A self-adjusting platinum surface for acetone hydrogenation

Significance

A collaborative approach containing reaction kinetics measurements under steady-state and transient conditions, electronic structure calculations employing density-functional theory, and microkinetic modeling on acetone hydrogenation is employed to provide insights into understanding the self-adjusting platinum surface for the hydrogenation of oxygenates over a wide range of reaction conditions. Elucidation of the repulsive interactions and the corresponding self-adjusting catalyst surface properties at various reaction conditions provides a basis to formulate rate expressions for heterogeneous catalytic processes used in the chemical industry.

Abstract

We show that platinum displays a self-adjusting surface that is active for the hydrogenation of acetone over a wide range of reaction conditions. Reaction kinetics measurements under steady-state and transient conditions at temperatures near 350 K, electronic structure calculations employing density-functional theory, and microkinetic modeling were employed to study this behavior over supported platinum catalysts. The importance of surface coverage effects was highlighted by evaluating the transient response of isopropanol formation following either removal of the reactant ketone from the feed, or its substitution with a similarly structured species. The extent to which adsorbed intermediates that lead to the formation of isopropanol were removed from the catalytic surface was observed to be higher following ketone substitution in comparison to its removal, indicating that surface species leading to isopropanol become more strongly adsorbed on the surface as the coverage decreases during the desorption experiment. This phenomenon occurs as a result of adsorbate–adsorbate repulsive interactions on the catalyst surface which adjust with respect to the reaction conditions. Reaction kinetics parameters obtained experimentally were in agreement with those predicted by microkinetic modeling when the binding energies, activation energies, and entropies of adsorbed species and transition states were expressed as a function of surface coverage of the most abundant surface intermediate (MASI, C₃H₆OH*). It is important that these effects of surface coverage be incorporated dynamically in the microkinetic model (e.g., using the Bragg–Williams approximation) to describe the experimental data over a wide range of acetone partial pressures.

A heterogeneous catalyst needs to adsorb the reactants, products, and reaction intermediates that cover 50% of the surface for optimal catalyst performance (1–4). Thus, the nature of the optimal catalyst should depend on the reaction of interest and the reaction conditions. We show in this paper, however, that platinum displays a self-adjusting surface that is active for hydrogenation of acetone over a wide range of reaction conditions because the surface binding energies are correlated with the surface coverage, such that the surface coverage remains near 50% (i.e., the effects of increasing partial pressure are compensated by weakening of surface binding energies). In addition, the inclusion of coverage effects in microkinetic models leads to predictions that explain empirically derived rate expressions that have been used effectively in the literature to describe experimentally observed reaction kinetics over a wide range of reaction conditions, based on assumptions of surface nonuniformity (e.g., such as incorporated in Tempkin adsorption isotherms) (5, 6). In this way, the surface containing uniform platinum sites behaves like a nonuniform surface in that it displays a range of surface binding energies (7). Moreover, these effects of surface coverage allow the surface properties of Pt to adjust to the reaction conditions, such that platinum-based catalysts remain highly active over a wider range of reaction conditions than would be achieved if the surface properties were independent of surface coverage.

The hydrogenation of carbonyl groups is an important reaction to produce chemicals in the fuel and chemical industries (8–10), and platinum-based catalysts have been widely used for these reactions (11–13). Although the adsorption of acetone has been investigated thoroughly (11, 14, 15), the binding of acetone and acetone-derived intermediates in the presence of hydrogen [i.e., competitive (8, 16) versus noncompetitive (17) site blocking] on partially covered surfaces during hydrogenation reactions has not been elucidated in detail.

In this study, we explored the effects of surface coverage on the kinetics of acetone hydrogenation on a 3 wt % Pt/SiO₂ catalyst. Reaction kinetics experiments were carried out in a flow reactor at atmospheric pressure, and the concentrations of the reactant and the product were measured by an online gas chromatograph equipped with a barrier discharge ionization detector (BID). We investigated the characteristics of the catalyst surface during reaction by measuring the reaction kinetics under steady-state and transient conditions (16, 18–20). Transient response profiles for desorption and steady-state isotopic transient kinetic analysis (SSITKA) experiments were used to elucidate further the effects of surface coverage on the surface properties at the hydrogenation conditions. To explain the observed reaction kinetics trends, we calculated the free-energy landscape on a clean Pt(111) surface and on the surface in the presence of a spectator species (i.e., C₃H₆OH*) using density-functional theory (DFT). Furthermore, we developed a microkinetic model to explore the significance of coverage-dependent energetics on the reaction kinetics using both Langmuir and Bragg–Williams approximations, in which the former approach neglects adsorbate/adsorbate interactions while the latter estimates these effects linearly in terms of the occupancy of the nearest-neighbor sites (5, 21).

Results

Experimental Reaction Kinetics Studies under Steady State.

At the experimental reaction conditions employed in this study (SI Appendix), we measured the turnover frequency (TOF) for acetone hydrogenation to be 0.17 s⁻¹ at 10% conversion. The apparent activation energy was determined to be 30 ± 3 kJ/mol in the temperature range of 333–373 K. The reaction orders with respect to the hydrogen and acetone partial pressures were 0.69 ± 0.19 and −0.36 ± 0.03, respectively. The positive hydrogen order indicates that hydrogen participates in kinetically relevant reaction steps, and the negative acetone order suggests that either acetone or an acetone-derived oxygenated intermediate is abundant on the platinum surface. These experimental findings are in agreement with literature data (8, 16).

Platinum-based catalysts are known to be active for hydrogenation of oxygenates over a wide range of reaction conditions. We measured the reactivity trend of acetone hydrogenation on Pt/SiO₂ with respect to acetone partial pressure (0.02–0.29 atm) for various acetone concentrations. Fig. 1 shows that the reactivity (navy diamonds) increases with decreasing acetone partial pressure. This behavior also suggests that the catalyst surface is blocked by acetone or hydrogenated-acetone intermediate, and its properties are self-adjusted based on the coverage of the most abundant surface intermediate.

Fig. 1.

Turnover frequency (s⁻¹) for acetone hydrogenation obtained from experimental measurements (navy diamonds) and from microkinetic models with the Bragg–Williams approximation (case 3, red circles) and using Langmuir isotherms (case 5, gray triangles) (left y axis). Coverages of C₃H₆OH* (orange rectangles) and H* (dark-cyan stars) estimated by the microkinetic model with the Bragg–Williams approximation (right y axis) with respect to acetone partial pressure. Experimental errors were found by calculating the SD between replicates. Error bars in the data obtained from the model represent the 95% confidence intervals.

Experimental Reaction Kinetics Studies under Transient Conditions.

To understand further the effects of surface coverage on the behavior of the Pt catalyst for acetone hydrogenation, we monitored the transient response of the catalyst. These experiments were carried out in a manner similar to traditional SSITKA techniques, in which the reaction is allowed to reach steady state before switching the initial feed mixture to a second feed stream containing an isotopically labeled reactant of interest. This method allows the catalyst surface to remain undisturbed during the switch. The transient response was then studied by monitoring the decay of the isopropanol product relative to that of an inert Ar tracer following a switch in feed in which the acetone reactant was either removed from the feed (desorption) or was substituted by a similarly structured ketone (e.g., 2-pentanone). The results of these transient response experiments are shown in Fig. 2. The area between the decay response of the argon tracer and the isopropanol product is used to estimate the surface residence time of intermediate species leading to the formation of isopropanol (τ_IPA) during both transients. Accordingly, surface residence times of 3.1 and 6.4 s were obtained from the desorption (Fig. 2A) and reactive (Fig. 2B) transients, respectively. The difference in residence times observed between both transients suggests that isopropanol is formed rapidly during both experiments, whereas further isopropanol is formed versus time during the SSITKA experiment. In this respect, species leading to isopropanol are formed during the SSITKA transient by displacing them from the surface after the introduction of the reactive species (from 2-pentanone), whereas these species become more strongly adsorbed on the surface as the coverage decreases during the desorption experiment. This behavior can be seen more quantitatively by estimating the number of surface species that lead to formation of isopropanol during the response of the catalyst for the two types of transients. The number of surface species that lead to formation of isopropanol was found to be four times larger for the reactive transient compared to the desorption case, indicating the larger extent to which adsorbed intermediates are removed from the surface by chemical displacement and highlighting the importance of surface coverage effects in the hydrogenation of acetone over platinum catalysts.

Fig. 2.

Normalized response profiles for an argon tracer (black line), acetone (red dotted line), and isopropanol (blue dashed line) during (A) desorption and (B) SSITKA transients. Reaction conditions: 10 mg 7.7 wt % Pt/α-Al₂O₃, 353 K, 1 atm, 20 cm³(STP)/min total flow rate.

DFT Calculations.

To explain our experimental observations, DFT calculations were performed using the Vienna ab initio simulation package (VASP) (22, 23). Exchange-correlation energies were obtained using the functional by Perdew, Burke, and Ernzerhof (PBE) (24) with a plane-wave cutoff of 600 eV. The semiempirical C₆/R⁶ term by Grimme (DFT+D2) (25) was added to correct for missing van der Waals interactions. The Pt(111) surface is modeled using a p(3 × 3) model with three atomic layers. Further computational details are described in SI Appendix. Based on results from DFT calculations, the hydroxypropyl pathway via the C₃H₆OH* intermediate was found to be the favored pathway for producing isopropanol from acetone [steps s1–s5].

$${\text{C}}_3{\text{H}}_6\text{O}+*\to {\text{C}}_3{\text{H}}_6\text{O}*,$$ [s1]

$${\text{H}}_2+2*\to 2\text{H*,}$$ [s2]

$${\text{C}}_3{\text{H}}_6\text{O}*+\text{H}*\to {\text{C}}_3{\text{H}}_6\text{OH}*+*,$$ [s3]

$${\text{C}}_3{\text{H}}_6\text{OH}*+\text{H}*\to {\text{C}}_3{\text{H}}_7\text{OH}*+*,$$ [s4]

$${\text{C}}_3{\text{H}}_7\text{OH}*\to {\text{C}}_3{\text{H}}_7\text{OH}+*.$$ [s5]

Gibbs free-energy values for elementary steps on Pt(111) for the hydroxypropyl pathway are shown in Fig. 3 (black pathway), along with the corresponding intermediate and transition-state structures. The calculated energetics on the clean surface (Table 1) agree well with recent reports (16, 26); however, the effects of surface coverage on these values have not been thoroughly investigated in the literature for competitive adsorption models. To account for coverage, reaction energetics for the hydroxypropyl pathway were recalculated in the presence of a C₃H₆OH* species $\left({\theta }_{\text{spectator}\left(\text{i}.\text{e}.,\text{initial}\hspace{0.17em}{C}_3{H}_{6}O{H}^{\ast }\right)}=1/9,{\theta }_{\text{C}3\left(\text{i}.\text{e}.,\text{final}\hspace{0.17em}{C}_3{H}_{6}O{H}^{\ast }\right)}=2/9\right)$, which was predicted by the microkinetic model to be the most abundant surface intermediate (MASI). In the presence of two spectator species (i.e., C₃H₆OH*), acetone can no longer bind to the p(3 × 3) surface model due to steric hindrance and adsorbs atop the C₃H₆OH* layer instead. Thus, the maximum ${\theta }_{\text{C}3}$ coverage that can be calculated using our DFT model is between 2/9 and 3/9. This behavior may justify the assumption of noncompetitive adsorption in previous microkinetic models that did not explicitly calculate coverage dependencies: pristine Pt(111) sites readily form C₃H₆OH*, but acetone cannot bind next to C₃H₆OH* due to steric hindrance. Thus, Pt(111) sites next to C₃H₆OH* can only bind hydrogen.

Fig. 3.

Gibbs free-energy pathways (PBE+D2) for acetone hydrogenation on the clean surface (black) and in the presence of a C₃H₆OH* species (red). The corresponding structures are shown below the plot using the following color code: C (black), H (white), O (red), and Pt (gray). Transition states are indicated by a dashed box.

Table 1.

DFT-calculated values on clean Pt(111) (denoted by ${\theta }_{C_3{H}_{6}O{H}^{\ast }}=0$) and on surfaces covered by an adsorbed C₃H₆OH* (denoted by ${\theta }_{C_3{H}_{6}O{H}^{\ast }}=0.11$)

Species/elementary step	Binding energy, kJ/mol
	${\theta }_{C_3{H}_{6}O{H}^{\ast }}=0$	${\theta }_{C_3{H}_{6}O{H}^{\ast }}=0.11$	$w_x$	${\mathrm{\Delta }}_{BW}$	$\mathrm{\Delta }{w}_x$	${\mathrm{\Delta }}_{LA}$
C3H6O*	−89	−85	32	—	24	—
H*	−266	−258	80	24	92	—
C3H6OH*	−282	−244	346	—	−72	−5
C3H7OH*	−99	−94	48	—	—	—
	Adsorption entropy, J/mol/K
C3H6O*	135	111	−222	—	—	—
H*	9	9	—	—	—	—
C3H6OH*	109	78	−276	47	161	−1
C3H7OH*	133	101	−285	—	—	—
	Activation energy, kJ/mol
Step s3	21	28	59	—	—	−5
Step s4	86	51	−304	−5	92	1
	Entropy of transition state, J/mol/K
Step s3	129	98	−284	—	−21	—
Step s4	123	82	−366	—	−61	−3

${\mathrm{\Delta }}_{BW}$ and ${\mathrm{\Delta }}_{LA}$ represent the adjustments made in the values for the clean surface in the microkinetic model with Bragg–Williams and Langmuir approximations, respectively. $w_x$ is the Bragg–Williams coverage coefficient obtained as shown in SI Appendix, Figs. S4–S7 and used in SI Appendix, Table S3, Eqs. 5–11. $\mathrm{\Delta }{w}_x$ is the adjustment in the DFT-calculated Bragg–Williams coefficient. A Pt(111) site occupied by a C₃H₆OH* species cannot have another C3 species binding in nearest-neighbor position due to steric hindrance. Thus, 0.11 ML is 1/3 of the saturation coverage of C₃H₆OH* on the catalyst surface, agreeing with the predicted coverage of C₃H₆OH* species in Fig. 1.

The adsorption of surface species was destabilized due to interactions with the MASI, thereby weakening the binding energies of adsorbed species (Table 1), as suggested in previous reports (27). However, while the binding energy of H* is only weakened by 9 kJ/mol in the presence of coadsorbed C₃H₆OH*, a second C₃H₆OH* species is destabilized by 31 kJ/mol. As a result, the first hydrogenation step becomes mildly endothermic in the presence of the spectator species (red pathway in Fig. 3). The presence of the MASI also leads to loss of entropy by inhibiting the translation and rotation of surface species. Furthermore, the energy barrier for the rate-limiting second hydrogenation step was lowered from 86 to 51 kJ/mol as the reaction became less endergonic (Fig. 3, red pathway, Table 1).

Microkinetic Modeling.

As described in the previous report (3), we used thermodynamic data for gas-phase species and properties of adsorbed species obtained through DFT calculations to construct a microkinetic model. At the typical reaction conditions of our experiments (353 K; H₂ partial pressure of 0.71 atm, acetone partial pressure of 0.29 atm), we used the microkinetic model to predict the overall TOF by first calculating the rate constants for the adsorption–desorption of reactants and product (step s1, s2, and s5) using collision theory (SI Appendix, Table S3, Eq. 1). For the hydrogenation steps (step s3 and s4), the forward rate constants were calculated using transition-state theory (SI Appendix, Table S3, Eq. 2). Applying the DFT-derived properties on the clean Pt surface (Table 1), the model significantly underestimated the hydrogenation activity, with negligible conversion of acetone (Table 2, case 1). Additionally, a slightly positive hydrogen order and a strongly negative acetone order were predicted when coverage effects were not considered. The predicted activation energy was 7 times larger than the experimental barrier, resulting in the low reactivity.

Table 2.

Reaction kinetics parameters obtained from experimental results (Exp) and predicted by the microkinetic models

Kinetic Parameter	Exp	Microkinetic model
		Case 1	Case 2	Case 3	Case 4	Case 5
TOF, s−1	(0.17 ± 0.01)	2.0·10−6	1.6·104	0.17	0.17	0.17
m	0.69 ± 0.19	0.26	0.47	0.56	0.68	0.55
n	−0.36 ± 0.03	−0.80	0.19	−0.29	0.53	−0.35
$\mathrm{\Delta }{\mathrm{H}}_{\mathrm{a}\mathrm{p}\mathrm{p}}^{‡}$	30 ± 3	238	71	38	37	39

Case 1: microkinetic model using clean surface properties; Case 2: microkinetic model using the properties on C₃H₆OH*-decorated Pt(111) surface; Case 3: microkinetic model with coverage-dependent properties using Bragg–Williams approximation; Case 4: microkinetic model with coverage-dependent properties using Langmuir approximation without ${\mathrm{\Delta }}_{LA}$ adjustments; Case 5: microkinetic model with coverage-dependent properties using Langmuir approximation) for 3 wt % Pt/SiO₂. Reaction orders are given for hydrogen (m) and acetone (n), and $\mathrm{\Delta }{H}_{app}^{‡}$ is the apparent activation energy in kJ/mol. Forward and reverse rate constants calculated by the microkinetic models for each case are given in SI Appendix, Table S5. Model-predicted reaction kinetics parameters were calculated at acetone conversion near 10%. Experimental error was found by calculating the SD between replicates. Reaction conditions: 125–250 mg Pt/SiO₂ mixed with 1.3–2.5 g α-Al₂O₃, T = 333–373 K, P = 1 atm, Feed: acetone (liq.) = 0.02–0.1 mL/min, H₂ = 25–300 cm³ (STP)/min, He = 0–117 cm³ (STP)/min.

A microkinetic model consistent with the observed reaction kinetics was obtained by including coverage effects employing the Bragg–Williams approximation. In this approach, the binding energies of adsorbed species, the activation energies of kinetically relevant elementary steps (step s3–s4), and the entropies of adsorbed intermediates and transition states are calculated dynamically as a function of the surface coverage by the C₃H₆OH* MASI using SI Appendix, Table S3, Eqs. 5–11. The Bragg–Williams coverage coefficients, calculated as the slope of the surface property versus coverage of the MASI using the values at ${\theta }_{{\text{C}}_3{\text{H}}_6{\text{OH}}^{\ast }}=0$ and ${\theta }_{{\text{C}}_3{\text{H}}_6{\text{OH}}^{\ast }}=0.11$ (Table 1 and SI Appendix, Figs. S4–S7), are represented as $w_{\text{x}}$ in the rate constant and equilibrium constant equations (SI Appendix, Table S3). Without any adjustment in the Bragg–Williams coefficients, the model predicted a larger TOF value compared to the experimental reactivity (Table 2, case 2).

To probe the importance of individual DFT-estimated values on the reaction kinetics of acetone hydrogenation, we carried out sensitivity analyses, where we calculate the change in the overall TOF with respect to the change in each DFT-predicted value (SI Appendix, Table S4). Based on the results from these analyses, we found that the binding energies of H* and C₃H₆OH*, the entropy of C₃H₆OH*, the activation energy of step s4, and the entropy of transition state for step s4 influence the reactivity predominantly. The most significant Bragg–Williams constants and the clean surface energies were then adjusted by $\mathrm{\Delta }{w}_x$ and ${\mathrm{\Delta }}_{BW}$, respectively, as listed in Table 1. This model predicted that the surface coverage by the hydrogenated acetone species (C₃H₆OH*) is equal to ∼0.29, which is in the range of the maximum C₃ coverage predicted by the DFT calculations.

The overall agreement between the experimental and model-predicted reaction kinetics is excellent when the effects of surface coverage are incorporated dynamically by the Bragg–Williams approximation (Table 2, case 3). We calculated the degree of rate control introduced by Campbell (28, 29) to identify the rate-determining steps by determining the change in overall reaction rate that results from an incremental increase in the forward and reverse rate constants of that step, while keeping the equilibrium constant for that step and all other rate constants constant. This analysis revealed that the first and second hydrogenations of acetone (steps s3 and s4) are kinetically significant, with the second being the step with the highest degree of rate control.

Most microkinetic models in the literature incorporate the effects of surface coverage statically, by adjusting the DFT-calculated values and then using the Langmuir isotherm, where the changes in the binding energy, activation energy, adsorbate entropy, and transition-state entropy are adjusted from the values on the clean surface, but these values are not dynamically correlated with the surface coverage. There have been several studies showing the importance of employing coverage-dependent energetics as a function of the MASI coverage in microkinetic models to accurately predict the reactivity (7, 27). Therefore, we employed this approach to describe acetone hydrogenation by using the surface properties calculated from the Bragg–Williams approximation at a surface coverage by the MASI of ${\theta }_{{\text{C}}_3{\text{H}}_6{\text{OH}}^{\ast }}=0.29$, and by holding these values to be constant, thereby employing static inclusion of coverage effects. Without any modification in these values, the model estimated the TOF similar to case 3; however, the reaction orders were significantly different compared to the experimentally obtained orders (Table 2, case 4), requiring further adjustments in the microkinetic model. By implementing the adjustments $\left({\mathrm{\Delta }}_{LA}\right)$ in Table 1, the TOF predicted by the model agreed well with the experimental reactivity, as well as the reaction orders and activation energy (Table 2, case 5). The surface coverage by the C₃H₆OH* MASI was 70%. The model with Langmuir approximation was sufficient to describe the observed reaction kinetics data for acetone hydrogenation; however, this model required larger adjustments $\left({\mathrm{\Delta }}_{BW}+\mathrm{\Delta }{w}_x+{\mathrm{\Delta }}_{LA}\right)$ compared to the adjustments made in case 3 $\left({\mathrm{\Delta }}_{BW}+\mathrm{\Delta }{w}_x\right)$.

In addition to the experimental reaction trends with respect to acetone partial pressure in Fig. 1, we also illustrated the reactivity predictions using the coverage-dependent microkinetic model with Bragg–Williams equations (case 3) and with the Langmuir model (case 5). Importantly, the trends displayed in Fig. 1 for the variation of the TOF with acetone partial pressure show close agreement between the experimental data and the predictions of the Bragg–Williams model (navy diamonds and red circles), indicating the importance of including the effects of surface coverage in the microkinetic model. Importantly, the microkinetic model of case 5 built using Langmuir isotherms does not describe the reactivity trends over a wide range of acetone partial pressures (gray triangles), even though the DFT-predicted parameters have been adjusted to account for effects of surface coverage. Specifically, whereas the Langmuir model can be formulated to describe the experimental data over a narrow range of reaction conditions (e.g., at the higher acetone partial pressures in Fig. 1), it predicts a decrease in TOF at lower acetone partial pressures, which was not observed in the experiments. Thus, we conclude that the effects of surface coverage on the surface energetics must be incorporated dynamically in the microkinetic model (e.g., using the Bragg–Williams approximation), in contrast to being incorporated statically (e.g., using Langmuir isotherms). In Fig. 1, we also show the coverages of C₃H₆OH* and of H* calculated by the Bragg–Williams model on the right y axis. The coverage of C₃H₆OH* slightly decreases while that of H* increases with decreasing acetone content in the reactor. These changes on the catalyst surface result in the activity enhancement at lower acetone pressures.

Conclusions

In conclusion, the elucidation of surface coverage effects for the calculation of reaction kinetics parameters is essential to understand and describe the reaction kinetics for hydrogenation of acetone over a wide range of reaction conditions. The presence of the C₃H₆OH* MASI affects the competitive binding of C₃H₆O*/C₃H₆OH*/H* differently, and these effects of surface coverage must be included dynamically in DFT-based microkinetic models (e.g., using the Bragg–Williams approximation), in contrast to being included statically (e.g., using Langmuir isotherms).

Methods

Experimental Method.

A Pt/SiO₂ catalyst was prepared by incipient wetness impregnation (30). The catalyst was packed in a stainless-steel tubular reactor. After in situ reduction of the catalyst, acetone hydrogenation was carried out in the flow-through system shown schematically in SI Appendix, Fig. S1. Reaction kinetics parameters were measured by flowing a mixture of H₂ and He and by introducing acetone to the reactor using a high-performance liquid chromatography (HPLC) pump at temperatures ranging from 333 to 373 K and 1 atm. The concentrations of acetone and 2-propanol were monitored by an online gas chromatograph (Shimadzu GC-2014) equipped with a BID. The TOF of 2-propanol production was calculated using the platinum surface site density of the fresh catalyst, determined by CO chemisorption.

Computational Method.

DFT calculations were conducted using VASP (22, 23) and implementing the projector augmented wave (PAW) method (31, 32). The functional by PBE was applied to determine exchange-correlation energies (25). van der Waals interactions were corrected using the semiempirical C₆/R⁶ term by Grimme (DFT+D2) (25, 33). Microkinetic models were developed in MATLAB as described in a previous report (3).

Further details of the experimental and computational methods are included in SI Appendix.

Code and Data Availability.

The computer code and data that support the plots within this paper and other findings of this study are available at the following DOI: 10.6084/m9.figshare.11338640 (34).