Hierarchical PtIn / Mg(Al)O Derived from Reconstructed PtIn-hydrotalcite-like Compounds for Highly E ﬃ cient Propane Dehydrogenation

: ThechallengesfacingpropanedehydrogenationaretosolvethePtsinteringandcarbondeposition. This paper provides a new way to disperse and stabilize Pt species and resist carbon deposition. Highly dispersed Pt species were topologically transformed from reconstructed PtIn-hydrotalcite-like precursors in a flower-like hierarchical microstructure. The lattice confinement of reconstructed hydrotalcite-like precursor is in favor of stabilizing the highly dispersed Pt species, and the hierarchical microstructure is an important factor to prolong its lifetime by enhancing tolerance to carbon deposition. In propane dehydrogenation, the propene selectivity decreases in the sequences of catalyst in flower-like > single-plate > block mass with small, flakeys. A propene selectivity of > 97% with a conversion of 48% at 600 ◦ C has been achieved over a flower-like PtIn / Mg(Al)O catalyst. Additionally, no visible Pt sintering can even be observed on this catalyst after a reaction time of 190 h. This strategy provides an effective and feasible alternative for the facile preparation of highly dispersed metal catalysts.

However, there are several open issues for this Pt-based catalyst under reaction conditions [3,7,8], such as Pt sintering, carbon deposition and propane cracking. Extensive research efforts have been made to inhibit deactivation and improve propene selectivity of Pt-based catalysts by doping promoters of Sn [9,10], In [11,12], Ga [13,14] and Cu [15] to form bimetallic catalysts. For example, Pt-In [16] and Pt-Ga [17,18] bimetallic nanoparticles (NPs) supported on calcined hydrotalcites (HT) show higher activity and selectivity in PDH, which has been attributed to the increased Pt dispersion and electron transfer from In or Ga to Pt [11,18]. Besides promoters, another method is to modify Al 2 O 3 support with ZnO [19], La 2 O 3 [20] or TiO 2 [21] or replace Al 2 O 3 support with ZnAl 2 O 4 [22] , ZSM-5 [23] or Mg(Al)O oxide [8,24]. Additionally, it has been verified that the single Pt atom sites on Pt-Sn/γ-Al 2 O 3 with a "sandwich structure" are more favorable for PDH reaction than Pt ensembles [25,26]. Gong et al. [15] also reported that the PtCu single atom alloys can enhance the propane conversion and propene selectivity simultaneously in PDH.
Whether as promoters or supports, an important aim of using them is to obtain highly and stably dispersed Pt-based catalysts so that improving PDH performance. However, it is still difficult to

Phase Structure and Reconstruction Study
In order to ascertain the crystal phase structure of all samples in details, the XRD patterns of precursors, calcined and reduced samples are shown together in Figure 1, and the analyzing results of XRD patterns of all samples before calcination are given in Table 1.
It can be seen that the XRD patterns of HT-F, HT-P and HT-M exhibit strong characteristic diffraction peaks in the HT phase [32], although their preparation processes are different. After calcination, followed by the impregnation of HT-FC, HT-PC and HT-MC with In 3+ ions, the characteristic diffraction peaks of the HT phase appear again in the XRD patterns of InHT-F, InHT-P and InHT-M, without other diffraction peaks. That suggests that the impregnation condition is beneficial to reconstructing the HT phase. A similar HT recrystallization phenomenon happened again after the impregnation of In-based metal oxides with Pt 4+ ions. For HT-M, the broad and well-defined diffraction peaks indicate that the obtained single HT phase was poorly crystallized. With the introduction of In 3+ ions and further Pt 4+ ions, the relative intensity of diffraction peaks of HT phase gives a decreasing tendency. From HT-F Catalysts 2019, 9,767 3 of 17 to InHT-F and HT-P to InHT-P, the shrinkage of peak intensities is more considerable than that from HT-M to InHT-M. This should be related to the decrease of HT crystallinity. At the same time, the average crystallite size in the c direction (Lc) also decreases with reconstruction and even drops from PtInHT-M to PtInHT-P and to PtInHT-F (see Table 1), which can be estimated by means of the Scherrer equation [33]. This indicates that the HT layer becomes thinner and thinner.
Catalysts 2019, 9, x FOR PEER REVIEW 3 of 18 intensities is more considerable than that from HT-M to InHT-M. This should be related to the decrease of HT crystallinity. At the same time, the average crystallite size in the c direction (Lc) also decreases with reconstruction and even drops from PtInHT-M to PtInHT-P and to PtInHT-F (see Table 1), which can be estimated by means of the Scherrer equation [33]. This indicates that the HT layer becomes thinner and thinner.  Additionally, the decreasing M 2+ /(M 3+ + M 4+ ) ratio can cause the increase of positive charge density in the layers, while the big radii of In 3+ (0.080 nm, the Shannon ionic radis [34]) and Pt 4+ (0.063 nm, Shannon ionic radius [34]) ions can broaden the distance of metal ions in the layers. These can  Additionally, the decreasing M 2+ /(M 3+ + M 4+ ) ratio can cause the increase of positive charge density in the layers, while the big radii of In 3+ (0.080 nm, the Shannon ionic radis [34]) and Pt 4+ (0.063 nm, Shannon ionic radius [34]) ions can broaden the distance of metal ions in the layers. These can make the octahedral cell shrink or expand and then influence the value of the lattice parameter a and c of HT phase, as shown in Table 1 the Pt and in species could be highly dispersed on the final metal oxides with close contact among them, and support [10].
After calcination and reduction, the HT phase disappears along with the occurrence of MgO (JCPDS file number 45-0946) and meixnerite (Mg 4 Al 2 (OH) 14 ·3H 2 O, JCPDS file number 35-0964) diffraction peaks in Figure 1. At the same time, there are not any metal phase diffraction peaks can be found in the XRD patterns, which means that Pt and/or in metal phases are exactly evenly-dispersed on the surface of HT-derived catalysts.
From TEM images of PtInHT-FR, PtInHT-PR and PtInHT-MR in Figure 2, it is difficult to find metal particles except thin nanosheets. That indicates that the metal particles are so small that the TEM cannot identify them. Therefore, it is reasonable to infer that metal atom clusters or single metal atoms are possibly formed on the surface of metal oxide nanosheets due to the low metal loading amount and the HT lattice confinement effect [10]. The EDS element-mapping images of PtInHT-FR further demonstrate the uniform distribution of Pt and in elements without aggregation, and support the probability of metal atoms or atom clusters type distributions [35].
Catalysts 2019, 9, x FOR PEER REVIEW 4 of 18 make the octahedral cell shrink or expand and then influence the value of the lattice parameter a and c of HT phase, as shown in Table 1. The change of lattice parameters a and c of the HT phase (Table  1) demonstrates that In 3+ and Pt 4+ ions have been introduced into the HT layer. It signifies that the Pt and In species could be highly dispersed on the final metal oxides with close contact among them, and support [10]. After calcination and reduction, the HT phase disappears along with the occurrence of MgO (JCPDS file number 45-0946) and meixnerite (Mg4Al2(OH)14·3H2O, JCPDS file number 35-0964) diffraction peaks in Figure 1. At the same time, there are not any metal phase diffraction peaks can be found in the XRD patterns, which means that Pt and/or In metal phases are exactly evenlydispersed on the surface of HT-derived catalysts.
From TEM images of PtInHT-FR, PtInHT-PR and PtInHT-MR in Figure 2, it is difficult to find metal particles except thin nanosheets. That indicates that the metal particles are so small that the TEM cannot identify them. Therefore, it is reasonable to infer that metal atom clusters or single metal atoms are possibly formed on the surface of metal oxide nanosheets due to the low metal loading amount and the HT lattice confinement effect [10]. The EDS element-mapping images of PtInHT-FR further demonstrate the uniform distribution of Pt and In elements without aggregation, and support the probability of metal atoms or atom clusters type distributions [35].  Figure 3 shows SEM images of three series of samples. It clearly depicts that the HT-F and corresponding derivative samples are constructed by huge ball-flower-like microstructure. For fresh HT-P and its derivative samples, perfect single-plate NPs are displayed in the SEM images. On the contrary, small, flakes are disorderly stacked into big block masses for HT-M and its derivative products. The calcination makes the nanosheets become rough and the impregnation-induced reconstruction makes them become smooth again. The loose surface for the calcined samples is related to the newly formed pores, which is the result of the decomposition and removal of interlayer  Figure 3 shows SEM images of three series of samples. It clearly depicts that the HT-F and corresponding derivative samples are constructed by huge ball-flower-like microstructure. For fresh HT-P and its derivative samples, perfect single-plate NPs are displayed in the SEM images. On the contrary, small, flakes are disorderly stacked into big block masses for HT-M and its derivative products. The calcination makes the nanosheets become rough and the impregnation-induced reconstruction makes them become smooth again. The loose surface for the calcined samples is related to the newly formed pores, which is the result of the decomposition and removal of interlayer anions and hydroxyl groups. It is clear that a large amount of small and ultrathin nanosheets grow in an orderly manner with ab-planes interdigitated perpendicular to the two lateral surfaces of recrystallizing original HT nanosheets for InHT-F and InHT-P. According to the XRD results in Figure 1, the reconstruction should also take place in InHT-M, although it is difficult to distinguish the small new nanosheets from the original flakes. In other words, this microstructure is indistinctive in the reconstructed HT-M series and instead severely aggregated nanoplates without any regular shape leave over. This microstructure should be the result of abundant hydroxyl groups on the edge sites and the basal plane of calcined HT plates [36]. The similar morphology can be seen in the SEM images of corresponding PtIn-based samples. In any case, there must be a strong interaction between the newly formed nanoplates and the initial huge templates, originating from the shared Al and other elements [36]. The strong interaction is beneficial to stabilize and disperse metal active species. The formation of ultrathin nanosheets is bound to decrease the average crystallite size in the c direction (Table 1) and weaken the diffraction peaks of HT phase in initial huge HT-F and HT-P template, which is consistent with the XRD results in Figure 1. As for the reduced samples, the basic morphology feature still can be maintained, but some broken particles can be found on the surface of single-plate PtInHT-PR sample. That indicates that the fully opening surface is detrimental to protect the newly formed nanosheets from destruction by an external force. anions and hydroxyl groups. It is clear that a large amount of small and ultrathin nanosheets grow in an orderly manner with ab-planes interdigitated perpendicular to the two lateral surfaces of recrystallizing original HT nanosheets for InHT-F and InHT-P. According to the XRD results in Figure 1, the reconstruction should also take place in InHT-M, although it is difficult to distinguish the small new nanosheets from the original flakes. In other words, this microstructure is indistinctive in the reconstructed HT-M series and instead severely aggregated nanoplates without any regular shape leave over. This microstructure should be the result of abundant hydroxyl groups on the edge sites and the basal plane of calcined HT plates [36]. The similar morphology can be seen in the SEM images of corresponding PtIn-based samples. In any case, there must be a strong interaction between the newly formed nanoplates and the initial huge templates, originating from the shared Al and other elements [36]. The strong interaction is beneficial to stabilize and disperse metal active species. The formation of ultrathin nanosheets is bound to decrease the average crystallite size in the c direction (Table 1) and weaken the diffraction peaks of HT phase in initial huge HT-F and HT-P template, which is consistent with the XRD results in Figure 1. As for the reduced samples, the basic morphology feature still can be maintained, but some broken particles can be found on the surface of single-plate PtInHT-PR sample. That indicates that the fully opening surface is detrimental to protect the newly formed nanosheets from destruction by an external force.  It can be seen that the slit-shaped mesopores are the common features of all samples. As for the recrystallizing single-plate samples, including InHT-P, PtInHT-P and corresponding calcined and reduced products, the reconstruction not only results in the formation of narrow slit-like pores due to the nanoplates' delaminating, but also leads to the appearance of honeycomb-like pores arising from the growth of ultrathin nanosheets perpendicular to the surface of the primary large singleplate nanoplates. Especially in the InHT-F and PtInHT-F series, the big wedge-shaped pore channels appear in the huge ball-flower-like microstructure, except the honeycomb-like pores. However, the similar size between the original flakes and newly formed nanosheets make the pore structures of the HT-M and reconstructed samples no different. Therefore, the obvious hierarchical pore feature can be assigned to recrystallizing flower-like and single-plate samples; and the pore shape and size of recrystallizing flower-like samples are more complicated than those of single-plate samples.

Morphology, Texture and Pt-In Interaction Analysis
Low-temperature N2 adsorption-desorption isotherms and pore-size distribution (PSD) curves were conducted to evaluate the textural properties of PtInHT-FC, PtInHT-PC and PtInHT-MC. Their plots are shown in Figure 4 and the corresponding SBET, Vp and dp are given in Table 2. According to the isotherm classification, all samples exhibit a type IV isotherm with H3 hysteresis loops, implying the presence of typical slit-shaped mesopores [8,24]. The PSD curves and dp values show that the pore size increases in the following order, PtInHT-FC < PtInHT-PC < PtInHT-MC, and the same trend can be found for their SBET values. It can be seen that the slit-shaped mesopores are the common features of all samples. As for the recrystallizing single-plate samples, including InHT-P, PtInHT-P and corresponding calcined and reduced products, the reconstruction not only results in the formation of narrow slit-like pores due to the nanoplates' delaminating, but also leads to the appearance of honeycomb-like pores arising from the growth of ultrathin nanosheets perpendicular to the surface of the primary large single-plate nanoplates. Especially in the InHT-F and PtInHT-F series, the big wedge-shaped pore channels appear in the huge ball-flower-like microstructure, except the honeycomb-like pores. However, the similar size between the original flakes and newly formed nanosheets make the pore structures of the HT-M and reconstructed samples no different. Therefore, the obvious hierarchical pore feature can be assigned to recrystallizing flower-like and single-plate samples; and the pore shape and size of recrystallizing flower-like samples are more complicated than those of single-plate samples.
Low-temperature N 2 adsorption-desorption isotherms and pore-size distribution (PSD) curves were conducted to evaluate the textural properties of PtInHT-FC, PtInHT-PC and PtInHT-MC. Their plots are shown in Figure 4 and the corresponding S BET , V p and d p are given in Table 2. According to the isotherm classification, all samples exhibit a type IV isotherm with H3 hysteresis loops, implying the presence of typical slit-shaped mesopores [8,24]. The PSD curves and d p values show that the pore size increases in the following order, PtInHT-FC < PtInHT-PC < PtInHT-MC, and the same trend can be found for their S BET values.   The broad PSD was in the mesopore range; big dp and SBET values of PtInHT-MC are related to unordered aggregation of small, flakey-like particles. As expected, the big, ordered nanosheets make PtInHT-FC and PtInHT-PC present the narrow PSD feature, and small dp and SBET values. The Vp, dp and SBET values of PtInHT-FC are far lower than that of PtInHT-PC especially, which is attributed to the fully open surface of PtInHT-PC relative to that of PtInHT-FC.
XPS analysis was applied to investigate the surface chemical state and relative concentration of In element on the reduced samples. The broad In3d5/2 and In3d3/2 peaks of catalysts can be deconvoluted into two peaks, respectively, in Figure 5. The deconvolution results of the corresponding spectra are listed in Table 3. The low bind energy (BE) contribution is assigned to metallic state In (In 0 ), and the high one is related to oxidation state In (In 3+ ). A slightly decreasing BE value of In 0 from PtInHT-FR to PtInHT-PR and to PtInHT-MR reflects the weakening electrons' transfer from metallic In to Pt species and the Pt-In interaction. Additionally, a slightly increased ratio of In 3+ /In 0 can be seen in Table 3, in accordance with the order of PtInHT-MR, PtInHT-PR and PtInHT-FR. This indicates that the flower-like hierarchical structure with porous channels is in favor of maintaining the In element in In 3+ ions rather than In 0 states, while the In 3+ ions in PtInHT-MR, with small, flakey NPs, can easily be reduced.  The broad PSD was in the mesopore range; big d p and S BET values of PtInHT-MC are related to unordered aggregation of small, flakey-like particles. As expected, the big, ordered nanosheets make PtInHT-FC and PtInHT-PC present the narrow PSD feature, and small d p and S BET values. The V p , d p and S BET values of PtInHT-FC are far lower than that of PtInHT-PC especially, which is attributed to the fully open surface of PtInHT-PC relative to that of PtInHT-FC.
XPS analysis was applied to investigate the surface chemical state and relative concentration of In element on the reduced samples. The broad In3d 5/2 and In3d 3/2 peaks of catalysts can be deconvoluted into two peaks, respectively, in Figure 5. The deconvolution results of the corresponding spectra are listed in Table 3. The low bind energy (BE) contribution is assigned to metallic state In (In 0 ), and the high one is related to oxidation state In (In 3+ ). A slightly decreasing BE value of In 0 from PtInHT-FR to PtInHT-PR and to PtInHT-MR reflects the weakening electrons' transfer from metallic In to Pt species and the Pt-In interaction. Additionally, a slightly increased ratio of In 3+ /In 0 can be seen in Table 3, in accordance with the order of PtInHT-MR, PtInHT-PR and PtInHT-FR. This indicates that the flower-like hierarchical structure with porous channels is in favor of maintaining the In element in In 3+ ions rather than In 0 states, while the In 3+ ions in PtInHT-MR, with small, flakey NPs, can easily be reduced.
It is worth mentioning that PtInHT-FC exhibits a higher surface Pt density than that of PtInHT-PC and PtInHT-MC, due to its low S BET and high D Pt , as listed in Table 2. In addition, the relative content of surface, metallic-state In increases in the order of PtInHT-FR, PtInHT-PR and PtInHT-MR (see Figure 5 and Table 3). Therefore, it can be inferred that the most of the surface In species have a chance to make close contact with Pt species and form plenty of strongly interacting Pt-In centers over PtInHT-FR. Accordingly, the Pt-In interaction over PtInHT-PR and PtInHT-MR should be decreased and weakened due to their low D Pt and high S BET . Additionally, too much In over PtInHT-MR could block Pt sites, thus cause a disadvantageous influence to the PDH reaction [8,24].
On the basis of the aforementioned results and preparation conditions, the formation of a ball-flower-like and single-plate hierarchical microstructure is tentatively proposed in Scheme 1.  It is worth mentioning that PtInHT-FC exhibits a higher surface Pt density than that of PtInHT-PC and PtInHT-MC, due to its low SBET and high DPt, as listed in Table 2. In addition, the relative content of surface, metallic-state In increases in the order of PtInHT-FR, PtInHT-PR and PtInHT-MR (see Figure 5 and Table 3). Therefore, it can be inferred that the most of the surface In species have a chance to make close contact with Pt species and form plenty of strongly interacting Pt-In centers over PtInHT-FR. Accordingly, the Pt-In interaction over PtInHT-PR and PtInHT-MR should be decreased and weakened due to their low DPt and high SBET. Additionally, too much In over PtInHT-MR could block Pt sites, thus cause a disadvantageous influence to the PDH reaction [8,24].
On the basis of the aforementioned results and preparation conditions, the formation of a ballflower-like and single-plate hierarchical microstructure is tentatively proposed in Scheme 1.

Catalytic Performance and Structure-Activity Relationship Discussion
To investigate the performance, three catalysts were evaluated in the PDH reaction. Figure 6 presents the propane conversion and propene selectivity against time for these catalysts. The

Catalytic Performance and Structure-Activity Relationship Discussion
To investigate the performance, three catalysts were evaluated in the PDH reaction. Figure 6 presents the propane conversion and propene selectivity against time for these catalysts. The homogeneous reaction is also tested in the blank tube and the propane conversion and propene selectivity as functions of time are shown in Figure S1. It was found that the conversion of propane is only around 1%, with a propene selectivity of around 65%. This indicates that the contribution of homogeneous reaction can be ignored. As shown in Figure S2, the selectivities to by-products (methane, ethane and ethene) are very low and can be ignored. In the PDH reaction, the initial propane conversions for PtInHT-FR, PtInHT-PR and PtInHT-MR are 6%, 44% and 28%, respectively. After a long induction period of 20 h, the propane conversion and propene selectivity of PtInHT-FR can be stabilized at above 40% and 97% with a maximum conversion of 48%. Until upon 190 h of time on stream, the conversion of PtInHT-FR only decreases to 36%. Despite a high initial conversion with a short induction period of around 5 h, a smooth stable period for around 50 h can be achieved by PtInHT-PR. A decreased propene selectivity cannot be ignored during the whole reaction process. As for PtInHT-MR, the reaction only can run for 20 h with moderate initial conversion; in the same period, the propene selectivity decreases severely with the rising propane conversion. After the reaction, the XRD patterns of used samples are shown in Figure 7. It can be found that the meixnerite phase is still retained in the used PtInHT-FU, which is supposed to be favorable for the PDH reaction [7]. Additionally, the change occurs from fresh PtInHT-FR having more MgO phase (see Figure 1 and Figure 7) while the used one has more meixnerite phase. This phase transfer of support could be the one cause of induction of performance during reaction. In addition, a new diffraction peak appears at around 26.0° in all the used samples, which belongs to the carbon deposits (JCPD file number 44-1644). The diffraction peak intensity of carbon deposits increases following the order: PtInHT-PU > PtInHT-MU > PtInHT-FU, which must be consistent with the amount of deposited carbon. After the reaction, the XRD patterns of used samples are shown in Figure 7. It can be found that the meixnerite phase is still retained in the used PtInHT-FU, which is supposed to be favorable for the PDH reaction [7]. Additionally, the change occurs from fresh PtInHT-FR having more MgO phase (see Figures 1 and 7) while the used one has more meixnerite phase. This phase transfer of support could be the one cause of induction of performance during reaction. In addition, a new diffraction peak appears at around 26.0 • in all the used samples, which belongs to the carbon deposits (JCPD file number 44-1644). The diffraction peak intensity of carbon deposits increases following the order: PtInHT-PU > PtInHT-MU > PtInHT-FU, which must be consistent with the amount of deposited carbon. As shown in Figure 8, a large amount of carbon deposits with different sizes and morphologies, including granule, stick, loop and so on, are formed on the surface of used catalysts after reactions. Moreover, even the morphology features of PtInHT-PU and PtInHT-MU disappear, due to the surface coverage and pore blockage by carbon deposits. However, the flower-like particles of PtInHT-FU are still well dispersed and recognizable after a long reaction time, reflecting the fact that the amount of carbon deposits on PtInHT-FU number fewer than those on PtInHT-PU and PtInHT-MU. The most important thing is that the carbon is only deposited on the partial pore channel mouth of PtInHT-FU, meaning the active sites in the pore channels have no chance to be covered by the deposited carbon. In order to determine the degree of carbon deposition during PDH reaction, TG experiments were conducted over the used catalysts. As shown in Figure 9, the mass loss of all samples must be related to the removal of deposited carbon. The amount of deposited carbon on the catalysts used increases following the order of PtInHT-FU, to PtInHT-MU and finally to PtInHT-PU. The low carbon amount is the most important factor to extend the lifetime of the flower-like PtInHT-FU. As shown in Figure 8, a large amount of carbon deposits with different sizes and morphologies, including granule, stick, loop and so on, are formed on the surface of used catalysts after reactions. Moreover, even the morphology features of PtInHT-PU and PtInHT-MU disappear, due to the surface coverage and pore blockage by carbon deposits. However, the flower-like particles of PtInHT-FU are still well dispersed and recognizable after a long reaction time, reflecting the fact that the amount of carbon deposits on PtInHT-FU number fewer than those on PtInHT-PU and PtInHT-MU. The most important thing is that the carbon is only deposited on the partial pore channel mouth of PtInHT-FU, meaning the active sites in the pore channels have no chance to be covered by the deposited carbon. As shown in Figure 8, a large amount of carbon deposits with different sizes and morphologies, including granule, stick, loop and so on, are formed on the surface of used catalysts after reactions. Moreover, even the morphology features of PtInHT-PU and PtInHT-MU disappear, due to the surface coverage and pore blockage by carbon deposits. However, the flower-like particles of PtInHT-FU are still well dispersed and recognizable after a long reaction time, reflecting the fact that the amount of carbon deposits on PtInHT-FU number fewer than those on PtInHT-PU and PtInHT-MU. The most important thing is that the carbon is only deposited on the partial pore channel mouth of PtInHT-FU, meaning the active sites in the pore channels have no chance to be covered by the deposited carbon. In order to determine the degree of carbon deposition during PDH reaction, TG experiments were conducted over the used catalysts. As shown in Figure 9, the mass loss of all samples must be related to the removal of deposited carbon. The amount of deposited carbon on the catalysts used increases following the order of PtInHT-FU, to PtInHT-MU and finally to PtInHT-PU. The low carbon amount is the most important factor to extend the lifetime of the flower-like PtInHT-FU. In order to determine the degree of carbon deposition during PDH reaction, TG experiments were conducted over the used catalysts. As shown in Figure 9, the mass loss of all samples must be related to the removal of deposited carbon. The amount of deposited carbon on the catalysts used increases following the order of PtInHT-FU, to PtInHT-MU and finally to PtInHT-PU. The low carbon amount is the most important factor to extend the lifetime of the flower-like PtInHT-FU.
After 190 h of reaction, it is still difficult to pick out metallic particles from the TEM images of PtInHT-FU in Figure 10. However, the Pt particles can be found on the surface of PtInHT-PU and PtInHT-MU. According to the particle size distribution of PtInHT-MU, the average particle size is higher for PtInHT-MU (5.5 nm) than that for PtInHT-PU (3.2 nm). That implies that the metallic particles on PtInHT-FU are more stable than those on PtInHT-PU and PtInHT-MU under a high temperature reaction, which has a vital influence for the performance of PDH. Certainly, the carbon deposits also can be easily found on the surface of all samples. After 190 h of reaction, it is still difficult to pick out metallic particles from the TEM images of PtInHT-FU in Figure 10. However, the Pt particles can be found on the surface of PtInHT-PU and PtInHT-MU. According to the particle size distribution of PtInHT-MU, the average particle size is higher for PtInHT-MU (5.5 nm) than that for PtInHT-PU (3.2 nm). That implies that the metallic particles on PtInHT-FU are more stable than those on PtInHT-PU and PtInHT-MU under a high temperature reaction, which has a vital influence for the performance of PDH. Certainly, the carbon deposits also can be easily found on the surface of all samples.   After 190 h of reaction, it is still difficult to pick out metallic particles from the TEM images of PtInHT-FU in Figure 10. However, the Pt particles can be found on the surface of PtInHT-PU and PtInHT-MU. According to the particle size distribution of PtInHT-MU, the average particle size is higher for PtInHT-MU (5.5 nm) than that for PtInHT-PU (3.2 nm). That implies that the metallic particles on PtInHT-FU are more stable than those on PtInHT-PU and PtInHT-MU under a high temperature reaction, which has a vital influence for the performance of PDH. Certainly, the carbon deposits also can be easily found on the surface of all samples. The activity induction should be closely related with the partial coverage of the active metal surface by In2O3 species [7]. We propose that the covered, active Pt particles tend to slowly migrate from the support onto In2O3 particles to build highly dispersed "sandwich structure" of the Pt-In2O3support under the reaction atmosphere [25,26]. Therefore the duration of induction should be determined by the coverage degree of In2O3 species.
Although the maximum conversion is up to 50%, the PDH reaction cannot be carried out after 50 h for PtInHT-PR and 20 h for PtInHT-MR due to the reactor blockage by a large amount of deposited carbon (see Figures 7−10), and the deposited carbon particles fully block the honeycomblike pores of PtInHT-PR and cover the block mass particles of PtInHT-MR. Fortunately, the deposited carbon over PtInHT-FR after reacting for 190 h is much less than that over others. It is interesting to note that the carbon is mainly deposited on the pore mouth of ball-flower-like PtInHT-FR, which protects the metallic sites on the interior honeycomb-like surface of wedge-shaped pores.
Obviously, the deposited carbon amount is inversely correlated to the corresponding propene selectivity. The ball-flower-like multi-level hierarchical microstructure with abundant wedge-shaped pores, and short-channel honeycomb-like pores, but not the slit-like pores, is in favor of propane and propene diffusion. Furthermore, the low SBET of PtInHT-FR facilitates the propene desorption and then decreases the likelihood of deep dehydrogenation and hydrogenolysis on metal sites. These can be used to explain the high selectivity and low carbon deposits of PtInHT-FR. On the contrary, plenty of carbon deposits and the sharply reduced selectivities of PtInHT-PR and PtInHT-MR arise from the propene concentrating on the surface with a high SBET value, which triggers carbon formation and accumulation.
Except for the influences of texture and morphology, the propene selectivity and catalytic stability can be improved, by partially transferring electrons from In to Pt [11]. PtInHT-FR presents a high surface In 3+ /In 0 ratio and BE value of In 0 , indicating that a high electron density of Pt and strong Pt-In interaction can be obtained (see Figure 5 and Table 3), which facilitates the desorption of propene and the migration of carbon precursors from the metal surface to the support [7,11]. Moreover, the suitable In 0 species can strengthen the Pt-In-support interaction, improve dispersion and prevent sintering of Pt particles (see Figures 2 and 10), stabilizing the catalytic performance of PtInHT-FR. Meanwhile, the overabundant In 0 species could bring about a disadvantageous influence to the PDH reaction. The high content In 0 species in PtInHT-PR and PtInHT-MR can not only block Pt active sites [8], but also weaken Pt-In interactions and lead to the Pt sintering (see Figures 2, 8 and  10). This is associated with the fact that the carbon deposits can easily form on the large-size Pt sites over PtInHT-PR and PtInHT-MR [37].
The regeneration and activation test for PtInHT-FU is shown in Figure 11. After each regeneration, the activity can be restored up to the level before regeneration, but decreases rapidly; almost no change in the selectivity can be observed. Although, the deposited carbon can be removed The activity induction should be closely related with the partial coverage of the active metal surface by In 2 O 3 species [7]. We propose that the covered, active Pt particles tend to slowly migrate from the support onto In 2 O 3 particles to build highly dispersed "sandwich structure" of the Pt-In 2 O 3 -support under the reaction atmosphere [25,26]. Therefore the duration of induction should be determined by the coverage degree of In 2 O 3 species.
Although the maximum conversion is up to 50%, the PDH reaction cannot be carried out after 50 h for PtInHT-PR and 20 h for PtInHT-MR due to the reactor blockage by a large amount of deposited carbon (see , and the deposited carbon particles fully block the honeycomb-like pores of PtInHT-PR and cover the block mass particles of PtInHT-MR. Fortunately, the deposited carbon over PtInHT-FR after reacting for 190 h is much less than that over others. It is interesting to note that the carbon is mainly deposited on the pore mouth of ball-flower-like PtInHT-FR, which protects the metallic sites on the interior honeycomb-like surface of wedge-shaped pores.
Obviously, the deposited carbon amount is inversely correlated to the corresponding propene selectivity. The ball-flower-like multi-level hierarchical microstructure with abundant wedge-shaped pores, and short-channel honeycomb-like pores, but not the slit-like pores, is in favor of propane and propene diffusion. Furthermore, the low S BET of PtInHT-FR facilitates the propene desorption and then decreases the likelihood of deep dehydrogenation and hydrogenolysis on metal sites. These can be used to explain the high selectivity and low carbon deposits of PtInHT-FR. On the contrary, plenty of carbon deposits and the sharply reduced selectivities of PtInHT-PR and PtInHT-MR arise from the propene concentrating on the surface with a high S BET value, which triggers carbon formation and accumulation.
Except for the influences of texture and morphology, the propene selectivity and catalytic stability can be improved, by partially transferring electrons from IN to Pt [11]. PtInHT-FR presents a high surface In 3+ /In 0 ratio and BE value of In 0 , indicating that a high electron density of Pt and strong Pt-In interaction can be obtained (see Figure 5 and Table 3), which facilitates the desorption of propene and the migration of carbon precursors from the metal surface to the support [7,11]. Moreover, the suitable In 0 species can strengthen the Pt-In-support interaction, improve dispersion and prevent sintering of Pt particles (see Figures 2 and 10), stabilizing the catalytic performance of PtInHT-FR. Meanwhile, the overabundant In 0 species could bring about a disadvantageous influence to the PDH reaction. The high content In 0 species in PtInHT-PR and PtInHT-MR can not only block Pt active sites [8], but also weaken Pt-In interactions and lead to the Pt sintering (see Figures 2,8 and 10). This is associated with the fact that the carbon deposits can easily form on the large-size Pt sites over PtInHT-PR and PtInHT-MR [37].
The regeneration and activation test for PtInHT-FU is shown in Figure 11. After each regeneration, the activity can be restored up to the level before regeneration, but decreases rapidly; almost no change in the selectivity can be observed. Although, the deposited carbon can be removed by a simple oxidation process (not the industrial condition), but the regeneration conditions can cause Pt sintering and decrease the Pt dispersion to a certain extent, which can result in the decrease of initial activity with cycles of regeneration [38,39].
Catalysts 2019, 9, x FOR PEER REVIEW 14 of 18 by a simple oxidation process (not the industrial condition), but the regeneration conditions can cause Pt sintering and decrease the Pt dispersion to a certain extent, which can result in the decrease of initial activity with cycles of regeneration [38,39].

Preparation
The flower-like precursor (HT-F) was prepared by a hydrothermal method. 0.82 g Mg(NO3)2·6H2O, 0.6 g Al(NO3)3·9H2O, 0.2 g SDS and 6 g urea were dissolved in 65 mL deionized water, then moved into a 100 mL Teflon autoclave. After aging at 100 °C for 20 h, the sample was filtered, washed with deionized water to neutrality and dried in air at 100 °C overnight.
Single-plate precursor (HT-P) was prepared using the same method as HT-F, except with no addition of SDS.
Block mass precursor (HT-M) was prepared by coprecipitation method. First, 17.09 g Mg(NO3)2·6H2O and 12.50 g Al(NO3)3·9H2O were dissolved in 100 mL deionized water. Another solution was prepared with 8 g NaOH and 10.6 g Na2CO3 dissolved in 100 mL deionized water. Then, two solutions were mixed drop-wise, with strong stirring at room temperature for 2 h. The pH value was controlled at about 10. Subsequently, the mixed solution was aged at 65 °C for 12 h. The resulting suspension was filtered, washed with deionized water to neutrality and dried in air at 100 °C overnight.
All the precursors were calcined for 4 h at 600 °C to obtain the corresponding calcined samples, named HT-XC (X represents morphology, either F, P or M).
The corresponding In-based precursors, InHT-X, were acquired by incipient wetness impregnation method for calcined samples HT-XC with In(NO3)3·xH2O aqueous solution at room

Preparation
The flower-like precursor (HT-F) was prepared by a hydrothermal method. 0.82 g Mg(NO 3 ) 2 ·6H 2 O, 0.6 g Al(NO 3 ) 3 ·9H 2 O, 0.2 g SDS and 6 g urea were dissolved in 65 mL deionized water, then moved into a 100 mL Teflon autoclave. After aging at 100 • C for 20 h, the sample was filtered, washed with deionized water to neutrality and dried in air at 100 • C overnight.
Single-plate precursor (HT-P) was prepared using the same method as HT-F, except with no addition of SDS.
Block mass precursor (HT-M) was prepared by coprecipitation method. First, 17.09 g Mg(NO 3 ) 2 ·6H 2 O and 12.50 g Al(NO 3 ) 3 ·9H 2 O were dissolved in 100 mL deionized water. Another solution was prepared with 8 g NaOH and 10.6 g Na 2 CO 3 dissolved in 100 mL deionized water. Then, two solutions were mixed drop-wise, with strong stirring at room temperature for 2 h. The pH value was controlled at about 10. Subsequently, the mixed solution was aged at 65 • C for 12 h. The resulting suspension was filtered, washed with deionized water to neutrality and dried in air at 100 • C overnight.
All the precursors were calcined for 4 h at 600 • C to obtain the corresponding calcined samples, named HT-XC (X represents morphology, either F, P or M).
The corresponding In-based precursors, InHT-X, were acquired by incipient wetness impregnation method for calcined samples HT-XC with In(NO 3 ) 3 ·xH 2 O aqueous solution at room temperature for 6 h. Next, PtInHT-XC catalysts were prepared by successive incipient impregnation method. Firstly, HT-XC was impregnated with an In(NO 3 ) 3 ·xH 2 O aqueous solution at room temperature for 6 h and later dried at 120 • C for 12 h to obtain In-based precursors InHT-X, and then these solids were calcined at 550 • C for 4 h; corresponding products were labeled InHT-XC. Finally, the same procedures and conditions were performed on InHT-XC impregnated with H 2 PtCl 6 ·6H 2 O, except that the time of impregnation was 2 h. After drying and calcination, the resulting solids were defined as PtInHT-X and PtInHT-XC, respectively. The loadings of Pt and In were 0.48 wt% and 1.38 wt%, respectively. After the reduction treatment under 5 vol% H 2 /N 2 at 600 • C for 2.5 h, these samples were labeled PtInHT-XR. And the used catalysts were marked as PtInHT-XU.
Low-temperature N 2 adsorption-desorption was performed at 77 K on a TriStar 3000 micromeritics apparatus (Micromeritics, Norcross, GA, USA) to collect the textural properties. Prior to measurements, the samples were outgassed under vacuum at 300 • C for 4 h, and then the specific surface area of samples was calculated by the Brunauer-Emmett-Teller (BET) method; the pore size distribution (PSD) was determined by the Barrett-Joyner-Halenda (BJH) method, used upon the adsorption branch of the isotherms.
Field emission scanning electron microscopy (FESEM) was carried out on a Hitachi S-4800 instrument (Tokyo, Japan) using a 3.0 kV electron beam.
Transmission electron microscopy and energy dispersive spectroscopy (TEM-EDS) were performed on a JEM-2100F field-emission transmission electron microscope (JEOL, Tokyo, Japan) with an accelerating voltage of 200 kV.
The X-ray photoelectron spectra (XPS) of catalysts were recorded on a Thermo Scientific Escalab 250Xi X-ray photoelectron spectrometer (Thermo Fisher Scientific, Waltham, MA, USA) using Al Kα (hv = 1253.6 eV) radiation. The BE was calibrated using the C1s level at 284.8 as an internal standard.
The dispersion degree of Pt (D Pt ) was calculated by CO chemisorption. CO chemisorption was conducted on AutoChem II 2920 analyzer (Micromeritics, Norcross, USA). A total of 100 mg of the samples were reduced under 10 vol% H 2 /Ar atmosphere at 600 • C for 2 h with a heating rate of 10 • C·min −1 . After reduction, the catalysts were swept with He, at 600 • C for 30 min, to remove H 2 , and then cooled down to 50 • C. Subsequently, the CO pulse chemisorption was carried out at 50 • C by injecting pulses of 5 vol% CO/He until CO was adsorbed to saturation.
Thermogravimetric analysis (TG) was carried on a DTG-50/50H thermal analyzer (PerkinElmer, Waltham, MA, USA) to determine the carbon amount over the used catalysts with a heating rate of 10 • C·min −1 from room temperature to 800 • C in air.

Catalytic Tests
Propane dehydrogenation (PDH) reaction was carried out in a fixed-bed reactor. The calcined sample (0.4 g) was placed into the stainless-steel tube and reduced to a flow of 5 vol% H 2 /N 2 at 600 • C for 2.5 h with a rising rate of 5 • C·min −1 . Afterwards, the PDH reaction was performed in a mixture of H 2 , N 2 and C 3 H 8 (H 2 :C 3 H 8 :N 2 = 7:8:35 (molar ratio) and WHSV = 3 h −1 ) at 600 • C. The regeneration test was performed at 600 • C, including oxidation in air for 2 h, sequencing purge with N 2 for 30 min and then reduction in 5 vol% H 2 /N 2 for 2 h. The propane and gas products were analyzed by an online gas chromatograph equipped with a FID detector and an Al 2 O 3 column. The propane conversion and propene selectivity were calculated as follows: Propane conversion(%) = propane, in − propane, out propane, in × 100% (1) Propene selectivity(%) = propene, out propane, in − propane, out × 100% (2)

Conclusions
In conclusion, the ball-flower-like, single-plate and block mass PtIn/Mg(Al)O catalysts were obtained by topological transformation and the reconstruction of HT precursors. The diversified morphology of catalysts was dependent on the conditions of synthesis. The metal ions of Pt 4+ and In 3+ can be introduced into the HT layer during the reconstruction process and the metallic species can be highly dispersed on the surface of Mg(Al)O. The ball-flower-like PtIn/Mg(Al)O catalyst exhibited high activity, excellent propene selectivity and superior durability as well as excellent resistance to the carbon deposition and Pt-sintering in the propane dehydrogenation to propene, which were mainly arising from its multi-level hierarchical microstructure with low specific surface area, and a large amount of strong and stable Pt-In interactions. It will be of great importance to revisit these catalysts and research the essential reasons for different catalytic performance in the PDH reaction by using more methods.