Munich, 19 July 2024 Our Ref.: SM 6938-01WO SOE/TWS Applicant: Symrise AG Serial Number: New patent application Symrise AG Mühlenfeldstraße 1, 37603 Holzminden, Germany Process for producing 4-hydroxyacetophenone The present invention lies in the field of cosmetics and, more specifically, concerns a process for the production of 4-hydroxyacetophenone (4-HAP). 4-hydroxyacetophenone (4-HAP) is a well-known large scale organic synthesis intermediate used in many life sciences syntheses including one of the industrial processes for paracetamol. A process for producing crude 4-HAP is known from EP 0 167 286 A1. In this process, phenol is acetylated to 4-HAP by contacting phenol with acetic acid or acetic anhydride. When acetic acid is the acetylating agent, per mole of phenol about 0.9 to 1.4 moles of acetic is used in the presence of about 20 to 50 moles of hydrogen fluoride, at a temperature of reaction of about 40 to 90°C, for a reaction period of about 10 to 120 minutes. When acetic anhydride is the acetylating agent, per mole of phenol, about 0.9 to 2.0 moles of acetic anhydride is used, in the presence of about 8 to 60 moles of hydrogen fluoride (HF), at a temperature of reaction of about 30 to 95°C for a reaction period of at least about 10 minutes. The process is said to result in a phenol conversion of at least about 80% and a reaction selectivity to crude 4-HAP of at least about 70%. As a drawback of the known process, the applicant observed that 90-94 % of the large HF excess can be recycled by distillation. However, 1.2 to 2 equivalents of HF remain complexed in the crude product and cannot be recycled by means of distillation. Even at p < 3 mbar, the required temperature leads to decomposition/polymerization of the crude product. This prevents the full recovery of HF, and requires its neutralization by about 2 equivalents potassium hydroxide (KOH), which in turn, has a significant cost impact. Furthermore, recycling of the resulting mixture of potassium acetate (KOAc) and potassium fluoride (KF) is difficult and elaborate. Moreover, to
achieve the quality and the olfactory specifications of cosmetic grade 4-HAP, crude 4-HAP has to undergo further purification. In addition, in an attempt to reduce risk, the EPA stipulates that HF is tried to be replaced by safer alternatives. On the other hand, 4-HAP can be produced batch wise from phenyl acetate and methanesulfonic acid. However, the known process has low productivity. The object to be solved by the present invention is the provision of a process for producing 4- HAP that overcomes at least some disadvantages of the known processes. In particular, it was the aim to circumvent the cost impact associated with the requirement of neutralization of HF using KOH and to obtain a more economic process for the production of 4-HAP, also in large scale. This object is achieved by a process as set forth in appended claim 1, among others. The dependent claims define preferred embodiments of the invention. According to the present invention, the process for the production of 4-HAP is based on Fries rearrangement of phenyl acetate in methanesulfonic acid (MSA). It comprises the steps of: a) providing a phenyl acetate source and a methanesulfonic acid (MSA) source; b) feeding the phenyl acetate source and the MSA source into a reactor; and c) allowing the phenyl acetate and the MSA (herein also referred to as “reaction mixture”) to react under conditions suitable for forming 4-HAP by Fries rearrangement. The formation of 4-HAP by Fries rearrangement follows reaction scheme (I):
A number of advantages are associated with the Fries rearrangement of phenyl acetate in MSA. In particular, the Fries rearrangement does not involve the use of HF, and does not require to remove/recycle a mix of KOAc/KF as compared to the known process. After feeding the phenyl acetate source and the MSA source into the reactor, water may be added.
Moreover, in accordance with the invention, step c) is carried out in a flow reactor. Thereby, productivity of the reaction can be varied (increased or decreased) as needed without affecting selectivity. When large amounts of product is needed, it is still possible to keep equipment (e.g. reactor, heat exchanger, mixing chambers, etc.) small. This is a great advantage over known batch processes which are not flexible and require handling of large volumes. Carrying out the reaction in step c) continuously simultaneously requires the feed to enter the reactor continuously and the reaction mixture to leave the reactor continuously. Moreover, the downstream processing may be carried out continuously as well, to keep the corresponding apparatuses small. Before feeding to the reactor, the phenyl acetate source and MSA source may be preheated separately, and the preheated streams combined in a mixer. At the end of the reaction time (residence time), the resulting reaction product may be cooled down and/or may be blended with water. This leads to quenching of the reaction and potential side reactions. The conversion of phenyl acetate is fast. Further, the risk that side-products are formed increases with progressing time. Therefore, it is preferred that the residence time in step c) ranges from 3 s to 5 min, preferably 15 s to 3 min, more preferably 30 s to 90 s. The comparably short residence time allows to reach a high productivity, yet keep equipment small. Advantageously, the isomeric ratio of 4-HAP to 2-HAP can be controlled via the reaction temperature and/or the MSA content fed into the reactor. According to a preferred embodiment of the invention, step c) is carried out at a temperature of at least 100°C, more preferably to at least 110°C, most preferably to at least 115°C. Further, step c) may be carried out at a temperature of at most 140°C, more preferably to at most 130°C, most preferably to at most 125°C. In particular, step c) may be carried out at a temperature ranging from 115° to 125°C. As is well-known to the skilled person, the temperature and residence time are dependent on each other. Lower temperatures tend to have a higher residence time, while higher temperatures tend to have a smaller residence time, until conversion is complete. The above temperature ranges are based on the finding that an improved isomeric ratio of 4-HAP to 2-hydroxyacetophenone (2-HAP) of 7.9 can be obtained at around 100°C reaction temperature. Therefore, lower temperatures appear to favor a higher isomeric ratio. However, already at a temperature of 100 °C the conversion of phenyl acetate was not quantitative anymore, at least not when using a residence time of 45 s, and a relatively high phenol content was found. At 120°C full conversion was reached. To avoid the complexity associated with recycling the phenol and the unreacted phenyl acetate, temperatures around 120 °C are therefore preferred. Depending on the reaction conditions, different ratios between the desired product 4-HAP and the undesired side products, 2-hydroxyacetophenone (2-HAP, 3) and phenol (PhOH) may be obtained. A high selectivity towards the desired 4-isomer can be achieved, when the MSA is applied in great excess, such as 8 or more, 9 or more, 10 or more, 11 or more, 12 or more, etc.,
equivalents MSA relative to the amount of phenyl acetate. On the other hand, lower amounts of MSA are favored from a perspective of process economics, such as 2 or more, 3 or more, 4 or more, 5 or more, 6 or more, 7 or more, or 8 or more, equivalents MSA relative to the amount of phenyl acetate. Accordingly, in the process of the invention, the molar ratio is generally not limited, and may in particular lie in any of the previous ranges, or in any of the following ranges, where the MSA source fed into the reactor includes the MSA in an amount ranging from 2 to 20, 4 to 18, 6 to 14, or 8 to 12, equivalents MSA relative to the amount of phenyl acetate. In particular embodiments, the MSA source fed into the reactor includes the MSA in an amount of 8 to 12 equivalents, relative to the amount of the phenyl acetate. Moreover, the water content of the MSA source has a bearing on the reaction outcome. In particular, water favors hydrolysis of the phenyl acetate to phenol, thereby leading to an undesired high phenol content in the reaction mixture. It is therefore preferred that the MSA source has a low water content. In a preferred embodiment, the MSA source is characterized by a water content of at most 3.0 wt.-%, preferably at most 2.5 wt.-%, more preferably at most 2.0 wt.-%, most preferably at most 1.5 wt.-%, as determined by Karl Fischer titration. The reaction product of step c) has preferably the following composition: In a preferred reaction product, the isomeric ratio of 4-HAP to 2-HAP is at least 4, preferably at least 5, more preferably at least 6, and/or the isomeric ratio of 4-HAP to 2-HAP is at most 100, preferably at most 50, more preferably at most 20, yet more preferably at most 10, most preferably, the isomeric ratio of 4-HAP to 2-HAP ranges between 6 and 10. Moreover, in a preferred reaction product, the 4-HAP is contained in an amount ranging from 70 to 95 wt.-%, preferably 75 to 90 wt.-%, more preferably 80 to 85 wt.-%. Furthermore, a preferred reaction product includes 2-HAP in an amount of less than 20 wt.-%, preferably less than 17 wt.-%, more preferably less than 15 wt.-%, most preferably less than 12.5 w.-%. In addition, a preferred reaction product contains phenol in an amount of less than 10 wt.-%, preferably less than 8 wt.-%, more preferably less than 6 wt.-%, most preferably less than 4 wt.- %. Because the reaction product contains large amounts of MSA, and minor amounts of 2-HAP and possibly phenol, it is preferred that the formed 4-HAP and/or MSA is/are respectively isolated/enriched/purified and recycled. The present invention further relates to downstream processing of the reaction product obtained in step c). Though the downstream process is described in relation to the above described
upstream process, the downstream process is to be understood to form a separate aspect of the invention. That is, a process for recovering, isolating and/or purifying 4-HAP from a mixture comprising 4-HAP and one or more of 2-HAP, MSA and phenyl acetate, wherein the mixture is preferably produced or producible in step c) of the (upstream) process of the invention, may be carried out as follows: According to a preferred embodiment of the invention, the formed 4-HAP and/or residual MSA is/are isolated from the reaction mixture as the reaction proceeds. Because the reaction is continuous, also the isolation is preferably (semi-) continuous. Preferably, the isolation involves one or more of the following: (i) liquid-liquid separation (solvent extraction), wherein the 4-HAP is enriched in an organic phase and the MSA is enriched in an aqueous phase; (ii) removing water and/or MSA by distillation , wherein the 4-HAP remains in the retentate, (iii) removing water and/or MSA from the 4-HAP by membrane separation such as nanofiltration or reverse osmose; (iv) direct crystallization of 4-HAP from the reaction mixture; and separating 4-HAP from 2-HAP by alkaline extraction, wherein the 4-HAP is enriched in an alkaline aqueous phase and the 2-HAP is enriched in an organic phase, optionally, wherein after phase separation the alkaline aqueous phase is acidified and the 4-HAP is back-extracted from the acidified aqueous phase into an organic phase. Dilution with water, optionally after cooling, prior to separation may avoid that highly concentrated and hot MSA enters the liquid-liquid separation unit, suppress side reactions and facilitate recycling. The outlet of the reactor may be coupled to the liquid-liquid separation unit, optionally via heat exchanger(s) and/or mixer(s). Preferably, the organic phase includes an organic solvent selected from the group consisting of methyl isobutyl ketone (MIBK), methyl tert-butyl ether (MTBE), heptane, toluene, p-menthane, 2,4-dimethyl octane, p-cymene, cyclopentyl methyl ether (CPME), octanol, butyl acetate, ethyl acetate, methyl isobutyl carbinol (MIBC) and combinations thereof, preferably MIBK. According to a further embodiment of the present invention, the liquid-liquid separation involves feeding the reaction mixture to a continuous extraction unit. A preferred continuous extraction unit is a continuous countercurrent flow extraction unit. Moreover, a preferred continuous extraction unit includes alternating mixer and settler units. Using a continuous countercurrent flow extraction unit, an (almost) complete separation of MSA and 4-HAP can be achieved, when the reaction product is diluted with water, when the extraction is carried out using MIBK as organic solvent, and when a backwash with water is included. High dilution is preferred due to the dependency of the partition coefficient from the dilution factor. Optionally, the process further comprises the step of separating 4-HAP from 2-HAP by liquid- liquid separation, preferably in the form of alkaline extraction from an organic solvent, and/or distillation. The alkaline extraction from an organic solvent results in an alkaline aqueous phase enriched in 4-HAP and an organic phase enriched in 2-HAP. After phase separation the alkaline
aqueous phase can be acidified and the 4-HAP back-extracted from the acidified aqueous phase into an organic phase. In an exemplary approach, the reaction mixture is diluted with water, made alkaline with NaOH, and then subjected to biphasic extraction using an organic phase such as MIBK or heptane. Subsequently, the aqueous phase is acidified and subjected to biphasic extraction using an organic phase such as MIBK or heptane. For reasons of economics, preferred processes further comprise the step of recycling at least a portion of the residual MSA and/or recycling at least a portion of the organic solvent. Recycling of MSA can, for instance, be achieved by dilution with water, extraction of crudes and distillation off of water. Preferred processes may further include purification of the isolated 4-HAP by one or more of precipitation, crystallization, melt crystallization and distillation. The process has thus far primarily be described starting from phenyl acetate as starting material. However, the process may further comprise the step of producing the phenyl acetate to be fed into the reactor by acetylation of phenol using acetic anhydride. It is also envisaged that the phenyl acetate is produced directly in the reactor, whereas the Fries arrangement occurs subsequently or simultaneously. In other words, the phenyl acetate source may comprise phenol and acetic anhydride, and the phenyl acetate is produced by acetylation of the phenol using the acetic anhydride in the reactor, or the phenyl acetate source may be phenyl acetate, which is produced by acetylation of phenol using acetic anhydride before feeding the phenyl acetate into the reactor. Phenyl acetate synthesis may be carried out by reacting phenol with acetic anhydride in the presence of sodium acetate (NaOAc). To this end, phenol may be charged to a reactor, NaOAc may be added, and the mixture may be heated, e.g., up to 100 °C. Then Ac2O may be added dropwise. After complete addition, the reaction mixture may be stirred at increased temperature, e.g., at 100 °C. After reaction is complete, crude phenyl acetate may be optionally purified by removing acetic acid (AcOH) using distillation and extraction using water and ethyl acetate (EtOAc). The phenyl acetate and/or the 4-HAP may be individually selected from substantially petrochemical, and substantially non-petrochemical origin, including mass balance approach. Mixtures of petrochemical and non-petrochemical phenyl acetate, such as defined by a mass ratio of petrochemical to non-petrochemical phenyl acetate ranging from 5% to 95% to 95% to 5%, 10% to 90% to 90% to 10%, 20% to 80% to 80% to 20%, 30% to 70% to 70% to 30%, etc. and/or mixtures of petrochemical and non-petrochemical 4-HAP, such as defined by a mass ratio of petrochemical to non-petrochemical 4-HAP ranging from 5% to 95% to 95% to 5%, 10% to 90% to 90% to 10%, 20% to 80% to 80% to 20%, 30% to 70% to 70% to 30%, etc. are specifically envisaged.
Moreover, the phenyl acetate source may include both petrochemical and non-petrochemical carbon, as would be formed when only one of the phenol and the acetic anhydride used for forming the phenyl acetate is of petrochemical origin, while the other one is of non-petrochemical origin. For example, the phenol may be petrochemical and the acetic anhydride may be non- petrochemical, or vice versa. Any mass ratio of petrochemical and non-petrochemical phenol and/or any mass ratio of petrochemical and non-petrochemical acetic anhydride is envisaged. In a preferred embodiment, the 4-HAP and/or the phenyl acetate is/are substantially non- petrochemical. A compound is considered “substantially non-petrochemical” herein, if at least 75 %, preferably at least 80 %, further preferably at least 85 %, particularly preferably at least 90 %, even further preferably at least 95 %, more preferably at least 97.5 % of its carbon is non- petrochemically-derived carbon, based on the total carbon content of the respective compound, as determined by radiocarbon dating. Particular preferred is that the 4-HAP and/or the phenyl acetate is/are non-petrochemical, i.e. all carbon of the respective compound is non- petrochemically-derived carbon, based on the total carbon content of the respective compound, as determined by radiocarbon dating. Methods for determining whether and which proportion of the carbon of a compound is petrochemically-derived and/or non-petrochemically-derived are known to a skilled person. In the framework of the present invention radiocarbon dating, e.g. according to ASTM D6866-16, is used. The method is based on the observation that the ratio of 12C and 14C is indicative for the non-petrochemical carbon content. Non-petrochemical phenyl acetate can, for example, be produced from non-petrochemical phenol and non-petrochemical acetic anhydride, following the above-mentioned route. Sources of non- petrochemical phenol are cyclic terpenes and terpenoids. Non-petrochemical phenol can be produced from lignocellulose-derived alkylmethoxyphenols. In other embodiments, the 4-HAP is substantially petrochemical. A compound is considered “substantially petrochemical” herein, if at least 75 %, preferably at least 80 %, further preferably at least 85 %, particularly preferably at least 90 %, even further preferably at least 95 %, more preferably at least 97.5 % of its carbon is petrochemically-derived carbon, based on the total carbon content of the respective compound. In this embodiment, it is particularly preferred that the 4-HAP and/or the phenyl acetate is/are petrochemical, i.e. all carbon of the respective compound is petrochemically-derived carbon, based on the total carbon content of the respective compound. Another aspect of the present invention pertains to a raw material comprising or consisting of 4- HAP produced by the process described herein. As mentioned above, the 4-HAP is substantially petrochemical or substantially non-petrochemical.
According to another aspect, a personal care product or composition, or pharmaceutical product or composition comprises the raw material of the invention. Yet another aspect of the invention concerns 4-HAP, obtainable or obtained by a process as described herein, wherein the 4-HAP contains at least 25 %, preferably at least 50 %, further preferably at least 70 %, particularly preferably at least 80 %, even further preferably at least 90 %, more preferably at least 95 % non-petrochemically-derived carbon, based on the total carbon content of the 4-HAP.
Experiments A preferred embodiment of the present invention relates to a continuous manufacturing process for 4-hydroxyacetophenone (4-HAP) through Fries rearrangement of phenyl acetate in the presence of excess methane sulfonic acid (MSA). The trials until achieving the invention included the establishment of an upstream and a downstream processing protocol based on continuous extraction to separate the product from the MSA. The process resulted in full conversion of phenyl acetate while leading to assay yields of 4-HAP, 2-HAP, and phenol of 82.5 %, 11.9 % and 8.1 %, respectively. The results could be improved further, especially with respect to the phenol content, when the MSA was dried with 0.15 equiv. of Ac2O (compared to phenyl acetate) at 90 °C for 30 min prior to the reaction. Using the pretreated MSA resulted in full conversion of phenyl acetate and assay yields of 4-HAP, 2-HAP, and phenol of 83.7 %, 10.7 % and 1.9 %, respectively. To demonstrate robustness, these conditions were tested in a prolonged run of almost 9 h, which resulted in the formation of 2.6 kg 4-HAP by means of a 30 mL reactor. The collected effluents were subsequently fed to a continuous extraction column. MIBK was identified as preferred extraction solvent when used as the dispersed phase. With this setting it was shown that 4-HAP (and 2-HAP) can be extracted almost quantitatively from the quenched reaction mixture leading to an organic phase which contains at least 20 wt.-% of 4-HAP (max. 27 wt.-%). If either a high enough MIBK feed or sufficient column length was used, only 20 mg 4- HAP per hour were lost to the raffinate. An aqueous back-wash was identified as a desirable step in the process to remove MSA completely from the extract as this acid would cause 4-HAP degradation in the subsequent distillation step. Specifically, the MIBK phase contained up to 0.5 wt.-% of MSA depending on the applied conditions, which could be reduced by a factor of 10 by introducing this back-wash. The feasibility of the Fries rearrangement (scheme (I), depicted above) of phenyl acetate (1) to 4- HAP (2) in the presence of methane sulfonic acid (MSA) was demonstrated. Depending on the tested reaction conditions, different ratios between the desired product 2 and the undesired side products, 2-hydroxyacetophenone (2-HAP, 3) and phenol (PhOH, 4), were obtained. A high selectivity towards the desired 4-isomer 2 was achieved, when at least 8 equivalents of MSA were applied. Consequently, efficient recycling of the latter is preferred to make this process economically viable. Further experiments aimed at development of a telescoped process for the synthesis of 4- HAP from phenol; and development of an efficient downstream processing protocol, including the isolation of 4-HAP (by crystallization and distillation) and recycling of MSA.
As described in more detail in the following, using a continuous flow synthesis of 4-HAP from phenyl acetate 12 molar equivalents of MSA at 120 °C and a residence time of 45 second resulted in (near) full conversion and an assay yield of 75% for product 2. MSA could be recovered and recycled at least six times. Although the selectivity towards 4-HAP was independent of the starting material (phenol or phenyl acetate), higher conversion (and yields) were achieved when the synthesis started directly from ester 1. Then, an aqueous work up procedure was developed, which allowed separation of the two isomeric hydroxyacetophenones and provided 4-HAP in 63% isolated yield. Fractional distillation of 4-HAP resulted in an off-white product. However, in view of the high boiling point and limited thermal stability of the product, a deep vacuum (< 0.1 mbar) was required for this distillation. Meeting such requirement may be difficult on production scale. Hence, a different isolation/purification procedure was sought-after. Based on a melting point difference of 100 °C between the isomers 2 and 3, a melt crystallization was looked into as a potential alternative. In a proof-of-principle experiment that used artificial mixtures of 2- and 4-HAP, it was possible to increase the 4-HAP content from 90 wt.-% up to 97.5 wt.-%. 1. Synthesis of phenyl acetate Phenyl acetate (PhOAc) synthesis was carried out by reacting phenol with acetic anhydride (Ac2O) in the presence of 10 mol% sodium acetate (NaOAc). To this end, phenol (50 g, 0.52 mol, 1.0 eq) was charged at 70 °C to the reactor. NaOAc (5.0 g, 60 mmol, 0.11 eq) was added and the mixture heated up to 100 °C. Then Ac2O (68 g, 0.67 mol, 1.3 eq) was added dropwise over a period of 120 min. After complete addition, the reaction mixture was stirred for another 7 h at 100 °C. The conversion of Phenol was monitored by TLC. After reaction was complete, crude phenyl acetate was purified from the reaction mixture by removing acetic acid (AcOH) using distillation at 70°C and extraction using water and ethyl acetate (EtOAc). Specifically, AcOH (40 g) was distilled off at 70 °C and 15 mbar. Crude product was cooled to room temperature, diluted with water (50 g) and extracted with EtOAc (100 g). After phase separation, the organic layer was washed with aqueous NaOH (50 g, 2 %) and water (50 g). The solvent (82 g EtOAc) was removed at 70 °C and 20 mbar to yield PhOAc (66.9 g, 98 % purity, 0.48 mol, 92.5 % yield). 2.1 Flow reactor setup A 3D-printed reactor was integrated into a flow setup as schematically depicted in Fig.1. Pumps were used to feed the chemicals. Phenyl acetate (PhOAc) synthesis was carried out by reacting phenol (PhOH) with acetic anhydride (Ac2O). PhOAc and MSA were preheated in heat exchangers. Then, the preheated streams were combined in a small mixer (not shown) and the joined effluent was reacted in a reactor. The resulting reaction mixture was cooled down. Afterwards, water was blended with the effluent and the diluted stream was cooled down. 2.2 Recovery setup
Dilution prior to extraction avoids that highly concentrated MSA enters the extraction column. Being a strong acid, MSA will be completely ionized in an aqueous solution. The heat of formation of the methanesulfonate ion in an aqueous solution is reported to be -163.79 ± 1.04 kcal/mol. Consequently, the solvation of MSA in water represents an exceptionally exothermic process (vide infra) and a significant temperature drifting during the extraction would arise. Hence, aqueous dilution and cooling prior to the continuous extraction ensure a safe and stable operation. The diluted stream was guided to a liquid/liquid extraction column using MIBK as organic solvent in which the formed 4-HAP was enriched along with side product 2-HAP. Distillation was carried out to separate 2-HAP from 4-HAP. The extraction solvents water and MIBK were fed by (peristaltic or HNP gear) pumps. The 4-HAP extract was collected at the top and the MSA raffinate at the bottom of the column. Recycle streams are shown. A residence time of 45 s is achieved in the installed 30 ml reactor when a total volumetric flow of 0.67 mL/s or 40 mL/min is used. Next, the total volumetric flow was converted into the total mass flow. Finally, the individual mass flows of PA and MSA were calculated considering the corresponding density of a pure compound and X its mass fraction, respectively. Taking into account the purity and stoichiometric ratio of the chemicals, the calculation resulted for the first experiment in flow rates of 6.28 g/min for PA (94 % purity by HPLC) and 50.51 g/min for MSA (purity 99 %). The application of PA with higher purity (99 %) results in a flow rate of 6.01 g/min and 50.91 g/min for MSA, respectively. In contrast to the process flow diagram shown in Fig.1, the outlet of the reactor was decoupled from the extraction column to enable the separate examination of flow conditions without the influence of additional extraction parameters. Details of tested reaction conditions and the corresponding assay results of the flow experiments are summarized in Table 1. The obtained results are discussed in the following paragraphs. Table 1: Summarized results of flow experiments with varied reaction temperature, concentration and quality of MSA.
3. Results 3.1 Varying the molar equivalents of MSA Starting material 1 was reacted in the presence of 12 molar equivalents MSA at 120 °C for 45 s. The effluent was diluted with equal molar amounts of H2O compared to MSA. Under these conditions full conversion of ester 1 was observed. The desired product 2 was formed with an assay yield of 82.4 % while the 2-isomer 3 was formed in 11.2 % resulting in an isomeric ratio of 7.4. Along with the rearrangement products 2 and 3, phenol (4) was formed as hydrolysis product in a yield of 10.9 % (Table 1, entry 1). The yield of 4-HAP was 82 %. Next, the influence of different amounts of MSA on the reaction outcome was tested (Table 1, entry 2-4). The reduction from 12 to 8 molar equivalents MSA resulted in minor changes of the reaction outcome. The yield for 4-HAP slightly decreased while the amount of phenol increased (Table 1, entry 2). A bigger change in the reaction numbers was observed, when the amount of MSA was reduced further to 6 molar equivalents. Under this reaction condition a diminished yield of 68.7 % for product 2 was obtained and only fractionally altered amounts of isomer 3 and phenol (4) (Table 1, entry 3). This result is explained by an undesired formation of tarry side-products at higher concentrations. In the presence of acid (and elevated temperatures), 4-HAP (2) tautomerizes to its corresponding enol, which opens the door for self-condensation reactions that finally lead to tar formation. The formation of such side products was not only observed in the collected effluent of the experiment but is also unmasked by the incomplete mass balance. Based on these findings it was concluded that a higher MSA content is favorable for the reaction outcome since a higher dilution helps controlling the tar formation. 3.2 Varying the reaction temperature Different reaction temperatures were applied because the regioselectivity of the Fries rearrangement is assumed to be temperature dependent. At lower temperatures the desired 4- isomer 2, the kinetic product, is preferably formed, whereas elevated temperatures favor the in
this case undesired thermodynamic 2-isomer 3. The first experiments were performed at 120 °C exclusively and resulted overall in a 4-HAP/2-HAP ratio of 7.4 (Table 1, entry 1-3). Consequently, the temperature was decreased to 100 °C in a next experiment to potentially increase the isomeric ratio. The performed experiment resulted in a high PA (1) conversion of 98.1 % and in assay yields of 70.8 % 4-HAP, 8.9 % 2-HAP, and 18.7 % phenol, respectively (Table 1, entry 4). Based on the assay results an improved isomeric ratio of 7.9 was obtained at a reduced reaction temperature. Therefore, a higher isomeric ratio can be assumed at even lower temperatures. However, already at a temperature of 100 °C the conversion of phenyl acetate (1) was not quantitative anymore at a fixed residence time of 45 s and a relatively high phenol content was found. In theory it would be possible to recycle the phenol and the unreacted phenyl acetate, but this endeavor will significantly increase the complexity of the DSP in the final process. A temperature at 120 °C ensures full conversion of the starting material 1. Additionally, the phenol content was tried to be reduced while increasing the yield of 4-HAP by chemical means. 3.3 Varying the MSA source Different MSA sources were tested. With a different MSA source (same purity 99%), we repeated the reaction conditions of the previous experiment (Table 1, entry 5 vs. 6). Surprisingly, the reaction outcome changed significantly when the new source of MSA was applied. The assay yields of 4-HAP and 2-HAP dropped to 65.9 % and 9.5 %, respectively, while the yield of phenol increased to 21.5 %. Repeating the experiment with different bottles of the same supplier resulted in the same reaction outcome and remained unchanged in a prolonged run of 40 min as well (Table 1, entry 6-8). These findings suggest a higher water content in the newly sourced MSA based on the fact that the supplier change alone led to a significant increase of the hydrolysis product phenol. Therefore, the water content of different MSA sources was determined by Karl-Fischer titration. Besides MSA sourced from Fluorochem and Sigma Aldrich, we also included CarlRoth as supplier for MSA (purity 99.5%). The results of the Karl-Fischer titration are listed in Table 2. Table 2: Water content of different MSA sources determined by Karl-Fischer titration. The analysis disclosed a water content of 2.6 wt% in the MSA sourced from Fluorochem. Based on this value 1.7 molar equivalents of water compared to phenyl acetate were fed during the flow experiments. In contrast, a lower water content of 2.2 wt% was found for the MSA sourced from Sigma Aldrich. This comparison shows that the increased phenol content in the experiments using Fluorochem’s MSA can indeed be linked to a higher water content. The additionally ordered MSA
from CarlRoth contained the lowest water content of 1.4 wt% and resulted in a reaction outcome (Table 1, entry 11) similar to the ones using Sigma Aldrich’s MSA. Consequently, these two sources were used for further experiments. 3.4 Varying the residence time Variation of the residence time was investigated. So far, the residence time was fixed in all experiments to 45 s. For a better understanding, the residence time was shortened to 30 s. In this case, the conversion of starting material 1 was still complete, and assay yields of 80.6 %, 11.9 %, and 10.7 % for 4-HAP, 2-HAP, and phenol were obtained, respectively (Table 3, entry 1). Compared to the benchmark reaction of 45 s (Table 3, entry 2) a decreased isomeric ratio of 6.8 was obtained. When doubling the initially used 45 s residence time, assay yields of 81.7 % for 4- HAP, 11.1 % for 2-HAP, and 6.1 % for phenol were obtained at a full conversion of phenyl acetate leading to an isomeric ratio of 7.3 (Table 3, entry 3). Consequently, only a minor improvement of the selectivity was achieved at the cost of the productivity. Therefore, we decided to keep using a residence of 45 s for the further investigation. Nevertheless, it was shown that the process is robust against variations of the residence time. Table 3: Reaction outcome varying the residence time. 3.5 Anhydrides as drying agent As mentioned above, the water content crucially influences the reaction outcome as it will determine the amount of undesired phenol by hydrolysis. Since neither changing the reaction temperature nor the residence time significantly influenced the phenol content, the water content of MSA was decided to be reduced by chemical means. Thus, the application of anhydrides as drying agents for MSA and selected acetic anhydride (Ac2O) and methane sulfonic anhydride (Ms2O) were tested for this purpose. MSA was dried by mixing the acid with a defined amount of the chosen anhydride for 30 min at 90 °C. Then, phenyl acetate was added and the reaction mixture was analyzed after 30 min. The results for Ms2O are listed in Table 4 and for Ac2O in Table 5. Table 4: Reaction outcome using different amounts of Ms2O as drying agent.
First, the benchmark was set using untreated MSA. Assay yields of 83.7 % for 4-HAP, 8.0 % for 2-HAP, and 5.7 % for phenol were obtained at 90 °C and a reaction time of 30 min (Table 4, entry 1). The benchmark could be slightly improved when a freshly opened bottle of MSA was used. In this case a similar yield of 83.8 % for 4-HAP was observed but the isomeric ratio was slightly increased from 10.5 to 10.9 (Table 4, entry 2). Then, dried MSA was used for the Fries rearrangement of starting material 1. The application of 0.66 equiv. Ms2O as drying agent resulted in an excellent isomeric ratio of 70.8 (Table 4, entry 3) and a low phenol content of 0.5 %. However, this result was achieved at the cost of an incomplete mass balance (76.3 %) and a reduced 4-HAP selectivity. Based on the significantly reduced amount of 2-HAP and the presence of a significant amount of tarry side products, we reasoned that consecutive reactions from 2-HAP are favored under these conditions. Therefore, we reduced the amount of anhydride in the next experiment ten-fold. In this case, an isomeric ratio of 11.8 was obtained along with a low phenol content of 3.2 % (Table 4, entry 4). Still, the mass balance is incomplete, and we reduced the amount of anhydride even further. When 0.01 equiv. Ms2O was used as drying agent a mass balance similar to the ones of the benchmark reactions is observed along with a 4-HAP yield of 85.2 %, but a phenol content similar to fresh MSA (Table 4, entry 5). Thus, we applied the conditions reported in literature that make use of 0.2 equiv. Ms2O compared to the starting material. Under these particular conditions, the isomeric ratio rose again to 16.3 and the phenol content decreased, but so did the mass balance. These batch experiments demonstrate that it is possible to decrease the water content in the MSA and therefore the phenol content. Along these lines, a compromise between a low phenol content and undesired tar formation was made. Table 5: Reaction outcome using different amounts of Ac2O as drying agent. Next, we evaluated if the same trends exist for another Ac2O. The application of varying amounts of anhydride (0.01-1.15 equiv.) results in a significantly decreased phenol content but also in
some cases in an incomplete mass balance (Table 5, entry 1 and 2), similar to Ms2O. When 0.01 equiv. Ac2O are used (Table 5, entry 3), an almost complete mass balance of 98.3 % and a reduced phenol content of 3.9 % were observed. Additionally yield and selectivity for 4-HAP were also increased to 86.4 %. The application of the literature conditions adapted to Ac2O resulted again in an incomplete mass balance (Table 5, entry 4). Surprisingly, an acceptable mass balance of 94.3 %, a yield of 86.1 % for 4-HAP as well as a low phenol content of 0.8 % were obtained when 0.1 equiv. of Ac2O compared to phenyl acetate were applied (Table 5, entry 5). This result demonstrated a good balance between drying the MSA and taming the side reactions. Additionally, the comparison between the two anhydrides revealed acetic anhydride to be slightly superior in terms of the reactions outcome, while being significantly cheaper than Ms2O. It is worth mentioning that overall the application of Ac2O resulted in a lower phenol contents compared to Ms2O. This could potentially point to a desired trapping of phenol as phenyl acetate by Ac2O, which is not possible for Ms2O. Table 6: Reaction outcome of flow experiments using Ac2O pretreated with MSA. For these reasons, 0.1 equiv. of acetic anhydride compared to phenyl acetate was used for drying MSA, but this time performing the Fries rearrangement under flow conditions. Using pretreated MSA under flow conditions significantly improved the yield of 4-HAP, reduced the phenol content compared to the bench mark (Table 6, entry 1 vs. entry 2.) and resulted in an acceptable mass balance. We repeated the flow experiments and increased the acetic anhydride amount to 0.2 equiv. compared to the starting material. Similar to the batch procedure, an incomplete mass balance was observed and the overall reaction outcome slightly deteriorated (Table 6, entry 3). The application of 0.01 equiv. Ac2O was not effective to improve the reaction outcome either. In this case a similar phenol content compared to the benchmark was observed (compare Table 6, entry 1 and 4). Since 0.1 equiv. of Ac2O resulted in an improved selectivity and 0.2 equiv. in the lowest phenol content, we applied 0.15 equiv. of Ac2O compared to the starting material as compromise. These conditions did not only result in a complete mass balance, but also resulted in an excellent reaction outcome of 86.8 % 4-HAP, an increased isomeric ratio of 7.7, and a low phenol content of 2.1 % (Table 6, entry 5). Based on these results, we decided to proceed using a temperature of 120 °C on the HTF side, a residence time of 45 s, and MSA sourced from CarlRoth or Sigma Aldrich that was pretreated with 0.15 equiv. Ac2O.
3.6 Long-runs Long-term stability of the process in the flow reactor was demonstrated. The results of the trials are listed in Table 7. In a first attempt we aimed for a production run of 2 h. In this case an almost quantitative conversion of PA of 99.5 % was achieved that led to assay yields of 80.8 % 4-HAP, 10.6 % 2-HAP, and 1.5 % phenol, respectively (Table 7, entry 1). Table 7: Reaction outcome of long-term runs.
These results equal an isomeric ratio of 7.7 and 673 g of crude 4-HAP. Additionally to the HPLC analysis of the effluent, we recorded the PhOAc temperature and all flow rates. As the temperature recording illustrates, a stable outlet temperature of approx. 100 °C can be found for PhOAc over the course of the reaction. However, the recording of the flow rates show an unstable MSA feed since peaks in the MSA flow rate are observed. This fluctuation can also be seen in the mass balance, which is incomplete. Since such fluctuations have not been observed during the optimization phase, we repeated the experiment twice. Similar results in terms of conversion and yield were observed (Table 7, entry 2 and 3), but the fluctuations in the MSA feed remained. Since the fluctuations appeared randomly and only over a time frame of 2-5 s, we hypothesized a potential cavitation was taking place in the pump head. If true, a slightly increased pressure on the MSA feeding vessel as well as a bigger feeding tube in direction of the pump head should circumvent the fluctuations. Thus, we exchanged the ⅛-inch tubing for a ¼-inch tubing and placed the MSA feed vessel at an elevated position. With these small adjustments it was possible for us to perform another run was performed over 8 h with stable reaction parameters. This can be seen by the consistency between set and measured feeds. Only in the case of the water feed 5 small fluctuations can be seen that are traced back to air bubbles. However, water is only used for diluting the effluent, the alternations in the water feed have no effect on the reaction outcome. In this long-run we were able to achieve complete conversion leading to yields of 83.7% 4-HAP (2.6 kg), 10.7 % 2-HAP, and 1.9 % phenol as well as an isomeric ratio of 7.8 and a good mass balance of 96.3%, respectively (Table 7, entry 4). Thus, the development of a stable flow system for the production of crude 4-HAP can be concluded. 3.7 Batch reaction (comparative example)
Phenol (20.1 g; 0.214 mol) was dissolved in 8 molar equivalents (163.9 g; 1.705 mol) of methane sulfonic acid. The solution was heated to 49°C and subsequently, acetic anhydride (26.2 g; 0.257 mol; 1.2 equiv.) was added over a period of 20 minutes. During the dosing, the heating was continued until the final temperature of 60°C was reached. The mixture was subsequently stirred at this temperature for another 130 minutes. After cooling to ambient temperature the reaction mixture was carefully diluted with water (170 g) and subsequently extracted three times with organic solvent (once with 135 g and twice with 75 g). The combined organic extracts were washed with water (twice with 60 g) and the subsequently the solvent was removed in vacuo. The 4-HAP was isolated from the obtained residue by means of fractional distillation at reduced pressure. Thereby, the pressure was slowly reduced to 0.1 mbar, and the bottom temperature was gradually increased to 180˚C. The melting point of 4-HAP (109˚C) required heating of the condenser in order to avoid solidification of the distillate in the condenser. The reaction yield of 4-HAP was 75%. The distillation yield of 4-HAP (sum of amount in fractions) was 90%. 4. Crystallization The collected effluent of the above mentioned flow process was stored for extraction experiments. It was observed that in most samples a solid crystalline-looking material, potentially crystallized 4-HAP, formed overnight. Therefore, a sample of the solid material was collected, blotted with a paper to free the solid as good as possible from attached mother liquor, and finally analyzed by HPLC. The analysis revealed the material to consist of up of 99.9 wt% 4-HAP and only traces of 2-HAP. This chance find opened the door for crystallization/precipitation procedures as an alternative to the planned extraction in the DSP. Therefore, we investigated the isolation of 4-HAP by crystallization/precipitation further. Further in respect of crystallization, reference is made to applicant’s international patent application WO 2023/170048 A1. 4.1 Crystallization from the effluent The amount of crystallized 4-HAP based on simple effluent storage was analyzed. Therefore, the amount of 4-HAP found in freshly collected effluent was compared to the amount of the supernatant solution after solid material formed. Overall, 8 samples collected in three individual flow experiments were analyzed. The results are shown in Table 8. All samples showed a solidified amount of 4-HAP between 34 and 37 % (Table 8, entry 1-2.4). Only the samples from a 100 °C-experiment (see Table 1, entry 4) differed. The effluent collected in this reaction revealed a solidified amount of approximately 21% for product 2 (Table 8, entry 3.2 and 3.3). The reason for the absence of solid product 2 in the first sample of the 100 °C-reaction (Table 8, entry 3.1) remains unclear, especially since equal amounts of 4-HAP were present in the fresh effluent (Table 8, entry 3.1-3.3). However, this particular sample indicates that the reaction still proceeds at room temperature, since the amount of 4-HAP increased and the remaining amounts of the
starting material disappeared (compare Table 1, entry 4). Moreover, the results of all experiments highlight that the majority of 4-HAP still remains in solution as the solubility of 4-HAP in the given MSA:H2O mixture of approximately 1:1 is too high. To increase the amount of solidified 4-HAP, additional crystallization experiments were performed. Table 8: Amount of 4-HAP crystallized out from the effluent upon storing. First, the effect of small amounts of additional water was investigated. Thus, 1 g of water was added to the effluent and the supernatant solution was homogenized by shaking. The next day, the solution was analyzed and a small amount of 0.77 %-points of additional solid 4-HAP was found (Table 9, entry 1). Next, we investigated the effect of sonication and colder storage temperatures on the amount of solid product 2. When the already solid containing effluent was sonicated for 1 minute and stored overnight at room temperature, we found that some of the solid 4-HAP was dissolved again (Table 9, entry 2). However, when the experiment is repeated but the sample was stored at 4 °C a significant additional amount of solid 4-HAP of 8.78 %-points was found (Table 9, entry 3). More solid material was expected since the overall solubility is known to decrease with decreasing temperature. Pursuing a complete isolation of 4-HAP by crystallization, we increased the water content in the next sample in a way that a MSA:H2O ratio of 1:3 was made. Under these conditions and additional storing overnight at room temperature 10.81 %- points of extra solidified 4-HAP were found (Table 9, entry4). Next, we were interested, if the solidification of 4-HAP was complete. Therefore, a sample that already contained solid product 2 was analyzed the next day again. In this case an additional amount of 1.34 %-points solidified. A recrystallization of the exact same sample led basically to the starting amount of 37.37 %. Thus, these experiments (entry, 5 and 6) show that the maximum amount of solidified product 2 at room temperature in a mixture of MSA:H2O is almost reached. As last experiment, we used the sample that initially did not contain solid 4-HAP. When this sample was seeded with 16 mg of the isolated material, the next day small quantities of 4-HAP crystals were found indicating that a seeding procedure can be used to initiate crystallization. However, roughly 50 % of 4-HAP remained in solution. A further increase of solidified product 2 would require the application of even larger amounts of water. Table 9: Additional 4-HAP crystals under various crystallization strategies.
4.2 Melt crystallization 4.2.1 Melt crystallization of artificial reaction mixtures including phenol To assess melt crystallization, we extended the artificial mixtures to also contain phenol; the hydrolysis product of the Fries rearrangement. As described in the previous chapter different amounts of phenol were formed depending on the water content of the MSA. Therefore, we corrected the artificial mixture for phenol in the first experiments. Two samples with ratios of 8:1:1 and 7:1:2 regarding the amounts of 4-HAP, 2-HAP, and phenol were prepared, respectively (compare Table 1). These mixtures were placed in an oil bath and heated until a homogenous melt was observed. The artificial mixture with a ratio of 7:1:2 was completely molten around 90 °C whereas a temperature of 110 °C was required to reach the same result with the 8:1:1 mixture. Then, the heating was turned off and the melts were kept stirring in the oil bath for slow cooling. In both cases one solid reaction mixture was observed below the melting temperature of the 4-isomer but above the melting temperature of the 2-isomer. This simple setup did not allow a separation of the isomers but demonstrates that a crystallization is potentially possible in the presence of varying amounts of different side-products (2-HAP and phenol). 4.2.2 Melt crystallization of a batch extract The concept of a melt crystallization was tested with an actual reaction mixture. A batch extraction was applied. First, the effluent was diluted further with water to achieve a MSA/H2O ratio of 1:2. Then, the aqueous phase was extracted thrice with MIBK. The HPLC analysis of both layers revealed that the product 2 and the side-products 3 and 4 were completely extracted into the organic phase. This experiment demonstrates the complete removal of organic material from the effluent by means of extraction. Consequently, this experiment was also the basis for the investigation of the continuous extraction process (vide infra). Then, the organic layer was concentrated under reduced pressure. After solvent evaporation a dark colored residue was obtained. The residue was transferred to the melt chamber of the melt crystallization setup, which consisted of a double-jacketed dropping funnel connected to a Liebig condenser. Both parts were heated individually, and the overall setup was tilted. To feed a melt
to the Liebig condenser, the melting chamber was heated to 110 °C; the temperature at which a homogenous melt in the previous experiments was observed. Then, we attempted the actual melt crystallization by feeding the melt dropwise to the condenser. We concluded that the melt crystallization is difficult at this stage in the DSP. However, it may be useful in the purification of 4-HAP at a later stage of the DSP. For instance, once the heavy side products described above are removed by a short path distillation only the isomeric mixture needs to be processed further. Then, the large difference in the melting points of 2-HAP (4-6 °C) and 4- HAP (104 °C) might prove beneficial. 4.3 Investigation of the continuous extraction Moreover, we focused on the development of a continuous extraction protocol to separate crude 4-HAP (as well as 2-HAP, phenol, and other organic side products) from the to be recycled MSA. 4.3.1 Solvent selection As potential extraction solvents, p-menthane, p-cymene, or 2,4-dimethyl octane would be the preferred choices. Heptane, cyclohexane, and toluene have been tested in batch extractions. However, the solubility of 4-HAP in them is marginal and their application caused precipitation. Being part of the same solvent class, similar solvation properties can be expected for p-menthane, 2,4-dimethyl octane, or p-cymene, excluding them as extraction solvents. An hypothesis that was underlined by testing p-cymene as solvent for pure 4-HAP. The attempt to dissolve 0.5 g 4-HAP in 5 g of p-cymene (2215-23-02-062) resulted in a 4-HAP content of only 0.2 wt% (HPLC) after the mixture was stirred for 24 h at ambient temperature. Thus, we moved onwards with the next solvents MIBK and MTBE. On the basis of initial data, MIBK seemed to be the better choice. Not only because it can dissolve up to 25-33 wt% 4-HAP at 20-30 °C, but also since it offers a partition coefficient for 4-HAP of 14, whereas the one for MTBE was determined to be 1. Accordingly, a more efficient extraction process on the basis of MIBK can be foreseen, and a successful application of MIBK as extraction solvent has already been demonstrated. 4.3.2 MIBK stability After MIBK was selected as extraction solvent, its stability in a highly acid environment on the basis of MSA was questioned. Being an α-CH acidic carbonyl compound the presence of acid may catalyze potential aldol reactions. To examine the stability, MIBK (as received) was stirred in the presence of 5 wt% MSA at ambient temperature. The accumulation of potential aldol products, and thus the stability of MIBK, was monitored over time by GC-MS. Already after 2 h non-neglectable signals for aldol products were observed by GC-MS. Their signals and thus their quantities kept growing over the monitored time frame of 72 h. Since the
formation took place in the presence of smaller quantities of MSA (5 wt%), this reactivity could present a problem for the extraction, where even larger amounts of MSA will be present. However, the effluent will not consist of pure MSA as it will be diluted with a significant amount water. Therefore, the influence of water on the formation of undesired aldol products was investigated next. The previous experiment was repeated, this time using wet MIBK that was prepared by stirring a biphasic mixture of MIBK and water at ambient temperature for 10 min. After phase separation, the wet MIBK was subjected to the stability test. Based on a water solubility of 19 g/L in MIBK, a maximal concentration of 2.3 wt% H2O in MIBK, and thus an excess of MSA towards water, is given. Like before, the reaction was monitored by GC-MS. No signal for aldol products could be detected by GC-MS after 2 h, whereas in the case of “dry” MIBK a significant amount was already present. After 72 h, a diminutive signal for aldol products was detected. These results indicate that water indeed tames the formation of aldol products and increases the stability of MIBK in an acid environment significantly. Moreover, an almost 1:1 ratio between MSA and water will be used in the column (vide infra). Thus, the reactivity of MSA will be reduced further. Finally, while in wet MIBK aldol products were detected after several hours of reaction time, the residence time in the extract in the column will be in the range of minutes, hampering the decomposition of MIBK further. Overall, the experiments so far justify selection of MIBK as extraction solvent. 4.3.3 Design of the continuous extraction process After sufficient stability of MIBK in an acidic aqueous environment had been demonstrated, we continued our endeavors designing a continuous extraction process. In theory, two separate extraction steps will be used. The first is the extraction of the effluent with MIBK. The second an aqueous back-wash of the extract to remove residual amounts of MSA. In this process, the freshly prepared effluent will be fed to the continuous extraction column. The exact feeding position will be determined during the investigation. The extraction solvent MIBK will be fed almost at the bottom of the column to maximize the number of extraction stages. Water will be fed at the top of the column for the potentially required back-wash for the same reason. As a result, it is planned to receive an aqueous MSA raffinate at the bottom of the column and the 4-HAP containing organic extract at the top of the column. Ideally the organic side products, e.g., 2-HAP, phenol, are also fully extracted into the MIBK phase. 4.3.4 Design of the continuous extraction column Based on the results so far, we decided to build a continuous extraction column consisting of alternating mixer and settler units. The column has an internal diameter of 4 cm and a length of approx.100 cm. The MIBK inlet was 7 cm and the water inlet 94 cm from the bottom. The extract overflow was located at 98 cm from the bottom. The effluent can be fed either at 25 cm, 50 cm,
or 75 cm from the MIBK inlet. The impellers used have a 6-blade geometry of 1 cm in height and 1.8 cm in diameter. Due to the acidic environment the impellers were made of Hastelloy C. The settler units were filled with 40 Raschig rings (5 mm height, 5 mm outer diameter, 3 mm internal diameter) to break the movement of the liquid due to agitation. 4.3.5 DoE study Employing a DoE study, the extraction process was then tried to be optimized. The feed rate for the effluent (MSA:H2O 1:1, collected over various flow experiments) was 10 mL/min. Then, we selected the ratio between the volumetric feeds of effluent and water as well as between MIBK and water as parameters for the DoE study. Thus, all feeds were linked with each other. Another parameter in the study was the position of the liquid-liquid boundary layer between raffinate and extract. This variation should allow to determine the influence of varying the number of theoretical extraction stages for both the 4-HAP extraction as well as the aqueous back-wash. This parameter was considered to be especially important since we wanted to combine both steps in a single extraction column. As response we selected the MSA content in the raffinate as well as the amount of 4-HAP lost to the raffinate compared to the initial amount of 4-HAP fed with the effluent. The tested extraction conditions and resulting responses of the DoE study are listed in Table 10. Table 10: Tested extraction conditions for the DoE study.
To have comparable results, samples have been taken once the steady-state for each individual extraction condition was reached. We decided to check for the steady-state by tracking the MSA content of the raffinate over time by IR. For the DoE we focused on the raffinate since the MSA should not only be recycled but should also contain as little organic material as possible. Samples of the raffinate were taken every 5 minutes. The MSA content reached a plateau that indicates the steady-state after 30 minutes. Once the steady-state was reached samples of extract and raffinate were taken and analyzed (Table 10). With this data in hand, the DoE model was fed and analyzed using the software package MODDE.
The analysis of all data shows no correlation between the selected parameters according to the model. However, the data include experiments, which have a theoretically impossible MSA extraction of over 100%. There are various reasons for these outliers. First, the IR calibration was done with solutions of pure MSA in water. Consequently, the presence of organic material in the raffinate, such as (side) products or traces of MIBK, might interfere with the analysis. Second, peristaltic pumps were used in the DoE experiments for the inlet- and outlet streams of the extraction column. Even though the calibration of the pumps was checked on a daily basis, it cannot be excluded that the volume fed per revolution changed for one or all pumps over the tested experiment. As a consequence, the calculations based on the set feed rates do not match anymore the actually used feeds. In case these outliers are excluded from the analysis, and at the cost of the reproducibility experiments, it appears that each parameter influences the extraction to some extent. Especially the ratio between the streams appears to have a significant influence on the outcome of the reaction. Even though the DoE study did not result in a straightforward optimization of the extraction process, the single experiments of the study reveal useful information that could be used for the further optimization (Table 10). First, it can be seen that even though the highest MSA content was approx. (only) 27 wt% in the raffinate, non- neglectable amounts of 4-HAP were lost to the raffinate. Since this MSA content should in principle result in a 4-HAP partition coefficient higher than 13 (compare 3.3.3.2.2), it indicates that the number of extraction steps is too small to ensure a complete extraction of 4-HAP. Consequently, the feed point of the effluent to the column should be significantly higher than the ones tested in the DoE study. Second, the best result in terms of lowest 4-HAP loss was obtained, when the MIBK feed was bigger than the water feed for the back-wash. Third, the relatively high water content resulted in a diluted MSA raffinate, which is unfavorable for the MSA recycling. 4.3.6 Batch extractions using actual effluent Additional experiments using small batch extractions have been performed. In these experiments actual effluent from the continuous process, based on a 1:1 w/w dilution of the reaction mixture with water, was applied and ergo led to a more representative simulation of the extraction column. To a defined amount of effluent was added just enough MIBK to form a second layer. After the mixture was vigorously shaken for 1 minute, both layers were analyzed. In this manner, the organic phase was enriched as far as possible in 4-HAP. Therefore, this sample of the organic layer can be considered to simulate the top of the extraction column. Repeating the experiment with equal amounts of effluent and MIBK should on the other hand represent a column section significantly closer to the MIBK feeding point. When actual effluent was used, the experiments revealed a potential enrichment of 4-HAP in the extract to up to almost 26 wt% at the top of the column while at the higher MIBK volume a 4-HAP content of 4.4 wt% was obtained. Based on these findings, it should be possible to efficiently extract 4-HAP and other organic material from the effluent as long as sufficient extraction stages are used. Extending this experiment, other MIBK volumes were tested as well to simulate the gradual enrichment of 4-HAP in MIBK over the
length of the column. In this way the extraction process was simulated and the partition coefficient of 4-HAP was more accurately determined using a representative matrix. A mean distribution coefficient of 9.8 (K value) was found for 4-HAP. Compared to the coefficients using pure material it is slightly reduced, but still high enough to potentially enable an effective extraction. Since actual effluent was used, the same procedure was applied to determine the distribution of 2-HAP and phenol in parallel. For 2-HAP a mean K value of 35.2 and for phenol of 25.2 was obtained. These coefficients indicate that once 4-HAP can be extracted efficiently, the side products will be too, leading in an ideal case to a raffinate that contains only trace quantities of organic material. 4.3.7 MIBK as continuous phase Unlike in the DoE study, we started a first extraction experiment with MIBK as continuous phase. After the column was filled with MIBK, the effluent as well as MIBK were fed with flow rates of 50 mL/min, and no aqueous backwash was used. Once the effluent was fed to the column, a colorful gradient slowly developed over time. After the system reached its steady-state the liquid-liquid boundary between extract and raffinate is clearly visible at the bottom of the column. Additionally the color intensity of the extract gradually increases towards the top of the column. This is already a visual indicator that the highest content of organic material is to be found at the top of the column. Such gradient was also visible when samples of raffinate and extract were taken over time. The results of the first extraction experiment building on the previously generated data is summarized in Table 11. Table 11: Results of the extraction using 50 mL/min for effluent- and MIBK feed using no back- wash, MIBK continuous phase. The applied conditions allowed a high MSA content in the raffinate of 51.9 wt%, while it contained only trace quantities of 4-HAP, and other organic material was not detectable by HPLC. The extract on the other hand revealed a 4-HAP content of 6.9 wt% and quantities of 0.54 wt% MSA. The amount of MSA was determined by titrating the extract with 0.1 N TBAOH (ntetrabutylammonium hydroxide, solution in a isopropanol/MeOH mixture). Additionally, the water content of the extract was determined by Karl-Fischer titration and shown to be 1.71 wt%.
This extraction experiment demonstrates an almost complete extraction of the organic material from the effluent if sufficient extraction stages and MIBK volumes are given. Without a backwash a high MSA content in the raffinate is achieved. However, smaller quantities of MSA are lost to the extract and still need to be removed prior to the next steps of the DSP. 4.3.8 MIBK as dispersed phase We tested if making MIBK the dispersed phase (again) would alter the extraction outcome. Therefore, we repeated the experiment, but started with a water-filled column. The results of the experiment are summarized in Table 12. In general, similar values to the previous experiment are obtained for the extract. In case of the raffinate, the results were further improved. At a similar MSA content only diminutive quantities of 4-HAP could be detected. The amount is so little that it equals a productivity or loss of 4-HAP in the raffinate of only 20 mg per hour. Next, we tried to increase the 4-HAP content in the extract, while reducing the MSA amount in it. To achieve this, we kept the conditions of 500 rpm, MIBK as dispersed phase, and an effluent feed rate of 50 mL/min. Table 12: Results of the extraction using 50 mL/min for effluent- and MIBK feed using no back- wash, MIBK dispersed phase. Since the next goal was to increase the 4-HAP content in the extract while reducing its MSA content, we tested first the limit of the extraction towards the 4-HAP content in the MIBK phase. Thus, we reduced the MIBK feed to 10 mL/min, still not using an aqueous back-wash. Already while monitoring the extraction process by IR, a high 4-HAP content revealed itself by crystal formation during analysis. The results of the extraction experiment are summarized in Table 13. A high 4-HAP content of 27.4 wt% being close to the maximal solubility, and a total organic content of 33.5 wt% were found in the MIBK extract. Table 13: Results of the extraction using 50 mL/min for effluent- and only 10 mL/min MIBK feed using no back-wash, MIBK dispersed phase.
Surprisingly, no MSA was detectable in the extract. The absence of MSA in the extract might be explained by the high organic content in the MIBK phase. Due to its high polarity and tendency to dissociate, MSA is completely forced into the aqueous phase. The analysis of the raffinate revealed a breakthrough of organic material into the aqueous phase. Consequently, there were either not sufficient extraction steps available to extract the high content of organic material completely and/or the organic phase was simply over-saturated. So far, effluent consisting more or less of a 1:1 ratio between MSA and water was used. As a next step we wanted to apply an aqueous back-wash of the extract using a water feed of 10 mL/min. To counter any dilution, the effluent was diluted with only 36.86 g/min water, instead of the previously used 47.47 g/min. Additionally, we increased the MIBK feed to 20 mL/min to handle the 4-HAP losses seen in the previous experiment. The results of this extraction are listed in Table 14. Table 14: Results of the extraction using 50 mL/min for effluent- and 20 mL/min MIBK feed, and 10 mL/min H2O as back-wash, MIBK dispersed phase.
The doubled volumetric feed of MIBK resulted in a 4-HAP content of 15.3 wt%. Due to the higher feed a reduced 4-HAP content was expected. With the additionally installed aqueous backwash, the MSA content was slightly reduced and determined to be 0.4 wt%. Since the used extraction setup only has one mixer unit for backwashing the extract, more stages should allow a better MSA removal. However, with the extraction column available for the experiments this would reduce the number of extraction units available for the effluent extraction. Consequently, an incomplete extraction would result or a significantly higher MIBK feed would be required in the current set-up. The MSA content in the raffinate dropped slightly to 49 wt% as a result of the used back-wash. This raffinate contained only minor quantities of 0.02 wt% 4-HAP. Even though a more concentrated effluent was applied, the temperature in this experiment remained between 25-27 °C, indicating no further temperature increase caused by dilution inside the column.
Since the backwash decreased the MSA content in the extract, but we could not use an additional extraction stage, we slightly increased the back-wash feed to 15 mL/min. To remove the last traces of 4-HAP we increased the MIBK flow rate once more to 25 mL/min. The results are listed in Table 15. With a slightly increased backwash rate it was possible to significantly reduce the MSA content in the extract to 0.09 wt%. Based on the higher water feed, the MSA content in the raffinate was slightly decreased to 42 wt%, but due to the increased MIBK flow rate, only traces of 4-HAP could be found in it. Table 15: Results of the extraction using 50 mL/min for effluent- and 25 mL/min MIBK feed, and 15 mL/min H2O as back-wash, MIBK dispersed phase.
In a further trial, we varied the feed rate of the effluent and decided to decrease it slightly to 45 mL/min and increase it in another experiment to 71 mL/min. Based on the previous results, a water feed of 15 mL/min was used to ensure sufficient backwashing. To find a balance between a high 4-HAP content in the extract and as little 4-HAP in the raffinate as possible, a MIBK feed of 20 mL/min was used as compromise. The results of both experiments are compared in Table 16. In case of a lower effluent feed rate of 45 mL/min, only 0.05 wt% MSA and 9.3 wt% 4-HAP were found in the extract, whereas an increased feed rate of 71 mL/min resulted in MSA- and 4- HAP contents of 0.21 wt% and 14.9 wt%, respectively. The results of the extraction experiments allow the following conclusion: an efficient extraction of 4-HAP from the aqueous effluent with MIBK is possible and the extract can be enriched to at least up to 20 wt% with 4-HAP. The losses of 4-HAP to the raffinate can be minimized as long as sufficient MIBK feed or number of extraction stages are used. The same is true for backwashing the extract with water to recover residual amounts of MSA. Such aqueous backwash prevents 0.5 wt% MSA to be lost with the extract. As soon as the extract is backwashed with water, the losses of MSA towards the extract can be reduced. The higher the water feed, the lower the residual amounts of MSA. However, this will result in a lower MSA concentration in the raffinate. Table 16: Results of the extraction using either 45 or 71 mL/min for effluent- and 20 mL/min MIBK feed, and 15 mL/min H2O as back-wash, MIBK dispersed phase.
5. Conclusions A continuous flow synthesis for the production of 4-HAP was developed. After several reaction parameters were analyzed, a residence time of 45 s, a reaction temperature of 120 °C on the HTF side of the reactor, and 12 molar equivalents of MSA were found to be preferred for the production of 4-HAP. These reaction conditions allowed full conversion of phenyl acetate leading to an assay yield of 82.5 % 4-HAP along with 11.9 % 2-HAP and 8.1 % phenol. Moreover, it was found that the water content of MSA has an effect on the reaction outcome. If MSA was dried with 0.15 equiv. Ac2O compared to PA prior to use, the reaction outcome was improved further. In this case assay yields of 4-HAP, 2-HAP, and phenol of 83.7 %, 10.7 %, and 1.9 % were obtained, respectively. These conditions were successfully applied in a long-run over > 8 h, resulting in the formation of 2.6 kg 4-HAP and a mass balance of 96.3 %. As a result of these findings, it can be concluded that it is preferred to apply MSA with a low water content (< 2 wt%, preferably <1 wt%). Ideally, such values are achieved in the MSA recycling process, but if not, addition of acetic anhydride may be considered for in situ drying. After establishing a continuous flow synthesis for 4-HAP, the effluents were used to investigate a continuous extraction of the organic products out of the aqueous effluent using MIBK. In these trials, we were able to determine some key values and parameters for the extraction. By optimizing the amount of effluent, water and MIBK the extraction could be steered towards full recovery of the MSA in a highly concentrated (50 wt% MSA) aqueous phase loosing negligible amounts of 4-HAP. Alternatively, an MIBK extract containing up to 27 wt% of 4-HAP with hardly any residual MSA present could be achieved in case less MIBK was used albeit that the raffinate still contained significant amounts of product in this case. Due to limitations of the used extraction column it was not possible to achieve both full recovery of 4-HAP in a concentrated organic phase as well as full recovery of the MSA in a concentrated aqueous phase at the same time. However, the results of the experiments provide a solid basis to design a column that is suitable for that
purpose; it should be noted that a low content of MSA in the organic phase prevents degradation of the product during distillation, and a backwash of the organic layer (e.g. in a second extraction column) is therefore preferred. Aging experiments on the diluted reaction mixture showed that 4-HAP can be directly obtained from the effluent in high purity by crystallization. Although only a modest yield of less than 50% could be obtained in this way it does show that crystallization is in principle an option to separate 4-HAP from 2-HAP at a later stage of the DSP if necessary. Thus, we can conclude that a continuous flow process followed by a continuous extraction is technically feasible to produce a crude 4-HAP stream in MIBK. 6. Experimental methods The amount of 2-HAP and 4-HAP was determined by HPLC using an Inertsil ODS-3150 x 4.6 5µm column. IR spectra were recorded on a Bruker Alpha II Compact FT-IR spectrometer using 32 scans, absorbance mode, a range of 3000 to 600 cm-1 and a resolution of 4 cm-1. The water content was determined by Karl-Fischer titration as follows. A 10 wt% solution of MSA in anhydrous pyridine was prepared. After complete addition of MSA, the reaction mixture was thoroughly shaken and the formed salts were removed by filtration using a syringe filter. The obtained pyridine solution was subjected to Karl-Fischer titration. The found water-content was corrected for the blank value of pyridine. The amount of MSA found in the MIBK extract was determined by titration. A sample (~5 mL) of the extract was added to 50 mL MIBK by means of a syringe and the exact sample amount was determined by weighing back the syringe. The solution was titrated with 0.1 N TBAOH solution in a isopropanol/MeOH mixture. The found amount was corrected for the blank value of MIBK.