation on freshly cleaved highly ordered pyrolytic graphite, and the results were aligned to its vacuum level work function at 4.45 eV. In all cases, between 12 and 13 nm of Pc was deposited on the sample to avoid the potential screening of photohole generated in the organic layer by the conductive substrate in ultrathin layer and to give a strong signal to noise ratio.[46] The densities of states corresponding to the EHOMO are represented by the peaks at the lowest binding energy of the valence band spectra for each compound (Figure ) and the corresponding ionization energies or EHOMO and the work functions determined by UPS (ΦUPS) and XPS (ΦXPS) can all be found in Table S3. The peaks in the UPS spectra, between 6 and 12 eV, correspond to the binding energies for the orbitals of the isoindoline groups, which are common for all MPcs.[46−48] The valence band structure is defined by the d-orbitals from the central metal chelated by four nitrogen groups and the corresponding binding energies are located between 0 and 6 eV.[46−48] In general, very little change in the valence structure is observed when substituting the chelating metal.[46−48] When studying the values, it becomes apparent that in some cases such as AlPc[16] and GePc, the substitution of fluoride for chloride results in a significant change in ΦUPS and ultimately EHOMO, whereas in the other cases, such as SiPc and GaPc, very little change was observed. For example, an increase of EHOMO of as much as ≈1.0 eV was observed between Cl-AlPc (EHOMO = −5.7 eV) and F-AlPc (EHOMO = −4.7 eV)[16] and as much as ≈1.5 eV was observed between Cl2-GePc (EHOMO = −5.8 eV) and F2-GePc (EHOMO = −4.3 eV), respectively. In comparison, however, very little difference was observed between Cl2-SiPc (EHOMO = −5.7 eV) and F2-SiPc (EHOMO = −5.6 eV) or between Cl-GaPc (EHOMO = −5.7 eV) and F-GaPc (EHOMO = −5.6 eV). Finally, it is important to note that ΦUPS ≈ ΦXPS, confirming the trend in values, with the exception of F-AlPc.

Figure 5

UPS spectra of both the chloro (blue line) and the fluoro (red line) (A) aluminum Pcs, (B) silicon Pcs, (C) gallium Pcs, and (D) germanium Pcs. The energy levels are aligned to its vacuum level work function, of highly ordered pyrolytic graphite, at 4.45 eV. As a comparison, Cl-AlPc and F-AlPc were taken from Lessard et al.[16]

Thermo Gravimetric Analysis

Thermo gravimetric analysis (TGA) was performed on both the chloro and the fluoro MPcs (Al, Ge, Ga, Si). The corresponding traces for the % mass loss and derivative of mass loss with respect to temperature can be found in Figure . What is interesting to note is that the mass loss profiles between the chloro and fluoro derivatives are uniformly different. For example, F-AlPc appears to exhibit no mass loss until roughly 625 °C, whereas Cl-AlPc begins to experience significant mass loss at 550 °C (Figure A).[16] These results indicate that the substitution of the axial chloride group for a fluoride group can have an effect on the decomposition temperature and possibly the sublimation temperature of these compounds.

Figure 6

Characteristic TGA spectra, where above is the % mass loss relative to temperature (°C) and below corresponds to the scaled derivative of the mass loss relative to temperature (°C) for the (A) Cl-AlPc and F-AlPc, (B) Cl2-SiPc and F2-SiPc, (C) Cl-GaPc and F-GaPc (D) Cl2-GePc and F2-GePc. As a comparison, Cl-AlPc and F-AlPc were taken from Lessard et al.[16]

Additional Process Improvement

After establishing the synthesis and characterization of this variety of fluoro MPcs, we were interested to further refine the process or consider alternative factors aside from DMF and CsF. We decided to further explore the process chemistry with the focal point of F2-SiPc, not just relevant to the dye and pigment industry but also because SiPcs have recently shown functionality in organic electronic devices such as OPVs[37] and OLEDs,[34] which are emerging technologies. Further justifying this aspect of the study, through kinetic Monte Carlo simulations based on experimental X-ray structures, Gali et al. identified that F2-SiPc is expected to exhibit ambipolar semiconducting behavior, with hole and electron mobilities lying in the range 0.1–1 cm2 V–1 s–1.[49]

For initial process optimization and improvement consideration, we considered alternative fluoride sources and solvents while taking into account safety of the reagents, potential environmental damage caused by the waste/by-products, and the cost of the reagents. Because of their abundance and low costs, fluorinated salts are good reactants for this synthetic process;[50] although they are still derived from HF, safety hazards are reduced as the acid is only handled in the large-scale facilities during salt production.

For this study extension, the following extra fluoride salts were considered along with CsF to execute the fluoride exchange of Cl2-SiPc: KF, TBAF, and (diethylamino) difluorosulfonium tetrafluoroborate (XtalFluor-E). As mentioned above, CsF, as compared to KF or NaF, is the most soluble in organic polar aprotic solvents (Tables S1.1–S1.5), providing better reaction conditions. Cesium has the lowest electronegativity of all nonradioactive elements and fluorine the highest, thus qualifying the noncovalent compound as an ideal fluorinating reagent for this reaction. Furthermore, fluoride reactivity in aprotic solvents is as follows: CsF > KF ≫ NaF. KF, with similar benefits as CsF, has also been explored as a promising fluorinating reagent.[51] Anhydrous TBAF has demonstrated nucleophilic aromatic substitution of chlorine elements in DMSO.[51] TBAF is a more environmentally sustainable fluorinating reagent as compared to CsF and KF[51] and was therefore also considered as a fluoride source. Finally, XtalFluor-E is a commercialized crystalline fluorinated salt that was designed to completely mitigate the production of HF, providing for safer reaction conditions comparatively. XtalFluor-E has successfully been used as a fluorinating reagent for alcohols and carbonyls, such as aldehydes and ketones.[52]

Three polar aprotic solvents were explored; DMF, N-methylpyrrolidine (NMP), and DMSO. Despite polar aprotic solvents being the most common for this type of reaction, research shows that the H-bonding network of a protic solvent can be manipulated through the use of additives to further optimize the production of halogenated MPcs.[50,53] Through hydrogen bonding, protic solvents, including water, aqueous solutions, and alcohol, bind with the anionic nucleophiles.[50] Solvent slurries (50% water) were therefore also explored. The use of water reduces the amount of solvent required and thus reduces the amount of environmental waste produced by this reaction. In addition, when the chloride atoms are released from Cl2-SiPc, it is possible that the presence of water will draw them out and act as a proactive purifying system. Lee et al. demonstrate successful nucleophilic fluoride reactions using an ionic liquid (IL).[50] ILs, such as 1-butyl-3-methylimidazolium tetrafluoroborate ([bmim][BF4]), are highly polarized ionic compounds, and can in fact help the dissolution of salts in organic solvents.[50] [bmim][BF4] is also known as a reusable/recyclable solvent, and thus was also explored for the synthesis of F2-SiPc.

First Screening of Conversion to F2-SiPc

We performed a series of fluoride exchange reactions under similar conditions, as identified in the Experimental Section, while substituting the solvents or the fluoride sources. The process conditions as well as the experimental results are compiled in Table . The success of the reaction was scanned by the presence of F2-SiPc in associated mass spectra. We found that CsF, KF, TBAF, and XtalFluor-E all successfully produce F2-SiPc in DMF and that CsF and TBAF can be used in DMSO. We also found that the use of [bmim][BF4] as a solvent also resulted in the formation of F2-SiPc. Of these, reactions 1, 2, 3, 5, and 9 showed no peaks for Cl2-SiPc, indicating full conversion (Table ). Reactions 8, 10, and 11, which include the use of a water slurry as a solvent, demonstrate no presence of F2-SiPc, but displayed the presence of F-SiPc+ fragment peaks, indicating potential incomplete conversion to F2-SiPc and the presence of the intermediate (F)(Cl)-SiPc.

Table 1

Fluoride Exchange Reactions for the Formation of F2-SiPc

sample	solvent	fluorine source	F-SiPca	F2-SiPcb	Xc
1	DMF	CsF	yes	yes	1.0
2	DMSO	CsF	yes	yes	1.0
3	[bmim][BF4]	CsF	yes	yes	1.0
4	DMF	KF	yes	yes	0.9
5	DMF	TBAF	yes	yes	1.0
6	DMF	XtalFluor-E	yes	yes	0.8
7	DMF/H2O	CsF	yes	yes	0.9
8	DMF/H2O/Aliquat	CsF	yes	no	0.7d
9	DMSO	TBAF	yes	yes	1.0
10	DMSO/H2O	CsF	yes	no	0.7d
11	DMSO/H2O	TBAF	yes	no	0.5d

Presence of F-SiPc fragment (m/z: 559.13) by mass spectrometry.

Presence of F2-SiPc fragment (m/z: 578.12) by mass spectrometry.

Conversion, X, defined as molar fraction of mass spectrometry area of F2-SiPc (m/z: 578.12) relative to Cl2-SiPc (m/z: 610.06).

Reactions where no F2-SiPc peak was identified by mass spectrometry, X = molar fraction of area of F-SiPc (m/z: 559.13) fragment relative to Cl2-SiPc (m/z: 610.06).

Toxicology and Environmental Considerations

NMP, DMF, and DMSO have very similar solvent properties, all being polar aprotic solvents with high boiling points and similar vapor pressures. From a toxicological point of view, DMSO is considered much safer than NMP and DMF. Unlike NMP, DMSO does not qualify as a reproductive toxin, nor is it under Proposition 65 in the United States as a developmental toxin.[54] DMSO also presents having much higher lethal doses than NMP, for example, having a dermal LD-50 of 40 g/kg rat as compared to NMP having a dermal LD-50 of 8 g/kg rabbit.[54] Having higher oral, dermal, and inhalation exposure lethal doses further justifies the use of DMSO. Unlike DMSO, NMP qualifies under the SARA Title III §313 Community Right-to-Know Program in Georgia, which requires reports of both routine and accidental chemical releases to be submitted to the Environmental Protection Agency and the state Emergency Response System.[54] Furthermore, NMP was recently investigated by the Government of Canada under the Chemicals Management Plan.[54] The government proposed that NMP is not harmful to the environment at the current levels of exposure; however, at higher levels of exposure, it is concerned about its environmental spread and developmental toxicity.[54] These investigations and levels of concern demonstrate the negative impact that these solvents can have on the environment. Using an environmentally friendly solvent is of high interest. As outlined, from a toxicological and ecological point of view, DMSO is the most suitable solvent available. It has a health hazard of 0, does not qualify as a reproductive toxin, and is not under investigation as an environmental hazard. On the basis of this research, DMSO is the least harmful solvent from among the group of polar aprotic solvents considered in this experiment.

Although the health hazards associated with [bmim][BF4] are more significant than those of DMSO, it can be considered a green solvent because of its ease of recyclability[55] and negligible evaporation because of very low vapor pressure of <0.000094 mmHg.[56] NMP, DMF, and DMSO qualify as volatile organic compounds and therefore if properly recycled the use of [bmim][BF4] could be beneficial. Certain ILs have been reused up to four times without losing their purity.[55] However, it is important to note that the precipitation of F2-SiPc from [bmim][BF4] required the need for dichloromethane and isopropanol, which do not have non-negligible environmental effects.[57,58]

The four fluoride sources, CsF, KF, TBAF, and XtalFluor-E, produced F2-SiPC peaks when reacted in DMF with Cl2–SiPc. Both CsF and TBAF underwent complete conversion, as compared to KF and XtalFluor-E, which both demonstrated the presence of Cl2-SiPc. All aforementioned reactions followed the same experimental design process, thus each having the same level of design simplicity. CsF and TBAF are slightly advantageous over KF and XtalFluor-E because of their display of complete conversion.

All four fluorinating reagents have negligible flammability and physical hazard. CsF, KF, and XtalFluor-E, under the Workplace Hazardous Materials Information System (WHMIS) classification, are D1B, indicating that they are toxic materials causing immediate and serious toxic effect through ingestion, skin absorption, and inhalation.[59−61] Furthermore, CsF falls under the WHMIS classifications D2A and D2B, indicating that it is a very toxic material causing other side effects including being a reproductive hazard, moderate skin irritant, and severe eye irritant.[60] TBAF is much less dangerous, categorized under the WHMIS classification E, indicating that it is a corrosive material.[62] All four fluorinated salts have similar storage conditions with the exception of XtalFluor-E, which requires refrigeration.[59−62] XtalFluor-E is a commercially developed fluorinating reagent, compatible with borosilicate glassware, and designed not to produce HF.[53] As compared to diethylaminosulfur trifluoride (DAST) and DeoxoFluor, previously designed fluorinating reagents, XtalFluor salts have better handling properties, a significantly reduced reaction with water, and a more stable decomposition temperature of 215 °C.[63] However, despite XtalFluor-E being popular because of mitigating the production of HF, under acidic conditions, this salt produces tetrafluoroboric acid, HBF4.[63] which is still a hazardous and highly corrosive material.[64,65] Furthermore, XtalFluor-E is designed to only be exposed to the atmosphere for short periods of time because of safety risks.[61] Taking into consideration the compounds’ individual safety risks and the production risk of HF and HBF4, TBAF appears to be the safest available compound for the production of F2-SiPc.

Very little ecological information is available for the fluoride sources. Fluorides exist in the environment most commonly as HF or very small particles.[66] These compounds typically remain air bound, settle to the ground, or are washed out by rain and incorporated into the hydrological cycle.[66] Fluoride typically accumulates in plants as well as animals that consume those plants.[66] It is important to control and mitigate the exposure of fluorine to the environment. All four fluorinating reagents pose similar risks to the environment.

We performed a preliminary cost analysis for both the solvent and the fluoride source. For a comprehensive analysis to be performed several factors that affect the cost of purchasing a reagent would be needed, such as (but not limited to) transportation costs, storage costs, process costs, taxes, intellectual property and licensing costs, and costs associated with safety and spill containment. Therefore, for our analysis we averaged the purchasing cost of four different chemical suppliers. We found that DMSO is roughly 2 and 11% more expensive than DMF and NMP, respectively; [bmim][BF4] on the other hand is roughly 12 000% more expensive than DMSO. Unless the [bmim][BF4] can be effectively recycled, we find it hard to believe that this can be a financially acceptable alternative. For the fluorinating reagents CsF and KF were both found to be roughly 60% more expensive than TBAF, whereas XtalFluor-E is roughly 1200% more expensive than TBAF. The elevated cost of XtalFluor-E is likely unacceptable unless the production of HF using TBAF is a significant problem, and a treatment of the waste stream is not possible.

Final Process Considerations and Opinions

Taking into account the experimentally verified conversion to F2-SiPc, the environmental and health considerations, and financial analysis, we proposed DMSO and TBAF as the optimized processes reagent and solvent for the conversion of Cl2-SiPc and F2-SiPc.

Although DMSO is not the cheapest available solvent, its costs are similar to both NMP and DMF and are predicted to decline as DMSO becomes more popular because of its nontoxic and biodegradable properties.[67] Furthermore, there are companies that offer their service to recover, refine, and convert DMSO into a reusable solvent for further production processes (Toray Fine Chemicals Co., Ltd.). As a reusable solvent, DMSO contributes to the reduction of chemical waste and environmental preservation. The global consumption of DMSO is expected to increase between 2016 and 2026.[67] DMSO is commonly used in many industries including agrochemicals, electronics, fine chemicals, coatings, and cleaning supplies.[67] Furthermore, the growth of the DMSO market will be further advanced by recent demand for it in pharmaceutical applications.[67] DMSO is used worldwide, with North America being the largest consumer.[67] These factors qualify DMSO as a readily available solvent, with relative certainty that the market will continue to grow and be supplied.

From an experimental standpoint, TBAF enabled complete conversion to F2-SiPc when reacted with Cl2-SiPc, unlike XtalFluor-E and KF. Furthermore, TBAF is the most cost-effective fluorinating reagent. From a toxicological and ecological point of view, TBAF is the most suitable fluorinating reagent available. TBAF is classified as an organometallic compound that is commonly used in research and industrial chemistry. The global TBAF market has grown significantly since 2016 and production costs are predicted to decrease in the future because of the application of alternate energy sources.[68]

With the availability of upstream raw materials and downstream demand for both DMSO and TBAF, this reaction design is feasible with a financially stable future. The availability of these products is also supportive of scaling up the process to an industrial level. In conclusion, the recommended process design for the production of F2-SiPc involves reacting Cl2-SiPc with TBAF in DMSO. This reaction optimizes safety factors and reduces negative environmental impact, while being a financially sustainable reaction with feasible promise to be maintained and scaled to an industrial level.

Conclusions

A series of fluorinated trivalent and tetravalent MPcs were synthesized using a straightforward one-step halide substitution reaction with CsF as a fluoride source. F-AlPc, F2-SiPc, and F2-GePc were synthesized by reflux in DMF for less than an hour, whereas the synthesis of F-GaPc required the addition of a crown ether to increase the solubility of CsF and reduce the solubility of CsCl. In all four cases, high purity and high yields were obtained with very simple purification techniques.

The absorbance properties of both the chloro and fluoro derivatives in a variety of solvents were determined and compared. In some cases, no changes were observed and in others, such as F-AlPc, a significant blue shift in the absorbance was observed compared to the chloro derivative. Electrochemical characterization of the chloro and fluoro derivatives was also performed, observing an increase in potential between 0.05 and 0.22 V for the fluoro derivatives compared to the chloro derivatives. The most significant change in oxidation potential was observed between F-AlPc and Cl-AlPc. UPS and XPS both identify, in some cases such as the change from Cl-AlPc to F-AlPc or Cl2-GePc to F2-GePc, a significant change in work function and EHOMO. However, very little change in work function or EHOMO was observed when comparing Cl-GaPc to F-GaPc or from Cl2-SiPc to F2-SiPc. TGA was performed on all MPcs. It was observed that the mass loss profile between the chloride and fluoride derivatives are different, indicating that the choice of halide substitution on the axial position changes the decomposition and sublimation temperature.

As a final consideration, we have concluded that the combination of TBAF and DMSO is the most optimized fluoride exchange reaction based on several factors including environmental impact, health and hazard on exposure, and economics. However, it must be noted that CsF as a fluoride source did result in a higher mass yield. This was supported by the successful results achieved from the fluoride exchange of Cl2-SiPc to F2-SiPc using CsF and TBAF in DMF and DMSO solvents, respectively.