The Catalytic Function of Phosphorus Enriched on the Surface of Vanadium‐based Catalysts in Selective Oxidations

Vanadium phosphates are established as the benchmark system for the selective oxidation of n‐butane towards maleic anhydride. By varying the phosphorus content on the surface of three V‐based catalysts with diverse performance, this study experimentally elaborates on the catalytic function of phosphorus. Contact time variation and cofeed studies revealed, that surface phosphates, deposited in sub‐monolayers via atomic layer deposition, significantly contribute to an increased product selectivity. Furthermore, our results suggest that the phosphorus particularly suppresses the consecutive combustion of the (re‐)adsorbed product. The recently introduced solid solution catalyst V1‐xNbxOPO4 with medium maleic anhydride selectivity could be tuned by the surface enrichment with phosphorus towards product selectivities of up to SMAN=60 %, under optimized alkane‐rich feed conditions. Therefore, POx‐V0.3Nb0.7OPO4 is introduced as promising catalyst, which is not based on vanadyl(IV) pyrophosphate, to access significantly higher MAN formation rates at increased alkane partial pressures of cn‐butane>10 %vol in n‐butane oxidation.


Introduction
A sustainable chemical industry builds upon high performing catalysts to efficiently produce chemicals attractive in both environmental and economic prospects. [1]Next to the development of novel processes based on heterogeneous catalysts, the optimization of existing ones is crucial to achieve current netzero targets. [2,3]In this context, the suppression of by-product formation not only aids a thorough utilization of resources, but also the reduction of greenhouse gas emissions. [2]The functionalization of hydrocarbon feedstock using oxidation reactions is a crucial part of multiple value chains in the chemical industry with production capacities of several million tons per year worldwide. [4,5]Promising catalytic systems have been developed and optimized producing olefins [6] or and oxygenates [7,8,9] from saturated alkanes, without fully understanding the key descriptors or property-performance relationships for a selective CÀ H activation with molecular oxygen. [10,11]he industrial role model for this reaction pathway is the selective oxidation of n-butane producing maleic anhydride (MAN) with more than 2 million tons produced per year. [4,12]The use of a linear alkane as a resource, which is cheaper and more available than the respective olefin, was enabled by the discovery of vanadyl(IV) pyrophosphate (VPP). [7,13,14,15]Although this catalyst was continuously improved throughout the years, it offers a yield limitation of around 65 % as parallel and consecutive oxidation towards combustion products (CO, CO 2 ) cannot be fully avoided. [12,13,16]][19] Long term studies revealed a leaching of phosphorus (P) from the catalyst during operation, which is induced by steam as by-product. [20]This phenomenon is directly connected to descending product selectivities, that suggest that a sufficient amount of P is required to run the reaction selectively. [12,21]This correlation between the P-content at the active surface of the bulk catalyst and its catalytic performance has been manifested in multiple studies, experimental and theoretical, that investigate on the crucial role of P. [22,23] Industrially, a gaseous compound like trimethyl phosphate, volatile at reaction temperatures, is added to the reaction feed to compensate the P-loss and maintain high MAN selectivities. [12,20,24]lthough there is a general consent on the importance of P, the actual function or the catalytic effect is still under debate. [12,25]Hypotheses vary widely from a crucial influence on the initial CÀ H activation by P [23,26] to an effect on the surface redox properties or the oxidation state of V. [27,28] The identification of relevant property-performance relationships is not only challenging due to the multifunctionality of the catalyst, but also because of a transient behaviour and surface dynamics under reaction conditions. [27,29,30]Therefore, the performance is not only a function of the catalyst alone, but also of the applied gas feed connected with a reaction-induced restructuring of the surface. [31]This leads to a gradient in the properties comparing the bulk and the active surface, like for instance a surface enrichment of P with a beneficial effect on the performance. [32]n situ characterisation further revealed the formation of a disordered surface layer [33] and showed that the catalytic functionality of VPP is based on a dynamic interplay between different phases or surface modifications with local surface states (V 4 + /V 5 + ). [18,34]henomenon beyond comprehension require a reduction in complexity to achieve a full understanding, for instance by using model systems.The presence of P was proven in multiple studies to tune the catalytic properties in different reactions on various V-containing catalysts. [17,35,36,37]PO x introduced to the surface of the simple binary oxide V 2 O 5 [37,38] could be demonstrated to elevate the performance in n-butane oxidation from full combustion towards increased MAN selectivities. [17,36,37]The surface decoration of bulk catalysts using atomic layer deposition (ALD) [39,40] as a solvent-free technique allows to elucidate on the influence of different elements without changing the basic features of the base material. [17]During this method, the element precursors are fed to the substrate in gas phase causing a reaction at the solid-gas interface, which leaves a surface terminated by the introduced precursor and ultimately as oxide species.The impact of one deposited ALD-cycle, which corresponds to a sub-monolayer, [41] on the (surface-)properties of the bulk catalyst and the catalytic performance reveals insights into the function and interaction of different elements on the active surface layer.
The presented paper features the experimental investigation of the functionality of phosphorus in vanadium phosphate catalysts that enables the selective oxidation of n-butane.Therefore, three V-based catalysts, with high (VPP), [12,15,42] medium (V 0.3 Nb 0.7 OPO 4 ), [43,44] and low (V 2 O 5 ) [17,37] selectivity towards MAN, are modified with a surface-deposition with PO x , using ALD.The interaction between PO x -species and vanadium (V) as redox active element is studied for both modified and unmodified catalysts in n-butane oxidation via temperature, contact time, and co-feed studies under conditions resembling those of industrial applications.A comparison between the three systems reveals that PO x not only influences CÀ H activation, but more importantly, it suppresses the consecutive oxidation of MAN towards CO x .As a result, this study experimentally confirms that a surface enrichment with PO x of V-based catalysts significantly influences the reaction network and kinetics in the selective oxidation of n-butane.This switch in performance is shown to be accompanied by an increase in acidity of the solid catalysts, while the chemical state of V at the near-surface is not affected.The tuned surface properties lead to significantly increased MAN selectivities on PO x -V 0.3 Nb 0.7 OPO 4 , a material which shows optimized performance under alkane-rich feed conditions that enables highly increased product formation rates, as this study will additionally elaborate on.

Results and Discussion
The V-based oxide V 2 O 5 and phosphate V 0.3 Nb 0.7 OPO 4 are obtained as single-phase catalysts (see XRPD patterns in Supporting Information Figure SI1).The VPP sample, synthesized according to patent literature, revealed the presence of a minor phase-impurity with vanadyl pyrophosphate as main phase.All synthesis protocols are adapted from literature and are briefly described in the Experimental Section. [15,43,44]Depending on the synthesis route, the samples differ in their specific surface area, varying between 6 and 50 m 2 g À 1 (see Figure SI2, and Table SI1 for details on the N 2 -physisorption experiments).Figure 1 shows that the n-butane conversion at a fixed temperature setpoint correlates with the accessible surface area, while a normalization to the area and the exact sample weights (see data set in the SI) reveals very similar reaction rates and therefore catalytic activity (see Figure SI3).In general, all three materials are capable to activate the feedstock forming the desired oxygenate MAN together with other acid products like acrylic and acetic acid and at the end of the line the combustion products CO and CO 2 .The reaction mechanism of this reaction is studied in much detail with the main focus on the formation of the product MAN on VPP. [16,45]The total oxidation towards CO x proceeds in two ways, directly from the alkane -in parallel -and from the (re-adsorbed) product MAN and the other acid products -consecutively. [16,46]Overall, the selected catalysts show a great deviation in their MAN formation rate, which is mostly restrained by product selectivity.As a result, much higher MAN selectivities can be achieved on the vanadium phosphate catalysts than the mono metallic counterpart (S MAN -V 2 O 5 < S MAN -V 0.3 Nb 0.7 OPO 4 < S MAN -VPP; see Figure 2, and Figure SI4 for the whole product spectrum).A variation of the contact time, as presented in Figure 2, gives an insight into the kinetics of the product formation and more importantly on the stability of the already formed product.V 2 O 5 forms MAN in low selectivities S MAN < 20 %, which is quickly consecutively converted to CO and CO 2 .At elevated temperatures and therefore higher alkane conversion this sample functions similar to a combustion catalyst.Nonetheless, the extrapolation to low conversion levels reveals the ability of the simple oxide to form MAN from n-butane, which is then quickly consecutively converted to CO x .The fact that the normalized catalytic activity is very similar to the phosphate catalysts, as shown in Figure 1, leads to hypothesis, that the presence of P does not predominantly contribute to CÀ H activation on Vbased oxidation catalysts.Next, the solid solution V 0.3 Nb 0.7 OPO 4 , which was recently introduced as novel material with a tunable performance for this reaction, [43,44] converts the n-butane with a higher selectivity towards MAN.Nevertheless, a significant consecutive combustion can be observed throughout a broad conversion-range varying the contact time.The maximum MAN selectivity is obtained by the state-of-the-art catalyst VPP.More importantly, this catalyst shows very stable product selectivity until an alkane conversion of around 80 % is reached.A further increase in conversion at longer contact times then leads to a cut in selectivity.
The much lower selectivity of V 2 O 5 compared to the phosphate catalysts suggests that the addition of P is crucial to enable a selective oxidation, while preventing full oxidation.However, the performance of heterogeneous catalysts is not only a function of the element combination, but rather an interplay between many factors like the stoichiometry, crystal structure, porosity to name only a few.With the wide range in performance as foundation, this work aims to reveal the actual role of P in an experimental approach by studying the effect of P added to the catalysts without changing their basic features.Therefore, a sub-monolayer of PO x is introduced to the surface of the three catalysts using ALD (details can be found in the Experimental Section).This method is a powerful tool to modify bulk catalysts towards potentially tuned performance, while structural integrity is maintained.The solvent-free method at the solid-gas interface enables the investigation of element interactions at the active surface of the catalysts.The introduction of additional elements during or after synthesis of bulk catalysts in more conventional ways can possibly lead to massive changes in the material-properties and the formation of new phases cannot always be avoided.
The effect of deposited PO x on V-based oxidation catalysts.Tris-(dimethylamino)-phosphin is introduced as ALDprecursor in gas-phase to the bulk catalysts, herein referred to as "base" catalysts, placed in a fixed bed reactor (see Experimental Section for details).The deposition is determined by the self-limiting saturation of the surface.In order to investigate the effect of the interaction of PO x on the surface of the bulk materials, one ALD-cycle is employed, which correspond to a sub-monolayer. [41]While in case of V 2 O 5 the termination layer of the catalyst is complemented with an additional element, in case of V 0.3 Nb 0.7 OPO 4 and VPP the surface is enriched with PO x .The second half-cycle of the ALD process includes a calcination of all samples, which leaves the surface with phosphorus oxide groups.The deposition of additional elements to bulk catalysts, like the introduction of PO x to V 2 O 5 , allows the quantification of overall deposition in reference to the whole bulk (ICP-OES, XRF) or specifically in the near-surface region (XPS).Accordingly, the overall element loading after one ALD cycle is at around 0.1 wt%, determined for PO x on V 2 O 5 by XRF in a previous study. [17]In case of a surface enrichment with PO x the determination or quantification of the deposited precursor is more challenging due to the fact, that it cannot be differentiated between the species already contained in the bulk and the elements introduced by the modification.However, during the surface modification the overall mass gain is monitored by an in situ magnetic suspension balance (see exemplary mass gain curve in Figure SI5).A weight-gain of the samples during the treatment is used to monitor the deposition of the termination groups (e. g.Δw = 2.2 wt% on PO x -V 0.3 Nb 0.7 OPO 4 after 1 ALD-cycle).Depending on the base catalyst and the used ALD-precursor, a certain change in the deposition amount can be expected due to the severe differences in surface area.Additionally, the reactivity of the surface and the availability of surface hydroxyl groups determine deposited mass during the ALD-process.More importantly, the deposition is in all cases terminated by a surface saturation, which ensures a reproducible process.Another characteristic of the surface deposition by ALD is a homogeneous element dispersion on the surface of bulk catalysts. [17]Moreover, the basic (mechanical-) features of the base catalyst are maintained during the modification step.Therefore, only a small deviation in the determined specific surface area or the specific pore volume is observed (see Table SI2 and Figure SI6 for details on the N 2 -physisorption experiments).In addition, the surface modification does not change the morphology of the samples and no additional phases are formed, due to the solvent free method at temperatures below the calcination temperature of the base catalysts. [36]ll samples, with and without the surface modification, are studied in the oxidation of n-butane following a fixed design of experiments (DoE) including a variation of the reaction temperature and contact time.The comparisons are based on the data obtained after the catalysts reach the maximum reaction temperature for the respective sample.Therefore, a possible formation phase does not disturb a clean comparison in performance.Additionally, continuous and non-reversible changes are monitored during the test procedures by applying the same conditions three times during the whole DoE.With these reference setpoints, the stability of the catalysts can also be investigated throughout a time on stream (TOS) of about 7 days (see Experimental Section for details).Figure 3 shows the catalytic impact of the surface deposition of a sub-monolayer of PO x on the activity and selectivity of the three catalysts during the temperature variation study (see Figure SI7 for details on the activity change and Figure SI8 for the complete product spectrum).In case of V 2 O 5 the addition of PO x to the surface favours both performance parameters with significantly higher MAN selectivities, while the conversion is slightly increased at a fixed temperature setpoint.The PO x -deposition on the two studied phosphate catalysts leads to a surface-enrichment with P (see Figure 3b and 3c), since it is one of the bulk elements.The solid solution V 0.3 Nb 0.7 OPO 4 converts n-butane in low but more stable quantities towards MAN.A selectivity of about S MAN = 20-25 % can be sustained also at high reaction temperatures.The deposition of PO x leads to a remarkable boost in product selectivity in the partial oxidation of n-butane of around S MAN = 50 %.This increase can be sustained throughout nearly the full range of alkane conversions.With this jump in performance, a promising novel V-phosphate based catalyst, PO x -V 0.3 Nb 0.7 OPO 4 , has been identified.Especially the capability to reach these high product selectivities at alkane conversions above 70 % could not be achieved by other bulk catalysts other than VPP. [43]Speaking of the benchmark system for this reaction, the effect of the surface enrichment with PO x has been studied as well.Figure 3c shows that for this base catalyst the surface modification only impacts the alkane conversion, while a stable product selectivity can be maintained.As a result, the performance is only moving on the S-X curve with the trend towards lower alkane conversions correlated to the surface enrichment with P on the surface of VPP.This effect is expected as the dynamic change of the P-content on the surface of this catalyst is discussed in much detail in literature. [27,30]In fact, the deposition of PO x follows the inverted activation process of the catalyst, in which surface V/P ratio varies under reaction conditions due to a loss in P connected to increasing activities.Therefore, a P-compound is co-fed to the reaction stream to sustain high product selectivity in industrial applications.Furthermore, the effect of reduced activity can be used in largescale applications to moderate temperature-profiles within the catalytic bed during operation. [12,20,24]The observed changes in the catalytic activity induced by the surface deposition are directly related to an effect on the active sites of the catalysts, as basic properties like the surface area are maintained during the process (see Table SI2).Consequently, on the surface of VPP the addition of P seems to only block the active sites responsible for the activation of n-butane, while the features for a selective conversion of the intermediate species are not interfered.In contrast to this, the improvements of the product formation on the base catalysts with a lower initial selectivity towards MAN emphasizes the beneficial catalytic function of sufficient amounts of P at the solid-gas interface of a V-based catalyst in selective oxidations.
In order to further elaborate on the cause of this effect, contact time variation studies of (PO x -)V 2 O 5 and (PO x -)V 0.3 Nb 0.7 OPO 4 were conducted.Following the complex reaction network, changes in the kinetics of the overall MAN formation by the surface modification give an insight into the tuned (surface-)processes.As already discussed, at elevated alkane conversions V 2 O 5 converts the feedstock completely into CO and CO 2 .A GHSV variation reveals, that this is not only due to a direct conversion from n-butane, but also and more importantly from a consecutive oxidation of MAN and acetic acid as intermediate products (see Figure SI9).The effect of PO xaddition to the surface of the catalyst does not only favour the overall product formation, but also increases the slope in the Y MAN -X n-butane plot, shown in Figure 4. Starting from the MAN selectivity of the first measurement point, the linear functions  indicating iso-selectivity within the plots.The change in slope of these lines represents the tuned potential to form MAN from n-butane.In contrast to this, the deviation between the following measurement points to the iso-selectivity line represents the degree the product selectivity suffers from consecutive combustion.For both base catalysts a lower MAN selectivity can be observed combined with a drastic drop in product formation at high conversions.For instance, the MAN yield drops to zero for V 2 O 5 , as all formed MAN is consecutively converted to CO x .However, the surface modified sample PO x -V 2 O 5 still shows consecutive combustion, but MAN yields close to the iso-selectivity line up until an alkane conversion of 30 %.In combination, this catalyst is capable of forming MAN also at n-butane conversions > 60 %, due to the surface modification.A similar behaviour can be observed for the (PO x -)V 0.3 Nb 0.7 OPO 4 samples.While the base catalyst alone shows quite drastic consecutive oxidation of the feedstock, the surface enrichment with PO x not only increases the overall MAN yield, but also significantly suppresses the consecutive combustion (see Figure 4b).
As discussed above, the contact-time variation is valuable tool to dig deeper into promoting effects in catalysis and investigate on the origin of the observed effects.In a complex reaction, multiple parallel and consecutive steps are mutually depending on each other, which at the end of the line specify the measured product distribution.For a clear confirmation of changing kinetics of single reaction pathways typically co-feed studies are conducted for every final or intermediate product.In order to further analyse the consecutive combustion of MAN formed from n-butane on the surface of V-based oxidation catalysts, the reaction product is co-fed to the reactant stream and the effect on the outlet concentrations are tested, comparing the base catalyst V 2 O 5 and the surface-modified counterpart.The MAN input was varied in four different concentrations and was monitored by GC-analysis of a bypass reactor, visualized as concentrations marked in yellow in Figure 5a.Then, the MAN outlet concentrations obtained over the two catalysts are shown in the same diagram, which indicates a severe consecutive combustion of the re-adsorbed product MAN on the unmodified V 2 O 5 .On the contrary, a further formation of MAN additional to the co-fed product can be measured on the PO x -modified V 2 O 5 .As the formed MAN can originate from both the partial oxidation of n-butane and from the co-feed, this study shows, that the consecutive combustion cannot be fully avoided.However, this comparison shows that the addition of PO x clearly suppresses a further oxidation of the product from the gas-phase (see Figure SI10 for the change in CO x concentrations).Figure 5b shows the product spectrum, which is obtained from the MAN co-feed.Therefore, the MAN is converted to both CO and CO 2 on unmodified V 2 O 5 .
Effect of surface modification on catalytic properties.The mechanistic understanding of the element deposition on porous solids via ALD has been studied in detail in literature. [41]he introduction of elements in low concentrations or submonolayers on the surface of a bulk catalyst, without changing its main features, can be used to study the actual functionality of the deposited element. [17]As the modification happens exclusively at the surface or solid-gas interface, a characterization of this effect is limited to surface-sensitive methods.In a recent publication it has been demonstrated how catalytic properties like the oxidation state of the redox active element V at the near-surface or the acidity can be tuned by the surface deposition with PO x or BO x . [36]The intensity of the effects is always a function of the used base material, the deposited element and the amount or ALD-cycle applied to the catalyst.The effect varies from a simple blockage of active, selective, or acid sites to a more complex electronic effect.
The impact of the PO x -deposition on the catalytic properties of the V-based oxidation catalysts, studied in this work, has been further analysed by surface-sensitive methods.The acidity of a solid catalyst is one descriptor for the interaction between the surface sites and the applied gas-phase, which has a strong influence on catalysis. [47]Partial oxidations are comprised by multiple reaction steps including adsorption processes for the CÀ H activation of the reactant and desorption processes to form the desired oxygenate in gas-phase.As a result, acidity is a valuable descriptor for catalytic performance.Figure 6 shows the results of the temperature-programmed desorption experiments using ammonia as probe molecule, which is commonly used study the surface acidity of solid catalysts. [48]The desorption signal, monitored by a TCD (see Experimental Section for details, and Figure SI11 for desorption curves), is used for an overall quantification of desorbed ammonia compared in the bar plot.Interestingly, the deposition of a submonolayer of PO x on both catalysts leads to a significant increase in quantified desorbed NH 3 .The described effect can be observed both before and after catalysis (see Figure SI11e for results of the fresh samples).The increase in acidity can be related to the tuned product selectivity, which can be determined for both samples.However, the difference in acidity cannot be correlated to catalytic activity, which is only enhanced on V 2 O 5 by the PO x deposition.Similar results were found in a study, in which P added to supported V-based reduction catalysts lead to an increased acidity and a decrease in catalytic activity. [49]Therefore, acidity is only one feature that can dictate catalytic functionality. [50]n order to monitor any possible effects on the chemical state of the elements at the near-surface by the modification, Xray photoelectron spectroscopy (XPS) was performed.First, all P-species are identified as phosphates, either deposited on the surface or already present in the catalyst bulk (see Figure SI12 for the XPS spectra).The V 0.3 Nb 0.7 OPO 4 exclusively comprise Nb 5 + detected via XPS, which is not influenced by the ALDmodification.In contrast, V is present in two different oxidation states V 4 + and V 5 + in all samples.A change in composition (V to Nb ratio) within the solid solution does lead to a significant switch in the V-oxidation state at the near-surface, which was identified as a precise descriptor for product selectivity in a previous study. [43]The modification of the surface with PO x , shown in this paper, does not lead to any significant changes in the oxidation state of V, based on the measurements of the spent samples, shown in Figure SI13.A slight oxidative effect towards V 5 + can be observed induced by the deposition of PO x on V 0.3 Nb 0.7 OPO 4 , comparing the fresh samples before and after modification.In another study it has been shown, that the first PO x -ALD cycle is capable of reducing near-surface V 5 + on V 2 O 5 .However, the calcination and treatment with oxygen for the removal of the ligands in the second half-cycle leads to the inverse effect and the initial oxidation state of V. [41] The V 2 O 5 used in this study is following this behaviour, as no significant effect on the chemical state can be determined through modification, nor testing.Since the comparison between the spent samples is closer to the actual reaction conditions, the surface enrichment with PO x on the surface of V 0.3 Nb 0.7 OPO 4 seems to have no effect on the electronic properties of nearsurface V-species.This suggests, that the promoting effect of surface PO x groups in catalysis is not related to a change the oxidation state of V, which is often correlated with performance.
Apart from the influence of the added surface groups on the bulk elements at the active surface, the deposition triggers a chemical bonding between the dosed precursor and the active surface-species.Based on the knowledge gained throughout the studies on VPP treated by a P-compound, a certain amount of P is necessary at the surface to suppress unwanted reaction pathways.One phenomenological descriptor for this effect is the site isolation, postulated by Grasselli as one of the seven pillars of oxidation catalysis. [51]According to multiple studies on the dynamic P-enrichment under reaction conditions on the surface of VPP, the presence of P leads to a steric isolation of the active V-species. [52]The deposition of PO x by ALD is externally simulating an enrichment process with PO x on the surface of the selected V-containing oxidation catalysts.According to the sample-composition at the near-surface determined by XPS (see Table SI3), the surface deposition of a submonolayer of PO x adds about 2 wt% of P to the surface of V 2 O 5 , which in this case shows no significant decline in the Vavailability at the surface (see Table SI4 and Figure SI14).On the contrary, the amount of V on V 0.3 Nb 0.7 OPO 4 drops from around 4.2 wt% to 3.4 wt%.This decline of 17 % of V indicates a shortage of V the redox active element at the surface of the catalyst.Therefore, a steric or geometric effect, isolating the Vatoms on the surface of the catalysts seems conceivable.In another perspective, the introduction of P by ALD can lead to a precise blocking of unselective V-sites in line with the tuned product selectivity.In case of VPP, these sites are possibly already saturated by PO x -groups, as a further enrichment with P only lead to decreasing activity with constant MAN selectivity (see Figure 3).Apart from the effects of the surface-modification, the results of the XPS studies, including both the fresh and the spent samples, indicate a certain change of the catalyst surfaces induced by the exposure to the reaction conditions.A variation in the oxidation states of the near-surface V-sites or a change in the surface composition, due to the reaction, shows similarities to the dynamic surface-restructuring observed on VPP.In this context, a certain mobility of the deposited PO xspecies seems conceivable.However, a complete migration to the catalyst bulk was not observed, as PO x -V 2 O 5 shows no leaching behaviour of P from the surface supported by a stable change in its catalytic performance.
Optimized alkane-rich feed conditions.The industrially applied partial oxidation of n-butane on VPP usually operates at low alkane concentrations right below the lower flammability limitation with air as the oxygen source (c n-butane = 2 %vol, c oxygen = 20 %vol). [7]Therefore, the productivity of the process is not only limited by the product selectivity, but also the overall utilization of the reactant flow in large scale fixed bed applications using multitube reactors. [24]These limitations are mainly caused by two reasons.First, the feed composition is based on the optimal operating point of VPP and increased alkane input-concentrations are mostly leading to a trade-off in terms of MAN selectivity. [53]Then, the mixture of n-butane in air is flammable between around c n-butane = 1.8 %vol and 7.8 %vol at a fixed oxygen concentration of c oxygen = 20 %vol. [54]As a result, an increase of the alkane concentration is only possible above the upper flammability limitation.At this point it must be noted that an operation in this range is feasible and applied for instance in fluidized bed processes. [7]Typically, novel or modified catalysts for this reaction are tested under the same reaction conditions used for the state-of-the-art system.Even though this benchmark comparison is necessary, the reaction conditions are possibly not ideal for other catalysts and, furthermore, increased reactant concentrations are highly desirable in terms of productivity.Still, the rise in the n-butane concentration would bring up further challenges to solve, as for example the heat deviation in industrial scale is already challenging to maintain even at these low alkane inputs.
In this study, the surface modified PO x -V 0.3 Nb 0.7 OPO 4 was identified as promising bulk catalyst in the partial oxidation of n-butane.Without an extensively optimized synthesis procedure, a bulk material with a high surface area is obtained, which activates the alkane feedstock even at mild temperatures.The surface enrichment with PO x via one ALD-cycle could be shown to enhance the MAN selectivity to S MAN = 50 % at high alkane conversions of X n-butane > 80 %, which could not be achieved for any other V-based bulk material except of VPP so far.Although the product selectivities are still superior on the benchmark catalyst, it must be kept in mind that VPP was optimized throughout decades of research to reach the current performance standards.Recently, the solid solution V 1-x Nb x OPO 4 was introduced as mixed-metal phosphate catalyst with tunable catalytic properties in the partial oxidation of n-butane. [43]While the system could be used to experimentally confirm the site isolation and the vanadium oxidation state as key properties for a selective partial oxidation, the material additionally showed unique characteristics under alkane-rich feed conditions.In contrast to VPP, the increased n-butane concentration above c n-butane > 10 %vol leads to both, a boost in the overall alkane consumption and a steadily increasing MAN selectivity.In the following, this effect has been investigated also for the PO x -modified V 0.3 Nb 0.7 OPO 4 by varying the feed composition.
Table 1 shows the impact of alkane-rich feed conditions on the catalytic performance comparing the base catalyst V 0.3 Nb 0.7 OPO 4 , the PO x -modified sample and VPP as reference.A change in the n-butane input by a factor of 5, while maintaining the other reaction parameters, leads to an increased overall n-butane consumption rate on all the three samples, but to a certainly different degree.This disparity becomes very clear by comparing the alkane conversion achieved at a fixed temperature setpoint.While VPP follows a more expectable trend with a decreased overall conversion, both unmodified and modified V 0.3 Nb 0.7 OPO 4 show an even greater alkane conversion, due to the increased reactant availability.This leads to the observed jump in the alkane consumption rate, by a factor of more than 4, compared to VPP, which is therefore not only a function of the difference in surface area (S BET,VPP = 26 m 2 g À 1 , S BET,VNbOPO4 = 52 m 2 g À 1 ).In general, the switch in the feed composition has a much greater effect on the reactant consumption than the product selectivity.Nevertheless, a certain tendency can be again observed with an increased MAN selectivity on the V 0.3 Nb 0.7 OPO 4 in comparison to a slight decrease in selectivity on VPP.The combination of both effects is illustrated in Figure 7, which shows the MAN formation rates at two different n-butane and oxygen concentrations (see Table SI5 for details).Hence, at low alkane concentrations and 20 %vol of oxygen the productivity to form MAN is superior on VPP.Whereas, a switch in the n-butane content makes the PO x -V 0.3 Nb 0.7 OPO 4 the most productive catalyst.The increased n-butane input leads to a jump in the alkane consumption rate (see Figure SI15) and accordingly to a highly increased MAN formation on the sample, which is much higher than the improvements observed for VPP at the same temperature.In this specific comparison based on the formation rate, the MAN selectivity plays a subordinated role, as less variation is observed by the switch in feed composition.Still, the product selectivity on both V 0.3 Nb 0.7 OPO 4 catalysts is slightly promoted by the alkane-rich feed conditions, while it drops in case of VPP (see Figure SI16 and Figure SI17 for the results of a temperature variation study).Consequently, the results of PO x -V 0.3 Nb 0.7 OPO 4 show a great potential enable an efficient n-butane oxidation Table 1.Variation of the feed composition: Change in catalytic performance (n-butane conversion, selectivity towards MAN, n-butane consumption rate, MAN formation rate) und alkane-rich feed conditions at c n-butane = 10 %vol at temperature setpoints of T = 325 °C and 350 °C to achieve similar conversions; Effect of an alkane-rich and oxygen-lean feed composition is shown in Table SI5  under alkane-rich feed conditions with highly increased MAN formation rates.
Once operating under alkane-rich feed conditions above the upper flammability limitation, a further increase in the reactant concentration is conceivable.At this point another bottleneck needs to be taken care of, namely the total consumption of oxygen, which would lead to unwanted side reactions.In a prior study, it has been shown that the reaction order of oxygen is close to zero on the solid solution catalyst. [43]Therefore, an increased oxygen supply or a staged feed of oxygen would be a solution to extend the boundary conditions for operation under shifted feed conditions.In a next study, the alkane input concentration was further increased towards c n-butane = 20 %vol.Again, to avoid full conversion of the reactants a comparably low reaction temperature is selected to study the effects in catalytic performance.While keeping all boundary conditions constant, like the steam and oxygen concentration and flow rates, the n-butane concentration is gradually increased in steps of Δ = 2 %vol until 20 %vol is reached.The impact on PO x -V 0.3 Nb 0.7 OPO 4 as well as VPP as a reference is illustrated by plotting the alkane consumption and MAN formation rate as a function of the alkane input concentration, as shown in Figure 8.As already discussed above, the change in the oxygen/n-butane ratio from 10 towards 2 leads to an increased alkane consumption and product formation on both catalysts.The severe difference becomes even more clear by comparing the incline in alkane consumption with the increasing n-butane feed.VPP shows an increased alkane consumption from below 2 mmol g À 1 h À 1 to slightly above 3 mmol g À 1 h À 1 at 10 %vol of n-butane.A further increase has a minor, but positive effect on the overall alkane consumption.At the same time, the selectivity towards the desired product MAN is slowly decreasing, which leads to only a very small improvement in the product formation, induced by the increased n-butane supply.In contrast to this, the surface modified solid solution PO x -V 0.3 Nb 0.7 OPO 4 converts the alkane in much higher rates under alkane-rich feed conditions (c n-butane > 10 %vol).Therefore, the jump in the n-butane consumption rate is much more severe and the continuous increase in alkane input leads to steadily rising rates.In combination, the product formation rate is improved by a factor of 6, as in this case also the selectivity towards MAN is increasing as well (see Figure SI16).The differences between the two catalysts becomes clear by comparing the slopes of the rate as a linear function of the n-butane input in the alkane-rich feed.To avoid the total conversion of oxygen, two different temperature setpoints are selected for the two samples.Due to the described opposite trends in catalytic activity, a comparable range in the n-butane conversion is achieved at the alkane-rich feed conditions (c n-butane = 10-20 %vol, see Table SI6).On the one hand, the improved slopes in both, the MAN formation and the alkane consumption rate on PO x -V 0.3 Nb 0.7 OPO 4 underlines the superior performance.On the other hand, the linearization shows a regime-change, as on both catalysts the reaction rates below the explosive range (c n-butane = 2 %vol) do not fit into these linear trends.
The unique characteristics of V 1-x Nb x OPO 4 under alkane-rich feed conditions were demonstrated for these catalysts in the full composition-range by varying the VÀ Nb ratio in a prior study. [43]he effect cannot be fully assigned to the difference in surface area, as in the mentioned study V 1-x Nb x OPO 4 compositions with even lower accessible surface area show a similar increase in the reaction rate.The discussed beneficial effect of the surfacedeposition of PO x is not affected by the change in feed concentrations and is therefore further enhancing the product formation rate (see Figure 7).This enables the access of potentially highly increased MAN formation rates in the selective oxidation of n-butane.Furthermore, only mild temperatures of < 350 °C are necessary to reach these high formation rates.However, the bottlenecks of the reaction are switching towards new limitation factors, for instance the much more prominent shortage of oxygen as reactant compared to the alkane.According to the stoichiometry of the reaction (see eq. 1), much more oxygen is required, which is only rarely limiting under reference feed conditions (c n-butane = 2 %vol, c oxygen = 20 %vol).As a result, without increasing the input of oxygen, which could be facilitated by feeding pure oxygen instead of using air as oxygen source, the n-butane  SI6 and the performance under alkane-rich feed conditions at different temperatures can be found in Figure SI17; flow rates are kept constant at a GHSV of 1000 h À 1 , 1 atm, O 2 = 20 %vol, H 2 O = 3 %vol.
conversion cannot be optionally increased by a rise in reaction temperature.This has a significant effect on the product reaction stream, as much more of the unconverted reactant is present.Currently, the industrially applied reaction operates at high n-butane conversions (X n-butane ~80 %), which leads to only a minor fraction of the alkane in the product stream.Depending on the resource cost, a separation of the product stream, after the MAN has been extracted by condensation, is not economically favourable.In the discussed alkane-rich feed conditions this calculation is mixed up by a much higher concentration of n-butane still present after the reaction, which makes the applications of for instance recycling streams or circulation processes interesting, thinking of an industrial scale.In the case of an exothermic reaction, a highly increased product formation rate is coupled with an additional generation of heat, that needs to be handled as well.
Despite of the mentioned aspects, which might need some more consideration in a larger scale, the here introduced catalyst PO x -V 0.3 Nb 0.7 OPO 4 shows a high potential to surpass the formation rates achieved by VPP.At this point, the process on VPP is still more efficient, considering the advantage in the selectivity towards MAN.However, V 0.3 Nb 0.7 OPO 4 possibly offers much more room for improvements, as beside of the surface modification many aspects are not necessarily optimized, as they have been done for VPP.Compared to other bulk catalysts, studied for partial oxidation and specifically for the selective oxidation of n-butane, the performance of PO x -V 0.3 Nb 0.7 OPO 4 is the closest to VPP measured so far.Similar to the state-of-the-art catalyst, this study emphasizes the crucial influence of the reaction gas-phase on the bulk oxidation catalyst.Another change in the reaction feed composition revealed that the absence of co-fed steam further promotes the catalytic performance specifically on PO x -V 0.3 Nb 0.7 OPO 4 with an increase in both conversion and selectivity towards S MAN = 60 % (see Figure SI18).A low steam feed concentration of 3 vol % supports the MAN formation on VPP, as without it, a slight decrease in selectivity is observed (see Figure SI19).On the contrary, these results suggest, that the process on PO x -V 0.3 Nb 0.7 OPO 4 could be tuned even more by removing the byproduct water out of the product stream.
Subsequently, the variation of the reactant concentration should be a much more prominent part of standardized performance tests, in order to reveal the full potential of novel or modified catalysts. [11,55]Furthermore, the harsh impact of the surface enrichment with only a sub-monolayer of PO x on the vanadium phosphate catalyst demonstrates the importance of the solid-gas interface for heterogeneous bulk catalysts.Therefore, surface sensitive methods like ALD are proven to be powerful tools to enhance catalytic performance.

Conclusions
Partial oxidations require multifunctional catalysts with the capability to activate the alkane feedstock, to form the desired oxygenate, to desorb it back to the gas-phase, and to prevent the consecutive combustion of the (re-)adsorbed product.Vbased oxide and phosphate bulk materials are known for decades to be a solution for these complex processes, while the actual functionalities of the catalysts are still under debate.In this work, we experimentally demonstrated the beneficial interplay between PO x and V at the near-surface of these catalysts.This led to a better understanding of the actual role of P in vanadium phosphate catalysts, which are applied in industry for many years.The deposition of small amounts of PO x as sub-monolayer on the surface of three catalysts with diverse catalytic behaviours via ALD revealed a significant increase in MAN selectivity on the catalysts with a low or medium initial selectivity.The formation of unwanted byproducts could be shown to originate not only from the direct combustion of the alkane, but also the secondary oxidation of already formed oxygenate products towards CO x .A variation in contact time and MAN cofeed studies suggest, that the presence of certain amounts of P actively suppresses the total combustion of the feedstock following these consecutive reaction pathways on V-based oxidation catalysts.A thorough characterization of the samples before and after the surfacemodification showed that the addition of PO x by ALD has no noticeable impact on the morphology, basic mechanic features, or the electronic environment of the redox active element V at the near-surface.On the contrary, surface-sensitive methods were used to relate the introduction of PO x to the surface of the bulk catalysts to an increase in acidity, which emphasizes the effect on the interaction between the active surface of the bulk catalysts and the applied gas-phase.A variation of the feed concentrations is used to further evaluate the potential of PO x -V 0.3 Nb 0.7 OPO 4 , a catalyst discovered in our study with very promising performance towards highly increased MAN formation rates.An increased n-butane supply of up to 20 %vol leads to a boost in the overall alkane consumption rate in combination with an improved MAN selectivity.The MAN formation could be pushed by a factor of around 6, compared to VPP where the alkane-rich feed conditions lead to only a small increase by a factor of 2, while the MAN selectivity drops in this case.Under optimized reaction conditions, PO x -V 0.3 Nb 0.7 OPO 4 converts n-butane towards MAN at a product selectivity of up to S MAN = 60 % at very high reaction rates, a benchmark which was never achieved by a VPP-free system so far.
This study highlights the importance of the gas-solid interface in the catalyst functionality of bulk oxide and phosphate catalysts.A surface enrichment with P of V-based oxidation catalysts was shown to be particularly beneficial for a selective oxidation and therefore a reduced combustion of resources.Our study demonstrates ALD to be a useful tool to enhance catalytic performance of bulk catalysts and, furthermore, to elucidate on the underlying principles of oxidation catalysis.Finally, the utilization of alkane-rich feed conditions (c n-butane > 10 %vol) highlights a potential way to access much higher productivities in n-butane oxidation.

Experimental Section
Catalyst synthesis and surface modification.The phosphate catalysts, namely V 0.3 Nb 0.7 OPO 4 , and VPP, were prepared in a sufficient batch size, according to synthesis-protocols briefly described in the SI.The binary oxide V 2 O 5 is commercially available (Dallan Galaxy Metal Material Co. Ltd.; CAS-Nr.1314-62-1).The mixed metal phosphate catalyst V 1-x Nb x OPO 4 [43,56,57] is accessed as solid solution through a solution combustion synthesis route. [58]The respective precursors are dissolved in water and used in a stochiometric ratio to obtain a V-lean orthophosphate structure V 0.3 Nb 0.7 OPO 4 .The VPP is synthesized according to patent literature following the organic route. [15]Phosphorus-oxide is deposited on the surface of the base catalysts using atomic layer deposition (ALD). [40,59]The overall deposition can proceed in multiple cycles with increasing loadings, while in this study only 1 ALD-cycle is performed, which could be shown to correspond to a submonolayer. [41,60]In the first ALD half cycle, tris-(dimethylamino)phosphin (Sigma-Aldrich, 97 %, room temperature) was introduced in a quartz tube fixed bed reactor containing 5 ml of the catalyst (T = 70 °C, 1 atm) using argon (99.999 %, 100 ml min À 1 ) as carrier gas.The cycle is self-limited once the surface saturation is achieved.The point of precursor-saturation is determined by online mass spectrometry.The mass gain during ALD was proven by an in situ magnetic suspension balance (see example in Figure SI5). [61]The second half cycle comprises the removal of the precursor ligands by the dosing of water in argon, leaving the surface with PO xgroups.In between, before and after the half-cycles the sample is purged with argon.Details about the experimental setup and the ALD process are described elsewhere. [41,61]After drying step in synthetic air (20 % O 2 in N 2 ) at T = 450 °C for 4 h, the modified samples are cooled down and sieved for catalyst testing.
Catalytic testing.Catalytic studies on the partial oxidation of n-butane were performed under industrially-relevant reaction conditions in a commercial 8-fold parallel test setup build by hte GmbH.The basic data set of every base catalyst and modified sample comprise a temperature and contact time variation study, following a fixed DoE.The feed composition is kept constant based on academic and industrial standards balanced with nitrogen (C 4 H 10 /O 2 /H 2 O: 2/20/3 %vol) at a gas space velocity of 2000 h À 1 and atmospheric pressure. [9]Sample volumes of V cat = 0.7/1.0 ml were inserted in stainless steel reactors with inner diameters of d inner = 7/ 10 mm embedded into an inert filling with steatite.To ensure the verified optimal mass and heat transfer, the catalyst powder is pressed (1 min, 1 ton, d pellet = 14 mm), crushed, and sieved before modification and testing with a final particle size fraction of 100-200 μm.After the surface modification via ALD the catalysts are only re-sieved before inserted into the test setup.Reaction temperatures are individually controlled in a range between T = 225 and 450 °C and monitored at three stages along the catalyst bed.The maximal temperature for every sample is determined by a conversion limitation of n-butane or oxygen, which should not exceed X � 85 %.The DoE of the basic data set includes a temperature ramp with five different setpoints.For the catalytic evaluation data was used, which was recorded after the catalysts once reach high performance, in order to avoid interference of possible formation phase effects.From the second part of the temperature ramp a setpoint is selected (ΔT = 25 K), at which an alkane conversion of around X = 15 % is reached.At this temperature the GHSV is varied at four different setpoints from 500 to 4000 h À 1 .In order to extract the effect of the surface-deposition with PO x , the modified samples were tested at the same temperature as the respective base catalyst during the contact time variation.The complete DoE results in a time on stream of the samples of around 7 days.Due to the repetition of reaction conditions (temperature ramp-up, ramp-down, GHSV variation), the catalytic data sets include a measurable descriptor for catalytic stability.Selected samples were additionally tested at different reactant concentrations including a variation of the n-butane (2, 10-20 %vol) or steam (3, 0 %vol) partial pressures.Details on the reaction conditions are provided in the data set within the SI.
The outlet concentrations of each reactor are monitored using an online gas chromatography system equipped with flame ionization and thermal conductivity detectors (Two GC 7890 A, Agilent).Next to the reactants, products like MAN, acrylic acid, acetic acid, acetaldehyde, alkenes, alkanes, CO, and CO 2 among others are identified and quantified.A closed carbon balance is calculated by comparing input and output streams.As an internal standard, Argon is used to eliminate a possible gas expansion.Moreover, a blank reactor is used for the online determination of the input reactant concentrations.Details on the technical setup and the evidence of the independence of mass and heat transfer limitations are described elsewhere. [16,56]The co-feed studies were conducted using a self-constructed transient kinetic test setup, which is comprehensively depicted elsewhere including all technical details. [9]The reaction product MAN was introduced as a steady state cofeed in four different concentrations (0, 0.2, 0.3, 0.5 %vol) to 0.6 ml of the catalysts (100-200 μm) inserted as a fixed bed into a reactor consisting of glassed line tubing (d inner = 4 mm).At the same time, the reaction conditions are kept constant (GHSV = 2000 h À 1 , T reaction = 420 °C, C 4 H 10 /O 2 /H 2 O/N 2 /He: 2/20/3/10/balance %vol), in accordance with those used in the performance tests.The conditions were selected based on the catalytic activity to ensure a low n-butane conversion of X n-butane = 10-15 %, at which a sufficient MAN selectivity is achieved.Raw-data evaluation was executed using the commercial software tool myhte towards performance descriptors (n-butane conversion X n-butane , selectivity S j , n-butane consumption rate r n-butane , product formation rate r j ) determined using a product-based approach (see eqs. 2-6 with i,j = reaction products, N C = number of carbon atoms, _ V Feed ¼feed velocity, V m = molar volume, m Cat = catalyst mass).The complete catalytic data set including exact catalyst weights is included in the SI.

X nÀ
N 2 -physisorption.Nitrogen physisorption analysis was conducted at 77 K (liquid N 2 ) using a Quadrasorb SI device manufactured by Quantachrome.Before the adsorption isotherm were recorded, the samples were degassed at T = 200 °C for 2 h.The apparent surface area was determined by the Brunauer-Emmett-Teller (BET) method in the linear P/P 0 = 0.05-0.3pressure range of fresh, spent, base and surface-modified catalysts.The apparent pore size distribution is calculated based on the nitrogen-desorption following the BJHmethod.
X-ray diffraction.For phase identification and purity control X-ray powder diffraction (XRPD) patterns were recorded at ambient temperature using an X'PERT Pro manufactured by PANalytical equipped with a scintillation detector (Cu Kα1 radiation, λ = 0.154 nm, 15 minutes exposure time in the angular range 10°� 2θ < 80°).
Elemental Analysis.The chemical state of the near surface region of the samples was studied by X-ray photoelectron spectroscopy (XPS).A Thermo Fisher Scientific K-Alpha with monochromatic AlÀ Kα X-ray source (1486.6 eV) was employed for the analysis, using a spot size of 200 microns, a pass energy of 200 eV for a survey, and 50 eV for high-resolution spectra.The samples were mounted on conductive carbon tape.The analysis of the spectra was done using the Avantage software.The C1s peak of adventitious carbon at 284.8 eV was taken as a reference of charge-shift correction for the measured spectra.For deconvolution of the V 4 + /V 5 + species, only the V 2p 3/2 region is used as the V 2p 1/2 overlaps with an O1s satellite (see Figure SI12). [41]The two peaks located at 518,2 eV and 516,8 eV, correspond to V 5 + and V 4 + , respectively. [36,43]No further oxidation states of vanadium species were detected.The ratio R XPS between both oxidation sates was calculated, as shown in eq. 7.
The spectra for Nb 3d region depict two peaks correspond to 3d 5/2 and 3d 3/2 core level of Nb 5 + in all samples.Figure SI12 shows exemplary spectra for the Nb 3d and the P 2p region.The two peaks located at 133,5 eV and 134,4 eV, correspond to 2p 3/2 and 2p 1/2 core level of P 5 + , respectively.Thus, P is present as phosphates in all samples.The O 1s region shows a main peak at 530,4 eV and a smaller one at 532,6 eV; corresponding to the O 1s peak of the metal oxide and possible organic contamination and/or residual products, respectively.
Ammonia temperature programmed desorption.For the NH 3 TPD experiments an AMI-300 setup was used, equipped with thermal conductivity detector (TCD), from Altamira Instrument.Selected samples (0.05 g) were inserted into a quartz U-tube fixed with quartz wool.After a drying step at T = 150 °C in He for 1 h, the catalyst is saturated with ammonia at T = 90 °C for 1 h.In order to check on the complete saturation, 20 gas pulses are dosed with a fixed amount of NH 3 in He (40 %vol) monitored by TCD.Afterwards the desorption proceeds in He flow until a temperature of T = 600 °C is reached.The quantification of desorbed ammonia is based on the TCD signal of the NH 3 -pulses after saturation of the sample (Pulses 15-20) and the exact catalyst mass.

Figure 1 .
Figure 1.Catalytic activity of the V-based oxidation catalysts V 2 O 5 , V 0.3 Nb 0.7 OPO 4 , and VPP: The n-butane conversion X 350 at a fixed temperature setpoint of T = 350 °C is shown together with the specific surface area S BET determined by N 2 -physisorption experiments; Additionally, the n-butane consumption rate normalized to the specific surface area is presented; see Figure SI3 for more details; 1 atm, C 4 H 10 /O 2 /H 2 O = 2/20/3 %vol, GHSV = 2000 h À 1 .

Figure 2 .
Figure 2. Consecutive combustion of MAN on the three different V-based oxidation catalysts V 2 O 5 , V 0.3 Nb 0.7 OPO 4 , and VPP: Selectivity towards MAN as a function of the n-butane conversion at a temperature setpoint of T = 375 °C; the complete product spectrum is shown in Figure SI4; 1 atm, C 4 H 10 /O 2 /H 2 O = 2/20/3 %vol, GHSV = 1000-8000 h À 1 .

Figure 3 .
Figure 3. Catalytic effect of the surface modification with PO x on V-based oxidation catalysts V 2 O 5 , V 0.3 Nb 0.7 OPO 4 , and VPP: Selectivity towards MAN of the bulk base catalyst and the PO x -modified sample as a function of the n-butane conversion at fixed temperature setpoints indicated within the diagram on a) V 2 O 5 , b) V 0.3 Nb 0.7 OPO 4 , c) VPP; see Figure SI7 for more details on the alkane conversion and Figure SI8 for the complete product spectrum; 1 atm, C 4 H 10 /O 2 /H 2 O = 2/20/3 %vol, GHSV = 2000 h À 1 .

Figure 4 .
Figure 4. Impact of surface modification with PO x on V 2 O 5 and V 0.3 Nb 0.7 OPO 4 : Yield towards MAN as a function of the n-butane conversion at fixed temperature setpoints indicated within the diagram while varying the contact time between 500 h À 1 and 4000 h À 1 on a) (PO x -)V 2 O 5 and b) (PO x -)V 0.3 Nb 0.7 OPO 4 ; for the change in the product spectrum see Figure SI9; 1 atm, C 4 H 10 /O 2 /H 2 O = 2/20/3 %vol.

Figure 5 .
Figure 5. MAN co-feed studies performed on PO x -modified V 2 O 5 compared to the unmodified base catalyst V 2 O 5 : (a) MAN concentration (inlet and outlet) as a function of the MAN cofeed demonstrates the consecutive combustion of the re-adsorbed product on V 2 O 5 (The change in outlet concentration of the reactant and other products is shown in Figure S10); (b) Products formed from the co-fed MAN at maximal concentration (~0.5 %vol) extracted from the change in the product concentrations by introducing the cofeed (see Figures S10b-d); the dotted line represents the MAN inlet at this setpoint, while the bar plots shows the change in outlet concentrations of MAN, CO and CO 2 comparing this setpoint to the measurement without cofeed; the stepwise introduction of the cofeed leads to no change in the CO x formation on PO x -V 2 O 5 , while on the unmodified V 2 O 5 the cofed MAN is converted to additional CO and CO 2 ; 2000 h -1 , 1 atm, C 4 H 10 /O 2 /H 2 O = 2/20/ 3 %vol, T = 420 °C, C 4 H 2 O 3 dosed in 4 concentrations between 0 and 0.5 %vol.

Figure 6 .
Figure 6.Quantification of desorbed ammonia in the NH 3 -TPD study: (a) Comparison between the total amounts of NH 3 desorbed from the base catalysts and the PO x surface modified catalysts after catalysis (spent); The samples were dried at T = 150 °C under He flow for 1 h, NH 3 flow for 1 h were performed until saturation of the samples at T = 90 °C; 20 NH 3 -pulses are monitored by the TCD to ensure saturation before the actual TPD measurement begins with a maximum temperature of T = 600 °C.

Figure 7 .
Figure 7. Variation of the feed composition towards alkane-rich feed conditions: MAN formation rate as a function of the input concentration of n-butane in the feed at two different oxygen concentrations of c oxygen = 20 % vol and 5 %vol on V 0.3 Nb 0.7 OPO 4 , PO x -V 0.3 Nb 0.7 OPO 4 and VPP at temperature setpoints of T = 325 °C and 350 °C; the change in conversion and the MAN selectivity is shown in Table SI5 and Figure SI16, the n-butane consumption rate is shown in Figure SI15; 1 atm, H 2 O = 3 %vol, GHSV = 2000 h -1 .

Figure 8 .
Figure 8. n-Butane consumption rate and MAN formation rate as a function of an elevated n-butane partial pressures above the upper explosion limitation between 10 %vol and 20 %vol on (a) PO x -V 0.3 Nb 0.7 OPO 4 and (b) VPP at a fixed temperature setpoints of T = 325 °C and 350 °C; the changes in the n-butane conversion and selectivity towards MAN under these conditions are shown in TableSI6and the performance under alkane-rich feed conditions at different temperatures can be found in FigureSI17; flow rates are kept constant at a GHSV of 1000 h À 1 , 1 atm, O 2 = 20 %vol, H 2 O = 3 %vol.