Gas phase dehydration of lactic acid to acrylic acid over alkaline-earth phosphates catalysts

Catalysis Today 226 (2014) 185–191

Contents lists available at ScienceDirect

Catalysis Today journal homepage: www.elsevier.com/locate/cattod

Gas phase dehydration of lactic acid to acrylic acid over alkaline-earth phosphates catalysts E. Blanco, P. Delichere, J.M.M. Millet, S. Loridant ∗ Institut de Recherches sur la Catalyse et l’Environnement de Lyon, IRCELYON, CNRS-Université Claude Bernard Lyon 1, 2 Avenue Einstein, F-69626 Villeurbanne Cedex, France

a r t i c l e

i n f o

Article history: Received 15 June 2013 Received in revised form 3 September 2013 Accepted 28 September 2013 Available online 27 October 2013 Keywords: Lactic acid Acrylic acid Gas phase dehydration Alkaline earth phosphates Acid–base balance

a b s t r a c t A series of alkaline-earth phosphates were prepared by co-precipitation method using sodium free or sodium containing precursors and evaluated for gas phase dehydration of lactic acid. The catalysts were characterized by BET measurements, X-ray diffraction, chemical analysis, XPS spectroscopy and both NH3 and CO2 -TPD. After checking the stability of catalysts under feed, it was shown that selectivity to acrylic acid strongly depended on reaction temperature but not on contact time. At temperature of 380 ◦ C, values ranging from 19 to 49% were measured for the different prepared catalysts. The highest value was reached with Ba3 (PO4 )2 (55% for C3 products) but selectivities rather close were obtained with different other phosphates suggesting kinetic limitation. Acid–base properties measurements revealed that alkaline-earth phosphates exhibited high proportion of acidic and basic sites with same weak strength. Furthermore, correlation between acrylic selectivity of alkaline earth phosphates and the acid–base balance were clearly established for the first time: selectivity was 50% for balance close to 1 and decreased by factor two increasing this parameter to 2. Finally, FTIR spectra of spent catalysts showed alkaline-earth lactates adsorbed over the catalysts which could be reaction intermediates for dehydration of lactic acid. © 2013 Elsevier B.V. All rights reserved.

1. Introduction Acrylic acid (AA) is a platform molecule used as building block to produce polymers and plastics. It is produced at 4.5 Mt/year by selective oxidation of propene, a petrochemical resource whose price is growing up quickly because of increasing demand and rarefaction of petroleum.

∗ Corresponding author. Tel.: +33 0472 445 300; fax: +33 0472 445 399. E-mail address: [email protected] (S. Loridant). 0920-5861/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.cattod.2013.09.059

With two functional groups, lactic acid (LA) is suitable for numerous applications like food industry, biopolymers such as PLA [1]. LA can be yielded by microbial fermentation processes [2,3] and by glycerol dehydrogenation [4]. As shown in the following scheme, LA can undergo dehydration to achieve AA but also decarbonylation/decarboxylation, condensation, hydrogenation, self-esterification that lead respectively to acetaldehyde (A) and COx , 2,3-pentanedione, propionic acid (PA), dilactide (DL) or polylactide (PL) [5]:

Acrylic acid is rarely obtained selectively from LA because of easy decarbonylation leading to acetaldehyde and CO. In the literature, several catalysts consisting chiefly of sulfates and phosphates

186

E. Blanco et al. / Catalysis Today 226 (2014) 185–191

of the group I and II metals, aluminum phosphates treated with an inorganic base, supported phosphates salts have been patented [6–8]. Yields to acrylic acid reaching 68% were obtained using modified zeolithes [9,10] but they suffered from coking and probably hydrothermal instability. Finally, calcium phosphates catalysts were studied. Han et al. [11,12] reported high AA selectivity reaching 75% over composite calcium phosphate in dehydration of methyl lactate. Recently, Ghantani et al. focused on calcium hydroxyapatites for LA dehydration and claimed 60% selectivity to AA [13]. Different studies allowed gathering important information on the reaction mechanism and on the key parameters for high performances. From study of the reaction mechanism over alkali-metal salts, Miller et al. proved that alkali lactate was formed during the reaction and concluded that this formation was key to obtain selectively AA [14–20]. Intramolecular dehydration of LA was proposed to occur under near supercritical water [21,22]. Aida et al. proposed that water must be in interaction with LA to perform dehydration [23]. Finally, Yan et al. who studied the effect of acidic and basic sites on NaY-zeolites modified by alkaline-earth cations, proposed that medium acidic and basic sites contribute to enhance LA dehydration [24]. The motivation of this work was to develop an alternative green route consisting in dehydration of lactic acid (LA). With that purpose, alkaline-earth ortho, pyrophosphates and hydroxyapatites were prepared and evaluated for gas phase dehydration of LA. Reaction parameters such as temperature and contact time were optimized for one catalyst to compare the performances of all the prepared catalysts. NH3 - and CO2 -TPD measurements were achieved to search for correlations with the selectivities. In particular, the influence of residual sodium cations on both the acido-base and catalytic properties was investigated. 2. Experimental 2.1. Catalysts preparations Sodium pyrophosphate (Na4 P2 O7 ), sodium hydrogenophosdiammonium hydrogenophosphate phate (Na2 HPO4 ), ((NH4 )2 HPO4 ), calcium nitrate (Ca(NO3 )2 ·4H2 O), strontium nitrate (Sr(NO3 )2 ), barium nitrate (Ba(NO3 )2 ) were purchased from Aldrich and used as precursors for catalysts preparations. The purity of these salts was 99%. A series of metals orthophosphates (MOP) and metals pyrophosphates (MPP) were prepared by coprecipitation. Since residual Na was detected for some catalysts prepared using sodium containing precursors, they were noted MOP-Na or MPP-Na. In a typical preparation of MOP catalyst, a solution of (NH4 )2 HPO4 or Na2 HPO4 at 0.2 mol L−1 was prepared. The pH of the solution was then adjusted to 9 adding NH4 OH (32%vol) or NaOH (0.5 M). Another solution of metal nitrate at 0.3 mol L−1 was added drop wise to the first solution maintaining constant the pH and stirring for 1 h. To prepare MPP catalysts, Na4 P2 O7 or (NH4 )2 HPO4 were used as phosphorus precursor. In the first case, a solution of metal nitrate at 0.3 mol L−1 was added drop wise to a solution of Na4 P2 O7 at 0.15 mol L−1 and the mixture was stirred for 1 h. Alternatively, a solution of (NH4 )2 HPO4 at 0.5 mol L−1 was prepared and the pH was adjusted to 10 using NH4 OH (32%vol) before adding drop wise a solution of metal nitrate at 0.5 mol L−1 while maintaining constant the pH and stirring for 1 h. Finally, CaPP and SrPP were obtained by precipitating (NH4 )2 HPO4 (0.2 M) and M(NO3 )2 (0.3 M) in formamide for 48 h. The low solubility of precursors in this solvent is assumed to slow down precipitate formation. For all the preparations, the precipitates were filtrated off, washed with deionized water, dried at 100 ◦ C and calcined

for 6 h under air-flow at a temperature determined from TGA measurement. As sodium-free and sodium containing calcium and strontium hydroxyapatites were obtained using the MOP or MPP protocols, they were respectively labeled MPOH and MPOH-Na. 2.2. Catalysts characterization TGA/DTA measurements of dried precipitates and used catalysts were achieved under air flow up to 800 ◦ C with a SETARAM TGA12 apparatus to determine respectively the calcination temperature and to characterize the organic matter deposited during reaction. Elementary analyses were obtained from ICP-OES spectra (ACTIVA/Jobin Yvon) after acidic dissolutions. The BET Specific Surface Areas (SBET ) were measured by nitrogen physisorption at −196 ◦ C using a Micromeritics ASAP 2020 instrument and applying the BET method. The samples were previously outgassed under vacuum of 10−3 Pa for 3 h at 300 ◦ C. Powder X-ray diffraction patterns were achieved on a Bruker D8 Advance A25 diffractometer equipped with a Ni filter (Cu K␣ radiation: 0.154184 nm) and a one-dimensional multistrip detector (Lynxeye, 192 channels on 2.95◦ ). The International Center for Diffraction Data (ICDD) library was used for phase identification and BRUKER TOPAS P program for quantification of the phases identified. FTIR spectra of self-supporting disks of powder dispersed in KBr were recorded at room temperature with a Vector 22 (Bruker) spectrometer. X-ray photoelectron experiments were carried out in a Kratos Axis Ultra DLD spectrometer. The spectra of the P 2p, C 1s, O 1s, Na 1s, Sr 3d, Sr 3p, Ba 3d, Ca 2p levels were recorded using the Al K␣ X-ray radiation (1486.6 eV), with pass energy of 20 eV and spot size aperture of 300 × 700 ␮m. The binding energies were calibrated using the C 1s band at 284.6 eV. Acid and base properties have been determined by TPD of NH3 and CO2 respectively. The measurements were achieved with BELCAT-M apparatus (BEL JAPAN, INC.). 100–200 mg of catalysts were pretreated at the calcination temperature for 1 h under He flow (50 mL min−1 , NTP). After cooling down to 100 ◦ C, NH3 and CO2 were adsorbed by flowing the catalysts under 5%NH3 -He or 5%CO2 He gas mixture for 30 min (30 mL min−1 , NTP) followed by He treatment at 100 ◦ C for 15 min to remove physisorbed molecules. The catalysts were then heated under He flow (50 mL min−1 , NTP) up to the calcination temperature with heating rate of 8 ◦ C min−1 . The acid and basic surface densities were calculated integrating the peaks areas from 100 to 450 ◦ C, using pulse calibrations and surface areas values. 2.3. Catalytic testing Dehydration of lactic acid was conducted in a fixed bed reactor (inner diameter 14 mm) operating at atmospheric pressure. A 20 weight% aqueous solution of lactic acid was fed using 307 HPLC pump (Gilson), vaporized at 170 ◦ C with home-made system and diluted with N2 . The vaporization temperature was determined from liquid vapor equilibrium simulated by the ProSim plus software (ProSim S.A.). The feed composition was LA/H2 O/N2 : 3/66/31. Before addition of the vaporized LA solution, the reactor was heated at the reaction temperature under N2 . The contact time was varied from 0.1 to 4.1 s (GHSV from 36,000 to 880 h−1 ). Condensable molecules were collected in a cold trap at −4 ◦ C and analyzed offline with a GC-2014 chromatograph (Shimadzu) equipped with AOC-20i auto injector, ZB-WAXplus (30 m, 0.32 mm) column and FID detector, while gas products, mainly CO, CO2 and N2 were analyzed online with the same chromatograph but using sampling valve, Carboxen 1000 column and TCD detector. The formulas used to calculate the conversions, products selectivity sets and carbon

E. Blanco et al. / Catalysis Today 226 (2014) 185–191

balances were the followings: Conversion (%) =

number of moles of LA fed − number of moles of LA in cold trap number of moles of LA fed

× 100

Selectivity (%) = K × with K =

number of moles of product × 100 number of moles of LA reacted

number of carbons of product number of carbons of LA

Carbon balance (%) =



products selectivity (%)

The numbers of moles of CO and CO2 were determined using N2 as internal standard whereas the other ones directly using the response factors of molecules. For products identification, GC × GC chromatograms were recorded using an Agilent 6890 apparatus with a liquid nitrogen cryogenic jet modulation (Zoex Corporation) coupled with a 5975B quadrupole mass spectrometer. The first column was a non-polar ZB1 column (30 m × 0.25 mm × 0.25 ␮m) and the second one, a polar ZB50 column (2 m × 0.1 mm × 0.1 ␮m). The NIST-MS 2011 database was used for identification. Complementary GC–MS analysis was achieved using a CPG HP6890 equipped with HPINNOWAX PEG capillary column and a HP-5973 mass detector. 3. Results and discussion 3.1. Physico-chemical properties The XRD patterns of the prepared calcium, strontium and barium phosphates are provided in Figures S1, S2 and S3 respectively and Table 1 summarized the crystalline phases identified by from ICDD database. All the powders were single phase or contained a phase present overwhelmingly. The CaPP and CaPP-Na catalysts were pure ␥-Ca2 P2 O7 (ICDD 00-015-0197) and ␤-Ca2 P2 O7 (tetragonal, ICDD 00-033-0297) respectively. CaPOH corresponded to hexagonal hydroxyapatite (ICDD 00-009-0432). The SrPP pattern was attributed to ␤-Sr2 P2 O7 (tetragonal, ICDD 00-013-0194) while the one of SrPP-Na corresponded to ␣-Sr2 P2 O7 (orthorhombic, ICDD 01-075-1490). The SrOP pattern corresponded to ␣-Sr3 (PO4 )2 (rhombohedral, ICDD 00-024-1008). SrPOH and SrPOH-Na were both a single phase of Sr5 (PO4 )3 OH (hexagonal, ICDD 04-007-5357) but the latter was more crystalline. No significant shift of bands was observed comparing their diffractogramms. In the case of BaPP, the pattern corresponded to ␴-Ba2 P2 O7 (hexagonal, ICDD 04-011-1688) but another minor phase was detected which is still unknown. For BaPP-Na, ␣-Ba2 P2 O7 (orthorhombic, ICDD 01-079-8010) and NaBaPO4 (monoclinic, ICDD 00-053-0376) were present in a 94:6 proportion. Finally, BaOP and BaOP-Na contained a mixture of Ba3 (PO4 )2 (rhombohedral, ICDD 04-015-1778) and Ba2 P2 O7 (orthorhombic, ICDD 01-0798010) phases in respective proportions of 87:13 for BaOP and 94:6 for BaOP-Na. Table 1 also compares the specific surface areas that ranged from 2 to 20 m2 g−1 for all the MPP and MOP catalysts. These surface areas did not depend on the calcination temperature since for instance a value of 14.5 m2 g−1 was measured for SrOP calcined at 750 ◦ C whereas values of 2.3 and 3.2 were obtained for the two BaPP samples calcined at only 500 ◦ C. It is worth noting that the two catalysts owning values close to 20 m2 g−1 were SrPP and CaPP,

187

which were prepared using formamide confirming the interest to use this solvent to slow down precipitation process. Specific surface areas of 82 and 88 m2 g−1 were measured for SrPOH and CaPOH, which is common for hydroxyapatites [25]. However, the surface area of SrPOH-Na was two times lower, which showed the presence of larger particles for this compound and certainly explained its higher crystallinity. M/P atomic bulk and surface ratios calculated from ICP analysis and XPS spectra respectively are listed in Table 1. The bulk stoichiometry was equal to the theoretical one for SrPP and CaPP and lower for the others catalysts for which it revealed the presence of additional phosphorus rich amorphous phase. For Ca and Ba phosphates, the XPS surface ratios were lower than the bulk ones suggesting that the additional phase was present as surface layer. Such feature is common to numerous phosphates [26,27] and could be due to phosphoric acid layer over the crystalline phases. It was also reported that prolonged exposure of calcium phosphates to the X-ray source resulted in even lower Ca/P values [28]. For strontium phosphates, the XPS surface ratios were systematically higher than the bulk ones. The same trend was reported for strontium hydroxyapatites [29]. It might be due to the fact that strontium rich faces were preferentially exposed on the surface or to surface segregation of strontium oxide. NH3 and CO2 -TPD curves of calcium, strontium and barium phosphates are provided in Figures S4 and S5 and Fig. 1 respectively. Most of the prepared catalysts owned mainly weak acid and basic sites with desorption temperatures ranging from 100 to 250 ◦ C. Their temperatures of maximal desorption (Tmax ) only varied from 109 to 150 ◦ C for acid sites and from 115 to 154 ◦ C for basic sites. It showed that alkaline earth phosphates owned sites with close strength. One exception can be noticed for CaPOH for which the Tmax values were 190 ◦ C for acid sites and 200 ◦ C for basic sites. Furthermore, it contained significant proportion of moderated sites with desorption temperature ranging from 250 to 450 ◦ C. Acid and basic moderated sites were also evidenced near 350 ◦ C for BaOP (Fig. 1) but present in relative small proportion. Furthermore, the NH3 and CO2 -TPD profiles were similar revealing that the catalysts contained high proportion of acidic and basic sites with same weak strength. These sites could be acid–base pairs or amphoteric species. The deduced surface densities of acidic sites were rather close to those of basic sites so that acid to base atomic balances ranged from 0.9 to 2.1 (Table 1). The very low number of basic sites measured for SrPOH could arise from bulk carbonatation during treatment under CO2 leading to erroneous measurement. In all cases, the Na-containing catalysts were less acidic than the ones without Na. The highest decreases were measured for CaPPNa and BaPP-Na which contained the highest quantities of Na. Such decreases suggested that Na+ cations replace some protons at the surface leading to lower densities of acid sites. Finally, no relationship between the acid–base balances and the XPS M/P ratios was found. 3.2. Catalytic properties 3.2.1. Products identification and blank testings As previously reported for different phosphates [12–14], the main reaction products were acrylic acid (AA), acetaldehyde (A), propanoic acid (PA) and COx . In this study, only small amounts of 2,3-pentanedione were detected in the condensates. Since additional peaks were observed on GC chromatograms, complementary analyses by GC × GC–MS were conducted to identify them. Dilactide (DL), hydroxyacetone, acetic acid and 2-butanone were evidenced. Hydroxyacetone and acetic acid was detected as impurities present in the LA solution whereas 2-butanone could be formed by aldol condensation of acetaldehyde [30]. Furthermore, two peaks with fragmentations corresponding to those of LA were

188

E. Blanco et al. / Catalysis Today 226 (2014) 185–191

Table 1 Physico-chemical properties of the prepared catalysts. Catalyst

CaPP CaPP-Na CaPOH SrPP SrPP-Na SrOP SrPOH SrPOH-Na BaPP BaPP-Na BaOP BaOP-Na a

Tcal (◦ C)

XRD identified phases

␥-Ca2 P2 O7 ␤-Ca2 P2 O7 Ca5 (PO4 )3 OH ␤-Sr2 P2 O7 ␣-Sr2 P2 O7 ␣-Sr3 (PO4 )2 Sr5 (PO4 )3 OH Sr5 (PO4 )3 OH ␴-Ba2 P2 O7 99% ␣-Ba2 P2 O7 + 1% NaBaPO4 93% Ba3 (PO4 )2 + 7% ␣-Ba2 P2 O7 97% Ba3 (PO4 )2 + 3% ␣-Ba2 P2 O7

500 500 500 500 600 750 500 500 500 500 600 500

SBET (m2 g−1 )

19.3 9.5 88.0 19.5 6.8 14.5 82.1 38.3 3.2 2.3 8.9 9.5

M/P

Na amount (at%)

ICP

XPSa

1.01 0.86 1.43 1.00 0.95 1.35 1.37 1.39 0.98 0.93 1.39 1.36

– 0.70 – 1.16 1.16 1.90 – 1.79 0.85 0.58 1.28 1.08

ICP

XPS

1.70

4.5

0.32

2.0

0.48

0.4

1.99

3.7

0.03

0.2

Acidity (␮mol m−2 )

Basicity (␮mol m−2 )

Acid–base balance

2.20 0.96 3.04 3.32 2.80 4.95 2.91 2.77 2.50 0.82 1.44 1.27

1.16 0.68 1.42 1.80 1.35 3.09 1.94 0.20 1.41 0.89 1.14 0.76

1.9 1.4 2.1 1.8 2.1 1.6 1.5 14.1 1.8 0.9 1.3 1.7

Determined from the Sr 3d to P 2p areas ratio.

Fig. 1. NH3 and CO2 TPD curves of the BaPP-Na, BaPP, BaOP-Na and BaOP catalysts.

observed. Using low ionization energy (25 eV) for GC–MS analysis, the two peaks were attributed to LA and lactoyllactic acid (PL, linear dimer). It was observed that the quantities of LA and DL decreased with residence time in the trap to the benefit of that of PL. Consequently, analyses of condensates were achieved immediately after ice trapping. A blank testing was performed at the vaporization temperature (170 ◦ C) with total flow of 46 mL min−1 in order to check the gas feed composition at the reactor entrance. The results displayed in Fig. 2 showed that LA vaporization led to formation of 24% of DL whereas it was present in negligible amount in the starting solution. The quantity of DL formed was proportional to the liquid flow (result not shown), which means that the thermodynamic equilibrium between AL and DL was reached after vaporization. Furthermore, blank testings achieved at 300 and 360 ◦ C showed some thermal degradations of DL into acetaldehyde and COx since selectivity evolved while conversion remained constant at 24% (Fig. 2). It indicates that DL was less thermally stable than AL. This latter one started to react above 360 ◦ C as shown by the increase of conversion to 32% at 380 ◦ C. Adding any catalyst, the quantity of DL after the reactor was systematically quite small compared to the one before. Therefore, DL was considered as reactant to calculate the conversions. 3.2.2. Effect of the reaction temperature The effect of the reaction temperature on the catalytic properties of CaPP-Na was investigated between 300 and 390 ◦ C at a

contact time of 2.7 s. The temperature was randomly varied and the experiment was achieved quickly (less than 2 h) to minimize casual contribution of deactivation. The evolutions of conversion, selectivity and carbon balance regards the temperature are plotted in Fig. 3. As expected, the conversion strongly increased with the temperature reaching 98% at 390 ◦ C. However, the carbon balances were poor below 350 ◦ C whereas they approached 100% above.

Fig. 2. LA conversions and products selectivity measured during blank testings achieved from 170 to 380 ◦ C, at total flow rate of 46 mL min−1 .

E. Blanco et al. / Catalysis Today 226 (2014) 185–191

189

Fig. 5. Evolution of LA conversion, products selectivities with the time on stream for the CaPP-Na catalyst at 350 ◦ C and contact time of 2.7 s. Fig. 3. Evolution of LA conversion, products selectivities and carbon balance with temperature at contact time of 2.7 s.

Furthermore, the AA selectivity increased from only 5% at 300 ◦ C to a maximum of 42% around 370 ◦ C and decreased above. The A selectivity raised more slowly from 12% at 300 ◦ C until 29% at 370 ◦ C and then remained constant. PA selectivity increased slowly between 300 ◦ C and 350 ◦ C from 5 to 13% and remained constant between 350 and 370 ◦ C to finally reach a maximum of 18% at 390 ◦ C. The evolutions of carbon balance and selectivity suggested that desorption of products was quite limited below 350 ◦ C. At the opposite, the AA selectivity dropped above 370 ◦ C to the benefit of those of PA and COx . The increase in PA selectivity was ascribed to consecutive hydrogenation of AA [17]. As a consequence of these observations, a temperature of 380 ◦ C was chosen to investigate the effect of the contact time. 3.2.3. Effect of the contact time The influence of the contact time on the catalytic performances of CaPP-Na catalyst at 380 ◦ C is shown in Fig. 4. While the conversion increased logarithmically, all the selectivity sets remained the same in the investigated range. It can be noticed that a different kinetic behavior was observed by Wadley et al. [17] for silicasupported sodium nitrate at 370 ◦ C since increase in the contact time from 0.5 to 4 s led to much lower AA yield as result of consecutive reaction leading to PA formation. In the present study, a slightly marked optimum of the AA selectivity was observed at 2.1 s and hence this time was chosen to compare most of the different

Fig. 4. Evolution of LA conversion, products selectivity with contact time at 380 ◦ C.

prepared catalysts. Additionally, the AA selectivity was 45% at 380 ◦ C and contact time of 2.1 s instead of 42% at 370 ◦ C and contact time of 2.7 s. Therefore, 380 ◦ C was selected as the optimal temperature to compare the catalysts. 3.2.4. Stability versus time on stream The evolution of LA conversion and AA selectivity measured at 350 ◦ C and contact time of 2.7 s are plotted in Fig. 5. In spite of small decrease during the first hours, the conversion was stable and no significant change of selectivity was observed for 22 h. The TGA curve of CaPP-Na achieved after reaction showed a small weight loss below 200 ◦ C due to physisorbed or weakly chimisorbed water (Figure S6). A second weight loss of ca 5% was observed from 200 to 450 ◦ C and corresponded to combustion of organic matter as revealed by the associated exothermic peaks at 307 and 331 ◦ C. Fig. 6 compares the IR spectra of CaPP and CaPP-Na before and after reaction. The bands observed for both catalysts in the ranges 400–500, 500–650, 850–1050, 1050–1250 cm−1 and at 726 cm−1 were assigned to ıs (PO3 ), ıas (PO3 ) bending, s (PO3 ), as (PO3 ) stretching and s (P O P) stretching vibrations respectively [31–34] and the ones at 1650 and 3400 cm−1 to bending and stretching vibrations of water molecules. Furthermore, two bands were observed at 1605 and 1740 cm−1 only for the spent catalysts. They were attributed to (C O) stretching vibrations of calcium lactate and lactic acid respectively [13,35]. Formation of lactates has also been reported for alkali containing catalysts after dehydration of lactic acid [16–18]. In this case, the (C O) stretching vibration was located near 1590 cm−1 . Interestingly, no band was observed at this wavenumber for CaPP-Na after reaction indicating that if sodium lactate was formed its relative quantity was small. 3.2.5. Catalytic performances The performances of the prepared catalysts are compared at 380 ◦ C in Table 2. For most of them, the carbon mass balance ranged from 95 to 101% ensuring the good quality of results. The lower balances for CaPP-Na, SrPP and SrPP-Na (89–92%) can be explained by small oligomerization of lactic acid and/or products, coke deposition in addition of uncertainty in GC analyses [14]. The three highest conversions were measured for the hydroxyapatites having the highest SBET values. However, very high conversions were also obtained for BaOP and BaOP-Na which exhibited surface areas of only ca 9 m2 g−1 . In the same mind, a higher conversion was measured for SrPP-Na compared to SrPP in spite of a lower SBET value. This underlined the difference of intrinsic catalytic activity of the phases present in the catalysts.

190

E. Blanco et al. / Catalysis Today 226 (2014) 185–191

Table 2 Comparison of performances of the prepared catalysts at 380 ◦ C. Catalyst

GHSV (h−1 )

Conversion (%)

Product selectivity (%) A

PA

AA

COx

Othersa

CaPP CaPP-Na CaPOH

5000 4500 2650

59 42 100

44 26 45

5 6 3

25 39 25

22 17 23

– 2 –

SrPP SrPP-Na SrOP SrPOH SrPOH-Na

1750

68 71 86 97 100

29 31 30 40 37

4 10 3 3 3

41 26 44 37 39

14 16 16 20 18

4 6 3 1 1

BaPP BaPP-Na BaOP BaOP-Na

1750

68 58 93 95

34 26 27 34

6 31 6 7

40 19 49 37

17 16 16 17

3 3 0 3

a

2,3-Pentanedione, acetic acid, hydroxyacetone, DL and 2-butanone.

A maximal AA selectivity of 49% was obtained for BaOP (55% for C3 products i.e. AA and PA). However, the AA selectivities measured for the three following catalysts (SrOP, SrPP, BaPP) were rather close suggesting kinetic limitation. One can noticed that the AA selectivity was only 25% for CaPOH, value, which was much lower than the one reported in the literature (60%) [13] and this in spite of close reaction temperature (380 instead of 375 ◦ C). This difference cannot be explained by the stoichiometry deviation of the CaPOH catalyst since it was shown that stoichiometry has very limited effect on AA selectivity for values varying from 1.3 to 1.9 [13]. AA selectivity appeared to depend neither on the nature of the cation nor on the type of phosphates (ortho vs pyro). Furthermore,

Fig. 6. Comparison of the IR spectra of the fresh, spent CaPP and CaPP-Na catalysts.

no correlation with the surface density of acid or basic sites was found. However, AA selectivity was the highest for balance close to 1.3 and decreased by factor two increasing this parameter to 2.1 (Fig. 7). To the best of our knowledge, it was the first time that such relation between AA selectivity and acid–base balance was clearly established. One can also notice that BaOP exhibited moderated acidic and basic sites of the same strength (Fig. 1) that could also favor high AA selectivity. Finally, no clear relation was established between A selectivity and the acid–base balance. Since sodium phosphate based catalysts were reported to be efficient for gas phase LA dehydration [14–18,36], the catalytic performances of sodium containing and sodium free alkaline earth phosphates have been compared. No direct effect of sodium on AA selectivity was evidenced since for instance it was high for CaPP-Na contrarily to SrPP-Na. Again, this result could not be explained by the strength of acid and basic sites since weaker sites were present for the two sodium containing catalysts (Figures S2 and S3). However, the acid–base balance was strongly decreased for CaPP-Na whereas it increased for SrPP-Na. These opposite effects could be attributed to different sites occupied by sodium cations leading to different modifications of acid–base properties. It confirmed the key role of the acid–base balance on AA selectivity. In that regard, Bautista et al. have proposed that dehydration reaction could occur through two E2 concerted mechanisms involving acid and basic sites [37]. In the case of BaPP-Na, high PA selectivity (31%) was obtained leading also to high C3 selectivity (50%). One particularity of this catalyst was to own acid–base balance lower than 1 which would favor hydrogenation of lactic acid or acrylic acid on basic sites.

Fig. 7. Correlations between the acrylic acid selectivity at 380 ◦ C and contact time of 2.1 s and acid–base balance determined by TPD measurements.

E. Blanco et al. / Catalysis Today 226 (2014) 185–191

According to Tam et al., decomposition of sodium lactate results in formation of PA and acetic acid above 350 ◦ C [18]. Since in this case the temperature of reaction was 380 ◦ C, all the Na-containing catalysts should exhibit higher PA selectivity. However, this behavior was only observed in the case of BaPP-Na and to a lesser extent for SrPP-Na. The C O stretching band observed at 1600 cm−1 by FTIR after reaction did not allow to conclude on the presence of sodium lactate. 4. Conclusion In this work, different alkaline earth phosphates were prepared by co-precipitation and evaluated for gas phase dehydration of lactic acid. Blank testings revealed formation of dilactide during vaporization that reacts over the catalysts. Then, different reaction parameters were varied. This study achieved with CaPP-Na revealed that (i) the reaction temperature had strong influence on AA selectivity and carbon balance and (ii) the selectivity depended weakly on the contact time and (iii) the catalyst was stable under feed. In optimized conditions, the highest conversions were measured for hydroxyapatites, which exhibited the highest surface areas whereas the best AA selectivities (40–50%) were obtained for ortho and pyrophosphates. The NH3 and CO2 TPD mirroring profiles revealed that alkaline earth phosphates contained high proportion of acidic and basic sites with same weak strength. The AA selectivity was the highest for acid–base balance close to 1 and decreased significantly increasing this parameter. Such correlation should allow sustaining the research of new efficient catalysts for dehydration of lactic acid. The presence of residual sodium cations, which is frequently encountered depending upon the used synthesis protocol for the phosphates based catalysts, was shown to decrease the acidity. Finally, alkaline earth lactates were shown to form during reaction and could correspond to intermediate species in the reaction pathway leading to AA. Acknowledgments This work was founded by the French agency, Agence Nationale de la Recherche, Program Chimie Durable – Industries – Innovation (CD2I), project GALAC (reference ANR-2010-CD2I-011-01). Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/ j.cattod.2013.09.059.

191

References [1] R. Datta, S.P. Tsai, P. Bonsignor, S. Moon, J. Frank, FEMS Microbiol. Rev. 16 (1995) 221–231. [2] M.A. Abdel-Rahman, Y. Tashiro, K. Sonomoto, Biotech. Adv. (2013), http://dx.doi.org/10.1016/j.biotechadv.2013.04.002. [3] M.A. Abdel-Rahman, Y. Tashiro, K. Sonomoto, J. Biotech. 156 (2011) 286–301. [4] F. Auneau, L. Sadr Arani, M. Besson, L. Djakovitch, C. Michel, F. Delbecq, P. Sautet, C. Pinel, Top. Catal. 55 (2012) 474. [5] Y. Fan, C. Zhou, X. Zhu, Catal. Rev. 51 (2009) 293–324. [6] R.E. Holmen, US 2,859,240, 1958. [7] C. Paparizos, S.R. Dolhyj, W.G. Shaw, US 4,786,756, 1988. [8] R.A. Sawicki, US 4,729,978, 1988. [9] P. Sun, D. Yu, Z. Tang, H. Li, H. Huang, Ind. Eng. Chem. Res. 49 (2010) 9082. [10] J. Zhang, Y. Zhao, M. Pan, X. Feng, W. Ji, C. Au, ASC Catal. 1 (2011) 32. [11] J.M. Lee, D.W. Hwang, Y.K. Hwang, S.B. Halligudi, J.S. Chang, Y.H. Han, Catal. Commun. 11 (2010) 1176–1180. [12] J.H. Hong, J.-M. Lee, H. Kim, Y. Kyu Hwang, J.-S. Chang, S.B. Halligudi, Y.-H. Han, Appl. Catal. A: Gen. 396 (2011) 194–200. [13] V.C. Ghantani, S.T. Lomate, M.K. Dongare, S.B. Umbarkar, Green Chem. 15 (2013) 1211–1217. [14] G.C. Gunter, D.J. Miller, J.E. Jackson, J. Catal. 148 (1994) 252–260. [15] G.C. Gunter, R.H. Langford, J.E. Jackson, D.J. Miller, Ind. Eng. Chem. Res. 34 (1995) 974–980. [16] G.C. Gunter, R. Cracium, M.S. Tam, J.E. Jackson, D.J. Miller, J. Catal. 164 (1996) 207–219. [17] D.C. Wadley, M.S. Tam, P.B. Kokitkar, J.E. Jackson, D.J. Miller, J. Catal. 165 (1997) 162–171. [18] M.S. Tam, G. Gunter, R. Craciun, D.J. Miller, J.E. Jackson, Ind. Eng. Chem. Res. 36 (1997) 3505–3512. [19] M. Tam, R. Craciun, D.J. Miller, J.E. Jackson, Ind. Eng. Chem. Res. 37 (1998) 2360–2366. [20] M. Tam, J.E. Jackson, D.J. Miller, Ind. Eng. Chem. Res. 38 (1999) 3873–3877. [21] W.S.L. Mok, M.J. Antal Jr., J. Org. Chem. 54 (1989) 4596–4602. [22] C.T. Lira, P.J. McCrackin, Ind. Eng. Chem. Res. 32 (1993) 2608–2613. [23] T.M. Aida, A. Ikarashi, Y. Saito, M. Watanabe, R.L. Smith Jr., K. Arai, J. Supercrit. Fluids 50 (2009) 257–264. [24] J. Yan, D. Yu, P. Sun, H. Huang, Chin. J. Catal. 32 (2011) 405–411. [25] C. Lamonier, J.-F. Lamonier, B. Aellach, A. Ezzamarty, J. Leglise, Catal. Today 164 (2011) 124–130. [26] I.-C. Marcu, I. Sandulescu, J.-M.M. Millet, Appl. Catal. A: Gen. 227 (2002) 309–320. [27] T. Tsuchida, J. Kuboa, T. Yoshioka, S. Sakuma, T. Takeguchi, W. Ueda, J. Catal. 259 (2008) 183–189. [28] C.C. Chusuei, D.W. Goodman, M.J. Van Stipdonk, D.R. Justes, E.A. Schweikert, Anal. Chem. 71 (1999) 149–153. [29] S. Ogo, A. Onda, Y. Iwasa, K. Hara, A. Fukuoka, K. Yanagisawa, J. Catal. 296 (2012) 24–30. [30] J.C. Serrano-Ruiz, J.A. Dumesic, ChemSusChem 2 (2009) 581–586. [31] M. Serghini Idrissi, L. Rghioui, R. Nejjar, L. Benarafa, M. Saidi Idrissi, A. Lorriaux, F. Wallart, Spectrochim. Acta Part A 60 (2004) 2043–2052. [32] V.P. Mahadevan Pillai, B.R. Thomas, V.U. Nayar, K.-H. Lii, Spectrochim. Acta Part A 55 (1999) 1809–1817. [33] N. Santha, V.U. Nayar, Spectrochim. Acta A 49 (1993) 47–52. [34] N. Khay, A. Ennaciri, J. Alloys Compd. 323/324 (2001) 800–805. [35] G. Cassanas, M. Morssli, E. Fabregue, L. Bardet, J. Raman Spectrosc. 22 (1991) 409. [36] J. Zhang, Y. Zhao, M. Pan, X. Feng, W. Ji, C.-T. Au, ACS Catal. 1 (2011) 32– 41. [37] F.M. Bautista, J.M. Campelo, A. Garcia, D. Luna, J.M. Marinas, Appl. Catal. A: Gen. 243 (2003) 93–107.