Atomically Dispersed Rhodium on TiO₂ for Tandem Hydrogenation–H/D Exchange of Cinnamic Acid

Abstract

An atomically dispersed rhodium on TiO₂ catalyst enables a tandem process, combining hydrogenative reduction with α,β-hydrogen–deuterium exchange of cinnamic acid, in which D₂O serves as the deuterium source. In contrast with previous reductive deuteration methods that yield only partially labeled 3-phenylpropanoic acids (D^α-inc.: ≤50%, D^β-inc.: ≤50%), this heterogeneous system delivers near-quantitative deuterium incorporation (D^α-inc.: 94%, D^β-inc.: 99%) under mild conditions, outperforming Rh nanoparticles and homogeneous Rh catalysts. Mechanistic studies indicate that α-C–H activation is the slowest transformation step within the overall process, owing to the exceptional C–H bond activation capability of the atomically dispersed catalyst; efficient α-C–H hydrogen–deuterium exchange is readily achieved. In addition, although catalyst recyclability is constrained by Rh aggregation, no Rh leaching is detected. This work provides a concise, operationally simple route to alkyl fully deuterated 3-phenylpropanoic acids (d₄-PA) and showcases the application of an atomically dispersed catalyst in tackling challenging deuterium-labeling transformations.

1. Introduction

Deuterium-labeled organic compounds play a crucial role across various fields, particularly in absorption, distribution, metabolism, and excretion (ADME) studies during drug development [1,2,3,4]. Selective replacement of hydrogen (H) with deuterium (D) at specific C–H bonds can enhance metabolic stability through the kinetic isotope effect, thereby allowing lower and/or less frequent dosing of drugs [5,6,7,8,9,10]. For example, deutetrabenazine—the first approved deuterated drug—exhibits an improved pharmacokinetic profile and reduced dosing frequency compared to its protium analogue [1,3]. Beyond pharmacology, deuteration has also found important application in materials science [11]. For example, deuterated organic light-emitting diode (OLED) materials have shown enhanced device stability and extend operational lifetimes by mitigating degradation associated with C–H bond cleavage [12]. This demand has been paralleled by the development of efficient synthetic methods that provide streamlined access to deuterated compounds.

Cinnamic acid (phenylpropenoic acid) (CA) is an abundant, low-cost α,β-unsaturated carboxylic acid (0.07 USD/g) [13,14]. Chemoselective reductive deuteration of its C=C bond adjacent to the carboxyl group provides a streamlined route to produce highly valued deuterated 3-phenylpropanoic acids [15]. So far, various strategies have been developed for the deuteration of cinnamic acid to produce deuterated 3-phenylpropanoic acids, including stoichiometric reductions with SmI₂ [16] or metallic sodium [17], homogeneous Pd- [18] or Rh-catalyzed protocols [19], heterogeneous Pd/C systems [20,21,22], and the latest photo- [23] and electrocatalytic variants [24]. However, these strategies typically afford only d₂ deuteration products (-HDC-CDH-) through C=C reductive deuteration with D sources but generally fail to achieve subsequent H/D exchange of the resulting d₂ alkyl carboxylic acids, leaving a great challenge to produce tetrasubstitution (d₄)-labeled products (Figure 1a).

The synthesis of tetrasubstituted (d₄) phenylpropionic acids (PA) requires a catalyst capable of both hydrogenating/deuterating the olefin and then efficiently activating the α- and β-Csp³–H bonds of the 3-phenylpropanoic acid intermediate. Building on our research in single-atom catalysis [25,26,27] and deuteration reaction [28,29], particularly our recent discovery that single-atom catalysts can achieve highly efficient benzylic Csp³–H H/D exchange [29], we report herein that an atomically dispersed Rh/TiO₂ catalyst fulfills these requirements. This heterogeneous atomically dispersed catalyst simultaneously hydrogenates the α,β-unsaturated carboxylic acid and incorporates deuterium at both α- and β-positions, converting cinnamic acid to a tetrasubstituted deuterium-labeled alkyl acid in one stroke (94% D-inc. at α and 99% D-inc. at β). During the reaction, no Rh leaching was detected, eliminating metal–product separation concerns and ensuring catalyst recyclability. By uniting reduction and labeling in a sample, durable heterogeneous atomically dispersed catalyst, this approach opens an efficient route to deuterium-enriched organic reagents that were previously difficult to access (Figure 1b).

2. Results and Discussion

Catalyst Synthesis and Characterization: The Rh/TiO₂ catalyst was prepared using a wet-impregnation method [30]. Typically, TiO₂ (P25) powder was impregnated with an aqueous solution of RhCl₃·3H₂O, and then dried and calcined at 400 °C in air. The resulting material was then reduced in 10 vol% H₂/He at 200 °C and is hereafter denoted as atomically dispersed Rh/TiO₂. For comparison, Rh nanoparticles (Rh/TiO₂-NPs) were prepared also by a wet-impregnation method similar to that used for the atomically dispersed Rh/TiO₂ catalyst, except with a higher Rh loading. The detailed synthesis procedure is described in Section 3.2 of the main text.

Inductively coupled plasma optical-emission spectroscopy (ICP-OES) showed a Rh concentration of 2054 mg L⁻¹, corresponding to a Rh loading of 0.21 wt%. The detailed results are provided in Table S1. X-ray diffraction (XRD, Figure 2a) showed no diffraction attributable to crystalline Rh, consistent with a highly dispersed and/or low-loading Rh phase. In addition, for both the bare support before calcination and the Rh/TiO₂ after calcination, the diffraction patterns show the expected anatase/rutile phases of P25 (anatase: 2θ ≈ 25.3° (101), 37.8° (004), 48.0° (200); rutile: 2θ ≈ 27.4° (110), 36.1° (101)). Transmission electron microscope (TEM) and aberration-corrected high-angle annular dark-field scanning transmission electron microscopy (AC-HAADF-STEM) further corroborated highly dispersion of Rh: low-magnification images revealed no Rh nanoparticles (Figure 2b,c and Figure S2), while high-magnification images displayed isolated bright spots corresponding to single Rh atoms or atomically dispersed sub-nanometer clusters containing a few Rh atoms (Figure 2d,e and Figure S2). Diffuse-reflectance infrared Fourier-transform spectroscopy of CO adsorption (CO-DRIFTS) provided additional evidence for atomically dispersed Rh. The spectrum exhibits two bands at ~2092 and ~2025 cm⁻¹ (Figure 2f), assigned to the symmetric and asymmetric stretches of Rh(CO)₂ dicarbonyl species, characteristic of linearly adsorbed CO on isolated Rh sites. The fact that no bridge CO adsorption at ~1860 cm⁻¹ may suggest that the Rh atoms in the sub-nano clusters are loosely distributed [30,31,32].

Catalytic Performance: Under H₂/D₂O conditions, a blank reaction without any catalyst and a control experiment with TiO₂ support as a catalyst were first performed. Neither hydrogenation product, PA, nor meaningful deuteration of the olefin (<5%) was observed. We then tested a homogeneous catalyst, RhCl₃·3H₂O. Although it hydrogenated CA to PA, deuterium incorporation at the α- and β-positions remained below 10%. These results underscore the limitations of TiO₂ and homogeneous Rh catalyst in achieving d₄-deuteration as TiO₂ alone is unable to efficiently hydrogenate, while the homogeneous Rh catalyst lacks the C–H activation strength required for effective deuteration.

On the other hand, Rh/TiO₂ not only hydrogenated CA but also delivered outstanding deuteration levels—94% at the α-position and 99% at the β-position of PA—underscoring the key role of atomically dispersed Rh sites. For comparison, Rh nanoparticle catalysts (Rh/TiO₂-NPs, the particles were 1.1 nm, please see Figure S3) achieved < 50% deuterium incorporation at either position, mirroring previous results with Pd/C and indicating poor H/D exchange activity toward the PA intermediate despite competent hydrogenation. The markedly superior performance of Rh/TiO₂ therefore highlights the necessity of atomically dispersed Rh for efficient tandem hydrogenation and H/D exchange reaction (Figure 3a).

To evaluate the stability of our Rh/TiO₂ catalyst, recycling experiments were carried out (Figure 3b). The Rh/TiO₂ catalyst demonstrated moderate reusability. In the first cycle, it efficiently catalyzed the hydrogenation of CA, achieving 95% deuteration at the β-position, although the α-position deuteration dropped to 80%. However, in the second cycle, deuteration at the α-position further declined to 40%, while β-position deuteration remained nearly unchanged.

To elucidate the origin of the performance loss, the spent catalyst was characterized (Figure 3b–d and Figures S4). Rh loading was examined via ICP–OES, and the results show that the Rh concentrations were 2054 mg L⁻¹ for the fresh catalyst and 2136 mg L⁻¹ for the spent one, corresponding to the same Rh loading of 0.21 wt% in both cases. This negligible difference in Rh content indicates that Rh leaching during the reaction was insignificant (Table S1), thereby verifying the heterogeneous nature of the process and ruling out Rh loss as the cause of the decreased α-deuteration level. Further, the post-reaction catalyst was characterized using TEM and CO-DRIFTS spectroscopy. TEM reveals the appearance of Rh nanoparticles (~2.7 nm), confirming aggregation of Rh (Figure 3c). Consistently, the CO-DRIFTS spectrum shows, in addition to the bands at 2092 cm⁻¹ and 2025 cm⁻¹ assigned to the symmetric and asymmetric stretches of positively charged gem-dicarbonyl Rh species, a bridged-CO band at 1853 cm⁻¹ characteristic of Rh nanoparticles/clusters (Figure 3d). Together, the TEM and CO-DRIFTS results provide clear evidence for the formation of Rh NPs in the spent catalyst.

These findings highlight that atomically dispersed Rh on TiO₂ is significantly more effective for the tandem hydrogenation–H/D exchange of cinnamic acid than Rh nanoparticles. The formation of Rh clusters during recycling dramatically reduces catalytic efficiency in the tandem hydrogenation–H/D exchange of cinnamic acid, particularly the H/D exchange reaction at the α-position, underscoring the critical importance of maintaining the single-atom nature of the catalyst to ensure high performance.

To gain deeper insight into the reaction process and mechanism, we conducted kinetic experiments (Figure 4a). We found that C=C hydrogenation was very rapid, with cinnamic acid fully hydrogenated within 60 min. β-Deuteration proceeded almost concurrently, while α-deuteration was significantly slower, reaching only 35% D-inc. after 60 min. These results indicate that deuterium incorporation at the β-position is more facile, whereas α-C–H activation is comparatively sluggish. For the synthesis of d₄-PA, this identifies α-C–H activation as the slowest transformation step.

Actually, the α-deuteration involves activation of the α-C–H bond adjacent to the carboxyl group, which is facilitated by Rh–O coordination at the metal–support interface. This demanding activation mode requires atomically dispersed Rh sites to provide sufficient Rh–O interfacial centers. Upon catalyst aggregation, part of these active sites is lost, leading to a decline in α-deuteration, consistent with the recycling results. In contrast, β-deuteration proceeds more readily because the benzylic C–H bond is weakened by resonance stabilization of the adjacent phenyl ring (≈88–90 kcal mol⁻¹ vs. 92–95 kcal mol⁻¹ for the α-C–H bond). The β-H/D exchange likely occurs via a relatively Rh–benzyl intermediate, accounting for the faster and more complete β-deuteration observed experimentally.

The effect of temperature on the tandem hydrogenation and H/D exchange was investigated (Figure 4b). At 60 °C, although hydrogenation remains efficient, deuterium incorporation is low (α: 26%, β: 50%). Raising the temperature to 80 °C improves β-deuteration to acceptable levels (β: 98%), but α-D-inc. only increases to about 60%. Optimal performance is reached at 100 °C, where both positions attain near-quantitative deuteration (α: 94%, β: 99%), consistent with the need for sufficient thermal energy to overcome the activation barrier for H/D exchange on the PA intermediate. Further increasing the temperature to 120 °C leads to a decline in D-inc., likely due to partial aggregation or structural evolution of the atomically dispersed Rh, which diminishes its activity.

To clarify the role of H₂ throughout the reaction, we investigated both the necessity of H₂ and the influence of its pressure (Figure 4c). We found that increasing the H₂ pressure from 1 atm to 3 and 5 atm led to a decrease in α-deuteration (from 98% to 85–88%), while β-deuteration remained essentially unaffected. This D-inc. drop in α-position is attributed to pressure-induced partial agglomeration of the single-atom Rh sites. However, when the reaction was carried out under a 10% H₂/N₂ gas mixture (1 atm), no formation of PA or any deuterated olefin was observed, indicating that a minimum threshold of hydrogen pressure is required for olefin hydrogenation to proceed. We further examined whether hydrogen pressure is also necessary for the H/D exchange of the intermediate PA. Under a nitrogen atmosphere, no deuteration of PA occurred, suggesting that hydrogen is also essential for facilitating H/D exchange (Figure 4d(i)). Combined with our earlier findings, these results suggest that hydrogen likely plays a role in promoting the dissociation of D₂O, which in turn enables the H/D exchange process.

In addition, we attempted to extend the reaction to cinnamic acids bearing different substituents. However, the atomically dispersed Rh/TiO₂ catalyst tended to aggregate under the reaction conditions, forming Rh clusters that promoted undesired over-hydrogenation of the substituted CA, such as those in –CF₃– or –OCH₃-substituted CA was observed, owing to the activating effect of these substituents on the aryl ring. In contrast, the para-Me-substituted cinnamic acid (3), in which the Me group exerts only a minimal activating effect on the aromatic ring, exhibited good performance, affording the deuterated product with 61% α-D-inc. and 97% β-D-inc. These findings indicate that the tandem hydrogenation–H/D exchange is effective for structurally stable aromatic substrates, while catalyst stability remains the key factor limiting broader substrate applicability. Ongoing work in our laboratory focuses on developing a more robust single-atom Rh catalyst to suppress aromatic ring hydrogenation and achieve an expanded substrate scope.

Finally, we directly assessed the H/D exchange performance of the intermediate PA under atomically dispersed catalytic conditions. The catalyst enabled efficient conversion of PA to d₄-PA (α: 95%, β: 95%), underscoring its capability to deuterate carboxylic acid derivatives through direct C–H bond activation—a transformation that is typically challenging for currently developed systems (Figure 4d(ii)).

3. Materials and Methods

3.1. Reagents and Product Characterization

All reagents were purchased from various chemical suppliers and were used without previous purification. Rhodium (III) chloride trihydrate was obtained from Sigma-Aldrich (St. Louis, MO, USA). Other reagents were commercially available from Energy Chemical (Beijing, China) and Leyan Chemical (Shanghai, China) and were used without further purification unless otherwise stated. The TiO₂ support (P25) was purchased commercially, with a nominal particle size of ~21 nm. ¹H NMR spectra were recorded at room temperature in DMSO-d₆ on a 400 MHz instrument.

3.2. Catalyst Preparation and Characterization Techniques

3.2.1. The Preparation Procedure for Rh Catalyst

Rh/TiO₂: Firstly, RhCl₃·3H₂O solution (7.6 mM, 3.9 mL) is impregnated onto a 2 g TiO₂ (P 25) support, stirred continuously for 4 h, and then left for 2 h. The catalyst was suction-filtered, washed with 300 mL of water, calcined at 400 °C for 2 h with a heating rate of 5 °C/min, and subsequently reduced under a 10% H₂/He atmosphere at 200 °C for 2 h.

The ICP-OES tests showed that the actual Rh loading of the Rh/TiO₂ was 0.21 wt%; please see Table S1.

Rh/TiO₂-NPs: Firstly, RhCl₃·3H₂O solution (7.6 mM, 46 mL) is impregnated onto a 2 g TiO₂ (P 25) support, stirred continuously for 4 h, and then left for 2 h. The catalyst was suction-filtered, washed with 300 mL of water, calcined at 400 °C for 2 h with a heating rate of 5 °C/min, and subsequently reduced under a 10% H₂/He atmosphere at 200 °C for 2 h.

ICP-OES tests showed that the actual Rh loading of the Rh/TiO₂ NPs was 2.5 wt%, please see the Table S1.

3.2.2. Details for the Catalytic Characterization Techniques

ICP-OES: The actual Rh loading, reported as weight percent Rh relative to total catalyst mass (wt%), was determined by ICP-OES after acid digestion (PerkinElmer Avio 550 Max, Waltham, MA, USA).

XRD: XRD patterns were collected at an Empyream instrument (Malvern Panalytical, Almelo, The Netherlands) equipped with a Cu Kα radiation source (λ = 0.15432 nm), operating at 40 kV and 40 mA. A continuous mode was used to collect data in the 2 θ range of 5° to 90°.

CO-DRIFTS: In situ diffuse reflectance infrared Fourier transform spectroscopy (DRIFT) spectra were collected at 25 °C with a Bruker Equinox 55 spectrometer (Bruker Optics, Ettlingen, Germany) equipped with a mercury cadmium telluride (MCT) detector at a resolution of 4 cm⁻¹ using 32 scans. The samples were put in a furnace, and before CO adsorption, the samples were reduced in situ at 200 °C with 10% H₂/He for 0.5 h, cooled to room temperature in helium, and the background spectrum was recorded. After that, 1% CO/He was introduced to the sample for 10 min, and the spectra were collected till the steady state. Then, pure helium was introduced to the sample to remove the extra CO gas, and the spectra of chemisorbed CO species were collected.

STEM: Scanning transmission electron microscopy (STEM) analyses were performed using a JEOL JEM-2100F (JEOL Ltd., Tokyo, Japan) operated at 200 kV. Before measurements, the samples were ultrasonically dispersed in ethanol, and a drop of the solution was put onto carbon films.

AC-HAADF-STEM: AC-HAADF-STEM images were obtained on a JEOL JEM-ARM200F (JEOL Ltd., Tokyo, Japan) equipped with a CEOS probe corrector. The sample was ultrasonically dispersed in ethanol, and the resulting solution was dropped on the carbon film supported by a copper grid.

3.3. General Procedures for the Reaction

A 10.0 mL reusable plastic vial was charged with CA (0.2 mmol), Rh/TiO₂ (106 mg), D₂O (2.0 mL), magnetic stirrer. The vial was transferred into a 10.0 mL autoclave. Once sealed, the steel reactor was flushed with H₂ gas with marked press (which mean the value read on the reactor’s pressure gauge relative to ambient atmospheric pressure) at room temperature (typically 1.0 atm; 3 or 5 atm for pressure-variation experiments), and then the reactor was heated to 100 °C and stirred at 400 rpm for 6 h in oil bath, during which the pressure showed minimal change. After reaction, the autoclave was removed from the oil bath and cooled to room temperature. The H₂ was carefully discharged, and the vial was removed from the autoclave. The crude media was washed with Ethyl Acetate (EA) (20 mL), centrifuged, and the combined solution was concentrated under a rotary evaporator. Further purification was not necessary. Isolated yield was given, and deuterium incorporation was determined by means of ¹H NMR spectroscopy.

3.4. Procedures for the Recycling Reaction

The Rh/TiO₂ catalyst was used in the reaction (see Section 3.3 for details). The spent Rh/TiO₂ catalyst was recovered and washed with ethanol (3 × 20 mL), dried in an oven at 60 °C for 12 h, and re-reduced in 10% H₂/He at 200 °C for 2 h. The regenerated catalyst was then reused for the next cycle, and the same procedure was repeated for subsequent recycles.

3.5. Procedures for the Kinetic Experiments

The Rh/TiO₂ catalyst was used under the standard conditions; except for the reaction time, which differed from the 6 h specified in Section 3.3—we conducted runs of 0.5, 1, 3, and 6 h—all other procedures were identical to those in Section 3.3.

4. Conclusions

In summary, using an atomically dispersed Rh/TiO₂ catalyst, we synthesize d₄-substituted phenylpropionic acid via a one-pot sequence in which the C=C bond of cinnamic acid is first hydrogenated, followed by H/D exchange at the otherwise inert α and β positions with D₂O—overcoming a traditionally challenging barrier. Mechanistic evidence shows that α-C–H activation is the rate-limiting step and requires isolated Rh–O interfacial sites, whereas β exchange proceeds faster through a benzylic pathway. This reactivity—near-quantitative α,β-deuterium incorporation under mild conditions—surpasses Rh nanoparticles and homogeneous Rh systems and highlights a fundamental principle of single-atom catalysis. Building on these insights, we are extending this single-atom strategy to achieve site-selective deuteration of more general carboxylic acids.