Elisabet
Pires
and
José M.
Fraile
*
Instituto de Síntesis Química y Catálisis Homogénea (ISQCH), Facultad de Ciencias, CSIC-Universidad de Zaragoza, Pedro Cerbuna 12, E-50009 Zaragoza, Spain. E-mail: josem.fraile@csic.es
First published on 2nd October 2020
The variation of the 31P chemical shift of triethylphosphine oxide in CDCl3 solution with a series of Brønsted acids at different molar ratios allows the determination of the value for the 1:1 species (δ1:1), which is much lower than the reported value at infinite dilution. This value correlates with the pKa of the acid in two zones, for acids stronger and weaker than TEPO–H+. The acid strength also controls the exchange rate in solution. The evolution of the chemical shift at high acid/TEPO molar ratios indicates the existence of a second TEPO–acid interaction, which is also dependent on the acid strength. This interaction is much more favorable in the case of a diacid, which shows chemical shift higher than expected for its pKa1 value.
The 31P NMR method with TEPO was soon adapted for the characterization of solids, creating a scale of global acidity.9 Although TEPO was initially used, in analogy with the AN determination, other trialkylphosphine oxides (PO), such as trimethylphosphine oxide (TMPO) or tributylphosphine oxide (TBPO), have also been utilized for this purpose, with the aim of getting information not only about acid strength, but also the acid type (Brønsted or Lewis), the location and the amount of acid sites.10,11
However, the deshielding effect on trialkylphosphine oxides is measured under very different conditions in both methods. In solution, the high dilution conditions and the extrapolation to calculate Gutmann's AN make the acid/PO ratio nearly ∞, considering the acid as a solvent, whereas on solids, the amount of PO used is close to or slightly below the stoichiometric 1:1 ratio. Surprisingly, the studies of acids in solution under similar quasi-stoichiometric conditions are rather scarce.
Myers et al. characterized BF3, TMSOTf (trimethylsilyl triflate) and mixtures of both in CDCl3 by using TEPO in amounts from 0.25 equivalents to a slight excess over the Lewis acids.12 Several interesting features were observed, such as the presence of several species with different Lewis acidity and the rather slow exchange of TEPO, as the 31P signals for free and coordinated TEPO did not collapse into one single peak at room temperature.
Koito et al. have studied the acidity of different Lewis acids, mainly metal triflates and chlorides, in D2O with acid/TMPO molar ratios in the range of 0.5 to 8.13,14 Several catalysts did not show any important deshielding effect on TMPO, showing that most of the TMPO was not coordinated to the Lewis acid. On the contrary, Sc(OTf)3, In(OTf)3 and ScCl3 showed an important deshielding effect, which is in agreement with their higher catalytic activity in water. The number of signals and the line width at different acid/TMPO ratios were used to determine the rather slow exchange rates.
Very recently, Diemoz and Franz have used a TEPO 31P chemical shift to create a scale of the hydrogen-bond activating effect of different organocatalysts, including alcohols, phenols, silanols, carboxylic acids, ureas, phosphoric acids and boronic acids.15 In this case, a catalyst/TEPO molar ratio of 3 was consistently used to get the saturation 31P chemical shift value. However, this study was restricted to rather weak acids.
Thus, a fair comparison of the 31P chemical shifts of analogous homogeneous and heterogeneous catalysts is difficult. This is due to the very different conditions in which the 31P chemical shifts are measured. In this manuscript, we present our results of 31P NMR spectroscopy in solution with TEPO as a probe molecule and different Brønsted acids that can serve as homogeneous analogues of heterogeneous catalysts.
For the preparation of the adducts, a solution of 250 mg of TEPO in anhydrous CH2Cl2 or CH3OH was prepared in a 25 mL volumetric flask. Then, p-toluenesulfonic acid or phenylphosphonic acid was weighed in a 10 mL round bottom flask, and the adequate volume of TEPO solution was added. The solution was stirred under argon for 30 min, and the solvent was removed by vacuum distillation. Dried KBr was added to the viscous liquid, and the solid was placed in a 4 mm ZrO2 rotor.
Solid state NMR spectra were recorded in a Bruker Avance III WB400 spectrometer with 4 mm zirconia rotors spun at the magic angle in N2 at 10 kHz. 31P NMR spectra (64 scans) were recorded using a 31P π/2 pulse length of 4.3 μs, with a spinal-64 proton decoupling sequence of 5 μs pulse length, and 30 s recycle delay. Pulses and chemical shifts were calibrated with (NH4)H2PO4.
Solvent | ANa | 31P δANa | 31P δ1:1b | pKac |
---|---|---|---|---|
a Gutmann's acceptor number and the corresponding value of 31P chemical shift of dissolved TEPO. b 31P chemical shift in CDCl3 for a TEPO–HA (1:1) species, calculated from the slope at low acid/TEPO molar ratio. c Data from ref. 16 unless otherwise stated. d From ref. 18. e From ref. 19. f Value for benzenesulfonic acid. g From ref. 20. | ||||
CHCl3 | 23.1 | 52.5 | — | — |
MeOH | 41.3 | 60.3 | — | — |
TFE | 53.8 | 65.6 | 54.6 | 12.4 |
AcOH | 52.9 | 65.2 | 56.5 | 4.76 |
HCOOH | 83.6 | 78.3 | 57.9 | 3.77 |
PhPO3H2 | — | — | 59.3 | 1.83d |
TFA | 105.5 | 87.6 | 62.8 | 0.23e |
MeSO3H | 126.1 | 96.4 | 74.3 | −1.92 |
pTosOH | — | — | 76.1 | −2.8f |
TfOH | 129.1 | 97.6 | 94.8 | −5.9 (−14.7g) |
Most of the pKa values (Table 1) are collected from the classical reference by Guthrie,16 with some exceptions, such as TFE,17 PhPO3H218 and TFA.19 A lower pKa value of TfOH (−14.7), based on more recent theoretical calculations20 has also been considered.
Methanol-d4 was first envisaged as a solvent due to the rather low solubility of some acids in CDCl3. However, the preliminary results obtained in methanol-d4 (Fig. S1, ESI†) indicated that this solvent was not suitable for this kind of measurement, as no shift was observed with weak acids, and no differences were observed among the strong acids, probably because the protonating species was [CD3ODH]+ in all of the cases. This effect is similar to that observed with Lewis acids and TMPO in D2O,14 with two general behaviors for highly active and poor catalysts. Thus, all of our studies were carried out in CDCl3, as a weakly polar medium, to simulate the acid–TEPO interaction on the solids in the absence of solvent.
In spite of the presumably fast acid–base equilibrium in solution, broad unsymmetrical 31P signals were obtained in some cases that became narrower single lines upon standing (some examples are collected in Fig. S2, ESI†). Hence, as a general procedure, the NMR samples were left to stand for a longer time (6–24 h) until a stable and reproducible 31P spectrum was obtained.
The evolution of the 31P chemical shift with the acid/TEPO molar ratio (spectra are collected in Fig. S3–S7, ESI†), in the case of weak acids, is represented in Fig. 1, showing a similar trend to the evolution of TMPO 31P δ in D2O with different M(OTf)x/TMPO molar ratios13 and TEPO with a different organocatalyst/TEPO ratio.15 The slope of the curve at low acid/TEPO molar ratios should be dependent on the acid strength (curve fits are collected in Fig. S8–S12, ESI†). Using that slope, the 31P δ values for a 1/1 molar ratio have been calculated (collected in Table 1, labeled as δ1:1), and they can be considered as the estimated chemical shifts for the different TEPO–HA (1:1) species. These values are in qualitative agreement with the pKa, but they are much lower than those obtained under the Gutmann's conditions for AN calculation (δAN). At increasing acid/TEPO molar ratios, the chemical shifts slowly increase linearly to converge with δAN. This result is in contrast with previous observations where the 31P chemical shift remained constant after the saturation of all of the TEPO.15 The slope of this second line is also higher for the stronger acids, although it was not determined for PhPO3H2 and TFE due to solubility problems in CDCl3. With this second straight line, it is possible to calculate a theoretical acid/TEPO ratio to get the 31P δAN. With both acetic and formic acids, molar ratios over 50 would be required to reach the 31P δAN, indicating that a very large excess of weak acid, that is a high dilution of TEPO, is necessary to reproduce the 31P δAN value. The molar ratio is significantly lower in the case of TFA (14.5), but a large excess is still required.
Fig. 1 Variation of the 31P chemical shift (ppm) with the acid/TEPO molar ratio in CDCl3: (○) TFE, (■) AcOH, (×) HCOOH, (△) PhPO3H2, (●) TFA. |
Fig. 2 Variation of the 31P chemical shift (ppm) with the acid/TEPO molar ratio in CDCl3 (A) and linear fitting of the two straight parts of the curves (B): (■) MeSO3H, (×) pTosOH, (▲△) TfOH. |
On the contrary, the behavior of TfOH was significantly different. The δ1:1 value (94.8 ppm) was very close to the δAN (97.6 ppm). Hence, the second slope was much lower than that of MeSO3H (Fig. 2B). Moreover, at TfOH/TEPO molar ratios slightly higher than 1, two signals were clearly detected (open and filled symbols in Fig. 2A), until a single signal was again formed at a molar ratio of 3 (Fig. 3).
The evolution of the chemical shifts in the spectra of the acids has been represented against the TEPO/acid ratio to better understand the changes from pure acid to the saturation with TEPO. In this regard, the proton donation can also be followed by 1H NMR in the case of AcOH, HCOOH and MeSO3H. The formyl proton shows a deshielding effect (Fig. 4A and Fig. S21, ESI†) with a linear relationship with the TEPO/acid ratio, which is in agreement with the increase in the electron density in the carboxylic moiety. On the contrary, the methyl groups of the methanesulfonic (spectra in Fig. S22, ESI†) and acetic acids suffer a shielding effect (Fig. 4B and C, respectively) when the TEPO/acid molar ratio increases. This is in agreement with the development of a negative charge (an effect observed in the spectra of the sulfonate and acetate salts21), although the variation is much more important in the case of MeSO3H. This is probably due to its stronger acidity and hence a larger proton donation degree.
The acidic proton appears in the 5–12 ppm range with variable linewidth (two examples are collected in Fig. S23 and S24, ESI†). However, no clear relationship could be found between the TEPO/acid molar ratio and the chemical shift or linewidth of the signal.
Finally, the 19F chemical shift of the fluorine atoms in TFA (spectra in Fig. S25, ESI†) and TfOH follow the same trend (Fig. 5) as the protons in AcOH and MeSO3H with a shielding effect at increasing amounts of TEPO. This indicates the formation of a salt, and a more intense effect in the case of the stronger acid TfOH. Moreover, a second signal is clearly detected in the spectrum of TfOH at the same TfOH/TEPO ratios, in which the 31P spectrum also shows 2 signals. This is in agreement with the formation of two different species with slow exchange rate.
δ = δ1:1x1:1 + δTEPOxTEPO = δ1:1r + δTEPO(1 − r) = δTEPO + r(δ1:1 − δTEPO) | (1) |
Fig. 6 Dependence of 31P δ1:1 with pKa (values in Table 1). Open symbol corresponds to data of HFIP from ref. 15. |
Thus, the weak acids are not strong enough to produce the protonation of TEPO. In addition, the nature of the TEPO·HA adduct can be envisaged as a hydrogen bond of different strengths, in the order TFE < AcOH < HCOOH < PhPO3H2 < TFA. In fact, from the data reported in the literature,15 the δ1:1 for 1,1,1,3,3,3-hexafluoro-2-propanol (HFIP) can be calculated (open symbol in Fig. 6), and it fits well with the expected value. The hydrogen bonding must generate only a very small positive charge on phosphorus (Scheme 1, equilibrium 1), and hence a very low deshielding effect, demonstrated by a low 31P Δδ. The acids stronger than TFA are able to transfer the proton to TEPO, developing a larger positive charge on phosphorus, and hence a more intense deshielding effect. From the δ1:1 values, two types of behavior can be considered. The superacid TfOH produces the complete proton transfer, forming an ion-pair with TEPO (Scheme 1, equilibrium 3). In addition, a complete deshielding effect is observed on phosphorus in the 1:1 species, as shown by a δ1:1 value that is very close to δAN. The strong acids MeSO3H and pTosOH produce a deshielding effect that is less intense than TfOH. This is due to an incomplete proton transfer (Scheme 1, equilibrium 2), but is much more intense than the weak acids.
A second parameter that allows for distinguishing between different behaviors is the 31P line width. The maximum value is always observed at an acid/TEPO ratio around 0.5, and it varies depending on the acid strength. The values are low (<10 Hz) with weak acids, indicating a fast exchange between the free and hydrogen-bonded TEPO. The maximum line width abruptly increases to values around 100 Hz when the acid strength reaches the TEPO pKa. This is probably due to a slower exchange of the proton-transferred TEPO. The slow exchange has also been described in the case of neutralization of Lewis acids with TMPO in D2O.13 From the maximum value of the line width, it decreases in a linear way with δ1:1 (Fig. 7), which is with increasing acid strength. This seems to allow a faster TEPO exchange, mainly in the case of the ion-paired [TEPO–H]+TfO−.
The evolution of the 31P δ at acid/TEPO molar ratios >1 (Fig. 1 and 2) seems to indicate the existence of a second reaction to form a TEPO–(HA)2 species (Scheme 1). In the case of weak acids, this reaction should be the formation of a second hydrogen bond (Scheme 1, equilibrium 1), leading to a minor increase in δ that requires a very high amount of acid to be detectable. On the other hand, MeSO3H and pTosOH are strong enough to produce a second proton transfer to TEPO (Scheme 1, equilibrium 2), which is able to reach the full deshielding effect at a rather low acid/TEPO ratio (4:1). As TfOH already produces the full deshielding effect with the first protonation equilibrium, the second TfOH molecule might be added to the PO bond, leading to a [Et3POTf]OTf species after water loss (Scheme 1, equilibrium 3). This species has only a minor deshielding effect on 31P with respect to [Et3POH]OTf, and it would also explain the slow exchange, as shown by the two signals of the 31P spectrum at high acid/TEPO molar ratio (Fig. 3), as well as the two signals in the 19F spectrum (Fig. 5B).
A single quite narrow TEPO 31P signal was obtained in all of the cases, with very low chemical shift anisotropy, as shown in the spectra at low spinning speed (up to 0.8 kHz, Fig. S28, ESI†). Then, the first conclusion was that the adducts were still present in the liquid phase on the solid surface. Even in the absence of solvent, a fast exchange of TEPO took place, averaging the signal. The obtained chemical shifts were compared with those observed in the solution phase (Fig. 8).
Fig. 8 Evolution of 31P chemical shift in solution (open symbols) and in solid phase (filled symbols) with acid/TEPO molar ratio: (■) pTosOH; (●) PhPO3H2. |
As can be seen, the chemical shifts at low molar ratio perfectly fit with the initial slopes of the curves in solution, showing the negligible effect of CDCl3 on the chemical shift under those conditions. On the contrary, the effect is significant at molar ratios higher than 1. Conversely, the second slope (molar ratio >1) is much lower than the first one in solution. In the solvent-free liquid phase on the solid, the evolution of the chemical shift is nearly linear up to higher molar ratios. The concentration of the sample after evaporation of the solvent seems to greatly favor the formation of the TEPO–(HA)2 species, irrespective from the acid strength.
As an example, the predicted value for a MeSO3H/HCOOH/TEPO ratio of 0.53:0.62:1 would be 66.6 ppm (0.53 × 74.3 + 0.47 × 57.9 + 0.15 × 0.33, where 0.33 is the slope of the second straight line in the HCOOH graphic), whereas the experimental value is 66.9 ppm.
However, the effect of a large excess of weak acid on the 1:1 species of the strong acid cannot be precisely predicted by the behavior of the pure weak acid. Thus, for a MeSO3H/HCOOH/TEPO ratio of 0.83:1.03:1, the predicted value is 71.8 ppm, whereas the experimental one is 73.7 ppm. Even more difference is observed at larger HCOOH excess (molar ratio of 0.55:11.1:1) between the predicted δ (70.4 ppm) and the experimental value (74.0 ppm). This seems to indicate that the effect of a second molecule of the weak acid is higher when it participates in a mixed 1:2 species with the strong acid, although this point deserves further investigation.
Fig. 9 Proposed structure for the TEPO·DMMA adduct and the analogous TEPO–(HA)2 species with weak acids. |
In solids with isolated acid sites, such as in the case of zeolite HZSM-5, the formation of the (TMPO)2–HA species has been proposed on the basis of 1H–31P HETCOR experiments and DFT calculations.25 However, on solids with high density of acid sites, the presence of neighbor acid sites may produce an effect similar to that observed with DMMA in solution, with the formation of the TEPO–(HA)2 species. This might explain the high dispersion of the 31P chemical shifts observed with the sulfonated carbons,26 leading to an uncertain assignment of the signals,27 a subject that requires further investigation according to these findings in solution.
Footnote |
† Electronic supplementary information (ESI) available: NMR spectra and fitted graphs for all the acids. See DOI: 10.1039/d0cp03812a |
This journal is © the Owner Societies 2020 |