Quantification of Electrophilic Activation by Hydrogen-Bonding Organocatalysts

Abstract

A spectrophotometric sensor is described that provides a useful assessment of the LUMO-lowering provided by catalysts in Diels–Alder and Friedel–Crafts reactions. A broad range of 33 hydrogen-bonding catalysts was assessed with the sensor, and the relative rates in the above reactions spanned 5 orders of magnitude as determined via ¹H- and ²H NMR spectroscopic measurements, respectively. The differences between the maximum wavelength shift of the sensor with and without catalyst (Δλ_max^–1) were found to correlate linearly with ln(k_rel) values for both reactions, even though the substrate feature that interacts with the catalyst differs significantly (ketone vs nitro). The sensor provides an assessment of both the inherent reactivity of a catalyst architecture as well as the sensitivity of the reaction to changes within an architecture. In contrast, catalyst pK_a values are a poor measure of reactivity, although correlations have been identified within catalyst classes.

1. Introduction

The field of small-molecule organocatalysis via noncovalent interactions has seen rampant growth over the past decade.¹ This area, which aims to mimic the mechanisms used by nature in enzyme catalysis, is attractive due to its potential for high catalyst tunability and substrate specificity, as well as obviating the use of metals. While such research has resulted in many catalysts operating through bifunctional mechanisms,² the primary interaction responsible for electrophile activation occurs through hydrogen-bonding to an acceptor moiety. The consequent LUMO-lowering results in rate enhancement. In comparison to metal-based systems (i.e., Lewis Acids),³⁻⁵ current metrics to estimate the reactivity of hydrogen bonding catalysts are ineffective. Although ΔpK_a values of the donor and acceptor may be used to infer hydrogen-bond strengths,⁶ this analysis fails to account for several important secondary interactions, including sterics, dual-activation, and binding geometry.⁷ As a result, the discovery of reactions compatible with hydrogen-bond catalysis is far outpacing understanding of catalyst interaction and mechanism. Indeed, while certain privileged organocatalyst motifs have been identified to be successful for several reaction types, rational design of these structures remains limited, relying on trial and error to achieve optimal reactivity and selectivity.

In a previous communication,⁸ we described preliminary results showing the utility of small organic chromophore S for the detection of hydrogen-bonding interactions by UV–vis spectroscopy for a small set of catalysts (Scheme 1). Herein, we assess the general utility of this colorimetric sensor as a predictive gauge for the relative reactivity of a broad range of organocatalysts, including several widely used motifs, with the goal of encompassing many different hydrogen-bonding arrays. An additional goal was to validate the sensor measurements across significantly different reaction profiles, particularly those involving noncarbonyl electrophiles. The lack of comprehensive rate data for a range of catalysts in different reactions is a barrier to understanding the factors controlling catalytic activation. As a consequence, the relative rates have been measured for an array of catalysts in two reactions with different groups that interact with the catalysts. The sensor signal has been analyzed with respect to reaction profile, catalyst structure, acidity, and acceptor preference. Our findings establish the sensor as a useful substrate “surrogate” for probing and gauging catalyst performance. Moreover, the data unequivocally establish that catalyst structure and binding mode are far more relevant to catalytic activity than acidity.

Scheme 1. UV–Vis Sensor Concept To Detect Weak Interaction.

2. Results and Discussion

2.1. Application of a Colorimetric Probe for Determining LUMO Activation

A key consideration in designing a method for measuring hydrogen-bond strengths is the very broad range of these noncovalent interactions (0.2–40 kcal/mol).⁹ Observing the very weak range of these interactions is a challenge with commonly employed spectroscopic techniques. For example, despite successful application in measuring Lewis Acid binding effects,¹⁰ preliminary NMR studies proved too insensitive for detecting the interactions of weak hydrogen-bonding catalysts with a carbonyl acceptor.¹¹

We proposed an alternative approach using the sensitivity of UV–vis absorption profiles, in which a change in electronic excitation of an acceptor chromophore occurs upon binding to a hydrogen-bond donor (Scheme 1). Specifically, imidazopyrazinone S displays solvatochromism with protic solvents as well as color changes with a small number of Lewis Acids.¹² We postulated that upon treatment with various hydrogen-bond donors, the carbonyl moiety of S would act as an acceptor moiety. The resulting hydrogen-bonding interaction would alter the electronic transition of the chromophore, detectable by simple UV–vis spectroscopy.

In line with this reasoning, treatment of sensor S in dichloromethane with various hydrogen-bond donors resulted in visible hypsochromic (blue) shifts (Figure 1). Importantly, compounds anticipated to be weaker donors, such as diphenylthiourea (1), yielded significant changes in sensor signal. Variation of binder concentration resulted in titration-like behavior, with a measurable end point upon saturation of sensor with catalyst. An array of catalysts (Chart 1), varying in structure and anticipated strength, was examined with the colorimetric sensor, and the Δλ_max upon saturation was determined.

Figure 1.

Response in the UV–vis spectrum of S upon increasing amounts of 12. [S] = 2.22 × 10^–5 M in CH₂Cl₂, [12] = 0 to 1.78 × 10^–4 M.

Chart 1. Hydrogen-Bonding Catalyst Structures Investigated.

DFT molecular orbital calculations were performed on bound and unbound sensor for selected hydrogen-bonding agents to gauge the orbital perturbation (Table 1). The calculated lowest energy transition accurately predicts the observed absorbance maximum for the free sensor. More importantly, the HOMO–LUMO energy gap was larger for all bound complexes, in accord with the empirically observed hypsochromic shift in Figure 1. An increased shift (lower λ_max) is predicted for binders of ostensibly greater strength (e.g., proton > benzoic acid > phenol).

Table 1. Calculated HOMO–LUMO Energies for Bound and Unbound Sensor Complexes^a.

binder	HOMO (eV)	LUMO (eV)	HOMO–LUMO (eV)	Calculated λmax (nm)
-	–4.23	–1.68	2.55	487
phenol	–5.11	–2.18	2.92	425
PhCO2H	–5.08	–2.02	3.05	407
proton	–9.43	–6.08	3.35	370

^aEnergies obtained from B3LYP/6-31G(d) optimized structures.

The UV absorption behavior of the sensor with the hydrogen-bonding agent can be represented as shown in Figure 2a. The lowest energy electronic transition may be ascribed to the n (HOMO) to π* (LUMO) transition, ΔE₁, corresponding to the measured λ_max. As supported by the above calculations, addition of a hydrogen-bonding agent stabilizes the ground state (HOMO) to a greater extent than the excited state (LUMO), i.e., ΔE₃ > ΔE₂. As a consequence, a hypsochromic shift is observed upon interaction of the sensor with the hydrogen-bond donors. For comparison, Figure 2b illustrates the energy diagram for a typical reaction with a hydrogen-bonding catalyst, in which catalysis is effected by LUMO-lowering of the electrophile (ΔE_a). We hypothesized that ΔE₃ – ΔE₂ is proportional to ΔE_a, i.e., the wavelength shift of the bound sensor•catalyst is proportional to the rate enhancement afforded in a reaction with the hydrogen-bonding catalyst.

Figure 2.

(a) Proposed energy diagram of the lowest energy electronic transition of the sensor upon interaction with catalysts of increasing strength, corresponding to the hypsochromic wavelength shift (Δλ_max). (b) LUMO-lowering of reactants via hydrogen-bonding catalysts, corresponding to increased reaction rates (k_rel).

2.2. Correlation of Binding with Sensor Wavelength Shift

As shown in Figure 1, a continuous wavelength shift was revealed upon saturation of the sensor with the catalyst. The lack of two distinct peaks in intermediate measurements containing both bound and unbound sensor indicates a rapid equilibration. Thus, plots of absorbance vs [catalyst] (see Figure 3 for an example with bisamidinium 12) were used to determine the binding constants (K_eq) for the sensor•catalyst complex. A significant range of blue shifts was observed for the different catalyst donors, ranging from ∼490 to 465 nm (Δλ_max ∼10–30 nm). In general, catalysts with larger Δλ _max values possessed much stronger binding constants. Since ΔE₁ is proportional to 1/λ_max, the energetics of the interaction of the sensor with the catalysts (ΔE₃ – ΔE₂) is proportional to 1/λ_max(sensor•catalyst) – 1/λ_max(sensor). Indeed, a good correlation of this inverse wavelength shift with ln(K_eq) was found (Figure 4). Note that in this plot, both axes are linearly proportional to energy terms: Δλ^–1 to the ΔE of the sensor electronic absorption, and ln(K_eq) to ΔG of sensor•catalyst formation. Importantly, this relationship establishes the observed wavelength shift as a reliable gauge for binding affinity of a catalyst to the sensor molecule.

Figure 3.

UV-titration curve of catalyst 12 in CH₂Cl₂ using [S] = 2.22 × 10^–5 M. Inlay: visible color change of sensor before (red) and after addition of 12 (yellow; [12] = 1.78 × 10^–4 M).

Figure 4.

Correlation between sensor wavelength shift and sensor•catalyst binding equilibrium constant. All titrations were performed with [S] = 2.22 × 10^–5 M in CH₂Cl₂.¹³

Using the sensor•catalyst wavelength shift as predictors of catalyst reactivity yields several noteworthy observations. Diol-based 30 (TADDOL) and silanol catalysts 31 and 32 afforded very weak shifts, despite application in numerous transformations, including Rawal’s seminal report on the asymmetric hetero Diels–Alder reaction.¹⁴ The greater λ_max shift of 32 compared to the related monosilanol 31 mirrors the increased reactivity of this silanediol scaffold, as elegantly reported by Mattson¹⁵ and Franz.^16,17 Benzoic acids and phenols spanned the intermediate range of sensor shifts, with trends clearly based on the electronic effects of aromatic substitution. Although these structures are not as commonly incorporated as hydrogen-bond catalysts, Schafmeister and co-workers have recently demonstrated the spiroligozyme catalyst 24, containing a carefully arranged carboxylic acid and phenol, as an effective ketosteroid isomerase mimic for the aromatic Claisen rearrangement.¹⁸N,N′-Diaryl thioureas and ureas, particularly those with multiple trifluoromethyl substituents such as Schreiner’s catalyst 4,¹⁹ afforded some of the largest sensor shifts, indicative of the immense utility of these structures in various organocatalysts.²⁰ The internally activated BF₂-urea 9 provided the largest shift within this class, in line with experimental reactivity data reported by Mattson and co-workers.²¹ Finally, formally cationic species, including guanidinium, amidinium, and Takenaka’s azaindolium 14(22) were the strongest binders, with wavelength shifts ranging from 26 to 34 nm (λ_max = 473–465 nm). Interestingly, one of the strongest noncationic binders was thiophosphoramide 16, possessing a pocket of three potential N–H donors. To date, this array has seen only limited use in organocatalysis.^23,24 Squaramide-containing scaffolds have yielded excellent results as hydrogen-bond activators;^2a,25 however, these compounds possess limited solubility, and are typically employed as heterogeneous catalysts. Representative squaramide 11, containing the common N-3,5-(CF₃)₂aryl and N′-alkyl array, was synthesized, and gave an apparent sensor end point of ∼480 nm. Due to its relative insolubility, an accurate binding equilibrium value could not be determined.

It is worth noting the experimental ease with which the sensor metric can be obtained. Compound S itself is easily obtained in 2 steps from commercial materials,^12a,26 and very little sensor or catalyst (particularly for strong catalysts) is necessary to obtain the wavelength shift. The titration experiment is largely insensitive to moisture, as illustrated by the poor binding observed in the sensor titration with water.

Applying the method of continuous variation to the sensor with catalyst 12 revealed a 1:1 binding stoichiometry with the sensor molecule (Figure 5).¹³ This observation is significant, since several other binding situations may be postulated, including donation of one catalyst molecule to several sensors (4 equivalent N–H bonds on 12, for example).

Figure 5.

Job plot analysis of catalyst 12 with sensor S showing 1:1 binding stoichiometry.

Benzoic acids and phenols offer useful templates to study electronic effects on sensor signal due to availability and well-understood behavior of aromatic substitution. Due to solubility limitations, ortho-substituted benzoic acids were studied rather than the para-substituted analogs. The electronic effects from substitution on the sensor interaction can be illustrated via a Hammett-type plot, as shown in Figure 6. As may be anticipated from the Brønsted catalysis law (see Section 2.6 for further discussion), increasingly electron-withdrawing substituents on these structures correlate with larger hypsochromic shifts of the sensor•catalyst complex. For both catalyst sets, highly linear relationships are evident with substituent σ parameters indicating that the wavelength shift provides an accurate readout of electronic perturbation on the hydrogen-bonding ability.

Figure 6.

Correlation of Hammett σ parameters for o-benzoic acids (σ_orthoⁱ)²⁷ and p-phenols (σ_para) with sensor•catalyst wavelength shifts.

Notably, the wavelength shifts seen with the sensor do not correspond directly with pK_a either in water (Figure 7a, R² = 0.0007) or DMSO (Figure 7b, R² = 0.0950). However, correlations are observed for closely related catalyst structures, wherein electronic perturbations modify the acidity of the donor moiety without introducing significant secondary effects. This observation provides potential for the sensor to estimate pK_a values within a series of related compounds. Persubstituted phenol 29 deviates from other phenolic catalysts, which may be attributed to the increased steric demand around the donating O–H bond.

Figure 7.

Plot of catalyst acidity in water (a) or DMSO (b) vs sensor•catalyst wavelength shifts. Correlation is only observed within closely related catalyst groups.

The correlation of the observed blue shift with the binding strength across a large range of hydrogen-bond donors proved the metric to be able to quantitatively detect these interactions. However, this finding does not necessitate a correlation with catalyst reactivity. In order for this correlation to occur, the sensor must be a good facsimile of the substrate that is undergoing reaction. Other factors, including alternate binding modes and steric effects, might come into play when a substrate interacts with hydrogen-bonding agent in a catalyzed reaction.

2.3. Comparison of Sensor Shifts with Hydrogen-Bond Catalyzed Diels–Alder Rate Data

To be a useful metric for the community, the sensor signal must correlate to empirically obtained rate enhancement via hydrogen-bond catalysis (Scheme 2). Myriad reaction profiles have been reported that are established to proceed via hydrogen-bond activation of the electrophile (LUMO-lowering activation). In order to best isolate the reactivity enhancement offered by the catalysts strictly due to hydrogen-bonding, we first targeted a reaction where the electrophile has only one possible point of interaction with the catalyst, and the nucleophile does not contain binding points (i.e., no heteroatoms). Additionally, the reaction should have minimal background rate and a method to easily analyze starting material and/or product concentrations.

Scheme 2. Sensor as a Surrogate for an Electrophilic Substrate.

The reaction of methyl vinyl ketone (MVK) with cyclopentadiene (Cp) offers a useful reaction platform that fulfills these criteria (Scheme 3), and has been used to gauge the relative strength of thiourea^19a and bisphenol^7b catalysts previously. Hydrogen-bonding to the ketone carbonyl accounts for catalysis in this reaction. Importantly, the binding in the sensor•catalyst complex is very similar to that of the MVK•catalyst intermediate as both interactions arise from a carbonyl acting as a hydrogen-bond acceptor.

Scheme 3. Diels–Alder Kinetic Study via ¹H NMR.

Systematic investigation of the Diels–Alder reaction of MVK and Cp with a variety of catalysts was performed under pseudo-first order conditions as described in Scheme 3. Kinetic data was acquired via continuous sampling (5 min intervals) by ¹H NMR spectroscopy, and each rate measurement was performed in triplicate. Relative rate constants, k_rel, were calculated as described in eq 1 from the observed pseudo-first order rate constant k′_obs and background rate k_background, and were normalized for catalyst concentration N. The resulting values directly provide the rate enhancement afforded by the catalyst.

1

As displayed in Figure 8, a plot of ln(k_rel) against the inverse sensor wavelength shift of 18 catalysts shows an excellent correlation. Catalysts with greater blue shifts when treated with the sensor show greater activity in the Diels–Alder reaction via correspondingly greater LUMO lowering of the ketone in the dienophile. More precisely, the change in energy of the sensor upon binding with the catalyst is proportional to the change in activation energy of the hydrogen-bond catalyzed Diels–Alder reaction.

Figure 8.

Correlation between the sensor•catalyst wavelength shift and catalyst rate enhancement in the Diels–Alder reaction between MVK and Cp.

The observed correlation establishes that, at least in this class of reaction, the sensor signal is a good indicator of LUMO-lowering ability of these small molecules as hydrogen-bond catalysts. Importantly, the results also indicate that the binding interaction of the sensor with catalysts is similar to that of methyl vinyl ketone with catalyst, i.e. the sensor is a useful gauge of carbonyl activation.

2.4. Comparison of Sensor Shifts with Hydrogen-Bond Catalyzed Friedel–Crafts Rate Data

The addition of various nucleophiles into nitroalkenes is one of the most widely used reaction motifs in hydrogen-bonding catalysts; it is often used as a measure of reactivity when developing and comparing novel catalyst structures.^{7d,15a,16,21a,22,28} To test the effectiveness of our sensor metric beyond the Diels–Alder reaction, we studied the Friedel–Crafts addition of N-methylindole (34) into nitrostyrene 35 (Scheme 4). Deuterated 35 was easily prepared via Henry condensation using d₃-nitromethane with the corresponding aldehyde. Again, the number of hydrogen bond acceptors is limited in this reaction. Nucleophile 34 provides a reasonably active coupling partner while minimizing potential catalyst interactions [pK_a(H₂O) N-methylindolium = −1.8].²⁹ Thus, the effects of hydrogen bonding catalysts on the activation of the styrene electrophile via binding to the nitro acceptor can be cleanly delineated.

Scheme 4. Friedel–Crafts Kinetic Study via ²H NMR.

Significantly, the catalyzed rates of this reaction allow comparison of the effects of hydrogen-bonding catalysts on carbonyl acceptors (the sensor and MVK) versus nitro acceptors (nitrostyrene 35) as outlined in Scheme 5. While binding geometries to carbonyl groups are anticipated to be similar, an analogous correlation may not be automatically presumed for a nitro group. In particular, the nitro group contains a formally delocalized negative charge across three atoms, and has been suggested to form κ²-activated intermediates with certain catalyst structural types such as squaramides and thioureas.^21b,21f,30

Scheme 5. Catalyst–Acceptor Binding for Sensor and Reaction Electrophiles.

As illustrated in Figure 9, kinetic reaction data was acquired using ²H NMR spectroscopy. Although initial ¹H NMR spectroscopic studies with proteo-35 provided usable data, employment of a deuterium label provides exceptional signal isolation. Moreover, overlap of signals from the catalyst is completely mitigated, since only substrate, product, and internal standard exhibit appreciable resonances. As a result, any catalyst can be readily analyzed.

Figure 9.

(a) Stacked ²H NMR plot and (b) kinetic profile for the Friedel–Crafts reaction shown in Scheme 4 using catalyst 12. Conditions: [34] = 1.33 M, [35] = 0.133 M, [37] = 0.133 M, [12] = 2.67 × 10^–4 M. The CDCl₃ peak arises from natural abundance in CHCl₃.

The Friedel–Crafts reaction was studied under pseudo-first order conditions using the broad series of hydrogen-bond catalysts shown in Chart 1. The relative catalytic strength of each catalyst was calculated according to eq 1, using averaged k′_obs values from duplicate trials. A plot of ln(k_rel) and inverse sensor wavelength shift is shown in Figure 10a. Similar to the Diels–Alder reaction, and predicted based on sensor shifts, cationic binders proved to be the most effective catalysts, followed by electron-deficient ureas and thioureas. Squaramide 11, employed here as a heterogeneous catalyst, provided moderate rate enhancement as predicted by its sensor wavelength shift. Thiophosphoramide 16 and sulfonamide 6 have previously been studied in a Friedel–Crafts reaction under nearly identical conditions by Shea and co-workers.^23a The k_rel values reported for these catalysts and 4 align closely with the values measured in this work.

Figure 10.

(a) Correlation observed between the sensor•catalyst wavelength shift and catalyst rate enhancement in the Friedel–Crafts reaction. (b) Correlations based on catalyst structure groups.

The overall correlation of the sensor wavelength shift with catalyst strength for the Friedel–Crafts reaction is good (R₂ = 0.84), but not as strong as that found in the Diels–Alder reaction (R₂ = 0.95). Analysis of this data suggests that the binding interaction of a hydrogen-bonding catalyst with the sensor carbonyl does not fully mimic that of a hydrogen-bonding catalyst with the activated nitroalkene. Closer inspection reveals that the wavelength shift is correlated even more strongly to catalyst strength within an isostructural catalyst series (Figure 10b). Specifically, the catalysts can be placed in four groups based on their general structure: benzoic acids, phenols, “Y-type” binders, and other N–H binders. The Y-type group consists of catalysts possessing two N–H donor groups separated by a single atom, such as ureas, thioureas, guanidines, bis-sulfonamides, etc. The other N–H binders include those catalysts that have more than one atom separating the donor array (squaramide 11, bisamidinium 12) or can only donate one N–H bond (benzotriazole 15, monoamidinium 13). Interestingly, the silanediol catalyst 32 exhibits reactivity falling nicely in line with benzoic acids, possibly due to a similar O–H geometry. This analysis quantitatively shows that Y-type structures are superior for nitroalkene activation. In contrast, phenols provided the least activation relative to their binding interaction with the sensor.

Overall, the sensor provides a good assessment of relative reactivity of hydrogen-bond catalysts in the Friedel–Crafts reaction. The presence of stronger correlations within catalyst structural classes is consistent with some catalysts activating nitroalkenes via a different mode (e.g., κ²-binding) that is not completely captured by the interactions of the catalysts with the carbonyl of the sensor molecule. This analysis underlines the complex nature of hydrogen-bonding, emphasizing that caution must be exercised in generalizing catalyst reactivity or selectivity from one reaction to another.

2.5. Unified Description of Reactivity versus Sensor Measurements

A general equation to describe catalyst strength for reaction r based on sensor response (Δλ^–1) is presented in eq 2. Parameter R_r (slope) represents the responsiveness of the rate per unit catalyst strength as determined by the sensor measurement. Parameter C_r (y-intercept) corresponds to inherent complementarity of the catalyst to the electrophilic reaction partner.

2

The parameter values (Table 2) for the reactions shown in Scheme 3 (Diels–Alder) and Scheme 4 (Friedel–Crafts) were obtained using the kinetic data from Figures 8 and 10, respectively. Comparison of the R_DA and R_FC values reveals that the Friedel–Crafts is more sensitive to catalyst strength. In other words, the same catalyst produces a greater relative rate enhancement for the Friedel–Crafts reaction than for the Diels–Alder reaction. Similarly, a given wavelength shift of the sensor by a catalyst will cause a greater reactivity change in the Friedel–Crafts vs the Diels–Alder reaction. On the other hand, the C_DA and C_FC values indicate the inherent complementarity of the electrophilic substrate with catalysts; greater complementarity translates to greater reactivity. Notably, the C values represent reactivity when there is no wavelength shift (no perturbation of the sensor by the catalysts) and represent the lower limit of LUMO activation afforded by the catalyst.

Table 2. Reaction Coefficients for Equation 3.

reaction	R	C
Diels–Alder	7.08	–6.36
Friedel–Crafts	9.59	–4.46

Catalyst group-specific coefficients r_r and c_r can be introduced (eq 3) to account for variation if the sensor binds the catalyst differently than the reaction electrophile. Due to the strong correlation of the sensor shift to relative rates independent of catalyst structure, coefficients are unnecessary for the Diels–Alder reaction (r_DA ≈ c_DA ≈ 1). As discussed in Section 2.4, the sensor does not completely model catalyst binding to the nitrostyrene acceptor of the Friedel–Crafts reaction. Accordingly, the slope and intercept data from Figure 10b were combined with the R and C values from Table 2 to afford coefficient values for r_FC and c_FC, respectively, as provided in Table 3.

3

Table 3. Catalyst Structure Coefficients for Friedel–Crafts Catalysis.

catalyst series	rFC	cFC
Y-type	1.00	0.80
Benzoic Acids	0.68	0.56
Phenols	1.06	1.39
Other N–H Binders	0.86	0.92

The coefficient values in Table 3 reveal general trends between the different catalyst structural types and rate in the Friedel–Crafts reaction. Again, r_FC values are a measure of responsiveness of a given catalyst architecture to a perturbation in sensor binding. For Y-type and phenolic catalysts, r_FC is noticeably higher than the other N–H binders, and particularly benzoic acids. Thus, for the same amount of wavelength shift, the phenol and Y-type catalysts provide greater increases in reactivity relative to the other N–H binders and benzoic acids. This observation indicates that the sensor can assess electronic effects in a catalyst series. Comparison of the Hammett effects on Δλ^–1 (Figure 6; ρ_acid = 5.8, ρ_phenol = 4.4) with those on ln(k_rel) (ρ_acid = 0.60, ρ_phenol = 0.33)¹³ provides support for this assertion; both measures show a stronger electronic effect for the carboxylic acid series.

On the other hand, lower c_FC values indicate the inherent complementarity of a given catalyst architecture. For example, the Y-type binders activate nitro electrophiles to a greater extent at a given wavelength shift relative to phenols or the other N–H binders. Interestingly, the c_FC value for benzoic acids would predict high catalytic activity relative to Y-type binders, but only in the weak binding regime (left side of plot). Due to the low r_FC value for benzoic acids, the trends invert such that Y-type binders are superior in the strong binding regime (right side of plot). Considering both terms together, the Y-type binders are both more complementary to the nitroalkene and more efficient at LUMO lowering, thereby providing superior reactivity.

2.6. Comparison of Catalyst Reactivity and Acidity: Brønsted Analysis

Acidity values have widely been used as a guiding principle in hydrogen-bond catalyst design, under the premise that a more acidic donor will form a stronger interaction and stabilize the buildup of anionic charge in the transition state to a greater extent. Indeed, several reports have observed increased activity with judicious electronic tuning of the donor hydrogen.^17c,38 However, even ostensibly subtle changes to catalyst structure can cause secondary factors to override the reliability of pK_a as a predictive measure, as demonstrated by Cheng’s recent study³⁹ on thiourea derivatives and even noted in the seminal work by Hine on mono- and bis-phenols.⁴⁰ Having proved the effectiveness of the sensor signal as a gauge for catalyst strength, we undertook a comparison with acidity to determine the similarities and differences between the two metrics.

Aggregate data for all catalysts spanning 3 orders of magnitude in reactivity for the Diels–Alder reaction and 4 orders in the Friedel–Crafts reaction is organized by increasing sensor wavelength shift in Table 4. Recent efforts by Schreiner⁴¹ and others⁴² provided accurate acidity values of common hydrogen-bond donors. The Brønsted catalysis equation (eq 4), which describes the relationship for the rate of an acid-catalyzed reaction with the pK_a of the acid,⁴³ was applied to this data.

4

Table 4. Sensor Shifts, Binding Constants, Relative Rate Data, and Acidity Values for Catalysts Investigated.

catalyst	λmax (nm)	Keq (M–1)	krel (Diels–Alder)	krel (Friedel–Crafts)	pKa (H2O)	pKa (DMSO)
H2O (33)	495.2	2.54	-	-	15.75a	32a
(R,R)-TADDOL (30)	490.8	1.57 × 101	0.023	-	∼16	28–30b
TBDMSiOH (31)	490.4	4.87	-	-	∼12c	-
Diphenylthiourea (1)	490.0	1.67 × 101	0.012	0.68	-	13.4d
Silanediol (32)	487.2	7.88 × 101	-	1.82	11.8c	-
(R)-BINOL (25)	487.0	3.18 × 101	0.034	0.35	10.28e	17.1e,f
4-t-Bu-phenol (26)	486.2	4.30 × 101	-	0.25	10.23g	∼18a
2-Me-BzOH (17)	486.2	7.04 × 101	-	3.21	3.91g	11.07h
BzOH (18)	485.4	8.35 × 101	0.086	1.72	4.20g	11.00h
Benzotriazole (15)	484.2	6.57 × 101	-	2.12	8.38g	11.9a
4-Br-phenol (27)	484.0	1.50 × 102	0.224	0.93	9.34g	-
(CF3)2-thiourea (2)	484.0	1.07 × 102	0.049	7.13	-	10.7d
2-Cl-BzOH (19)	482.6	2.15 × 102	-	5.81	2.94g	9.70h
Sulfonamide (5)	482.2	1.04 × 103	0.337	36.4	-	<12.9a,i
4-NO2-phenol (28)	481.0	6.05 × 102	0.733	4.94	7.14g	10.8a
F5-phenol (29)	480.8	4.92 × 102	-	2.82	5.53j	-
(CF3)2-urea (6)	480.4	1.05 × 103	0.724	24.1	-	16.1d
Squaramide (11)k	480.0	-	-	12.5	-	11.5–12.0l
Spiroligozyme (24)	480.0	6.30 × 102	-	-	4–5	-
2-NO2–BzOH (20)	480.0	8.27 × 102	0.669	22.8	2.17g	8.66h
(CF3)3-thiourea (3)	479.8	4.52 × 102	0.442	47.0	-	9.5d
(CF3)4-urea (8)	479.8	9.38 × 103	-	120.0	-	13.8d
(CF3)3-urea (7)	479.6	2.52 × 103	-	73.4	-	14.9d
3,5-NO2–BzOH (21)	478.0	1.02 × 103	-	-	-	-
F5–BzOH (22)	477.2	1.41 × 103	1.04	22.2	1.75m	-
(CF3)4-thiourea (4)	477.0	1.77 × 103	1.22	130.6	-	8.5d
Thiophosphoramide (16)	475.0	3.11 × 103	1.23	291.1	-	-
BF2-urea (9)	474.4	2.65 × 104	-	603.5	-	7.5n
Guanidinium (10)	473.2	1.84 × 104	2.81	1446	10.12g	14o
MonoAmidinium (13)	473.0	3.34 × 103	3.69	89.1	11–12g	12.8–13.6p
TFA (23)	470.0	5.49 × 103	-	-	0.23g	-
Azaindolium (14)	469.8	-	12.2	2604	4.6g	6q
BisAmidinium (12)	465.0	3.22 x105	35.1	2763	11–12g	12.8–13.6p
PPTS	-	-	0.190	72.7	5.2a	3.4a
Diphenylphosphate	-	-	0.951	26.9	1.29g	-
CSA	-	-	8.53	103.3	–0.6a	1.6a

^aRef (31).

^bRef (32).

^cRef (17c).

^dRef (41).

^eRef (33).

^fValue for 2-naphthol.

^gRef (29).

^h.Ref (27).

ⁱValue for the less acidic N-phenylmethanesulfonamide analogue.

^jRef (34).

^kTitrations and reactions were performed heterogeneously.

^lRef (42a).

^mRef (35).

ⁿRef (38c).

^oRef (36).

^pRef (37).

^qRef (22).

Figures 11 and 12 display the results for selected catalyst series in the Diels–Alder and Friedel–Crafts reactions, respectively.⁴⁴ These plots prove the linear free energy relationships (LFERs)⁴⁵ between catalyst acceleration and acidity among catalysts of very similar structure. In general, these LFERs exhibited a narrow range of α values (0.39–0.47; Table 5), indicating a similar degree of hydrogen-bonding in the transition states and that these reactions are not proceeding through formal protonation (α = 1). Slightly higher values found for benzoic acids in the Diels–Alder (α = 0.57) indicate a greater degree of proton transfer in the transition states consistent with the ionic nature of benzoic acids. Lower values for ureas in the Friedel–Crafts (α = 0.31) indicate a lesser degree of proton transfer in the transition states in line with the covalent nature of the N–H bonds.

Figure 11.

Brønsted catalysis plot for the Diels–Alder reaction, demonstrating LFERs for closely related catalyst groups.

Figure 12.

Brønsted catalysis plot for the Friedel–Crafts reaction demonstrating LFERs for closely related catalyst groups.

Table 5. Brønsted α-Values for Different Catalyst Structural Types from Figures 11 and 12.

		α
catalyst series	solvent	Diels–Alder	Friedel–Crafts
Thioureas	DMSO	0.41	0.47
Ureas	DMSO	-	0.31
Benzoic Acids	DMSO	-	0.39
Benzoic Acids	H2O	0.57	0.46
Phenols	H2O	0.39	0.39

More significantly, these figures clearly demonstrate the inherent limitations of estimating catalyst strength using acidity metrics. A pertinent example can be found in the thiourea and urea catalysts. Based solely on pK_a measurements, thioureas would be predicted to provide much higher activity. In practice, the opposite is observed where ureas exhibited greater (6, 7 vs 2, 3) or similar (8 vs 4) activation of the nitro group compared to their thiourea analogues, despite the much greater acidity of the thioureas (4–5 orders of magnitude difference). Highly reactive catalysts not belonging to a clearly defined series, including common Brønsted acids, are included in Figure 11 and further highlight the disparity between acidity and activity. Interestingly, Takenaka’s azaindolium catalyst 14 displayed much higher activity than PPTS, despite similar acidities and similar pyridinium–H donor moieties. Taken together, these observations reinforce the risks of using pK_a measurements to estimate reactivity.

3. Conclusions

In conclusion, a sensor is described that provides an assessment of the reactivity over 3–4 orders of magnitude for 33 hydrogen-bonding catalysts. Useful correlations are obtained between the wavelength shifts that catalysts cause to the pyrazinone sensor and the rate-determining steps in Diels–Alder and Friedel–Crafts reactions. As a result, only the wavelength shifts upon saturation of the sensor with catalysts need to be measured vs the more time-consuming titration studies. Consequently, the sensor may also find use as a rapid means for measuring pK_a values in series of related compounds.

In contrast to established acidity (pK_a) values, the sensor wavelength shift is a highly predictive metric for the relative reactivity of catalysts encompassing a broad range of structures and strengths. Notable acidity-activity disparities include cationic catalysts (low acidity, high reactivity), and benzoic acids (high acidity, low reactivity), the strengths of which are more accurately gauged by their interaction with the sensor. Overall, the sensor is a superior surrogate for the diverse electrophiles (enone and nitroalkene) used compared to water and is better able to assess secondary interactions. The data collected was used to formulate the relationship described in eq 3, which provides a direct means of assessing the reactivity of a catalyst in a given reaction using the sensor signal. The resulting parameters also reveal relationships between substrate•catalyst binding, catalyst-induced LUMO-lowering, and catalyst structure.

Investigation of additional catalyzed reactions with the sensor, empirically or computationally, has the potential to expand eq 3 to achieve quantitative predictive power across a large range of reaction platforms and catalysts (hydrogen bonding, Brønsted acid, and Lewis acid). Further studies to achieve this goal are underway.

Supporting Information Available

Experimental procedures, characterization of new compounds, UV-titration data, and Diels–Alder and Friedel–Crafts rate data. This material is available free of charge via the Internet at http://pubs.acs.org.

The authors declare no competing financial interest.

Funding Statement

National Institutes of Health, United States