Pd supported on defective TiO2 polymorphic mixtures: Effect of metal-support interactions upon glycerol selective oxidation

Resul s in Enginee ing 16 (2022) 100737 A ailable online 31 Oc obe 2022 2590-1230/© 2022 The Au ho s. Published by Else ie B.V. This is an open access a icle unde he CC BY-NC-ND license (h p://c ea i ecommons.o g/licenses/by- nc-nd/4.0/). Pd suppo ed on de ec i e TiO 2 polymo phic mix u es: E ec o me al-suppo in e ac ions upon glyce ol selec i e oxida ion Ma ías Gas ´ on Rinaudo a , * , Ana Ma ía Bel ´ an b , Asunci´ on Fe n´ andez c , Luis Edua do Cadús a , Ma ia Roxana Mo ales a a Ins i u o de In es igaciones en Tecnología Química (INTEQUI-CONICET), Uni e sidad Nacional de San Luis (UNSL), Facul ad de Química Bioquímica y Fa macia, Almi an e B own 1455, Capi al, 5700, San Luis, A gen ina b Depa amen o de Ingenie ía y Ciencia de los Ma e iales y del T anspo e, Escuela Poli ´ ecnica Supe io , Uni e sidad de Se illa, Vi gen de ´ A ica 7, 41011, Se illa, Spain c Ins i u o de Ciencia de Ma e iales de Se illa, (CSIC-Uni . Se illa), A da. Am´ e ico Vespucio 49, 41092, Se illa, Spain ARTICLE INFO Keywo ds: Ti anium oxide polymo phs Glyce ol selec i e oxida ion Palladium-based ca alys Me al-suppo in e ac ion Mechanochemical ac i a ion ABSTRACT Palladium ca alys s suppo ed on de ec i e mixes o ana ase, TiO 2 (II) and u ile c ys alline phases, p e iously ob ained by high-ene gy ball milling, we e syn hesized and es ed o glyce ol selec i e oxida ion. A deep cha ac e iza ion o hese unusual ma e ials was ca ied ou o elucida e ca aly ic and physicochemical ea u es. Elec on densi y ans e om suppo o me al o ice e sa, depending on he polymo phs p esen , could no only al e palladium pa icle sizes and i s su ace oxida ion s a e bu also educibili y and oxygen mobili y o ca alys s. Fu he mo e, acid-base p ope ies achie ed also in luenced ca aly ic ac i i y unde mild condi ions o liquid-phase glyce ol oxida ion. A con e sion o 94% and a selec i i y o glyce ic and lac ic acids o 48% and 22% espec i ely we e ob ained o he Pd ca alys suppo ed on mechanochemically ac i a ed ana ase. The p esence o se e al polymo phs in a me al oxide suppo could he e o e bene i o handicap ca aly ic cycle o a pa icula eac ion. Me al-suppo in e ac ions play a key ole in he e ogenous ca alys s and hus he a ional design o suppo s comes on he scene. 1. In oduc ion Wi h he ac ual end o biomass-based uels, alo iza ion o by- p oduc s ob ained in hese p ocesses becomes an in e es ing al e na- i e o p omo e g een and sus ainable echnologies [1,2]. In case o biodiesel, di ec subs i u e o diesel, he inc easing demand obse ed in he las yea s is also accompanied by aw glyce ol as a su plus by-p oduc (ob ained in o de o 10 w %). This si ua ion leads o he need o glyce ol con e sion in o alue-added p oduc s, o inc ease ca - bon balance and imp o e he economy o biodiesel p oduc ion [3,4]. Fo una ely, he enewable na u e and high- unc ionali y o glyce ol has a ac ed he in e es o scien i ic communi y in i s ans o ma ion [3,5, 6]. Among di e en ou es, glyce ol selec i e oxida ion is an a ac i e al e na i e due o low-cos , a ailabili y and eco- iendly o igin o ai o oxygen used as co- eac an , gi ing expensi e oxygena ed p oduc s such as dihyd oxyace one, glyce ic and lac ic acids as p oduc s [4,7–9]. Glyce ic acid is cu en ly used as pla o m o p oduce amino acids such as se ine bu also o ea skinca e diso de s due o i s p o en biological ac i i y [10]. In case o lac ic acid, i is widely used in di e en a eas including ood, pha maceu ical and ex ile indus ies [2]. Howe e , i s mos signi ican applica ion is as p ecu so o he p oduc ion o poly- lac ic acid, which is conside ed as one o he mos p omising bioplas ics o subs i u e pe oleum-de i ed plas ics such as polys y ene, poly- p opylene, and polye hylene e eph hala e [11]. Liquid-phase glyce ol selec i e oxida ion unde basic condi ions and using suppo ed ca alys s includes a complex eac ion ne wo k (Scheme 1) ha in ol es se e al pa hways o ob ain co esponding aldehyde, ke one and ca boxylic acids [8,9]. As s a ed in he li e a u e, a high pH commonly achie ed by NaOH addi ion is needed o he ini ial dep o ona ion o glyce ol and u he dehyd ogena ion o glyce aldehyde and/o dihyd oxyace one, ollowed by oxida ion o acid p oduc s [3,8,12]. Among di e en s eps in bo h solu ion and o e ca alys ’s su ace, se e al au ho s ag eed ha oxygen is indi ec ly in ol ed du ing oxida ion, emo ing elec ons om he su ace and egene a ing hyd oxyl g oups o close ca aly ic cycle [8, 12,13]. Acco dingly, o achie e good con e sion and selec i i y, he selec ion o an ac i e phase and suppo gains g ea ele ance. In his ma e , mono and bime allic ca alys s based on Pd, P and Au suppo ed on di e en me al oxides and ca bon ma e ials ha e been ex ensi ely * Co esponding au ho . E-mail add ess: [email p o ec ed] (M.G. Rinaudo). Con en s lis s a ailable a ScienceDi ec Resul s in Enginee ing jou nal homepage: www.sciencedi ec .com/jou nal/ esul s-in-enginee ing h ps://doi.o g/10.1016/j. ineng.2022.100737 Recei ed 16 Augus 2022; Recei ed in e ised o m 15 Oc obe 2022; Accep ed 28 Oc obe 2022 Resul s in Enginee ing 16 (2022) 100737 2 used in liquid-phase oxida ion eac ions, whe e bo h dispe sion and pa icle size o me al pa icles a e conside ed o be he dominan ac o s o a good ac i i y and selec i i y [14,15]. Howe e , he ole o suppo s is no limi ed o ac as me al pa icles ca ie s [16–18]. Se e al imes, he in e ace be ween me al and suppo becomes a be e ac i e si e han he me al i sel and hus, an app op ia e selec ion o a ca alys suppo could educe he quan i y o expensi e me als needed (e.g. noble me als) [3,15,16,19–21]. In case o glyce ol oxida ion, me al ox- ides wi h edox p ope ies a e known o ha e highe mobili y o adso bed hyd oxyl and oxygen species a he in e ace be ween me al and suppo [8]. Mo eo e , me al-suppo in e ac ion is a well-known phenomenon a ibu ed o elec onic and geome ic e ec s ha induce cha ge ans e be ween me al pa icles and suppo [15,19,22]. These in e ac ions oge he wi h acid-base na u e o suppo migh no only enhance he adso p ion capaci y o ac i e si es bu also inhibi he sin- e ing o me al pa icles du ing oxida ion [3,17,23]. A sho coming e- mains in he li e a u e when i comes o unde s anding suppo ’s con ibu ion on ca aly ic pe o mance, e.g. when de ec i e me al oxides and di e en polymo phs a e p esen in a gi en suppo . A a ional design including he suppo is he e o e c ucial o ob ain sus ainable me al-suppo ed ca alys wi h high ac i i y, eusabili y and selec i i y o desi ed p oduc s [16,22,24,25]. Conside ing se e al me al oxides, TiO 2 is a ecognized suppo o me als which shows di e en p ope ies depending on he polymo phs p esen in i s s uc u e [16]. Since mos eac ions in he e ogeneous ca alysis ake place on he su ace o me al ac i e phase and/o bounda ies be ween me al and suppo , he p esence o a ious c ys- alline phases could impac on ca aly ic p ope ies [15,16,26]. To he bes o ou knowledge, ca alys s wi h modi ied TiO 2 suppo s o med by de ec i e s uc u es and unusual polymo phic mix u es ha e no been used o glyce ol selec i e oxida ion un il now. Hence, we p opose he s udy o palladium ca alys s suppo ed on mixes o ana ase, TiO 2 (II) and u ile phases ob ained by high-ene gy ball milling in a p e ious wo k [27] o e alua e he ole o hei di e en physicochemical p ope ies and me al-suppo in e ac ions in liquid-phase glyce ol selec i e oxida ion. 2. Expe imen al 2.1. Ca alys s p epa a ion We imp egna ion me hod was used o deposi palladium on h ee di e en TiO 2 suppo s ob ained in a p e ious wo k [27]. 2 mL o a 0.1 M aqueous solu ion o HCl (37.0 w %, Aned a) wi h app op ia e quan- i ies o PdCl 2 (99.9%, Sigma Ald ich) was added o each i ania suppo in o de o achie e palladium loadings o 0.25 w % on a me al basis. Reco e ed solids we e d ied in an o en unde acuum a 70 ◦C o e - nigh . Then, samples we e calcined in a mu le u nace unde ai a 500 ◦C o 4 h using a hea ing amp o 10 ◦C min −1 o elimina e he es s o me al p ecu so . Final Pd suppo ed ca alys s we e labeled Pd/Ti x Scheme 1. Reac ion ne wo k o glyce ol oxida ion. M.G. Rinaudo e al. Resul s in Enginee ing 16 (2022) 100737 3 whe e x co esponds o he milling ime o p e iously syn hesized suppo s. 2.2. Cha ac e iza ion 2.2.1. X- ay di ac ion (XRD) X- ay di ac ion was pe o med using a Rigaku Ul ima IV di ac- ome e ope a ed a 20 mA and 30 kV wi h a Cu K α (λ =0.15405 nm) adia ion lamp. S ep scan da a we e eco ded a a s ep wid h o 0.02◦ and a coun ing ime o 2 s. 2.2.2. Raman spec oscopy Raman spec oscopy was ca ied ou using an A + lase exci a ion (514 nm) a ying he sou ce powe and a esolu ion o 3 cm −1 (Model RAMAN-Renishaw - In ia Raman Mic oscopy). 2.2.3. Scanning elec on mic oscopy (SEM-EDX) Mo phology o he ca alys s was s udied wi h a scanning elec on mic oscope coupled o Elec on Dispe si e X- ay Spec oscopy sys em (LEO 1450 VP SEM). Samples we e deposi ed on aluminum sample holde s and spu e ed wi h gold be o e analysis. 2.2.4. T ansmission Elec on Mic oscopy (TEM) T ansmission Elec on Mic oscopy (TEM) s udies we e pe o med in a FEI TALOS F200S mic oscope ope a ing a an accele a ing ol age o 200 kV. Fo TEM analyses, samples we e dispe sed in e hanol by ul a- sound be o e d opping hem in a holey ca bon ilm on a coppe g id. Samples we e s udied by High-Angle Annula Da k-Field (HAADF) im- aging, Scanning T ansmission Elec on Mic oscopy (STEM) and Ene gy- dispe si e X- ay Spec oscopy (EDS). 2.2.5. Tex u al analyses Tex u al p ope ies we e e alua ed by B unaue -Emme -Telle (BET) me hod using ni ogen adso p ion a 77 K in a Mic ome i ics Gemini V equipmen . Ca alys s we e p e iously ou gassed o e nigh a 250 ◦C unde N 2 . 2.2.6. Induc i ely coupled plasma – op ical emission spec oscopy (ICP- OES) Pd con en was checked in an ICP- Va ian-Ashile equipmen a e diges ion o he ca alys s in a mix u e o HNO 3 , HCl and HF acids. 2.2.7. X- ay pho oelec on spec oscopy (XPS) XPS spec a we e eco ded wi h a SPECS Mul i echnique ins umen equipped wi h a dual X- ay sou ce o Mg/Al and a PHOIBOS 150 hemisphe ic analyze . A pass ene gy o 30 eV and an Al anode ope a ed a 200 W was used. The p essu e was kep unde 1 ×10 −9 mba du ing all measu emen s. Spec a we e ob ained o he C 1s, O 1s, Ti 2p and Pd 3 d egions. C 1s binding ene gy (284.6 eV) was used as in e nal e e - ence o co ec peak posi ions. XPSPeak 4.1 ee so wa e was used o da a p ocessing. 2.2.8. Tempe a u e p og ammed measu emen s H 2 -Tempe a u e P og ammed Reduc ion (H 2 -TPR) and O 2 -Tempe - a u e P og ammed Deso p ion (O 2 -TPD) we e ca ied ou in a home- made equipmen using a qua z ubula eac o and a TCD de ec o unde he same condi ions epo ed in a p e ious s udy [27]. 2.2.9. 2-P opanol decomposi ion es Acid-base p ope ies we e s udied h ough 2-p opanol decomposi- ion es in a ixed-bed s ainless-s eel ubula eac o . 50 mg o each ca alys and suppo , wi h pa icle sizes be ween 300 and 600 μ m, we e i s ly p e ea ed in A (150 mL min −1 ) o 30 min a 290 ◦C and a - mosphe ic p essu e. Reac ion es s we e pe o med in oducing 7.5 ol % isop opanol (99.5%, Sin o gan) o he eac o , using A as ca ie and he same ope a ional condi ions o 120 min. P oduc s we e analyzed in an online gas ch oma og aph (Buck Scien i ic 910) equipped wi h an FID de ec o and a silica capilla y GC column (SPB1). 2.3. Ca aly ic es s 2.3.1. Reac ion condi ions Glyce ol selec i e oxida ion was ca ied ou in a 50-mL s eel Pa ype eac o equipped wi h a p essu e gauge, o en and con olle s o empe a u e and agi a ion. 20 mL o 0.15 M glyce ol (99.7%, Biopack) aqueous solu ion, sodium hyd oxide (≥97.0%, Biopack) o each NaOH: glyce ol mola a io o 2 and 200 mg o ca alys (glyce ol:Pd heo e ical mola a io o 600) we e added in o he eac o . Mix u e was hea ed o 100 ◦C unde 1000 pm o agi a ion speed and 1 ba o pu e oxygen o 4 h. 2.3.2. P oduc s analyses P oduc s analyses we e pe o med in an HPLC equipped wi h an Aminex HPX-87H column (300 ×7.8 mm) and an UV diodes a ay as de ec o . A 0.005 M solu ion o H 2 SO 4 (98.0%, Aned a) was used as eluen a 30 ◦C and a low a e o 0.6 mL min −1 . Iden i ica ion and quan i ica ion o p oduc s was ca ied ou by compa ison wi h ex e nal calib a ion cu es using s anda d solu ions o each chemical: glyce ol (99.7%, Biopack); lac ic acid (Pha maceu ical Seconda y S anda d, Me ck), D-glyce ic acid sodium sal (≥95.0%, Me ck); oxalic acid (99.5%, Ucb); o mic acid (85.0%, The mo Scien i ic). An aliquo o each eac ion sample was i s ly il e ed and neu alized by adding a ixed quan i y o H 2 SO 4 solu ion be o e ch oma og aphic analysis. 3. Resul s and discussion Ma e ials p esen ed he ein consis o a complex ma ix ha includes palladium pa icles and di e en a angemen s o TiO 2 polymo phs wi h de ec i e s uc u es induced by ball-milling. An exhaus i e cha ac e - iza ion is he e o e equi ed o comp ehend i s p ope ies and how hey could impac on ca aly ic beha io . 3.1. XRD analyses I is known ha he na u e o suppo s can highly a ec he shape and dispe sion o ac i e species, imp o ing he s abili y o ca alys s du ing eac ion [16,28]. Hence, XRD analyses we e pe o med o e al- ua e he c ys alline s uc u e o suppo ed ca alys s. We obse ed (Fig. 1 A) ha samples we e o med by di e en mix u es o i ania poly- mo phs. Pd/Ti5 showed in ense e lec ions associa ed o ana ase phase (PDF 00-21-1272) bu also wo b oad and low-in ensi y e lec ions a angles o 27.54◦and 31.53◦a ibu ed o (110) u ile e lec ion (PDF 00-21-1276) and (111) high-p essu e TiO 2 (II) e lec ion (PDF 00-72-0021) espec i ely. These obse a ions con i med he p esence o an almos pu e ana ase phase in his ca alys , as expec ed o i s suppo [27]. On he o he hand, Pd/Ti45 exhibi ed an inc ease in bo h TiO 2 (II) and u ile e lec ions bu a dec ease in ana ase e lec ions. Pd/Ti120 also showed he e lec ion a ibu ed o TiO 2 (II) phase and he appea ance o se e al high-in ensi y u ile e lec ions. Ob ained esul s e ealed he di e en composi ion o i ania phases p esen in he h ee ca alys s. Fu he mo e, b oad signals obse ed we e expec ed due o la ice de- ec s and small pa icles p esen in milled suppo s [29]. Palladium con en o ca alys s (0.25 w %) could be conside ed ou o he de ec ion limi s o XRD equipmen [26]. Howe e , an incipien signal o (101) PdO e lec ion (PDF 00-06-0515) a 34.04◦was de ec ed in Pd/Ti5 ca alys . This beha io sugges ed he p esence o la ge palladium pa - icles deposi ed on his ca alys , also suppo ed by a u he TEM anal- ysis. In addi ion, oxidized species o palladium we e likely o be o med since all ca alys s we e calcined in ai a 500 ◦C [30]. The absence o his signal in Pd/Ti45 and Pd/Ti120 ca alys s could be associa ed o smalle pa icles inely dispe sed on he suppo s, a o ed by he g ea e numbe o de ec s and he combina ion o polymo phs p esen in hese suppo s M.G. Rinaudo e al. Resul s in Enginee ing 16 (2022) 100737 4 [25,27,30–32]. We also made a ough calcula ion o a e age c ys alli e sizes o he ca alys s by means o Sche e equa ion and he ela i e in ensi ies o he mos in ense e lec ion o each c ys alline phase, (101) ana ase, (110) u ile and (111) TiO 2 (II), as sugges ed by Dulian e al. [33]. We obse ed (Table 1) a educ ion end acco ding o suppo s’ milling ex en and he pa icle e inemen caused by mechanochemical p ocess [27,34]. These esul s we e also in line wi h he inc ease in S BET alues (Table 1) obse ed in a subsequen analysis. We ound smalle c ys alli e sizes o Pd/Ti5 and Pd/Ti45 compa ed o he alues ob ained o espec i e suppo s (66 and 49 nm espec i ely) [27]. This ac was associa ed o he combined e ec o acid medium du ing p epa a ion, palladium deposi ion and calcina ion a 500 ◦C which could modi y i ania s uc u es. Being ana ase he main phase p esen in bo h ca a- lys s, less s able pa icles o suppo may be deagglome a ed and e-c ys alized du ing syn hesis s eps. In case o Pd/Ti120, he c ys alli e size inc eased compa ed o he suppo (23 nm), showing mo e in ense e lec ions, which could be associa ed o he s abili y o u ile and c ys al g ow h du ing calcina ion [27]. We also es ima ed he ca alys s composi ion (Table 1) using he mos in ense B agg e lec ion o each i ania phase and he equa ions epo ed by Du a ci e al. [35,36]. Re- sul s showed he unusual mixes o i ania phases p esen in he suppo s [27] wi h g ea e quan i y o ana ase phase in case o Pd/Ti45 and mo e e iden ly in Pd/Ti120. These esul s could be associa ed o palladium addi ion and u he calcina ion, which is known o s abilize ana ase phase, as epo ed by o he au ho s [37]. Mo eo e , o ma ion o s able in e me allic s uc u es be ween Ti om ana ase and palladium could induce a e e se ans o ma ion om TiO 2 (II) and/o u ile o ana ase, inc easing he quan i y o he la e [37]. E en hough he mix u e o c ys alline s uc u es and impu i ies is a d awback when using mechanochemical and he mal s eps o syn hesis, he exis ence o g ain bounda ies be ween di e en phases o i ania could no only bene i cha ge mobili y bu also se e as ene ge ic si es o eac an s and/o p oduc s, imp o ing he ca aly ic cycle du ing e- ac ion [38]. 3.2. Raman, mo phological and ex u al analyses Raman spec oscopy (Fig. 1 B) was pe o med o sus ain he esul s om XRD. High-in ensi y bands obse ed a 142, 397, 515 and 638 cm −1 we e a ibu ed o ana ase phase while bands a 171, 315, 427 and 525 cm −1 we e associa ed o TiO 2 (II) phase [27,39,40]. Addi ionally, signals obse ed a 430, 445 and 608 cm −1 we e ela ed o u ile phase [41,42]. Once again, Raman esul s also con i med i ania phases p e- sen in he ca alys s, wi h no e idence o palladium species due o i s low concen a ion and/o high dispe sion. Fig. 1. XRD spec a (A) and Raman spec a (B) o ca alys s. Table 1 Weigh ac ion (%) o i ania polymo phs, a e age c ys alli e size and ex u al p ope ies o he ca alys s. Ca alys Ana ase w % TiO 2 (II) w % Ru ile w % A e age c ys alli e size (nm) S BET (m 2 g −1 ) Po e diame e (nm) a To al Po e Volume (cm 3 g −1 ) b Pd/Ti5 99.2 0.3 0.5 54.2 10.0 20.9 0.05 Pd/Ti45 48.4 43.3 8.3 44.1 15.3 26.0 0.10 Pd/Ti120 23.0 49.4 27.6 42.1 17.2 8.8 0.04 a BJH adso p ion b anch a e age po e diame e . b Quan i y o N 2 adso bed a ela i e p essu e o 0.98. M.G. Rinaudo e al. Resul s in Enginee ing 16 (2022) 100737 5 Tex u al and mo phological p ope ies a e also impo an o unde - s and ca alys s’ su ace a chi ec u e. F om ex u al analyses, we obse ed speci ic su ace a ea (S BET ) alues (Table 1) in acco dance wi h he milling ime o each i ania suppo . Conside ing he measu emen e o o ±1 m 2 g −1 , alues we e in he o de o he ones ob ained o he suppo s [27] e en a e palladium addi ion and calcina ion. These e- sul s migh indica e ha nei he he small quan i y o palladium no he syn hesis condi ions blocked and/o collapsed suppo s’ po es [15]. I is wo h highligh ing ha su ace a eas ob ained we e lowe han ypical alues o me als suppo ed on oxides (a leas 50 m 2 g −1 ). Howe e , hese alues we e adequa e enough o dispe se small quan i ies o palladium (0.25 w %), as obse ed by STEM ollowing analyses. Po e diame e s a ied in he ange 9–21 nm (Table 1) wi h sligh ly highe alues in case o Pd/Ti5 and Pd/Ti45 compa ed o he suppo s [27]. This could indica e ha po es’ mo phology changed a some ex en du ing ca alys s syn hesis, pa icula ly in he case o suppo s wi h g ea e quan i y o ana ase phase, and in line wi h p e ious XRD dis- cussion [43,44]. Addi ionally, o al po e olumes loca ed in he ange 0.04–0.1 cm 3 g −1 (Table 1) as expec ed o he low po osi y o i ania [26,40]. These alues we e also highe han espec i e ones o Pd/Ti5 Fig. 2. SEM mic og aphs (10 kx) and Pa icle Size Dis ibu ion (PSD) his og ams o ca alys s: Pd/Ti5 (A), Pd/Ti45 (B) and Pd/Ti120 (C). M.G. Rinaudo e al. Resul s in Enginee ing 16 (2022) 100737 6 and Pd/Ti45 suppo s, which could also be a ibu ed o he ca alys s p epa a ion [44]. In spi e o hese changes, S BET alues we e no a ec ed. I is impo an o no ice ha po e sizes ob ained we e highe enough o anspo big eac an molecules, such as glyce ol (a e age molecule size o 0.62 nm) and hence eac ion a e dec ease due o in- e nal di usion is imp obable o happen [3,45]. Fu he mo e, low po- osi ies obse ed may indica e ha S BET alues co esponded almos en i ely o he ex e nal su ace a ea [3], which could induce he p e - e en ial loca ion o palladium species on ex e nal su ace o suppo s, a o ing he access o eac an s o s a he mechanism o eac ion [3]. Mo phology o ca alys s was s udied by SEM. Pd/Ti5 sample (Fig. 2 A) showed pa icles wi h sphe ical shape and diame e s be ween 120 and 220 nm. Pd/Ti45 ca alys (Fig. 2 B) p esen ed g ape-like clus e s o med by a couple o pa icles, wi h a e age diame e s be ween 100 and 200 [46]. These agglome a ion migh be a consequence o elec o- s a ic e ec on ine pa icles (wi h inc eased su ace ene gy) as epo ed by o he au ho s [46,47]. Smalle sizes we e obse ed in case o Pd/Ti120 (Fig. 2C) wi hin a ange o 80–180 nm. Resul s we e in line wi h he ones obse ed o he suppo s, wi h a educ ion in pa icle sizes when inc easing milling ime [27]. Mo eo e , he smalle pa icle sizes co ela ed well wi h he inc ease in S BET alues (Table 1) om Pd/Ti5 o Pd/Ti120. 3.3. ICP-OES and STEM-EDS analyses Pd loadings, measu ed by ICP-OES, we e ound in he ange 0.2–0.36 w % (Table 2) compa ed o he heo e ical 0.25 w %. Di e - ences we e a ibu ed o expe imen al e o s du ing syn hesis and/o diges ion o he samples. STEM and EDS analyses we e ca ied ou o ha e an insigh in o he sizes o palladium species. I was di icul o disc imina e be ween Pd and/o PdO pa icles and he suppo s’ pa icles due o he small me al quan i y and/o i s high dispe sion (Fig. 3A–D), p ecluding he measu ing o an s a is ical a e age size o me al domains. Howe e , we could in e a he e ogeneous me al dis ibu ion since some Pd pa icles in he ange o 4–5 nm bu also a ound 20 nm we e ound in Pd/Ti5 ca alys . Mo eo e , we could sca cely obse e me al pa icles in he ange o 4–6 and 5–10 nm o Pd/Ti45 and Pd/Ti120 espec i ely [48]. These obse a ions sus ained p e ious XRD esul s, whe e PdO e lec- ion could only be seen in Pd/Ti5, and hus bigge me al pa icles could be expec ed in his ca alys . Fu he mo e, i has been ound ha se e al solids submi ed o mechanochemical ac i a ion exhibi ed an imp o ed so p ion capaci y o di alen hea y me als om solu ion [49,50]. This beha io could also be ex apola ed o he case o Pd 2+ ca ions ( om PdCl 2 p ecu so ) p esen in he imp egna ion solu ions used in his s udy o add Pd on i ania suppo s. The quan i y o de ec s ob ained in he suppo s by milling p ocess, and hei dis ibu ion, could he e o e ac as ac i e cen e s o he adso p ion o me al pa icles and inc ease he dispe sion o me allic domains on he su ace [28,49,50]. Along wi h his beha io , smalle a e age me al domains could also be expec ed when using low me al amoun s, as epo ed by o he au ho s [51]. Pa icula ly, he he e ogenous dis ibu ion o me al pa icles ound in Pd/Ti5 could be a ibu ed o he sligh mechanochemical ac i a ion o he suppo and he majo p esence o ana ase c ys al phase. Mo eo e , Debecke e al. [28] ha e demons a ed ha a suppo composed o a mix u e o TiO 2 c ys al s uc u es (ana ase and u ile) dic a es he mo phology and dispe sion o Ru species owing o la ice ma ching and enhanced s abili y (impeding sin e ing) upon annealing. Hence, di e en me al-suppo in e ac ions could be achie ed acco ding o he c ys alline phases p esen in he ca alys s, changing bo h dispe sion and oxida ion s a e o palladium pa icles [28]. Subsequen ly, an XPS s udy was pe o med o elucida e su ace chemis y o each ca alys . 3.4. XPS analyses A he Ti 2p le el, Pd/Ti5 p esen ed Ti 2p 3/2 and Ti 2p 1/2 double s (Fig. 4 A and Table 3) wi h binding ene gies (BE) alues close o he ones epo ed o Ti 4+ in ana ase (458.59 eV and 464.31 eV espec i ely) [52]. These signals we e shi ed o highe BE compa ed o he suppo s (458.37 and 464.04 eV espec i ely), indica ing ce ain deg ee o me al-suppo in e ac ion a e palladium addi ion [27]. Ti 2p 1/2 could also be decon olu ed in o a low-in ensi y signal a ibu ed o Ti 3+ . In case o Pd/Ti45 and Pd/Ti120, double s p esen ed BE wi h alues close o he ones epo ed o Ti 4+ in u ile (458.46 eV and 464.23 eV espec i ely) [52]. Fo bo h cases hese signals could be mainly assigned o Ti 4+ om TiO 2 (II) phase since he wo ca alys s con ained be ween 40 and 50% o his c ys alline phase. Mo eo e , signals shi ed o lowe BE a e palladium addi ion when compa ed o he suppo s. This could be a ibu ed o a pa ial educ ion o Ti 4+ o Ti 3+ gene a ing oxygen acancies, as epo ed by o he au ho s [15]. In line wi h his obse a- ion, bo h ca alys s also p esen ed a low-in ensi y signal loca ed a ound 460.3 eV and assigned o Ti 3+ a Ti 2p 1/2 le el. A he Pd 3 d le el, ob ained signals we e decon olu ed in o ou peaks (Fig. 4 B and Table 3) co esponding o Pd 0 ( e e ence alues o 334.7 eV o Pd 3d 5/2 and 340.4 o Pd 3d 3/2 ) and PdO ( e e ence alues o 336.2 eV o Pd 3d 5/2 and 341.4 o Pd 3d 3/2 ). We selec ed Pd 3d 5/2 signals o in e p e a ion o palladium oxida ion s a e (Table 3). In case o Pd/Ti5, signals shi ed o lowe BE han he alues epo ed o Pd 0 and PdO, indica ing me al-suppo in e ac ion e en in non- educed ca alys s. Fo Pd/Ti45 and Pd/Ti120, hese signals shi ed o he oppo- si e di ec ion (highe BE) also indica ing a me al-suppo in e ac ion, bu di e en om Pd/Ti5. Shi s o highe BE o Pd 0 ha e been usually associa ed o smalle pa icles o Pd [22]. This is in line wi h ou STEM obse a ions, we e Pd/Ti45 and Pd/Ti120 showed smalle me allic do- mains han Pd/Ti5. Fo Pd/Ti5, he shi s o highe BE o Ti 2p and lowe BE o Pd 3 d could indica e an elec onic ans e om suppo o me al, as obse ed by Namdeo e al. [3] o Pd/TiO 2 ca alys s. This beha io was a ibu ed o a possible o ma ion o TiPd x O s uc u es whe e palladium oxida ion s a e could be mo e nega i e han ze o [3,37]. Due o he high elec onega i i y o Ti, a cha ge ans e om suppo o me al could be expec ed, gene a ing an elec on cloud owa ds palladium [3]. In line wi h his explana ion, Pd/Ti5 ca alys p esen ed he highes con en o Pd 0 (34.7%). Acco ding o Mu a a e al. [53], Pd 0 –PdO in e aces a e mo e ac i e in C–H dissocia ion o me hane han PdO o me allic Pd. In ou case, he p esence o Pd 0 /PdO in e aces in he ca alys s could be bene icial o glyce ol oxida ion in liquid phase in e ms o OH − adso p ion and glyce ol dehyd a ion o glyce aldehyde in e media e [8, 12]. The opposi e shi s obse ed o Pd/Ti45 and Pd/Ti120 o lowe BE in Ti 2p and highe BE in Pd 3 d could be associa ed o an elec onic ans e om small me al pa icles o suppo , gene a ing oxidized species (Pd δ+ wi h δ ≥2) and/o Pd/PdO x in e aces [3,54]. Dodson e al. [15] epo ed a highe con en o hese species in u ile phase. In ou case, bo h ca alys s sha e a g ea quan i y o TiO 2 (II) phase (43.3 and 49.4 w % espec i ely) and an upwa d end in u ile (8.3 and 27.6 w % espec i ely) which migh be he cause o he opposi e elec on ans e compa ed o he almos pu e ana ase p esen in Pd/Ti5. These esul s shed a ligh on he impo ance o g ain bounda ies in poly- mo phic ans o ma ions, which could change he di ec ion o elec on densi y ans e be ween me al and suppo , a ec ing specia ion and ca aly ic p ope ies o ma e ials. Inc eased numbe o de ec s, and con en o u ile and TiO 2 (II) phases in he suppo s milled o longe Table 2 XPS and ICP-OES analyses o ca alys s. Ca alys a %, heo e ical a %, XPS w %, heo e ical w %, ICP-OES Pd/Ti Pd/Ti Pd Pd Pd/Ti5 0.19 0.39 0.25 0.20 Pd/Ti45 0.51 0.29 Pd/Ti120 0.64 0.36 M.G. Rinaudo e al. Resul s in Enginee ing 16 (2022) 100737 7 Fig. 3. STEM images and EDS analyses: Pd/Ti5 (A* and B), Pd/Ti45 (C) and Pd/Ti120 (D). *Rep oduced wi h pe mission om e . [43]. M.G. Rinaudo e al. Resul s in Enginee ing 16 (2022) 100737 8 ime would lead o mo e elec on-de icien palladium species (oxida ion s a es highe han ze o) due o elec on mig a ion om palladium o i anium, as can be seen in Pd 0 /(Pd 0 +Pd δ+ ) end (Table 3) [14,15,17]. Ca ionic species o palladium and/o Pd/PdO bounda ies ob ained in he h ee ca alys s a e expec ed o accele a e he mechanism o glyce ol oxida ion [3,8]. Addi ionally, om he poin o iew o ans e -induced elec os a ic in e ac ions, he sin e ing o posi i ely cha ged palladium pa icles demands mo e ene gy han ze o alence palladium pa icles [17]. The e o e, he a ailabili y o ca ionic species on he su ace may also s abilize he ca alys s unde eac ion cycles, p e en ing a high a e o sin e ing. Pd/Ti a % (Table 2) we e in all cases highe han he heo e ical calcula ion o he bulk (0.19 a %) bu simila when a ec ed by S BET , which sus ained he dispe sion o Pd on he su ace o suppo s [16]. O 1s le el was decon olu ed in o wo signals (Fig. 5 and Table 3) associa ed o la ice oxygen (O la ) (ca. 529.9 and 529.7 eV o ana ase and u ile espec i ely) and oxygen adso bed on oxygen acancies (O ads ) (ca. 531.09 and 531.22 eV o ana ase and u ile espec i ely) [52]. In case o Pd/Ti5, O la showed a BE close o he alue epo ed o ana ase while in Pd/Ti45 and Pd/Ti120 his alue was close o he BE o u ile, in acco dance wi h he c ys alline phases coexis ing in he ca alys s. Also, Ti/O la a omic a ios (Table 3) we e in all cases lowe ha he heo e ical (0.5) which may gi e e idence o ce ain numbe o oxygen acancies p esen in he ca alys s. In line wi h hese alues, Pd/Ti5 showed he highe O ads quan i y, 16.26%, while Pd/Ti45 and Pd/Ti120 showed alues in he o de o 10%. These con en s we e in he same o de as obse ed in he suppo s [27], wi h a sligh a ia ion asc ibed o palladium addi ion and u he calcina ion. These low alues o O ads achie ed a e milling we e p e iously explained in e ms o oxygen acancies mig a ion om he su ace o he bulk in o de o induce polymo phic ans o ma ion du ing he syn hesis o suppo s [27]. Dis inc i e p ope ies obse ed in he h ee ca alys s migh ha e a decisi e ole on ac i i y and selec i i y du ing glyce ol oxida ion, whe e cha ge and oxygen mobili y a e key ac o s. In o de o emphasize suppo ’s milling e ec , we decided o p epa e and analyze a ca alys composed o 0.25 w % palladium using p is ine TiO 2 suppo (ana ase, S BET o 8 m 2 g −1 ) [27] and labeled i Pd/Ti. A he Ti 2p le el we ob ained signals in he o de o epo ed ones o ana ase phase wi h no subs an ial shi s in binding ene gies. In case o O 1s we obse ed an O ads quan i y o 23%, displaying some deg ee o su ace Fig. 4. XPS spec a o Ti 2p (A) and Pd 3 d (B) o ca alys s. Table 3 XPS analyses o ca alys s. Ca alys Binding Ene gy (eV) a %, XPS a . a io, XPS Ti 2p 3/2 (Ti 4+ ) Ti 2p 1/2 (Ti 4+ ) Ti 2p 1/2 (Ti 3+ ) O 1s (O la ) O 1s (O ads ) Pd 3d 5/2 (Pd 0 ) Pd 3d 5/2 (Pd δ+ ) Pd 0 /(Pd 0 + Pd δ+ ) O ads /(O la + O ads ) Ti/O la Pd/Ti5 458.79 464.48 460.56 529.73 531.06 334.37 336.04 34.70 16.26 0.45 Pd/Ti45 458.38 464.06 460.33 529.65 531.22 334.32 336.21 30.33 10.10 0.42 Pd/ Ti120 458.33 464.02 460.40 529.58 531.07 334.42 336.18 20.04 10.42 0.42 M.G. Rinaudo e al. Resul s in Enginee ing 16 (2022) 100737 9 oxygen acancies, in ag eemen wi h i s suppo [27]. Mo e in e es - ingly, decon olu ed Pd 3d 5/2 signals ga e alues o 338.23 and 339.46 eV, indica ing he only p esence o ca ionic palladium species [55]. Since no subs an ial shi s in Ti 2p signals we e obse ed, we could deduce a negligible me al-suppo in e ac ion in his ca alys , whe e Pd δ+ species would be a consequence o su ace oxygen in p is ine i ania. These esul s allowed us o ema k ha e en a sho milling ime could p o oke he suppo ’s ‘ac i a ion’ and bene i me al-suppo in e ac ion, inducing elec on mig a ion acco ding o he de ec s in o- duced and he TiO 2 polymo phs achie ed. To deepen in he s udy o oxygen mobili y and oxygen species p e- sen in he ca alys s, o i al impo ance in oxida ion, we ca ied ou O 2 -Tempe a u e P og ammed Deso p ion (O 2 -TPD) expe imen s. 3.5. O 2 -TPD analyses Ouyang e al. [56] epo ed di e en oxygen species o O ads signals including oxygen om he su ace o palladium pa icles and oxygen om he su ace and subsu ace o suppo s. Deso p ion signals (Fig. 6) ound below 100 ◦C we e ela ed o oxygen species weakly bonded o he su ace, as epo ed by o he au ho s [57,58]. On he o he hand, deso p ion signals loca ed a ound 450 ◦C ha e been associa ed o su - ace la ice oxygen om he suppo [27]. These signals we e obse ed a highe empe a u es when compa ed o he suppo s, which could be explained by he me al-suppo in e ac ion expe ienced a e palladium addi ion, making oxygen elease mo e di icul . Signals in he ange 174–220 ◦C we e a ibu ed o oxygen species adso bed on oxygen a- cancies [59]. These oxygen species a he su ace could be in ol ed in he o ma ion o oxygena ed palladium species, wi h s onge bonds, ha he e o e deso b a highe empe a u es. This ac could explain he disappea ance o se e al low- empe a u e signals ha we e obse ed in he suppo s [27]. Also, he highe in ensi y and lowe empe a u e o he deso p ion peak a 174 ◦C obse ed in Pd/Ti5 could be associa ed o he g ea e quan i y o oxygen acancies exhibi ed by XPS (Table 3). Fu he mo e, acco ding o Ouyang e al. [56] signals loca ed close o 200 ◦C could be a ibu ed o molecula oxygen adso bed on palladium pa icles, since hey obse ed hem in Pd black e e ence. Signals nea 350 ◦C could be associa ed o he deso p ion o oxygen om inely dispe sed PdO while he signal a 575 ◦C, only obse ed in Pd/Ti5, migh be ela ed o oxygen om la ge PdO pa icles, as epo ed by Huang e al. [58]. These obse a ions also ha e a good co ela ion wi h he palladium pa icles in e ed by STEM, wi h la ge sizes o Pd/Ti5 and smalle sizes o Pd/Ti45 and Pd/Ti120, leading o mo e oxygen-ac i e species in he o me . Ca alys s wi h highe oxygen mobili y a e hen expec ed o p omo e glyce ol selec i e oxida ion, in e ms o oxygen supply and/o egene a ion o he ac i e su ace [12,30,60]. 3.6. H 2 -TPR analyses G ain bounda ies and in e aces be ween i ania phases and/o palladium could inc ease cha ge mobili y and hus educibili y o he ca alys s [32]. H 2 -TPR measu emen s we e pe o med o u he s udy me al-suppo in e ac ions, ha play a majo ole in ca aly ic eac ions ha p oceed ia edox cycle [53]. S ong signals ound in he ange 533–578 ◦C (Fig. 7) we e associa ed o he educ ion o su ace Ti 4+ Fig. 5. XPS spec a o O 1s o ca alys s. Fig. 6. O 2 -TPD o ca alys s. M.G. Rinaudo e al.