Tetraalkyl Hydroxymethylene-bisphosphonate and Dialkyl 1-Diphenylphosphinoyl-1-hydroxy-ethylphosphonate Derivatives by the Pudovik Reaction and Their Rearranged Products

The reaction of diethyl α-oxoethylphosphonate and diethyl oxobenzylphosphonate with diethyl phosphite, dimethyl phosphite, and diphenylphosphine oxide affords, depending on the substrates and conditions (nature and quantity of the amine catalyst, temperature, and solvent), the Pudovik adduct and/or the corresponding >P(O)–CH–O–P(O)< product formed by rearrangement. The nature of the substituent on the central carbon atom (a methyl or phenyl group) influences the inclination for the rearrangement. The asymmetric products (either adducts or rearranged species) with different P(O)Y functions (Y = RO or Ph) exhibit interesting NMR features.


Introduction
Hydroxymethylene bisphosphonic (dronic) acid derivatives are of importance due to their beneficial effect in the treatment of bone diseases [1][2][3][4]. Their synthesis, starting from the corresponding substituted acetic acid and phosphorus trichloride/phosphorous acid, was thoroughly investigated [1,5,6]. Another approach is the Pudovik reaction (vide infra), the original version of which involves the addition of dialkyl phosphites onto the carbonyl carbon of oxo compounds, such as aldehydes and ketones [7,8].
In 1956 McConnell and Coover described that the solvent-free reaction of diethyl α-oxoethylphosphonate with diethyl phosphite using diethylamine as the catalyst afforded the adduct tetraethyl α-hydroxy-α-ethylidenebisphosphonate [9]. The product was claimed to be identified by boiling point, refractive index, elemental analysis and IR spectral data. Six years later, Fitch and Moedritzer proved that the corresponding phosphonate-phosphate was formed in the above experiment [10]. The principal proof was the 31 P-NMR spectrum, as the species under discussion exhibited two δ P shifts at −1.3 and 20. 2 Nicholson and Vaughn prepared an analogous oxophosphonate-dimethyl phosphite adduct at 0 • C using diethyl ether as the solvent, and 0.05 equivalents of dibutylamine as the catalyst [11]. After thirty years, Vepsäläinen and Turhanen and their co-workers prepared a series of symmetrical and asymmetrical tetraalkyl 1-hydroxyethylidene-1,1-bisphosphonates, making use of the Nicholson method. The products obtained in variable yields (16-92%) are shown in Figure 1. There was no mention of the rearrangement [12]. A few tri-, P,Pdi-, P,P-di-, and mono esters of 1-hydroxyethylidene-bisphosphonic derivatives were also described [13]. Later on, both the diethyl α-oxoethylphosphonate-diethyl phosphite adduct and the related rearranged product could be selectively prepared in a solvent-free microwaveassisted manner, by applying a temperature of 120 °C for 20 min together with 5% of diethylamine (or dibutylamine) catalyst, or 50% dibutylamine additive, respectively [14,15]. Hammerschmidt and co-workers investigated the stereochemical course of the α-hydroxyphosphonate-phosphate rearrangement in general, and they proved that this transformation proceeds with the retention of configuration [16,17].
It was a challenge for us to investigate the possibilities for the fine tuning of this reaction, and to expand its scope by synthesizing mixed derivatives comprising the (EtO)2P(O)CZ(OH)P(O)Y2 scaffold, where Y = alkoxy or Ph and Z = Me or Ph.

Results and Discussion
In the first case, diethyl α-oxoethylphosphonate obtained by the Arbuzov reaction of acetyl chloride and triethyl phosphite, was subjected to the Pudovik reaction with diethyl phosphite in the presence of diethylamine, or dibutylamine under different conditions (Table 1). Besides the expected tetraethyl α-hydroxy-methylphosphonate (2), the rearranged product (3) was also formed. Performing the addition in diethyl ether at 0 °C for 8 h and applying Et2NH in quantities of 5%, 20% and 40%, the ratio of products 2 and 3 was 100-0, 98-2 and 87-13, respectively (Table 1, entries 1-3). Bu2NH was also a suitable catalyst: in the presence of 5% amine, the adduct 2 was obtained in a selectivity of 99% (Table  1, entry 4). Then, in the hope of shifting the product ratio towards favoring the rearranged species 3, we carried out the reactions under solvent-free conditions at higher temperatures. Applying 40% Et2NH at 120 °C and 135 °C for 20 min, the ratio of species 2 and 3 was 66-34 and 51-49, respectively (Table 1, entries 5 and 6). On further heating at 135 °C for 3 h, the ratio of the two components (2 and 3) practically reversed (66-34% changed to 32-68%) ( Table 1, entry 5 and footnote "b"). Bu2NH (5%) remained selective with respect to the addition, as after a reaction at 120 °C for 20 min, the ratio of products 2 and 3 was 94-6 (Table 1, entry 7). In order to shift the ratio in favor of the rearranged product 3, the experiment that comprised boiling the reagents in toluen, in the presence of 40% Et2NH for 7 h was the best choice (Table 1, entry 8). It is noteworthy that, for a similar reaction performed in the presence of only 20% Et2NH for 5 h, the ratio of the components reversed, as the ratio of adduct 2 to rearranged product 3 was 88-12 (Table 1, entry 9). Products 2 and 3 were obtained from the most successful experiments (Table 1, entries 1 and 8, respectively) in a yield of 86% and 75%, respectively, by column chromatography. It is noted that product 2 is the tetraethyl ester of etidronic acid, which is a first-generation dronic acid. Similar rearrangements were also observed during the base-promoted transformation of β-hydroxyphosphine oxides [18]. Later on, both the diethyl α-oxoethylphosphonate-diethyl phosphite adduct and the related rearranged product could be selectively prepared in a solvent-free microwave-assisted manner, by applying a temperature of 120 • C for 20 min together with 5% of diethylamine (or dibutylamine) catalyst, or 50% dibutylamine additive, respectively [14,15]. Hammerschmidt and co-workers investigated the stereochemical course of the α-hydroxyphosphonatephosphate rearrangement in general, and they proved that this transformation proceeds with the retention of configuration [16,17].
It was a challenge for us to investigate the possibilities for the fine tuning of this reaction, and to expand its scope by synthesizing mixed derivatives comprising the (EtO) 2 P(O)CZ(OH)P(O)Y 2 scaffold, where Y = alkoxy or Ph and Z = Me or Ph.

Results and Discussion
In the first case, diethyl α-oxoethylphosphonate obtained by the Arbuzov reaction of acetyl chloride and triethyl phosphite, was subjected to the Pudovik reaction with diethyl phosphite in the presence of diethylamine, or dibutylamine under different conditions (Table 1). Besides the expected tetraethyl α-hydroxy-methylphosphonate (2), the rearranged product (3) was also formed. Performing the addition in diethyl ether at 0 • C for 8 h and applying Et 2 NH in quantities of 5%, 20% and 40%, the ratio of products 2 and 3 was 100-0, 98-2 and 87-13, respectively (Table 1, entries 1-3). Bu 2 NH was also a suitable catalyst: in the presence of 5% amine, the adduct 2 was obtained in a selectivity of 99% (Table 1, entry 4). Then, in the hope of shifting the product ratio towards favoring the rearranged species 3, we carried out the reactions under solvent-free conditions at higher temperatures. Applying 40% Et 2 NH at 120 • C and 135 • C for 20 min, the ratio of species 2 and 3 was 66-34 and 51-49, respectively (Table 1, entries 5 and 6). On further heating at 135 • C for 3 h, the ratio of the two components (2 and 3) practically reversed (66-34% changed to 32-68%) ( Table 1, entry 5 and footnote "b"). Bu 2 NH (5%) remained selective with respect to the addition, as after a reaction at 120 • C for 20 min, the ratio of products 2 and 3 was 94-6 (Table 1, entry 7). In order to shift the ratio in favor of the rearranged product 3, the experiment that comprised boiling the reagents in toluen, in the presence of 40% Et 2 NH for 7 h was the best choice (Table 1, entry 8). It is noteworthy that, for a similar reaction performed in the presence of only 20% Et 2 NH for 5 h, the ratio of the components reversed, as the ratio of adduct 2 to rearranged product 3 was 88-12 (Table 1, entry 9). Products 2 and 3 were obtained from the most successful experiments ( Table 1, entries 1 and 8, respectively) in a yield of 86% and 75%, respectively, by column chromatography. It is noted that product 2 is the tetraethyl ester of etidronic acid, which is a first-generation dronic acid. Similar rearrangements were also observed during the base-promoted transformation of β-hydroxyphosphine oxides [18].  It was a challenge for us to prepare the hydroxymethylene bisphosphonates with mixed ester functions. Therefore, diethyl α-oxoethylphosphonate (1) was reacted with dimethyl phosphite in diethyl ether at 0 °C for 8 h in the presence of 5 and 20% Et2NH. To our surprise, the outcome of these two experiments was quite different: while in the first case, the expected adduct 4 was the major component (94%) ( Table 2, entry 1), in the second case, the rearranged products (5-1 and 5-2) predominated (71% and 17%), ( Table 2, entry 2). The Pudovik reaction was much less selective with dimethyl phosphite than with the diethyl counterpart (compare entry 2 of Table 2 with entry 2 of Table 1). The application of 5% Bu2NH led also to a selective addition, as adduct 4 was exclusively formed ( Table 2, entry 3). The use of 40% Et2NH in diethyl ether at 0 °C for 8 h led to the predominant formation of the rearranged products 5-1 (76%) and 5-2 (21%) all together 97% ( Table  2, entry 4). At the same time, the application of 5% DBA without any solvent at 120 °C for 20 min afforded the adduct 4 in a selectivity of 79% (Table 2, entry 5). To promote the rearrangement, the reaction of α-oxophosphonate (1) with dimethyl phosphite was performed in boiling toluene in the presence of 20% Et2NH for 5 h. Indeed, the rearranged products (5-1 and 5-2) predominated at 85% (Table 2, entry 6).
One can see that the nature of the amine (DEA or DBA) and its quantity (5-40%) have a major impact on the outcome of the reaction of diethyl α-oxoethylphosphonate (1) and dialkyl phosphite, while the temperature, as well as the use or lack of solvent, has a lesser effect. Adduct 4 and the mixture of rearranged products (5-1 and 5-2) were prepared from the best experiments ( It was a challenge for us to prepare the hydroxymethylene bisphosphonates with mixed ester functions. Therefore, diethyl α-oxoethylphosphonate (1) was reacted with dimethyl phosphite in diethyl ether at 0 • C for 8 h in the presence of 5 and 20% Et 2 NH. To our surprise, the outcome of these two experiments was quite different: while in the first case, the expected adduct 4 was the major component (94%) ( Table 2, entry 1), in the second case, the rearranged products (5-1 and 5-2) predominated (71% and 17%), ( Table 2, entry 2). The Pudovik reaction was much less selective with dimethyl phosphite than with the diethyl counterpart (compare entry 2 of Table 2 with entry 2 of Table 1). The application of 5% Bu 2 NH led also to a selective addition, as adduct 4 was exclusively formed ( Table 2, entry 3). The use of 40% Et 2 NH in diethyl ether at 0 • C for 8 h led to the predominant formation of the rearranged products 5-1 (76%) and 5-2 (21%) all together 97% ( Table 2, entry 4). At the same time, the application of 5% DBA without any solvent at 120 • C for 20 min afforded the adduct 4 in a selectivity of 79% (Table 2, entry 5). To promote the rearrangement, the reaction of α-oxophosphonate (1) with dimethyl phosphite was performed in boiling toluene in the presence of 20% Et 2 NH for 5 h. Indeed, the rearranged products (5-1 and 5-2) predominated at 85% (Table 2, entry 6).
One can see that the nature of the amine (DEA or DBA) and its quantity (5-40%) have a major impact on the outcome of the reaction of diethyl α-oxoethylphosphonate (1) and dialkyl phosphite, while the temperature, as well as the use or lack of solvent, has a lesser effect. Adduct 4 and the mixture of rearranged products (5-1 and 5-2) were prepared from the best experiments (Table 2, entries 3 and 4) by column chromatography in yields of 87% and 76%, respectively. Products 4, 5-1, and 5-2 exhibited 31 P-NMR spectra comprising doublet patterns for each signal (see below and Experimental).  Products 4, 5-1, and 5-2 exhibited 31 P-NMR spectra comprising doublet patterns for each signal (see below and Experimental).
Then, diethyl oxobenzylphosphonate (6) was reacted with diethyl phosphite in the presence of 5% Et2NH and 5% Bu2NH in diethyl ether at 0 °C for 8 h. In the first case, exclusively the diethyl phosphonobenzylphosphate (8) formed by rearrangement of the Pudovik adduct (7) was present in the mixture (Scheme 1). It is noteworthy that when using Bu2NH as the catalyst, a mixture of 20% hydroxy-bisphosphonate (7) and 80% phosphono-phosphate (8) was obtained. Interrupting the Bu2NH-enhanced reaction of oxophosphonate (6) with of diethyl phosphite after 3 h (at 0 °C), the ratio of the components was reversed as there was 95% of adduct 7 and 5% of phosphono-phosphate (8)  Then, diethyl oxobenzylphosphonate (6) was reacted with diethyl phosphite in the presence of 5% Et 2 NH and 5% Bu 2 NH in diethyl ether at 0 • C for 8 h. In the first case, exclusively the diethyl phosphonobenzylphosphate (8) formed by rearrangement of the Pudovik adduct (7) was present in the mixture (Scheme 1). It is noteworthy that when using Bu 2 NH as the catalyst, a mixture of 20% hydroxy-bisphosphonate (7) and 80% phosphono-phosphate (8) was obtained. Interrupting the Bu 2 NH-enhanced reaction of oxophosphonate (6) with of diethyl phosphite after 3 h (at 0 • C), the ratio of the components was reversed as there was 95% of adduct 7 and 5% of phosphono-phosphate (8)   Products 4, 5-1, and 5-2 exhibited 31 P-NMR spectra comprising doublet patterns for each signal (see below and Experimental).
Then, diethyl oxobenzylphosphonate (6) was reacted with diethyl phosphite in the presence of 5% Et2NH and 5% Bu2NH in diethyl ether at 0 °C for 8 h. In the first case, exclusively the diethyl phosphonobenzylphosphate (8) formed by rearrangement of the Pudovik adduct (7) was present in the mixture (Scheme 1). It is noteworthy that when using Bu2NH as the catalyst, a mixture of 20% hydroxy-bisphosphonate (7) and 80% phosphono-phosphate (8) was obtained. Interrupting the Bu2NH-enhanced reaction of oxophosphonate (6) with of diethyl phosphite after 3 h (at 0 °C), the ratio of the components was reversed as there was 95% of adduct 7 and 5% of phosphono-phosphate (8)  Complete rearrangement followed the Pudovik reaction of oxobenzylphosphonate (6) with dimethyl phosphite using 5% of the secondary amine catalyst in diethyl ether solution at 0 • C (8 h) (Scheme 2). Due to the asymmetry of the intermediate 9, the dialkyl phosphonobenzyl phosphate was formed as two isomers (10-1 and 10-2). Using Et 2 NH the yield was 81%, while the isomeric ratio was 80-20%. phosphonobenzyl phosphate was formed as two isomers (10-1 and 10-2). Using Et2NH the yield was 81%, while the isomeric ratio was 80-20%.
The assignment of 5-1 and 10-1, as well as 5-2 and 10-2 as the major and minor isomers, respectively, is tentative. However, considering the electrophilicity of the P atom in the (MeO)2P(O) and the (EtO)P(O) groups, the former is assumed to be more capable of being attacked by the hydroxy group, as the MeO substituent is somewhat less electrondonating than the EtO one. At the same time, steric effects may also play a role in the observed selectivity. The adduct with a C-Ph substituent (9) has an increased inclination for the rearrangement than the adduct with a C-Me unit (4) due to electronic effects.
In the final part of our experimental work, α-oxophosphonates 1 and 6 were reacted with diphenylphosphine oxide. While the interaction of α-oxoethylphosphonate (1) with diphenylphosphine oxide at 0 °C in the presence of 20% diethylamine in diethyl ether afforded the Pudovik adduct 11 (Scheme 3). The similar reaction of oxobenzylphosphonate (6) led to a 1:1 mixture of the two possible rearranged products; phosphinoyloxybenzylphosphonate (13-1), and phosphinoylbenzylphosphate (13-2) (Scheme 4). The ratio of isomers 13-1 and 13-2 was in accordance with the electrophilicity of the Ph2P(O) and (EtO)2P(O) moieties during rearrangement. However, due to the more sterically hindered Ph2P(O) moiety, the ratio of the isomers is more balanced (40-60%) compared to that of 10-1 and 10-2 (20-80%), due to the way that formation of a P-O bond occurs with the liberation of 385 kJ mol -1 [19].
Again, the adduct containing a phenyl group on the central carbon atom (as in 12) seemed to be less stable compared to the methyl analogue 11. Scheme 3. The reaction of diethyl α-oxoethylphosphonate (1) with diphenylphosphine oxide.  The assignment of 5-1 and 10-1, as well as 5-2 and 10-2 as the major and minor isomers, respectively, is tentative. However, considering the electrophilicity of the P atom in the (MeO) 2 P(O) and the (EtO)P(O) groups, the former is assumed to be more capable of being attacked by the hydroxy group, as the MeO substituent is somewhat less electron-donating than the EtO one. At the same time, steric effects may also play a role in the observed selectivity. The adduct with a C-Ph substituent (9) has an increased inclination for the rearrangement than the adduct with a C-Me unit (4) due to electronic effects.
In the final part of our experimental work, α-oxophosphonates 1 and 6 were reacted with diphenylphosphine oxide. While the interaction of α-oxoethylphosphonate (1) with diphenylphosphine oxide at 0 • C in the presence of 20% diethylamine in diethyl ether afforded the Pudovik adduct 11 (Scheme 3). The similar reaction of oxobenzylphosphonate (6) led to a 1:1 mixture of the two possible rearranged products; phosphinoyloxybenzylphosphonate (13-1), and phosphinoylbenzylphosphate (13-2) (Scheme 4). The ratio of isomers 13-1 and 13-2 was in accordance with the electrophilicity of the Ph 2 P(O) and (EtO) 2 P(O) moieties during rearrangement. However, due to the more sterically hindered Ph 2 P(O) moiety, the ratio of the isomers is more balanced (40-60%) compared to that of 10-1 and 10-2 (20-80%), due to the way that formation of a P-O bond occurs with the liberation of 385 kJ mol -1 [19].
The assignment of 5-1 and 10-1, as well as 5-2 and 10-2 as the major and minor isomers, respectively, is tentative. However, considering the electrophilicity of the P atom in the (MeO)2P(O) and the (EtO)P(O) groups, the former is assumed to be more capable of being attacked by the hydroxy group, as the MeO substituent is somewhat less electrondonating than the EtO one. At the same time, steric effects may also play a role in the observed selectivity. The adduct with a C-Ph substituent (9) has an increased inclination for the rearrangement than the adduct with a C-Me unit (4) due to electronic effects.
In the final part of our experimental work, α-oxophosphonates 1 and 6 were reacted with diphenylphosphine oxide. While the interaction of α-oxoethylphosphonate (1) with diphenylphosphine oxide at 0 °C in the presence of 20% diethylamine in diethyl ether afforded the Pudovik adduct 11 (Scheme 3). The similar reaction of oxobenzylphosphonate (6) led to a 1:1 mixture of the two possible rearranged products; phosphinoyloxybenzylphosphonate (13-1), and phosphinoylbenzylphosphate (13-2) (Scheme 4). The ratio of isomers 13-1 and 13-2 was in accordance with the electrophilicity of the Ph2P(O) and (EtO)2P(O) moieties during rearrangement. However, due to the more sterically hindered Ph2P(O) moiety, the ratio of the isomers is more balanced (40-60%) compared to that of 10-1 and 10-2 (20-80%), due to the way that formation of a P-O bond occurs with the liberation of 385 kJ mol -1 [19].
Again, the adduct containing a phenyl group on the central carbon atom (as in 12) seemed to be less stable compared to the methyl analogue 11. Scheme 3. The reaction of diethyl α-oxoethylphosphonate (1) with diphenylphosphine oxide.  phosphonobenzyl phosphate was formed as two isomers (10-1 and 10-2). Using Et2NH the yield was 81%, while the isomeric ratio was 80-20%.
The assignment of 5-1 and 10-1, as well as 5-2 and 10-2 as the major and minor isomers, respectively, is tentative. However, considering the electrophilicity of the P atom in the (MeO)2P(O) and the (EtO)P(O) groups, the former is assumed to be more capable of being attacked by the hydroxy group, as the MeO substituent is somewhat less electrondonating than the EtO one. At the same time, steric effects may also play a role in the observed selectivity. The adduct with a C-Ph substituent (9) has an increased inclination for the rearrangement than the adduct with a C-Me unit (4) due to electronic effects.
In the final part of our experimental work, α-oxophosphonates 1 and 6 were reacted with diphenylphosphine oxide. While the interaction of α-oxoethylphosphonate (1) with diphenylphosphine oxide at 0 °C in the presence of 20% diethylamine in diethyl ether afforded the Pudovik adduct 11 (Scheme 3). The similar reaction of oxobenzylphosphonate (6) led to a 1:1 mixture of the two possible rearranged products; phosphinoyloxybenzylphosphonate (13-1), and phosphinoylbenzylphosphate (13-2) (Scheme 4). The ratio of isomers 13-1 and 13-2 was in accordance with the electrophilicity of the Ph2P(O) and (EtO)2P(O) moieties during rearrangement. However, due to the more sterically hindered Ph2P(O) moiety, the ratio of the isomers is more balanced (40-60%) compared to that of 10-1 and 10-2 (20-80%), due to the way that formation of a P-O bond occurs with the liberation of 385 kJ mol -1 [19].
Again, the adduct containing a phenyl group on the central carbon atom (as in 12) seemed to be less stable compared to the methyl analogue 11. Scheme 3. The reaction of diethyl α-oxoethylphosphonate (1) with diphenylphosphine oxide. Again, the adduct containing a phenyl group on the central carbon atom (as in 12) seemed to be less stable compared to the methyl analogue 11.
Adducts 2, 4, and 11, as well as rearranged products 3, 5, 8, 10, and 13 were characterized by 31 P, 13 C and 1 H-NMR, as well as with HRMS data. With the exception of compounds 2, 3 and 4, all other species are new derivatives.
The 31 P-NMR spectral data were in accord with the different types of compounds synthesized. The symmetrical tetraethyl Pudovik adducts 2 and 7 revealed a signal at δ P 20.3 and 17.6, respectively, while the diethyl-dimethyl derivative 4 appeared at δ P 20.0 and 22.8 with a 2 J PP of 35.0 Hz. The adduct with (EtO) 2 P(O) and Ph 2 P(O) functions (11) had signals at δ P 21.8 and 29.1 with a 2 J PP of 22.4 Hz.

General
The 31 P, 13 C, 1 H-NMR spectra were taken on a Bruker DRX-500 spectrometer operating at 202.4, 125.7, and 500 MHz, respectively. The couplings are given in Hz. LC-MS measurements were performed with an Agilent 1200 liquid chromatography system, coupled with a 6130 quadrupole mass spectrometer equipped with an ESI ion source (Agilent Technologies, Palo Alto, CA, USA). High-resolution mass spectrometric measurements were performed using a Thermo Velos Pro Orbitrap Elite hybrid mass spectrometer in positive electrospray mode.
For the 31 P, 13 C and 1 H-NMR spectra of the compounds prepared see Supplementary Materials.

Conclusions
In summary, the Pudovik reaction between α-oxophosphonates (ZC(O)P(O)(OEt) 2 , Z = Me or Ph) and Y 2 P(O)H reagents (Y = EtO, MeO, and Ph) may lead to the corresponding adducts (Y 2 P(O)C(OH)ZP(O)(OEt) 2 ) and/or their rearranged versions. The outcome mostly depended on the Z substituent, the quantity of the dialkylamine (DAA) catalyst, and, to a lesser extent, on the nature of the DAA and Y substituents, as well as on the temperature and the solvent. In a few cases, time also had an influence on the course of the reaction. In cases where Z = Me, the adducts were the primary products, but with suitable modifications the reactions could be tuned to yield the rearranged derivatives. At the same time, in cases where Y = Ph, the corresponding adducts were only intermediates that were converted spontaneously to their rearranged versions. This phenomenon was explained by electronic factors. In reaction with dimethyl phosphite and diphenylphosphine oxide, the rearranged species comprised two isomers.