Air Regeneration of Ethanol-Laden Pellet NaY-SiO₂ and Pt/NaY-SiO₂: Effects of Air Flow Rate on Pt Morphology and Regeneration Efficiency

Abstract

Regeneration process and adsorbent performance were investigated by a fixed-bed adsorber at 300 °C. Surface species, zeolite structure, and Pt morphology were characterized by FT-IR, XRPD and EXAFS, respectively. Performance test results indicated that ethanol adsorption capacity of Pt/NaY-SiO₂ is about 2.5 times that of NaY-SiO₂. After regeneration, adsorption-capacity loss is 2.5 and 43%, respectively, for Pt/NaY-SiO₂ regenerated at superficial velocity of 13.2 (PtR_(HF)) and 5.3 cm/min (PtR_(LF)); in contrast, it is 8 and 21%, respectively, for NaYR_(HF) and NaYR_(LF). The appearance of absorption bands in the CH stretching region (υ_CH) of the IR spectra characterizing the regenerated NaY-SiO₂ suggested that the adsorption-capacity loss for NaY-SiO₂ was mainly caused by the deposition of carbonaceous species formed in regeneration, which cannot be burned off readily at 300 °C. In contrast, no υ_CH bands have been observed for the IR spectra of PtR_(HF) and PtR_(LF), indicating that Pt helps to burn off carbonaceous species. However, Pt agglomeration was observed in TEM and EXAFS for Pt/NaY-SiO_2(LF). The appearance of a υ_CO band at about 2085 cm⁻¹ of the IR spectra characterizing PtR_(LF) suggested that Pt agglomeration was induced by CO adsorption. The growth of Pt particles decreases the ethanol adsorbed on Pt together with the conversion of ethanol to ethoxides and aldehyde, leading to a decrease of adsorption capacity.

1. Introduction

The effluents from industrial processes often contain VOCs (volatile organic compounds), which pose a threat to human health and the environment [1,2,3]. For low VOC concentration cases, a three-step process can be used [4]. In this process, the VOCs are first concentrated by adsorption at low temperature until the breakthrough occurs. The adsorbent is regenerated by passing a small quantity of air or other gas to generate a VOC concentrated process stream, which is converted to innocuous CO₂ in an incinerator. To improve the process, catalytic components can be added to promote both adsorption and incineration, and the process steps are integrated into a single reactor. In the integrated catalytic/adsorption process, fixed bed reactors containing catalytic adsorbents are employed to continuously adsorb VOCs from the effluents in one bed and convert VOCs on the catalytic adsorbent into benign carbon dioxide and water in the other [5,6,7]. Compared with the conventional adsorber-incinerator, the system is relatively simple, high in VOC removal efficiency, and low in operation and maintenance cost. However, for the catalytic/adsorption process, extra fuel input for air regeneration is still necessary and is a key criterion for assessing the process efficiency. In order to reduce the use of extra fuel, the catalytic adsorbent should have dual functions, which serves to adsorb VOCs rapidly and oxidize the adsorbed VOCs catalytically and completely at modest temperatures. In addition, the catalytic adsorbent should be hydrothermally stable to allow repeated adsorption and regeneration cycles [8].

NaY-SiO₂ extrudate was used as an adsorbent and was prepared by blending NaY powder with silica gel, and followed with kneading, extruding and calcination [9]. NaY zeolite was chosen because: (1) this material is non-flammable with high thermal stability, allowing high-temperature regeneration; (2) the surface area is high (>500 m²/g), which allows metal or metal oxides to be dispersed on the surface, thereby enhancing catalytic oxidation activity; and (3) Na form zeolite may suppress the formation of coke precursors during the regeneration [9,10,11]. Noble metals, such as Pt, have catalytic oxidation activity for VOC [7,11,12,13,14,15] and were incorporated into NaY-SiO₂ by means of impregnation techniques to prepare the catalytic adsorbents.

In the previous paper, the goals were to investigate the effects of Pt on the adsorption performance of Pt/NaY-SiO₂ and explain how Pt enhances the adsorption capacity while decreasing the adsorption rate [14]. The experimental results indicated that the adsorption capacities of NaY-SiO₂ in removing ethanol from water-containing air stream are greatly enhanced by the addition of Pt and the active Pt clusters on NaY play the main roles in catalytically converting the adsorbed ethanol to ethoxides and aldehyde. Those Pt-adsorbed species could spill over to displace sodium and acid site bound water, leaving the Pt clusters for further adsorption, and thus increasing ethanol adsorption capacity.

In the regeneration, the air flow rate is the key parameter. The flow rate has to be sufficient so as to: (1) supply enough oxygen for combustion; (2) minimize the resistance of component transfer; (3) carry away the heat generated to prevent overheating the bed and to form a sharp moving combustion band; and (4) to prevent the morphology change of catalytic adsorbents and the corresponding performance-deterioration. Hence, the goals of this study are to study the effects of flow rate on the regeneration performance and to investigate the roles of Pt in increasing regeneration efficiency. Moreover, the importance of sufficient air in successful regeneration and how air supply in regeneration affects Pt morphology will be reported as well.

An empirical model was used to describe experimental breakthrough curves, and the parameters of the model were used to access the regeneration efficiency. The surface-bound species in regeneration were characterized by DRIFT (diffuse-reflectance infrared Fourier transform). The morphology of Pt clusters, the oxidized Pt species formed in regeneration, and the interactions between Pt and support were characterized by EXAFS (extended X-ray absorption fine structure) spectroscopy. The characterization results combined with the parameters estimated from the model were used to rationalize the change of Pt morphology in regeneration and its corresponding effects on regeneration efficiency.

EXAF measurements performed in transmission mode present a much higher signal-to-noise ratio than those in fluorescence mode; thus, high Pt–Pt shell contributions, oxidation of Pt, and Pt–support interactions would not be affected significantly by noise. To facilitate EXAF study, a high Pt precursor dosage of 1 wt. % was used to prepare Pt/NaY-SiO₂, so that EXAF can be measured in transmission mode [16]. However, because of the rather high metal loading, Pt clusters are more likely to be aggregated. For industrial application, noble metal loading is designed to provide sufficient catalytic oxidation activity to ensure no carbon residue deposited on the catalytic adsorbent even after the deterioration of adsorption capacity due to the destruction of the zeolite frameworks with several regeneration cycles. Hence, optimal Pt loading should be investigated further.

2. Results and Discussion

2.1. Performance Tests

The dynamics of fixed-bed adsorption (breakthrough curve) for the fresh and air-regenerated NaY-SiO₂ and Pt/NaY-SiO₂ are shown in Figure 1 and Figure 2, respectively. The stoichiometric dynamic adsorption capacity was calculated by integrating the area above the whole breakthrough curve. The results indicated that the adsorption capacity of NaY-SiO₂ was decreased 8 and 21% after the 300 °C regeneration at high (v_s = 13.2 cm/min) and low flow rates (v_s = 5.3 cm/min), respectively; the samples were noted as NaYR_(HF) and NaYR_(LF), respectively, for high and low flow rates. The adsorption capacity was only decreased 2.5% for Pt/NaY-SiO₂ regenerated at high flow rate (PtR_(HF)), whereas it was 43% at low flow rate (PtR_(RF)).

As shown in Figure 1, only ethanol was detected for NaY-SiO₂. In contrast, for Pt/NaY-SiO₂, both ethanol and a small amount of acetaldehyde were detected at the end of the adsorption cycle (Figure 2). Acetaldehyde is formed from the oxidative dehydrogenation of ethanol catalyzed by Pt clusters. Much lower acetaldehyde formation was detected for PtR_(LF), as opposed to that for PtR_(HF) and for fresh Pt/NaY-SiO₂, suggesting loss of Pt activity in low-flow-rate air regeneration.

During the performance tests, the adsorption/reaction zone moves through the adsorber bed; adsorbent is saturated behind the adsorption/reaction zone and is free of adsorbate in front of the zone. The length of the adsorption/reaction (A/R) zone is a function of the mass transfer rate of the ethanol adsorbate from the gas stream to the active sites of the catalytic adsorbent and is represented by the steepness of breakthrough curve. In cases where the mass-transfer rate is decreased, the adsorption/reaction zone broadens and the breakthrough curve becomes less steep. As shown in Figure 1 and Figure 2, the breakthrough curve for Pt/NaY-SiO₂ is not as steep as that for NaY-SiO₂, indicating that other steps were involved besides ethanol adsorption when Pt is present. Reported in our previous paper, ethanol could be adsorbed on Pt clusters first, catalytically dissociated to ethoxide, and then diffused the adsorbed ethanol and/or ethoxide to the adsorption sites of NaY [14]. Since all these steps take time, the mass transfer rate for Pt/NaY-SiO₂ is thus lower than that for NaY-SiO₂.

As shown in Figure 2, the adsorption capacity and mass transfer rate of PtR_(HF) are close to those of the fresh Pt/NaY-SiO₂, while the adsorption capacity of PtR_(LF) is lower but the mass transfer rate is slightly higher (Figure 2). The results are consistent with the loss of Pt activity for PtR_(LF) suggested by the decrease of acetaldehyde formed in the catalytic adsorption process.

2.2. Model for Breakthrough Curve

In order to assess the effects of air flow rate on the efficiency of regeneration quantitatively, a model for breakthrough curves was proposed. The differential material balance for a fixed adsorption bed is formulated as:

$$-D_l\frac{{\partial }^{2}C}{\partial z^2}+u\frac{\partial C}{\partial z}+\frac{\partial C}{\partial t }+\frac{{\rho }_p\left(1-\epsilon \right)}{\epsilon }\frac{\partial q}{\partial t}=0$$

(1)

$$t=0, q_r=q^0, z=0, C=C^0$$

(2)

In Equation (1), D_l (cm²·s⁻¹) is axial dispersion coefficient; ε is bed void fraction; c (mg/cm³) is gas phase adsorbate concentration in bulk flow; z (cm) is axial distance; u (=v_s/ε, cm/s) is interstitial velocity; t (min) is elapsed time; ρ_p (g/cm³) is apparent adsorbent density, and q (mg adsorbate/g adsorbent) is the adsorption capacity per unit mass of adsorbent [17,18].

In addition, the rate of mass transfer across the pellet (dq/dt) is equal to ethanol adsorbed on the catalytic adsorbents and is formulated by an empirical equation, ∂q/∂t = ηkaC(q0 − q); where C is ethanol concentration in gas phase, q is adsorption capacity, q0 is stoichiometric adsorption capacity, ka is adsorption constant, and η is a ratio of adsorption rates with component transfer resistance to adsorption rate without resistance [14].

Since the characteristic time, t, for adsorbate moving from the bed inlet to the outlet is much longer than the residence time of the transporting gas stream, L/u (L: length of adsorption bed), pseudo-steady-state approximation, ∂C/∂t = 0, was assumed for Equation (1). Because L/d_p in this study is about 50, the axial dispersion term of Equation (1) is set at zero. After neglecting the unsteady and axial dispersion terms, Equation (1) can then be directly integrated [19]. The analytical solution is given as:

$$\frac{C\left(z,t\right)}{C^0}= \frac{\exp \left(\eta k_a{C}^{0}t\right)}{\exp \left(\frac{\eta k_a{q}^0{\rho }_{p}z}{u_{s }}\right)+\exp \left(\eta k_a{C}^{0}t\right)-1},u_s=\frac{\epsilon }{\left(1-\epsilon \right)} u$$

(3)

Since the exponential terms in Equation (3) are much greater than (1), the equation can be further simplified as:

$$\frac{C\left(L,t\right)}{C^0} \approx \frac{\exp \left(\tau t\right)}{\exp \left(\xi \right)+exp\left(\tau t\right)}, \xi =\frac{\eta k_a{q}^0{\rho }_{p}L }{u_{s }}, \tau =\eta k_a{C}^0$$

(4)

The stoichiometric adsorption capacity (q⁰) was used to evaluate the efficacy of regeneration, while η was used to characterize diffusion limitation. Since Ka is a constant, lumped ηKa can thus be used to evaluate the relative component transfer resistance before and after regeneration, although we were unable to decouple η and Ka.

The parameters in Equation (4) were estimated by fitting the breakthrough curves with least-squares method, and the results are summarized in

Table 1. Ethanol adsorption capacities and effectiveness factor obtained from breakthrough curves.

Adsorbent	τ, 1/min	ξ	q0, mg/g	(η × ka), L/mg·min
NaY-SiO2	0.172	13.21	57.8	2.29
NaYR(HF)	0.158	11.25	53.4	2.10
NaYR(LF)	0.163	9.94	45.8	2.17
Pt/NaY-SiO2	0.0573	10.84	142.2	0.76
PtR(HF)	0.0627	11.58	138.7	0.83
PtR(LF)	0.0831	8.97	81.1	1.11

Note: Adsorbent samples were tested at 38 °C with total GHSV of 450 h⁻¹ in flowing 40,000 ppm of ethanol with 700 ppm water moisture in air.

.

2.3. Regeneration Tests

As shown in Figure 3a,b, Pt catalytically oxidizes the surface species on NaY-SiO₂; hence, the total amount of CO₂ evolved from Pt/NaY-SiO₂ is about 6 times that from NaY-SiO₂ for both high and low air flow rate (

Table 2. Ethanol desorption and total oxidation capacities obtained from ethanol desorption and CO₂ formation profiles in air regeneration.

Adsorbent	Desorption mg/g	Total Oxidation mg/g
NaYR(HF)	30.8	14.0
NaYR(LF)	30.9	6.0
PtR(HF)	46.2	84.7
PtR(LF)	91.0	35.0

). There are two peaks appearing in the CO₂ evolved spectra of PtR_(HF) and PtR_(LF). The results suggest at least two different kinds of species on Pt/NaY-SiO₂. The peak appearing at low temperature (about 150 °C for PtR_(HF)) could be attributed to the oxidation of ethanol and acetaldehyde, while that at high temperature (300 °C) could be the characteristic peak of the combustion of carbon residue, which were formed from the oligomerization of ethanol-derivative species, such as ethylene and crotonaldehyde, conversion of acetone to mesityl oxide, and hydrocarbon decomposition [20,21]. Ethylene could be formed from dehydration of ethanol catalyzed by BrØnsted acid sites inside NaY and/or the decomposition of diethylether (DEE) [22]. Crotonaldehyde could be formed from aldol condensation of two molecules of acetaldehyde, which was formed from the dehydrogenation of ethanol. Acetone could be produced through aldol condensation of acetate, which could be formed by the oxidation of acetaldehyde [20]. In our previous study, characteristic peaks of ethoxides species have been detected by IR and EXAFS further suggested the coordination of ethoxide with Pt clusters via bridging hydrogen of ethanol (Pt^HOC₂H₅) [14]. Inferred from the paper reported by Ferencz et al., carbon was suggested to be formed from the decomposition of Pt–bound CH₃ species, presumably, through acetaldehyde and acetyl species [20].

The oxidation of ethanol in regeneration was strongly influenced by air flow rate. As shown in Figure 3b, the peak characterizing the oxidation of ethanol is shifted from 150 °C to about 290 °C and the total amount of CO₂ decreased by about 60% when superficial velocity was decreased from 13.2 to 5.3 cm/min (Table 2). In addition, a significant number of partial oxidation species, CO, were also formed at a low flow rate besides CO₂; in contrast, only trace amounts of CO were detected at a high flow rate (Figure 3b).

Concomitant with the oxidation of the adsorbed ethanol, ethanol and ethanol-derivative species were desorbed from Pt/NaY-SiO₂ and NaY-SiO₂. The main ethanol-derivative species for Pt/NaY-SiO₂ is acetaldehyde, and that for NaY-SiO₂ is diethylether (DEE). In our previous paper [9], protonated ethanol (C₂H₅OH₂^δ+) was detected in FT-IR spectroscopy characterizing ethanol adsorbed on NaY-SiO₂ at room temperature. Inferring from the literature that the protonated ethanol is more susceptible to elimination or substitution reaction [22], we suggested that DEE could be catalytically formed from the reaction of the protonated ethanol on NaY with undissociated ethanol during air regeneration. Ethylene could thus be produced by ethanol dehydration and/or DEE cracking. However, this could be due to quick oligomerization and/or strong adsorption, no significant amount of ethylene was detected in the effluent gas.

Based on the carbon balance calculated from the breakthrough curve (Figure 1) and profiles of evolved species (Figure 3a and Figure 4a), about 22 and 38% of ethanol-derivative species (Table 2) were estimated to be deposited on NaY-SiO₂ after air regeneration at high and low flow rates, respectively. These strongly adsorbed carbon residues are defined as coke. The paraffinic coke could be formed from the ethylene oligomerization and aromatics from cyclization followed by hydrogen transfer reactions [21,23].

Total oxidation, partial oxidation, dehydration, and desorption of the adsorbed ethanol are the competitive processes occurring in air regeneration. The addition of Pt to NaY-SiO₂ increases the oxidation rate and thus decreases the amount of ethanol desorption (Figure 4a,b). In addition, instead of DEE formed on NaY-SiO₂, acetaldehyde was formed from the dehydrogenation of Pt–associated ethoxy-group; as shown in Figure 4b, only trace DEE was detected in high flow rate regeneration. However, this could be due to insufficient air supply for oxidative dehydrogenation reaction; small amounts of DEE were also detected in low flow rate regeneration of Pt/NaY-SiO₂.

For pure NaY-SiO₂, ethanol is adsorbed on the BrØnsted acid site to form a hydrogen-bonded ethanol monomer (C₂H₅OH₂^δ+). This hydrogen-bonded monomer then reacts with an oncoming ethanol to form DEE directly (Figure 5a). The monomer intermediate could also adsorb the oncoming ethanol to form ethanol dimer, which indirectly produces DEE by elimination of water molecule (Figure 5b) [22].

Oxidative dehydrogenation of ethanol to acetaldehyde via ethoxy species may require the oxide surface to acquire a negative charge or hydroxyl group to abstract hydrogen from ethanol. For Pt/NaY-SiO₂ regenerated at high flow rate with sufficient oxygen, surface platinum-oxide was detected by EXAFS (Section 2.5). The oxide with negative charge facilitates the adsorption of ethanol, which dissociates heterolytically to yield an ethoxide bound to Pt and a proton bond to oxide (Figure 5c). The surface Pt–bound ethoxide intermediate then was dehydrogenated to acetaldehyde [24,25]. In contrast, for regeneration at low flow rate, no significant surface platinum-oxide species have been detected by EXAFS (Section 2.5). In this case, Pt clusters could be oxidized to form transient surface Pt–O, which catalytically dehydrogenate ethanol to form acetaldehyde. This transient Pt–O was quickly reduced by H₂ and/or CO in insufficient oxygen conditions, so that no Pt–O could be observed using EXAFS. The acetaldehyde could also be formed via the abstraction of hydrogen from Pt bound ethanol, Pt^HOC₂H₅, to form ethoxide intermediate, followed by abstraction of hydride from the ethoxide intermediate. The hydride was associated with surface proton to form hydrogen molecules (Figure 5d) [26].

In the presence of platinum, the affinity of ethanol to Pt–O may be higher than that to NaY, resulting in a shift of reaction toward oxidative dehydrogenation. However, when oxygen supply is insufficient, some of ethanol was adsorbed on NaY and dehydrated to produce DEE.

2.4. Characterization of Surface Species on NaY-SiO₂ and Pt/NaY-SiO₂ by FT-IR

There is residual coke deposited on the regenerated NaY-SiO₂, as shown by the IR spectra (Figure 5a), consistent with lower adsorption capacity of the regenerated NaY-SiO₂.

The presence of absorption bands in the CH stretching region of paraffinic groups (3000–2800 cm⁻¹) after in situ air treatment of EtOH–laden NaY-SiO₂ (Figure 6A) indicated the formation of coke on NaYR_(HF) and NaYR_(LF) [27]. In contrast, no absorption bands in this region being observed for Pt/NaY-SiO₂ suggests the roles of Pt in inhibiting coke formed on Pt/NaY-SiO₂ during regeneration. Since dehydration reactions and combustion (oxidation) reaction are competitive reactions in air regeneration, combustion of ethanol catalyzed by Pt minimizes the formation of coke precursors together with oligomerization reactions. In addition, facilitated by exothermic heat associated with the ethanol combustion, coke formed on NaY-SiO₂ can be catalytically oxidized to CO₂ quickly.

There were no aromatic-coke characteristic absorption bands in the range between 3000 and 3200 cm⁻¹ observed, suggesting that cokes formed on NaY-SiO₂ are mostly paraffinic (Figure 6A) [27]. The absorption bands at 2974 cm⁻¹ and 2842 cm⁻¹ were assigned to asymmetric stretching (ν_sym) and symmetric vibration (ν_asym), respectively, for CH₃, and the absorption bands at 2930 cm⁻¹ (ν_sym) and 2894 cm⁻¹ (ν_sym) were assigned to ν_sym and ν_asym, respectively, for CH₂. As shown in Figure 5a, band intensities in the CH stretching region (3000–2800 cm⁻¹) decreased with increasing air flow rate, whereas the relative intensity between CH₃ and CH₂ (I_CH3/I_CH2) did not vary significantly. Since I_CH3/I_CH2 is an indication on the chain length of the carbonaceous species, the results suggested that increasing air flow rate in regeneration reduces carbonaceous species formation due to the increase of thermal combustion reaction (Table 2) but does not affect chain propagation reaction.

Regeneration of Pt/NaY-SiO₂ showed the formation of CO in addition to CO₂ at low flow rate, while only trace CO was observed at high flow rate (Figure 6B). As expected, no characteristic peaks in the region of carbonyl stretching (υ_CO) were observed for PtR_(HF) (Figure 6B). In contrast, the υ_CO characteristic peak at 2085 with a shoulder at 2100 cm⁻¹ was observed for PtR_(LF) (Figure 6B). CO molecules have a stretching frequency of 2143 cm⁻¹. The shift of the peak to a lower frequency for PtR_(LF) suggests the adsorption of CO on Pt clusters. When CO is adsorbed on Pt clusters, electron back-donation transfer occurs from the d-orbital of Pt to anti-bonding 2π* orbital of the linearly coordinated CO molecule leading to a decrease of carbon-oxygen bond order together with υ_CO frequency [28].

The band peaking at 2085 cm⁻¹ is the same as that for CO adsorbed on Pt/NaY at room temperature reported in the literature [29]. The shift of the CO absorption bands to higher frequencies for the shoulder (2100 cm⁻¹) may be due to the interaction of foreign gases, O₂, with the chemisorbed CO. This effect was attributed to a decrease of the electron density at the Pt surface, which decreased the back-donation from the metal to the CO π* orbital, and thus increased the υ_CO frequency [30].

2.5. Effects of Air Flow Rate on the Structure of Pt/NaY-SiO₂

Even though the regeneration temperature (300 °C) is far lower than the zeolite damage temperature (600 °C) [31], the NaY framework might be destroyed due to hot spot and water formed in the combustion reaction. Hence, XRPD was used to examine any thermal destruction of NaY in air regeneration.

The comparison of synchrotron XRPD patterns for fresh Pt/NaY-SiO₂, PtR_(HF), and PtR_(LF) are shown in Figure 7. Since there are no significant differences in XRPD patterns between fresh Pt/NaY-SiO₂, PtR_(HF), and PtR_(LF), the results indicate that NaY structure was intact after regeneration; in other words, regeneration flow rate has no effects on zeolite frameworks. Lower adsorption capacity for PtR_(LF) should be attributed to other factors and Pt-morphology change was expected to be the most possible one. Since the Pt crystal size is too small to be characterized by synchrotron XRPD, the morphology of Pt clusters, as well as Pt–support interactions, were then characterized by EXAFS.

As shown in Figure 8, the k³-weighted Pt–Pt phase- and amplitude-corrected Fourier transforms show peaks appearing at 2.7, 3.9, 4.7, and 5.7Å, consistent with the metal-metal distance of bulk Pt clusters; k³-weighted Fourier transform minimizes low-Z (Z is atomic number) contributions, such as Pt–O; hence, the appearance of Pt–Pt characteristic peaks was not influenced by these contributions [16].

The Pt–Pt characteristic peak intensities for fresh Pt/NaY-SiO₂ are lower than those for PtR_(LF), while they are slightly higher than PtR_(HF). Since the amplitude of the Fourier transformed EXAFS function reflects the metal particle size [16,32], the higher amplitude intensity for PtR_(LF) suggests Pt sintering at low flow-rate regeneration. The results were confirmed by TEM images; as show in Figure 9a,c, the comparison of TEM images between Pt/NaY-SiO₂ and PtR_(LF) clearly shows that the average Pt particle size of PtR_(LF) is much bigger than that of Pt/NaY-SiO₂. In contrast, TEM images of Pt/NaY-SiO₂ and PtR_(HF) were difficult to differentiate. The results suggested that the lower amplitude intensity of PtR_(HF) (Figure 8) was attributable to the loss of Pt–Pt contribution caused by the formation of oxidic Pt species. With sufficient oxygen, surface Pt atoms could be oxidized concomitantly with the combustion reaction. The formation of oxidic Pt in high flow rate regeneration was confirmed by analyzing low-Z contributions of EXAFS.

A normal k² (2.68 < k < 14.3 Å⁻¹)-weighted Fourier transformation without correction was performed and the major contributions (Pt–Pt_{1st shell} + Pt–O) were isolated by inverse Fourier transformation in the range 1.73 < r < 3.32 Å. With the difference technique, a Pt–Pt was calculated and then subtracted from the isolated data. A Pt–Pt contribution was calculated that agreed as well as possible with the experimental results in the high k range (7.0 < k < 14.0 Å⁻¹); the low-Z contributions in this range are small. After the subtraction, the residual spectrum was expected to represent the contributions from Pt oxides and/or Pt–support interactions.

For the fresh Pt/NaY-SiO₂, the imaginary of Pt–O phase-corrected Fourier transform of the residual spectra displayed a peak at 2.1–2.2 Å (Figure 10). This peak was assigned to the interactions between platinum and zeolite oxygen [33,34]. In comparison with the residual spectrum of fresh Pt/NaY-SiO₂, a broad peak with shorter bond distance appears at 2.0–2.1 Å for PtR_(HF). The broad peak could be a lumped peak contributed by α–PtO₂ with a Pt–O bond distance of 2.05 Å [35] and Pt–support interactions.

As opposed to the residual spectra of PtR_(LF) and fresh Pt/NaY-SiO₂, the intensity of the residual spectrum for PtR_(LF) is much smaller, suggesting that oxidic Pt species formed on PtR_(LF) are insignificant. In addition, because the Pt–O contribution for PtR_(LF) is rather small and is greatly interfered with by noise, parameter estimation for this contribution was not attempted.

The structural parameters of fresh Pt/NaY-SiO₂ and PtR_(HF) were estimated by a nonlinear least-squares multiple-shell fitting routine [36]. The structural parameters are listed in

Table 3. Summary of EXAFS analysis results.

Shell	N	R		1000 × Δσ2	ΔE0
		(Å)		(Å2)	(eV)
EtOH.Pt-NaY
Pt–O	0.7 ± 0.1	2.17 ± 0.01		1 ± 1	−16 ± 1
Pt–Pt	7.4 ± 0.2	2.77 ± 0.01		3 ± 1	−2 ± 1
PtR(HF)
Pt–O	2.2 ± 0.1	2.06 ± 0.01	9 ± 1		−6 ± 1
Pt–Pt	6.4 ± 0.1	2.76 ± 0.01	3.6 ± 0.2		−4.3 ± 0.1
PtR(LF)
Pt–Pt	10.4 ± 0.1	2.76 ± 0.01	5.0 ± 0.1		−2.1 ± 0.1

Note: N, the coordination number for the absorber-backscatterer pair; R, the average absorber-backscatterer distance; Δσ², the difference in Debye-Waller factors between sample and standard; ΔE₀, the inner potential correction.

and the comparisons of experimental and calculated Pt–O are shown in Figure 10.

After air regeneration at high flow rate, the average Pt–O bond distance decreased from 2.17 to 2.06 Å, the coordination number increased from 0.8 to 2.2, and the Pt–Pt coordination number decreased from 7.4 to 6.4 (Table 3), confirming the oxidation of Pt clusters concomitant with the combustion of ethanol and its derivatives. In contrast, after regeneration at low flow rate, the Pt–Pt coordination number increased from 7.4 to 10.4. These results, combined with the increase of higher Pt–Pt shell intensity (Figure 8), suggest the growth of Pt clusters with morphology resembling bulk Pt.

The average Pt particle size on NaY-SiO₂ determined by the use of first-shell Pt–Pt coordination number (CN_Pt–Pt) of EXAFS measurements [37] is 1.3 and 3.3 nm, respectively, for fresh Pt/NaY-SiO₂ and PtR_(LF), while that determined from TEM images is about 1.5 and 3.8 nm, respectively (Figure 9a,c). The larger Pt particle size determined by TEM could be because: (1) the coupling between CN and the relative Debye-Waller factor (Δσ) of EXAFS equation causes the inherent uncertainty of coordination numbers (CN) estimation; and (2) Pt clusters with sizes less than 1 nm are difficult to resolve by TEM [38].

Oxidic Pt species were formed in Pt clusters during air regeneration with high flow rate. Because the formation of these oxidic species decreases Pt–Pt contributions, we are unable to estimate Pt particle size by the use of the correlation of CN_Pt–Pt with particle size reported in literature [37]. However, it is shown in TEM images that the average Pt particle size of PtR_(HF) is close to that of fresh Pt/NaY-SiO₂ (Figure 9a,b), suggesting that only few or no Pt clusters were aggregated in high flow regeneration.

2.6. Effects of Air Supply Rate on Regeneration Efficiency

For efficient and complete regeneration of the adsorbent, adequate supply of oxygen is necessary. Based on stoichiometry calculation, without desorption, it takes about 45.6 mmole O₂ per 5 g of ethanol-laden Pt/NaY-SiO₂ adsorbent. Besides total oxidation, partial oxidation, and ethanol desorption also took placed in air regeneration. In high flow rate regeneration, an air supply of 61.2 mmole is enough for ethanol total oxidation; notwithstanding sufficient oxygen supply, about one-third of ethanol was desorbed from Pt/NaY-SiO₂, concomitant with total oxidation. However, in low flow rate regeneration, corresponding to low air supply (24.5 mmole) and low partial pressure of oxygen, more ethanol was desorbed, more acetaldehyde was produced, and some CO was also formed. The formation of CO greatly deteriorates regeneration efficiency.

The mechanism for the aggregation of Pt clusters remains to be elucidated. However, CO co-adsorbed on Pt clusters, besides O₂ in the regeneration at low flow rate, suggests a role of CO adsorption. When CO is adsorbed on less stable (smaller) Pt clusters, back-donation effects may lead to a decrease in Pt–support interactions. Consequently, the affinity of this mobile CO-adsorbed Pt species for more stable (larger) Pt clusters may be greater than that of the mobile Pt species to support. The decreased metal–support interactions lift migrating Pt clusters, causing them to float over the support [39]. The enhancement of surface diffusivities due to floating of Pt clusters promotes the coalescence of Pt clusters. The CO-adsorption-assisted Ostwald ripening [40] induces the change of Pt morphology, characterized by the coalescence of Pt clusters in 3-dimensional (Figure 8).

In our previous paper, we reported that ethanol bonds to bulk fcc Pt clusters of Pt/NaY-SiO₂ through bridging hydrogen [14]. Facilitated by the adsorption heat, some of the adsorbed ethanol molecules, Pt^HOC₂H₅, were converted to ethoxide and acetaldehyde in breakthrough performance tests. In comparison with ethanol, these ethanol-converted species could have higher affinity for NaY-SiO₂, and thus could spill over to displace sodium and acid site-bound water and leave the Pt clusters for further adsorption. This catalytic adsorption property of Pt leads to an increase of adsorption capacity. The aggregation of Pt clusters in the regeneration at low flow rate, however, deteriorates the catalytic adsorption function of Pt, leading to a decrease of adsorption capacity.

3. Materials and Methods

3.1. Preparation of PtO₂/NaY and NaY Pellet

Impregnation technique was used to prepare NaY-supported PtO₂ (PtO₂/NaY) powder. Zeolite powder (Sodium Silicoaluminate, Na₂O:Al₂O₃:5.0 SiO₂:xH₂O, GRACE Davison) of 60 g was put in a flask and pretreated by heating to 120 °C under vacuum. Platinum salt solution was prepared by dissolving 1.92 g H₂PtCl₆.6H₂O (Uni region Bio-Tech) in 44.6 g de-ionized water. The solution was then added dropwise while shaking the vacuum flask. The slurry was then dried in vacuum at 150 °C for 2 h and calcined in air at 450 °C for 4 h. The resulting material was noted as PtO₂/NaY powder. To make pellet PtO₂/NaY, a silica powder of 30 g (T600, PPG Industries) was mixed with a 3 N nitric acid solution to form a silica colloid. 60 g PtO₂/NaY powders were placed in a bowl of a small kneader. The blades of the kneader were rotated while the small portions of the silica colloid were being added to cause the powder to be converted to a paste. The paste was extruded from an extruder to make cylinder extrudate of 2 mm in diameter. The PtO₂/NaY extrudate were then cut to form pellets of about 6 mm in length. The pellets were pre-dried at 120 °C for 8 h to remove water and calcined under air at 450 °C for 4 h. The resulting pellet sample was noted as PtO₂/NaY-SiO₂. The same procedure was used to prepare pellet NaY (NaY-SiO₂).

3.2. Performance Tests

The dynamics of fixed-bed adsorption for PtO₂/NaY-SiO₂ and NaY-SiO₂ were performed in a stainless-steel tube with an inside diameter of 2.2 cm and length of 45 cm. The reactor was heated electrically and temperature controlled by a PID temperature controller with three K-type thermal couples at the outer wall of the reactor. The reactor was packed with 5.0 g adsorbents mixed with inert ceramic (sphere of about 2.0–5.0 mm diameter) in a ratio of 1:10 by volume, and a gradient packing method was used so that the adsorbent bed would have a nearly uniform temperature in air regeneration and the wall and bypassing effects would be minimized. Prior to adsorption tests, NaY-SiO₂ were heated in flowing air (50 mL/min, volumetric flow rate) at 150 °C for 1 h. For PtO₂/NaY-SiO₂, the sample was pre-reduced by treatment with H₂ at 300 °C for 1 h and was noted as Pt/NaY-SiO₂. After the samples were cooled down to room temperature (about 35 °C), performance tests were carried out. The total flow rate of the reaction mixture (78 mg of ethanol with 0.52 mg water moisture in L of air) was 50 mL/min. Mass flow controllers (AeraFC-7700C) were used for accurate and stable control of gas flow rates. Effluents from the reactor were collected and analyzed by use of GC periodically.

After the saturation, the ethanol-laden adsorbents were regenerated by heating the adsorption bed with a ramping rate of 10 °C/min to 300 °C and maintained for 2 h in flowing air. In order to study the effects of air flow rate on regeneration efficiency, air superficial velocity (v_s), defined as volumetric flow rate of that fluid divided by the cross-sectional area of adsorption column, was set at 13.2 (high flow rate) and 5.3 cm/min (low flow rate), respectively. The NaY-SiO₂ regenerated at high flow rate was notes as NaYR_(HF), and that at low flow rate NaYR_(LF).

The performance of the regenerated Pt/NaY-SiO₂ and NaY-SiO₂ were tested by the same steps as those of the fresh one. For X-ray absorption spectroscopy measurements, Pt/NaY-SiO₂ samples after air regeneration were purged with N₂ for 2 h. After N₂ purge, the regenerated samples were removed from the adsorption column and transferred to a dry box for storage. The samples regenerated at high and low flow rates were noted as PtR_(HF) and PtR_(LF), respectively.

The GC used for analysis of ethanol and its secondary products was a Shimadzu Model GC-14B equipped with a capillary column (length: 60 m; ID: 0.32 mm; Film: 0.5 μm; stationary phase: HP-INNOWAX, Restek Corp., Bellefonte, PA, USA) and an FID detector. The column was operated at 40 °C to 200 °C with ramping rate of 10 °C/min and a flow rate of 50 mL/min of dry He. Oxidation species, CO and CO₂, were analyzed by GC-14B equipped with a unibead-packed column of 3 m × 3 mm i.d. and a TCD detector.

3.3. Synchrotron XRPD

Powder X-ray diffraction was performed at the BL01C2 beamline of the National Synchrotron Radiation Research Center (NSRRC, Hsinchu, Taiwan), with the wavelength of 0.9535 Ǻ (13.0 keV). The ring of NSRRC was operated at energy 1.5 GeV with a typical current 300 mA with top-up injection mode. The synchrotron X-ray was produced from a 5.0 T superconducting wavelength shift magnet. Beyond the prefocusing mirror, a double crystal monochromator, which used the Si (111) plane to yield the monochromatic beam, was followed by a refocusing toroidal mirror. The sample was sealed in the capillary. During the X-ray exposure, the sample was kept at fast spin in order to increase the orientations of powders. Two-dimensional diffraction patterns were recorded by a Mar345 imaging plate system, with a sample-to-detector distance of 300 mm. Diffraction angle was calibrated with silver Behenate and Si powders (NBS640b) standards. One-dimensional XRPD (X-ray powder diffraction) profiles were integrated from selected fan-like areas of the symmetrical 2-D powder rings using the Fit2D program [41].

3.4. Characterization of Catalyst Samples by Infrared Spectroscopy

Diffuse reflectance infrared Fourier transform spectra (DRIFT) characterizing ethanol adsorbed on the catalytic adsorbents were recorded with a Shimadzu IR Prestige-21 (Kyoto, Japan), equipped with an MCT detector. The cell was connected to a vacuum/gas-handling manifold for in situ treatment. The pellet samples were crushed and made into powder. For PtO₂/NaY-SiO₂, the powder sample was reduced by flowing H₂ at 300 °C for 1 h. For NaY-SiO₂, the sample was pretreated by flowing air at 150 °C for 1 h. After the pretreatments, the samples were cooled to room temperature, and then air (flowing at 10 mL/min at 1 atm) containing 40,000 ppm ethanol and 7000 ppm (mole) water was introduced into the cell and maintained for about 1 h. After the saturation of ethanol adsorption, air with a superficial velocity of about 13 and 5 cm/min, respectively, was introduced into the cell. The cell was heated to 300 °C with ramping rate of 10 °C/min and maintained for about 2 h. After the samples were cooled to room temperature, the cell was evacuated to a pressure of about 0.1 to 1 torr and IR spectra with a 2 cm⁻¹ spectral resolution were recorded.

3.5. X-ray Absorption Spectroscopy

All X-ray absorption spectra of Pt L_III edge (11,564 eV) were measured on the beam line BL17C at the Synchrotron Radiation Research Center (SRRC) in Hsinchu, Taiwan, with storage ring energy 1.5 GeV and a beam current between 120 and 200 mA. A Si (111) double-crystal monochromator was employed for energy selection, and mirrors rejected higher harmonic radiation. The transmission geometry was arranged using gas-filled ionization chambers to monitor the intensities of the incident and transmitted X-ray beams. For fresh Pt/NaY-SiO₂, PtO₂/NaY-SiO₂ pellets were pulverized, pressed into wafer, and loaded in to the EXAFS cell, allowing treatment in flowing gases prior to the measurement. The reduction conditions were exactly the same as those for performance tests.

For PtR_(HF) and PtR_(LF), the samples were also pulverized and pressed into wafer in a dry box, and then transferred and loaded into the EXAFS cell in a N₂-filled glove bag.

A full spectrum for the Pt L_III absorption edge was obtained over energy levels from 11,364 to 12,764 eV. To ensure reliability of the spectra, Pt metal foil was also monitored to evaluate the stability of the energy scale for each measurement. Data reduction and data analysis were performed with the XDAP code developed by Vaarkampet et al. [42]. Standard procedures were used to extract the EXAFS data from the measured absorption spectra. The pre-edge was approximated by a modified Victoreen curve [16,43] and the background was subtracted using cubic spline routines [16]. Normalization was performed by dividing the data by the height of the absorption edge at 50 eV above the edge [16,42]. The EXAFS contributions for Pt–Pt were analyzed with phase shifts and backscattering amplitudes obtained from EXAFS data for Pt foil. Pt–O were analyzed with phase shifts and backscattering amplitudes calculated from FEFF8 [44].

4. Conclusions

The EXAFS and FT-IR results suggest that ethanol on Pt/NaY-SiO₂ appears to be adsorbed catalytically, and proceeds through ethoxide to aldehyde. Those Pt adsorbed species could spill over to displace sodium and acid site bound water, leaving the Pt clusters for further adsorption, and thus increase ethanol adsorption capacity to about 2.5 times that of NaY-SiO₂ [14]. Moreover, in comparison with NaY-SiO₂, Pt/NaY-SiO₂ offers higher VOC destructive efficiency in air regeneration. During air regeneration, long-chain paraffinic coke was formed on NaY-SiO₂, leading to lower regeneration efficacy. In contrast, the species adsorbed on Pt/NaY-SiO₂ were catalytically oxidized to CO₂ and H₂O at high flow rate (v_s = 13.2 cm/min); the adsorption capacity loss after regeneration was 2.5% for Pt/NaY-SiO₂, in contrast to 8% for NaY-SiO₂. However, despite the fact that reducing air flow rate in regeneration is an effective way to decrease auxiliary-fuel utilization, aggregation of Pt clusters was observed in low-flow-rate regeneration (v_s = 5.3 cm/min). Hence, for industrial application, regeneration efficiency loss caused by the Pt aggregation should be taken into account in exploring the optimal regeneration conditions. Mesoporous silica materials might be useful for adsorbing VOCs, specifically, for large molecule. Trapping Pt clusters in the pores of mesoporous material, instead of the deposition of Pt on silica binder and/or the external surface of NaY particles, may confine Pt clusters in pores during regeneration and thus suppress the aggregation of Pt clusters. In addition, Pd–Pt bimetallic interactions may help Pt to anchor on zeolite or mesoporous materials, thereby, inhibiting Pt migration in regeneration. For long-term commercial operation, sufficient air supply to minimize CO-induced Pt aggregation should be employed. Moreover, optimal Pt loading and/or oxychlorination-redispersion [39] should be considered to ensure no coke is deposited on the catalytic adsorbent even after the destruction of the zeolite frameworks with several regeneration cycles.