Vapor-Phase Deoxygenation of Lactic Acid to Biopropionic Acidover Dispersant-Enhanced Molybdenum Oxide CatalystXinli Li,†,∥ Jun Pang,‡,∥ Ju Zhang,‡ Chunyu Yin,† Weixin Zou,§ Congming Tang,*,† and Lin Dong§

†School of Chemistry and Chemical Engineering, Chongqing University of Technology, Chongqing 400054, People’s Republic ofChina‡Chemical Synthesis and Pollution Control Key Laboratory of Sichuan Province, China West Normal University, Nanchong,Sichuan 637002, People’s Republic of China§Jiangsu Key Laboratory of Vehicle Emissions Control, Center of Modern Analysis, Nanjing University, Nanjing 210093, People’sRepublic of China

*S Supporting Information

ABSTRACT: Propionic acid obtained from fermentation-derived lacticacid has been appreciated since propionic acid is mainly used as a foodpreservative, satisfying a natural food idea. Vapor-phase deoxygenation oflactic acid to biopropionic acid over dispersant-dispersed molybdenumoxides was investigated in this work. It was found that different dispersantsdisplayed different performances, and the N element in dispersants had apositive effect. MoO3 was soon reduced to MoO2 under the in situhydrogen atmosphere, and the latter played an important role in catalyticconversion of lactic acid to propionic acid. The discriminating experimentsrevealed that propionic acid was formed mainly through directdeoxygenation of lactic acid (main path) and not hydrogenation of acrylicacid as an intermediate (minor path). Furthermore, only in situ hydrogenwas efficient, and external hydrogen was hardly efficient during catalyticreaction. Under the base-free conditions, catalyst offered excellent activityand durability and efficiently reduced the acid-treatment section in product separation.

1. INTRODUCTION

Propionic acid is mainly used as a preservative for food andanimal feed, and an intermediate and solvent for chemicalsynthesis. As used for the former, it displays more excellentbacteriostatic effect in epiphyte and mildew at pH < 6 than thetypical preservative of benzoic acid. For that reason, it is widelyused as a food preservative for grain, bread, cake, cheese, andmeat, etc.1−4 For the latter, it is used in esterification of alcoholto synthesize its corresponding ester and also used in solventfor various reactions. At present, its production mainly relieson the oxidization of propanal obtained from ethylenehydroformylation,5−10 hydrocarboxylation of ethylene,11,12

carbonylation of ethanol,13 hydrogenation of acrylic acid,14,15

and hydration of acrylonitrile,16−18 and these processes forpropionic acid production are almost all based on petroleumfeedstock, thus attributed to nonrenewable resources. Thesearch for new processes for production of propionic acidbased on sustainable feedstock is extremely urgent. Lactic acid(LA) as sustainable feedstock has been produced on a largescale by corn fermentation. In addition, rich and inexpensivebiomass materials such as cellulose,19 wheat straw,20 sugars,21

and sorbitol22 have also been used to produce LA in recentpublications. More importantly, LA instead of petroleummaterial is used to produce propionic acid, which is known asbiopropionic acid and is appreciated popularly in use as a

preservative for food and animal feed since it can well meet theconcept of healthy food.In pioneering work, homogeneous precious metal complexes

based on Pt, Pd, and Ir were used to catalyze conversion of LAto propionic acid under hydrothermal conditions.23 Amongthese precious metal complexes, PtH(Pet3)3 offered the bestcatalytic performance (around 50% yield of propionic acid) at250 °C in water at pH = 2. Korstanje et al.4 subsequentlydeveloped non-noble molybdenum complexes to catalyze thedeoxygenation of LA into propionic acid, achieving 41% yieldof sodium propionate in the presence of alkali reagent(NaOH). Heterogeneous catalysis which has a merit of easyseparation between catalyst and product is becoming attractive.Earlier, Pt/Nb2O5 was used to catalyze conversion of LA topropionic acid, achieving 25% propionic acid yield at 350 °Cand 50 bar H2 pressure.

24 Very recently, Co powder was usedas a catalyst and Zn powder as reductant for deoxygenation ofLA to propionic acid under the liquid−solid reaction mode at250 °C.25 Under the optimal reaction condition, 58.8%propionic acid yield was achieved. Differentiating from

Received: September 26, 2018Revised: November 22, 2018Accepted: December 6, 2018Published: December 19, 2018

Article

pubs.acs.org/IECRCite This: Ind. Eng. Chem. Res. 2019, 58, 101−109

© 2018 American Chemical Society 101 DOI: 10.1021/acs.iecr.8b04713Ind. Eng. Chem. Res. 2019, 58, 101−109

Dow

nloa

ded

via

NA

NJI

NG

UN

IV o

n M

ay 2

2, 2

019

at 0

6:29

:52

(UT

C).

Se

e ht

tps:

//pub

s.ac

s.or

g/sh

arin

ggui

delin

es f

or o

ptio

ns o

n ho

w to

legi

timat

ely

shar

e pu

blis

hed

artic

les.

rhythmic reaction in the above work, we used continuous gas−solid reaction to check the Fe and its oxides on activity for LAconversion to propionic acid.26,27 By comparison, Fe3O4 wasfound to act as an active species and offered an acceptablepropionic acid yield of ∼50%.Although the precedent reports on LA conversion to

propionic acid have made considerable progress, the reactionmechanism is unclearly known and catalytic efficiency is stilllow. Ongoing efforts are still required in order to achieveindustrial application. In the present work, molybdenum oxideswere used as a catalyst in the fixed bed reactor fordeoxygenation of LA to propionic acid, and the reactionmechanism was further revealed by characterization of catalystand discriminating experiments.

2. EXPERIMENTAL SECTION2.1. Materials. L-(+)-Lactic acid (80 wt %, fermentation

type) is supplied by Henan Jindan Lactic Acid TechnologyCo., Ltd. and is used for the production of propionic acidwithout further purification. Acetaldehyde, acrylic acid,propionic acid, acetic acid, 2,3-pentanedione, and isopropanol,together with hydroquinone, are purchased from SinopharmChemical Reagent Co., Ltd. Acrylic acid, propionic acid, aceticacid, 2,3-pentanedione, and acetaldehyde are used for gaschromatograph reference materials, and isobutanol is utilizedas internal standard material. Molybdenum trioxide, ethanol,H2O2, and chemical dispersants such as PVP (poly-(vinylpyrrolidone)), Tween 80 (polysorbate 80), and P123(PEO−PPO−PEO) are purchased from Energy Chemical Co.Natural dispersants such as yolk and albumen are purchasedfrom a supermarket.2.2. Preparation of Catalysts. Dispersant-dispersed

MoO3 was prepared as followed. In a typical experiment,0.005 mol of molybdenum trioxide (MoO3) was fully dispersedin 15 mL of mixed solution of H2O2−CH3CH2OH (1:1 byvolume) under a stirring state for 30 min at room temperature.Next, the resultant solution was heated to 60 °C and turned toyellow color. Subsequently, the dispersant (0.18 g) was addedto the resultant solution, and stirring continued for 1 h. Then,the resultant solution was slowly evaporated to remove thesolvent, and the dispersant-dispersed MoO3 was obtained .2.3. Catalyst Characterization. Powder X-ray diffraction

measurement was conducted on a Dmax/Ultima IV diffrac-tometer operated at 40 kV and 20 mA with Cu Κα radiation.The FT-IR spectra of the catalysts were recorded in the rangeof 500∼4000 cm−1 on a Nicolet 6700 spectrometer. Redoxproperties of the samples were estimated by H2-TPR on aFinesorb-3010 Instrument. The sample (ca. 50−60 mg) waspurged with dry He (50 mL/min, purity > 99.999 vt%) at 250°C for 1.0 h, followed by reducing the furnace temperature to30 °C, and switching to a flow of 10.0 vt% H2/He to executeH2-TPR in the range of 30−650 °C at a rate of 10 °C/min. X-ray photoelectron spectroscopy (XPS) measurements wereperformed using a Thermo Scientific K-Alpha spectrometer,equipped with a monochromatic small-spot X-ray source and a180° double focusing hemispherical analyzer with a 128-channel detector. Spectra were obtained using an aluminumanode (Al Kα = 1486.6 eV) operating at 72 W and a spot sizeof 400 μm. Survey scans were measured at a constant passenergy of 200 eV and region scans at 50 eV. The backgroundpressure was 2 × 10−9 mbar, and during measurement 3 × 10−7

mbar argon because of the charge compensation dual beamsource. Data analysis was performed using CasaXPS software.

The binding energy was corrected for surface charging bytaking the C 1s peak of contaminant carbon as a reference at284.6 eV.

2.4. Catalyst Evaluation. The synthesis of propionic acidfrom lactic acid over the catalysts was carried out in an up−down tubular quartz reactor with a 4 mm inner diameteroperated at an atmospheric pressure. The catalyst (ca. 500 mg,20−40 meshes) was placed in the middle of the reactor, andquartz wool was placed in both ends. First, the catalyst waspretreated at the required reaction temperature (ca. 390 °C)for 1.0 h under N2 with high purity (0.1 MPa, 1.2 mL/min).The feedstock (20 wt % solution of LA) was then pumped intothe reactor (LA aqueous solution flow rate, 1.6 mL/h) anddriven through the catalyst bed by nitrogen. The contact timeof reactant over the catalyst is estimated according to eqs 1 and2.28The liquid products were condensed using an ice−waterbath and analyzed off-line using a SP-6890 gas chromatographwith a FFAP capillary column connected to a FID.Quantitative analysis of the products was carried out by theinternal standard method using isobutanol as the internalstandard material. GC-MS analyses of the samples wereperformed using an Agilent 5973N mass selective detectorattachment. The reaction tail gas was analyzed using GC with apacked column of TDX-01 connected to TCD detector. Theconversion of LA and the selectivity toward propionic acid orother byproducts were calculated according to eqs 3−5.

= WF

CT(1)

=Fn

tLA

(2)

CT, contact time; g·h/mol; W, catalyst mass (g); F, LA feedflow rate, mol/h; nLA, the moles of lactic acid passed over thecatalyst, mol; t, LA feed time, h.

=‐

×Xn n

nconversion ( )/% 1000 1

0 (3)

=−

×Sn

n nselectivity ( )/% 100p

0 1 (4)

= × ×Y X Syield ( )/% 100 (5)

where n0 is the molar quantity of LA fed into the reactor, n1 isthe molar quantity of LA in the effluent, and np is the molarquantity of lactic acid converted to propionic acid or otherbyproducts such as acetic acid, acetaldehyde, and acrylic acid.

3. RESULTS AND DISCUSSION3.1. Effect of Dispersants on Catalytic Performance.

According to early work,4 molybdenum species has beenevaluated in activity for deoxygenation of LA to propionic acid,and its complexes displayed far more activity than its oxides(ca. MoO3) in reactive distillation. But its oxides have lowerprice and more stability than its complexes. Improving itsactivity for deoxygenation of LA to propionic acid,undoubtedly, displays a potential prospect. In this work, wechose water, instead of organic solvent, to dilute LA and fixed-bed reactor to substitute for reactive distillation, and evaluatedthe activity of its oxides. It was noted that MoO3 displayed alow activity for selective deoxygenation of LA to propionic acid(propionic acid yield, 1%) in previous work although LAconversion (72%) was acceptable.4 However, its activity was

Industrial & Engineering Chemistry Research Article

DOI: 10.1021/acs.iecr.8b04713Ind. Eng. Chem. Res. 2019, 58, 101−109

102

largely enhanced by using the vapor-phase fixed-bed reactor inthe presence of water instead of organic solvent. For example,propionic acid yield was up to 43.4%, and LA conversion wasalso up to 97.1% (shown in Figure 1).

It is expected that high dispersion of molybdenum oxideprecursor can improve catalytic performance. Here, we chosefive typical dispersants to disperse MoO3 in H2O2−ethanol andevaluated their activity (also shown in Figure 1). For LAconversion, it decreased slightly with addition of any one of thedispersants by comparison with blank experiment withoutusing dispersant. But, almost all samples treated withdispersants offered higher propionic acid yield except forTween 80. Especially, the albumen-dispersed catalyst has giventhe best propionic acid yield of 60.8% up to now, exceeding theblank experiment by 17.4%. In order to illustrate the positiverole of dispersants, we evaluated the catalytic performance ofstarting material MoO3, and the result was shown inSupporting Information Table S1. It is evident that thecatalytic performance of starting material MoO3 is lower thanthat of the blank sample and far lower than the samples withdispersants. These results indicated dispersant-enhancedselective deoxygenation reaction of LA to propionic acid.

Further observation showed that different dispersantsdisplayed different performances and increased in the followingorder: Tween 80 < P123 < Yolk < PVP < Albumen. Bycomparison in elements constructed of dispersants, the latterthree dispersants with N element gave better results than theformer two dispersants, indicating the N element displayed apositive effect in dispersing MoO3 in H2O2−ethanol. Moreimportantly, bio-albumen is greener, safer, and moreecofriendly for preparing catalyst for production of ediblepropionic acid and propionates from bio-lactic acid.In order to further understand the difference of catalytic

performance of samples treated with different dispersants,XRD, FT-IR, XPS, and H2-TPR measurements were used tocharacterize these samples, and the results were shown inFigures 2−5. From Figure 2a, the blank sample displayeddifferent diffraction peaks from those of the starting material(MoO3), suggesting that the MoO3 structure changed in theH2O2−C2H5OH solution. Actually, MoO3 interacted withH2O2 to form H2MoO5 (MoO3·H2O2), agreeing with thestandard sample H2MoO5 (JCPDS No.41-0359). With furtheraddition of dispersants, the characteristic diffraction peaks alsowere different from those of the blank sample, suggesting thatsample structures continuously changed. By comparison withXRD standard cards, the characteristic diffraction peaksmatched well with MoO3·2H2O (JCPDS No.16-0497).These suggested that highly dispersed MoO3 with dispersantsinteracted with H2O, instead of H2O2, to form MoO3·2H2O,agreeing with standard sample MoO3·2H2O (JCPDS No.16-0497). The observation on XRD of samples treated with/without dispersants showed that dispersants can well disperseMoO3 in H2O2−C2H5OH solution. So it is reasonable thatcatalytic performance relates to the dispersity of MoO3 in LAdeoxygenation to propionic acid. From Figure 3a, thecharacteristic absorption bands almost remained consistent atabsorption bands of >900 cm−1, while the absorption bands ofsample Tween 80 and sample P123 were different from othersat absorption bands of 500−700 cm−1. Slight diversity of IRspectra reflected the influence of dispersant in MoO3 precursorpreparation. From Figure 3b, the sample treated with Tween80 displayed two adjacent absorption bands at around 1626and 1716 cm−1, being different from other samples. Thissuggested that organic species covered the surface of thecatalyst, resulting in lower activity.XRD patterns of the spent catalyst (Figure 2b) showed

identical diffraction peaks at 2θ = 26.1°, 37.1°, and 53.4°,matching well with the standard sample of MoO2 (JCPDS No.

Figure 1. Effect of dispersant on deoxygenation of LA to propionicacid: catalyst, 0.48−0.50 g, 0.38 mL; carrier gas N2, 1.2 mL/min; feedflow rate, 1.6 mL/h; LA feedstock, 20 wt %; reaction temperature, 390°C; TOS, 1−2 h (sample 1), 3−4 h (sample 2), 5−6 h (sample 3), 7−8 h (sample 4), and 9−10 h (sample 5). LA conversion is lactic acidconversion average value for five samples; PA yield is propionic acidyield average value for five samples.

Figure 2. XRD patterns of dispersant-treated MoO3 (a, before reaction; b, after reaction).

Industrial & Engineering Chemistry Research Article

DOI: 10.1021/acs.iecr.8b04713Ind. Eng. Chem. Res. 2019, 58, 101−109

103

32-0671). Another piece of evidence from XPS character-ization can also demonstrate that the spent catalysts belong toMoO2 phase. For example, Mo characteristic peaks occurred atbinding energies of 229.6 and 232.8 eV, respectively, indexedto Mo 3d5/2 and 3d3/2 in MoO2 (Figure 4b). These resultspotentially indicated MoO2 as an active species fordeoxygenation of LA to propionic acid. In order to furtherverify this view, we tested the activity of the spent catalyst, andthe results were shown in Figure S1. The spent catalyst almostoffered activity identical with that of fresh catalyst. It isreasonable from this experimental evidence that MoO2 isdemonstrated as an active species.

H2-TPR profiles of fresh samples (Figure 5) can also supportthese activity data. Blank sample and Tween 80-treated samplehardly give an evident reduction peak in the range of 370−475°C by comparison with other samples, suggesting that they aredifficultly reduced in 10 vt% H2−He mixture. The reducibilityof Mo species may relate to its dispersibility. In the catalystpreparation, starting MoO3 displayed a low dispersion in blanksample and Tween 80-treated sample. For that reason, theyoffered a poor activity since the low valence molybdenum isjust active species.

3.2. Reaction Temperature. According to the comparisonof dispersants for catalytic deoxygenation of LA to propionicacid, albumen has been found to be an excellent dispersant.Therefore, the effect of reaction temperature on catalyticperformances such as LA conversion and product selectivitywas investigated over the albumen-dispersed MoO3, and theresults were shown in Figure 6. In the wide range of 270−410°C, LA conversion drastically changed with an enhancement ofreaction temperature (Figure 6a). In low temperature, ca. 270°C, LA conversion is low, only 28.8%, suggesting that LA isstable during the contact time of 140 g·h/mol over thiscatalyst. As reaction temperature increased from 270 to 300°C, LA conversion jumped to 87.1%, and then it increasedslightly with further increase of reaction temperature.According to previous work,26 propionic acid formationdepended on reaction temperature, and enhancing temperaturefavored formation of propionic acid. In this work, we alsoobserved that propionic acid selectivity increased from 36.2%to 62.8% with an increase of reaction temperature from 270 to390 °C (Figure 6b). Furthermore, observing in the range of300−390 °C, LA conversion remained >87%, and propionic

Figure 3. FT-IR spectra of dispersant-treated MoO3 (a, before reaction; b, after reaction).

Figure 4. Mo 3d and O 1s binding energies of the used MoO3 treated with PVP and albumen.

Figure 5. H2-TPR profiles of samples treated with differentdispersants (before reaction).

Industrial & Engineering Chemistry Research Article

DOI: 10.1021/acs.iecr.8b04713Ind. Eng. Chem. Res. 2019, 58, 101−109

104

acid selectivity changed drastically from 37.1% to 62.8%,indicating that enhancing temperature can efficiently suppressthe parallel side reactions, and the deoxygenation of LA topropionic acid gradually dominated. As reaction temperatureincreased to 410 °C, propionic acid selectivity decreased to44.7%, but higher than that obtained over iron oxide at 400 °C(propionic acid selectivity, 19.5%).27 These results suggestedthat dispersant-dispersed molybdenum oxides have moreselective conversion of lactic acid to propionic acid thanprevious catalysts. As a result, 390 °C was chosen as an optimalreaction temperature considering LA conversion and propionicacid selectivity.3.3. LA Concentration. Over the albumen-dispersed

MoO3, effect of LA concentration on reaction performanceof LA deoxygenation to propionic acid was further evaluated,and the results were shown in Table 1. Contact time (W/F)

decreased with an increase of LA concentration, suggestingthat the residence time of LA molecules over the surface ofcatalyst shortened. From the data shown in Table 1, LAconversion decreased slightly, and always remained >95% withan increase of LA concentration from 10 to 40 wt %, indicatingcontact time was enough for LA conversion at any one of theseLA concentrations. Product distribution was substantiallyaffected by LA concentration. For example, propionic acidselectivity was low, only 35.6% at LA concentration of 10 wt %,

and it increased to 62.8% at LA concentration of 20 wt %.These results indicated that side reactions dominated at lowLA concentration, and LA deoxygenation to propionic aciddominated at high LA concentration. It is known that thehydrogen source for LA deoxygenation to propionic acid isoriginated from inner reaction systems such as LA decarbox-ylation, acetaldehyde hydration, and LA steam re-forma-tion.4,27 As LA concentration was low, hydrogen productionwas insufficient for LA deoxygenation to propionic acid,resulting in a low propionic acid selectivity. At enhancing LAconcentration, more hydrogen was produced that could beutilized in deoxygenation of LA to propionic acid andsuppressed other side reactions. As a result, propionic acidselectivity increased at high LA concentration. In addition,enhancing LA concentration can also increase propionic acidproductivity without increasing device devotion from the viewof industry.

3.4. Reaction Mechanism. In order to fully understandthe mechanism on LA converted to propionic acid, a series ofexperiments are designed and performed as follows (Table 2).

It mainly contains two paths for LA conversion to propionicacid (Figure 7). In path I, acrylic acid obtained throughdehydration of LA was believed to be an intermediate andfurther hydrogenated to propionic acid. It is easily acceptedthat according to previous investigations, the reaction on LAconverted to acrylic acid occurred at the present reactiontemperature of 390 °C.29−35 Next, the generated acrylic acidhydrogenated with hydrogen to form propionic acid. In orderto verify this path, we directly used acrylic acid as substrateinstead of LA and performed hydrogenation reaction of acrylicacid with external hydrogen at identical reaction conditions.

Figure 6. Reaction temperature vs catalytic performance (a, LA conversion; b, product selectivity): catalyst, 0.50 g, 0.38 mL; carrier gas N2, 1.2mL/min; feed flow rate, 1.6 mL/h; LA feedstock, 20 wt %; TOS, 1−2 h (sample 1), 3−4 h (sample 2), 5−6 h (sample 3), 7−8 h (sample 4), and9−10 h (sample 5). Lactic acid conversion is average value for five samples; product selectivity is average value for five samples. LA, lactic acid; PA,propionic acid; AC, acetic acid; AD, acetaldehyde; AA, acrylic acid.

Table 1. Effect of LA Concentration on Deoxygenation ofLA to Propionic Acid

selectivity (%)

LAconcn(wt %)

W/F(g·h/mol)

LAconversion

(%) PA AC AD AA others

10 281 98.3 35.6 12.5 32.5 7.4 1215.4 182 97.6 43.8 10.3 24.4 6.2 15.320 140 96.8 62.8 9.2 20.6 3.9 3.530 93 96.5 57.8 7.4 21.2 3.1 10.540 70 95.8 58.8 7.3 22.2 3.2 8.5

aCatalyst: 0.50 g, 0.38 mL; carrier gas N2, 1.2 mL/min; LA feed flowrate 1.6 mL/h, reaction temp. 390 °C; TOS, 1−2 h (sample 1), 3−4 h(sample 2), 5−6 h (sample 3), 7−8 h (sample 4), and 9−10 h(sample 5). bLA conversion: lactic acid conversion is average value forfive samples; product selectivity: product selectivity is average valuefor five samples. cLA, lactic acid; PA, propionic acid; AC, acetic acid;AD, acetaldehyde; AA, acrylic acid.

Table 2. Discriminating Experiments on Reaction Pathsa

entry feedstock/carrier gasPA yield(%)

1 20 wt % LA/N2 60.82 11.57 wt % acrylic acid/H2 14.03 14.45 wt % LA/H2 59.84 11.12 wt % acrylic acid + 8.88 wt % formic acid/N2 15.25 12.20 wt % LA + 7.8 wt % formic acid/N2 62.06 12.41 wt % LA+ 7.59 wt % acetaldehyde/N2 65.1

aConditions: catalyst, uncalcined MoO3, 0.2 g; feed flow rate, 1.6mL/h; carrier gas, 1.2 mL/min; reaction temperature, 390 °C; molarratio of hydrogen/LA (or acrylic acid) > 1.2:1.

Industrial & Engineering Chemistry Research Article

DOI: 10.1021/acs.iecr.8b04713Ind. Eng. Chem. Res. 2019, 58, 101−109

105

Propionic acid yield of 14.0% (entry 2) was obtained throughcalculation based on acrylic acid, far less than that obtainedfrom LA as substrate (∼60%, in entry 1). This result at leastdemonstrated that, with the adoption of external hydrogen asthe hydrogen source, hydrogenation reaction proceeded withdifficulty compared to the present catalyst. For that reason, wegive up providing external hydrogen to utilize in situ hydrogenfor hydrogenation of acrylic acid to propionic acid. Here,formic acid was chosen as the source of in situ hydrogen, andthe propionic acid yield (15.2%, entry 4) was enhanced slightlyin comparison with that from external hydrogen. Thus, theseresults suggested whether external hydrogen or in situhydrogen hydrogenated should be used as the present catalystfor conversion of acrylic acid to propionic acid.In path II, the direct hydrogenolysis of C−OH bond in LA

to propionic acid naturally became possible. It is known thatpropionic acid is obtained without providing external hydro-gen, indicating that the hydrogen for hydrogenolysis of C−OHbond in LA originated from LA steam re-form reaction.Evidently, the in situ hydrogen from LA steam re-form reactioncan be well utilized in LA hydrogenolysis. Furthermore,improving the in situ hydrogen formation rate has anopportunity to enhance hydrogenolysis of LA to generate

propionic acid. Here, we chose two precursors of formic acidand acetaldehyde to generate hydrogen more easily than LA.Indeed, propionic acid yields were enhanced, 1.2% and 4.3%,respectively (shown in entries 5 and 6). In addition, externalhydrogen was used as hydrogen source for hydrogenolysis ofLA to propionic acid. Unfortunately, the propionic acid yielddecreased 1% (shown in entry 3). These results indicated thatexternal hydrogen was unfavorable for direct hydrogenolysis ofLA and the in situ hydrogen was favorable for this reaction inthe presence of molybdenum oxide catalyst.Based on the discussion of reaction paths, the possible

reaction mechanism was proposed. Unlike other catalyticreactions with induction period, MoO3 was soon in situconverted to MoO2 in the hydrogen atmosphere from LAsteam re-form reaction, and efficiently catalyzed the LAconversion to propionic acid. In order to verify this view,XRD and FT-IR measurements were used to investigate thechange of catalyst with time on stream (Figure 8). Within 1 hon stream, the characteristic diffraction peaks indexed to MoO3disappeared, and new characteristic diffraction peaks indexedto MoO2 occurred (Figure 8a). With further increase ofreaction time, the catalyst always remained in the state ofMoO2. In other words, Mo existed as +4 valence. Similarly, theFT-IR fingerprint region in 500−750 cm−1 also showedobvious discrimination between before and after reaction(Figure 8b), according to XRD results. Lower valence has torelate to strong reduction atmosphere, while the reactionatmosphere was constructed with a great deal of water steamand a small amount of hydrogen lacks its correspondingreduction circumstance. Furthermore, H2-TPR was used tocharacterize the reducible ability of catalyst with time onstream (Figure 8c). Compared with fresh sample (beforereaction), the spent sample displayed a newly relative

Figure 7. Reaction paths for LA conversion to propionic acid.

Figure 8. Structure, property, and catalytic performance vs time on stream (a, XRD; b, FT-IR; c, H2-TPR; d, catalytic performance).

Industrial & Engineering Chemistry Research Article

DOI: 10.1021/acs.iecr.8b04713Ind. Eng. Chem. Res. 2019, 58, 101−109

106

temperature peak of hydrogen consumption centered at 325°C, and the high temperature peak of hydrogen consumptioncentered at 425 °C decreased in peak intensity, indicating anenhancement effect of reducibility. As such, the change ofcatalyst reducibility has been almost completed within 1 hreaction. Figure 8d showed the performance of catalyst withtime on stream. The induction period was hardly observed;instead, excellent initial activity and stability were observed.The observation of activity agreed well with the change ofcatalyst in structure and property during the catalytic reaction.Back to the reaction mechanism, a catalytic cycle was

proposed in Figure 9. As characterized with XRD, MoO3 was

reduced to MoO2 in the presence of in situ H2. The obtainedMoO2 deoxygenated the C−OH group (α-hydroxyl group) inLA, and it again come back to MoO3. As a result, LA wasconverted to propionic acid (product) due to its deoxygena-tion. Then MoO3 was again reduced to MoO2 by in situhydrogen and began a new catalytic cycle.3.5. Stability. In order to evaluate the prospect of catalyst

in industrial application, long-term stability of heterogeneouscatalyst has to be checked.36−38 In this work, the stability ofalbumen-dispersed MoO3 was investigated at 390 °C and theLA feed flow rate of 10 mL/h (corresponding contact time =22.5 g·h·mol−1), and the results were depicted in Figure 10. LA

conversion retained about 95% within 40 h on stream, and itdecreased slightly. For example, the reaction time lengthenedto 43 h on stream, and LA conversion was also up to 91%.Compared with iron oxide in previous reports,26,27 LAconversion in this work was somewhat better than that inthe previous reports. Similarly, propionic acid yield (53−61%)also retained an excellent stability with time on stream. Interms of PA yield, the result was by far better than previousreports.4,27 It is also noted that the LA feed flow rate of 10mL/h (corresponding contact time = 22.5 g·h·mol−1) used inthis work is far more than previous reports on LA conversioninto other chemicals such as acrylic acid and 2,3-pentanedione,indicating that deoxygenation reaction of LA is faster thanother conversions of LA, and rich active sites for deoxygena-tion of LA exist on the surface of the catalyst.29,30,33,39−41 Inorder to fully understand the stability of active sites of catalystat low LA conversion (∼30%), the catalyst was reduced from0.5 to 0.16 g and remained with the LA feed flow rate (shownin Figure S1). It was found that catalyst still offered anexcellent stability around 80 h on stream.

4. CONCLUSIONSBio-propionic acid was efficiently produced through LAdeoxygenation over molybdenum dioxide. It was found thatpropionic acid was formed through not hydrogenation of anacrylic acid intermediate derived from dehydration of LA butdirect deoxygenation of LA using molybdenum dioxide to formmolybdenum trioxide. The in situ hydrogen was efficientlyutilized to draw an oxygen atom from molybdenum trioxideand was left in H2O molecule during the catalyticdeoxygenation of lactic acid. As such, molybdenum trioxideagain came back to the state of molybdenum dioxide andbegan a new catalytic cycle. Encouragingly, at high LA feedflow rate of 10 mL·h−1, the catalyst offered a satisfactorystability within 45 h on stream.

■ ASSOCIATED CONTENT*S Supporting InformationThe Supporting Information is available free of charge on theACS Publications website at DOI: 10.1021/acs.iecr.8b04713.

Supplementary catalytic performance information(PDF)

■ AUTHOR INFORMATIONCorresponding Author*Tel.: +86 23 62563250. E-mail: tcmtang2001@163.com.ORCIDCongming Tang: 0000-0001-5199-8286Lin Dong: 0000-0002-8393-6669Author Contributions∥X.L. and J.P. contributed equally to this work.NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTSThis work was supported by the Initial Scientific ResearchFoundation of Chongqing University of Technology (GrantNos. 2017ZD47 and 2017ZD48), the Chongqing EducationCommission Youth Project (Grant No. KJQN201801109), theSichuan Province Science and Technology DepartmentApplication Foundation Project (Grant No. 2017JY0188),

Figure 9. Proposed reaction mechanism over the molybdenumcatalyst.

Figure 10. Stability of catalyst (LA, lactic acid; PA, propionic acid):catalyst, 0.38 mL; albumen-dispersed MoO3, 0.5 g; carrier gas N2, 1.2mL/min; reaction temperature, 390 °C; LA feedstock, 20 wt %; feedflow rate, 10 mL/h.

Industrial & Engineering Chemistry Research Article

DOI: 10.1021/acs.iecr.8b04713Ind. Eng. Chem. Res. 2019, 58, 101−109

107

and Opening Foundation of Jiangsu Key Laboratory of VehicleEmissions Control (Grant No. OVEC 032).

■ REFERENCES(1) Maier, E.; Kurz, K.; Jenny, M.; Schennach, H.; Ueberall, F.;Fuchs, D. Food preservatives sodium benzoate and propionic acid andcolorant curcumin suppress Th1-type immune response in vitro. FoodChem. Toxicol. 2010, 48, 1950−1956.(2) Coblentz, W. K.; Bertram, M. G. Effects of a propionic acid-based preservative on storage characteristics, nutritive value, andenergy content for alfalfa hays packaged in large round bales. J. DairySci. 2012, 95, 340−352.(3) Coblentz, W. K.; Coffey, K. P.; Young, A. N.; Bertram, M. G.Storage characteristics, nutritive value, energy content, and in vivodigestibility of moist, large rectangular bales of alfalfa-orchardgrass haytreated with a propionic acid-based preservative. J. Dairy Sci. 2013, 96,2521−2535.(4) Korstanje, T. J.; Kleijn, H.; Jastrzebski, J. T. B. H.; KleinGebbink, R. J. M. Biopropionic acid production via molybdenum-catalyzed deoxygenation of lactic acid. Green Chem. 2013, 15, 982−988.(5) Huang, L.; Xu, Y. D. Synergy of ruthenium and cobalt in SiO2-supported catalysts on ethylene hydroformylation. Appl. Catal., A2001, 205, 183−193.(6) Zapirtan, V. I.; Mojet, B. L.; van Ommen, J. G.; Spitzer, J.;Lefferts, L. Gas phase hydroformylation of ethylene using organo-metalic Rh-complexes as heterogeneous catalysts. Catal. Lett. 2005,101, 43−47.(7) Ha, H. N. T; Duc, D. T.; Dao, T. V.; Le, M. T.; Riisager, A.;Fehrmann, R. Characterization and parametrical study of Rh-TPPTSsupported ionic liquid phase (SILP) catalysts for ethylene hydro-formylation. Catal. Commun. 2012, 25, 136−141.(8) Diao, Y.; Li, J.; Wang, L.; Yang, P.; Yan, R.; Jiang, L.; Zhang, H.;Zhang, S. Ethylene hydroformylation in imidazolium-based ionicliquids catalyzed by rhodium-phosphine complexes. Catal. Today2013, 200, 54−62.(9) Navidi, N.; Thybaut, J. W.; Marin, G. B. Experimentalinvestigation of ethylene hydroformylation to propanal on Rh andCo based catalysts. Appl. Catal., A 2014, 469, 357−366.(10) Liu, J.; Yan, L.; Ding, Y.; Jiang, M.; Dong, W.; Song, X.; Liu, T.;Zhu, H. Promoting effect of Al on tethered ligand-modified Rh/SiO2catalysts for ethylene hydroformylation. Appl. Catal., A 2015, 492,127−132.(11) Zoeller, J. R.; Blakely, E. M.; Moncier, R. M.; Dickson, T. J.Molybdenum catalyzed carbonylation of ethylene to propionic acidand anhydride. Catal. Today 1997, 36, 227−241.(12) Chepaikin, E. G.; Bezruchenko, A. P.; Leshcheva, A. A.Catalytic mono- and polycarbonylation of ethylene into propionicacid derivatives and alternative polyketones. Kinet. Catal. 1999, 40,313−321.(13) Zhang, Q.; Wang, H.; Sun, G.; Huang, K.; Fang, W.; Yang, Y.Promoting effect of CeO2 addition to Ni/AC catalyst for vapor phasecarbonylation of ethanol to propionic acid. Catal. Commun. 2009, 10,1796−1799.(14) Zhu, S.-F.; Yu, Y.-B.; Li, S.; Wang, L.-X.; Zhou, Q.-L.Enantioselective Hydrogenation of a-Substituted Acrylic AcidsCatalyzed by Iridium Complexes with Chiral Spiro AminophosphineLigands. Angew. Chem., Int. Ed. 2012, 51, 8872−8875.(15) Li, Y.; Dong, K.; Wang, Z.; Ding, K. Rhodium(I)-CatalyzedEnantioselective Hydrogenation of Substituted Acrylic Acids withSterically Similar beta,beta-Diaryls. Angew. Chem., Int. Ed. 2013, 52,6748−6752.(16) Wang, Y.; Liu, H. F.; Toshima, N. Nanoscopic naked Cu/Pdpowder as air-resistant active catalyst for selective hydration ofacrylonitrile to acrylamide. J. Phys. Chem. 1996, 100, 19533−19537.(17) Li, J.; Chen, J.; Wang, Y.; Luo, G.; Yu, H. Hydration ofacrylonitrile to produce acrylamide using biocatalyst in a membranedispersion microreactor. Bioresour. Technol. 2014, 169, 416−420.

(18) Li, J.; Yang, L.; Ding, X.; Chen, J.; Wang, Y.; Luo, G.; Yu, H.Visual study of mass transfer characterization in the process ofbiological catalytic hydration of acrylonitrile using pendant dropmethod. RSC Adv. 2015, 5, 79164−79171.(19) Wang, Y. L.; Deng, W. P.; Wang, B. J.; Zhang, Q. H.; Wan, X.Y.; Tang, Z. C.; Wang, Y.; Zhu, C.; Cao, Z. X.; Wang, G. C.; Wan, H.L. Chemical synthesis of lactic acid from cellulose catalysed bylead(II) ions in water. Nat. Commun. 2013, 4, 2141.(20) Zhang, Y. M.; Chen, X. R.; Luo, J. Q.; Qi, B. K.; Wan, Y. H. Anefficient process for lactic acid production from wheat straw by anewly isolated Bacillus coagulans strain IPE22. Bioresour. Technol.2014, 158, 396−399.(21) Esposito, D.; Antonietti, M. Chemical conversion of sugars tolactic acid by alkaline hydrothermal processes. ChemSusChem 2013, 6,989−992.(22) Ramirez-Lopez, C. A.; Ochoa-Gomez, J. R.; Gil-Rio, S.; Gomez-Jimenez-Aberasturi, O.; Torrecilla-Soria, J. Chemicals from biomass:synthesis of lactic acid by alkaline hydrothermal conversion ofsorbitol. J. Chem. Technol. Biotechnol. 2011, 86, 867−874.(23) Odell, B.; Earlam, G.; Cole-Hamilton, D. J. Hydrothermalreactions of lactic acid catalysed by group VIII Metal complexes. J.Organomet. Chem. 1985, 290, 241−248.(24) Serrano-Ruiz, J. C.; Dumesic, J. A. Catalytic processing of lacticacid over Pt/Nb2O5. ChemSusChem 2009, 2, 581−586.(25) Huo, Z. B.; Xiao, J. F.; Ren, D. Z.; Jin, F. M.; Wang, T.; Yao, G.D. Chemoselective synthesis of propionic acid from biomass andlactic acid over a cobalt catalyst in aqueous media. Green Chem. 2017,19, 1308−1314.(26) Zhang, Y.; Li, X.; Sun, L.; Tang, C. Facile preparation of Fe3O4from Fe2O3 via reduction of mixed vapors containing lactic acid andwater: catalytic deoxygenation of lactic acid into propionic acid.Chemistry Select 2016, 1, 5002−5007.(27) Li, X. L.; Zhai, Z. J.; Tang, C. M.; Sun, L. W.; Zhang, Y.; Bai, W.Production of propionic acid via hydrodeoxygenation of lactic acidover FexOy catalysts. RSC Adv. 2016, 6, 62252−62262.(28) Sad, M. E.; Gonzalez Pena, L. F.; Padro, C. L.; Apesteguia, C.R. Selective synthesis of acetaldehyde from lactic acid on acid zeolites.Catal. Today 2018, 302, 203−209.(29) Zhang, L. L.; Theng, D. S.; Du, Y. H.; Xi, S. B.; Huang, L.; Gao,F.; Wang, C.; Chen, L. W.; Borgna, A. Selective conversion of lacticacid to acrylic acid over alkali and alkaline-earth metal co-modifiedNaY zeolites. Catal. Sci. Technol. 2017, 7, 6101−6111.(30) Yan, B.; Tao, L. Z.; Mahmood, A.; Liang, Y.; Xu, B. Q.Potassium-ion-exchanged zeolites for sustainable production of acrylicacid by gas-phase dehydration of lactic acid. ACS Catal. 2017, 7, 538−550.(31) Lyu, S.; Wang, T. F. Efficient production of acrylic acid bydehydration of lactic acid over BaSO4 with crystal defects. RSC Adv.2017, 7, 10278−10286.(32) Zhang, X. H.; Lin, L.; Zhang, T.; Liu, H. O.; Zhang, X. F.Catalytic dehydration of lactic acid to acrylic acid over modified ZSM-5 catalysts. Chem. Eng. J. 2016, 284, 934−941.(33) Lari, G. M.; Puertolas, B.; Frei, M. S.; Mondelli, C.; Perez-Ramirez, J. Hierarchical NaY zeolites for lactic acid dehydration toacrylic acid. ChemCatChem 2016, 8, 1507−1514.(34) Guo, Z.; Theng, D. S.; Tang, K. Y.; Zhang, L. L.; Huang, L.;Borgna, A.; Wang, C. Dehydration of lactic acid to acrylic acid overlanthanum phosphate catalysts: the role of Lewis acid sites. Phys.Chem. Chem. Phys. 2016, 18, 23746−23754.(35) Nafe, G.; Lopez-Martínez, M. A.; Dyballa, M.; Hunger, M.;Traa, Y.; Hirth, T.; Klemm, E. Deactivation behavior of alkali-metalzeolites in the dehydration of lactic acid to acrylic acid. J. Catal. 2015,329, 413−424.(36) Zhang, P.; Guo, Y.-J.; Chen, J.; Zhao, Y.-R.; Chang, J.; Junge,H.; Beller, M.; Li, Y. Streamlined hydrogen production from biomass.Nature Catalysis 2018, 1, 332−338.(37) Zhang, C.; Wang, T.; Ding, Y. J. Oxidative dehydrogenation oflactic acid to pyruvic acid over Pb-Pt bimetallic supported on carbonmaterials. Appl. Catal., A 2017, 533, 59−65.

Industrial & Engineering Chemistry Research Article

DOI: 10.1021/acs.iecr.8b04713Ind. Eng. Chem. Res. 2019, 58, 101−109

108

(38) Behrens, M.; Studt, F.; Kasatkin, I.; Kuhl, S.; Havecker, M.;Abild-Pedersen, F.; Zander, S.; Girgsdies, F.; Kurr, P.; Kniep, B. L.;Tovar, M.; Fischer, R. W.; Norskov, J. K.; Schlogl, R. The active site ofmethanol synthesis over Cu/ZnO/Al2O3 industrial catalysts. Science2012, 336, 893−897.(39) Li, X. L.; Zhang, Y.; Chen, Z.; Cao, P.; Zou, W. X.; Tang, C.M.; Dong, L.; Wang, Y. Sustainable production of 2,3-pentanedione:Catalytic performance of Ba2P2O7 doped with Cs for vapor-phasecondensation of lactic acid. Ind. Eng. Chem. Res. 2017, 56, 14437−14446.(40) Li, X. L.; Sun, L. W.; Zou, W. X.; Cao, P.; Chen, Z.; Tang, C.M.; Dong, L. Efficient conversion of bio-lactic acid to 2,3-pentanedione on cesium-doped hydroxyapatite catalysts with bal-anced acid-base sites. ChemCatChem 2017, 9, 4621−4627.(41) Sun, L. W.; Li, X. L.; Tang, C. M. Comparative study oncatalytic performance of the production of 2,3-pentanedione fromlactic acid condensation over SiO2-supported alkali metal nitrates.Acta Phys.-Chim. Sin. 2016, 32, 2327−2336.

Industrial & Engineering Chemistry Research Article

DOI: 10.1021/acs.iecr.8b04713Ind. Eng. Chem. Res. 2019, 58, 101−109

109