Vous êtes sur la page 1sur 12

~

A
PA
LE
IY
D
CP
AT
L
SS
I
A: GENERAL

ELSEVIER

Applied Catalysis A: General 127 (1995) 165-176

Chemoselective catalytic oxidation of glycerol with


air on platinum metals
Rfgis Garcia, MichNe Besson *, Pierre Gallezot
Institut de Recherches sur la Catalyse-CNRS, 2 Avenue Albert Einstein, 69626 Villeurbanne Cedex, France

Received 1 December 1994; revised 14 February 1995; accepted 14 February 1995

Abstract

The transesterification of triglycerides extracted from oilseeds to obtain biodiesel fuel yields up to
14% by weight of glycerol as by-product. The liquid-phase oxidation of glycerol with air on platinum
catalysts was investigated to prepare valuable oxidation products such as glyceric acid or dihydroxyacetone. The effect of the pH (pH range 2-11) and of different metal catalysts was studied. The
selectivity to glyceric acid can be as high as 70% at 100% conversion on Pd/C at pH 11. On Pt/C
catalyst, glyceric acid was still the main product (55% selectivity) ; but the deposition of bismuth on
platinum particles orientates the selectivity towards the oxidation of the secondary hydroxyl group to
yield dihydroxyacetone with a selectivity of 50% at 70% conversion.
Keywords: Dihydroxyacetone; Glyceric acid; Glycerol; Oxidation; Palladium, Platinum

1. Introduction
The transesterification with methanol of triglycerides extracted from oilseeds
(rapeseed, sunflower) yields so-called biodiesel, which may attain a production of
750 000 tonnes / year by 1995 [ 1 ]. Since this reaction yields up to 14 wt.- % glycerol
as by-product, a surplus of ca. 100 000 tonnes/year of this chemical would be added
to the present market of ca. 610 000 tonnes/year, and new outlets would have to
be found to absorb this surplus. Projects are under way to convert glycerol into
ester or ether derivatives to obtain various marketable products, e.g. biodegradable
surfactants [2]. Since glycerol 1 is already a highly functionalized molecule compared to hydrocarbons from petrochemistry, an advantageous outlet would be to
use it as feedstock for the production of valuable oxygenates.
*Corresponding
72445399.

author. E-mail mbesson@catalyse.univ-lyonl.fr,

0926-860X/95/$09.50 1995 Elsevier Science B.V. All fights reserved


SSDI O 9 2 6 - 8 6 0 X ( 9 5 ) 00048-8

tel. (+33)

72445358, fax. (+33)

R. Garcia et al. /Applied Catalysis A: General 127 (1995) 165-176

166

This investigation was directed at designing chemoselective catalysts to orientate


the glycerol oxidation reaction towards either the oxidation of the primary alcohol
functions, to give glyceric acid 2 (GLYA), or the oxidation of the secondary alcohol
function, giving dihydroxyacetone 3 (DHA) and hydroxypyruvic acid 4 (HPYA).
OH

OH

OH

OH

To date, the market for these chemicals is limited because of their high cost,
firstly because they are produced from glycerol (which presently has a high and
fluctuating cost), and secondly because their syntheses are expensive. In fact,
dihydroxyacetone and hydroxypyruvic acid are obtained by low-productivity fermentation processes, and glyceric acid by potentially polluting oxidation processes
with mineral acids [ 3]. Beside its use as a tanning agent in the cosmetics industry,
dihydroxyacetone at a lower market price could be more widely used as synthon
in organic synthesis [4]. Thus, dihydroxyacetone and hydroxypyruvic acid are
possible starting materials for OL-serine synthesis [ 5].
Platinum metals are more selective for the oxidation of primary alcohols than
for the oxidation of secondary alcohols [ 6,7] but their selectivity to the latter can
be greatly improved by associating them with p-electron metals, particularly heavy
metal atoms of Group IV (lead) and V (bismuth). In that way, for the oxidation
of gluconic acid, it was shown that the ratio SIS6, where $2 is the selectivity to
oxidation of the secondary alcohol next to the carboxylic group and $6 the selectivity
to oxidation of the terminal primary alcohol, increases from 0.086 on a Pt/C catalyst
to 12.4 on the same Pt/C catalyst after addition of lead atoms [8,9]. Kimura et al.
[10,11 ] have recently published a study on glycerol oxidation showing that the
addition of bismuth to platinum catalysts greatly improves selectivity towards the
secondary alcohol. In that way, a 20% yield in dihydroxyacetone (DHA) was
obtained in a batch reactor at 30% conversion on a l%Bi-5%Pt/C catalyst [ 10],
and a 30% DHA yield at 40% conversion was obtained in a fixed-bed catalytic
reactor with a 0.6%Bi-3%Pt catalyst supported on granular charcoal [ 11 ].
Taking into account our previous studies on oxidation of glyoxal [ 12,13] and
glucose [ 14,15 ], the oxidation of aqueous glycerol solutions with air was conducted
on platinum or palladium catalysts obtained by specific preparation methods.

2. Experimental
2.1. Glyceroloxidation procedure
Oxidation of aqueous solutions of glycerol was carried out at atmospheric pressure in a thermostated glass reactor of 500 ml equipped with a stirrer, a gas supply

R. Garcia et al. /Applied Catalysis A: General 127 (1995) 165-176

..... I

167

t. .....

Fig, 1. Scheme of the reaction vessel and attachments. 1. stirrer; 2. thermometer; 3. condenser; 4. air inlet; 5.
flowmeter; 6. temperature regulator; 7. pH electrode; 8. pH meter; 9. pen recorder; 10. pump; 11. scale; 12. 02
electrode; 13. pO2 meter

system, an oxygen electrode (Ingold) and a pH electrode (Radiometer). The


reactor lay-out is depicted in Fig. 1. The catalyst was suspended in 300 ml of water
under a nitrogen atmosphere and the suspension was heated to 333 K whilst stirring
continuously at 1200 rpm. Glycerol was then added and, following a delay of 10
min, air was bubbled through the slurry at 0.75 ml rain- 1. The initial concentration
of the aqueous glycerol solution was 1 tool 1- l ( 10 wt.-%). The reaction medium
was maintained at a constant pH by addition a 30% sodium hydroxide solution
using a pump controlled by a pH meter. The temperature, pH and oxygen pressure
in the liquid phase were continuously recorded. Samples of the reaction medium
were taken at various time intervals and analyzed to follow the conversion and

168

R. Gareia et al. /Applied Catalysis A: General 12 7 (1995) 165-176

product distribution. Analysis was carried out using high-performance liquid chromatography (HPLC) with UV and refractive index (RI) detectors after separation
on an ion exclusion column (Alltech OA-1000). The concentration of bismuth in
the solutions following reaction were measured by atomic absorption spectroscopy.

2.2. Catalyst preparation and character&ation

Palladium catalysts were prepared by impregnation of active charcoal (CECA


50S) with acidic solutions of PdC12- ions at 298 K for 5 h. The reduction was
carried out at 273 K by dropwise addition of 37% aqueous solution of formaldehyde
followed by dropwise addition of 30% KOH solution. The Pd/C (5.5 wt.-% Pd)
catalyst was then dried at 333 K under vacuum and was not subsequently reduced
prior to use. A monometallic platinum Pt/C catalyst (5 wt.-% Pt) was prepared in
the same way by impregnation with an acidic solution of PtC16- ions.
Bimetallic PtBi/C catalysts were prepared by three different techniques, viz.:
( 1) Platinum was loaded on the active charcoal previously functionalized by NaC10
treatment to create surface carboxylic acid groups. Ion exchange of the protons of
these carboxylic acid groups, by Pt(NH3)42+ ions, was performed as described
previously [12,13]. Catalyst Pt/C (7 wt.- % Pt) thus obtained was loaded with
bismuth. The catalyst was suspended in a glucose solution under a nitrogen atmosphere; then the required volume of a solution of BiONO3 in hydrochloric acid ( 16.2
g 1-1 of Bi) was added to the suspension whilst continuously stirring. During all
these operations nitrogen flow was maintained and the temperature was held at 313
K. Under these conditions, bismuth was reduced by a redox surface reaction and
bimetallic particles of homogeneous composition were formed as shown previously
[8,9]. Two samples, PtBi/Cex containing 1.2 and 4.6 wt.-% Bi, respectively, were
obtained. (2) Catalyst 5% Pt/C, prepared by impregnation, was loaded with bismuth as described above. Catalysts PtBi/Ci containing 5 wt.-% of platinum and
1.1 or 2.7 wt.-% of bismuth were obtained. (3) Catalysts PtBi/Ccoi were prepared
by coimpregnation of the support with acidic solutions of H2PtC16 and BiC13 at 278
K for 5 h. The reduction was carried out by addition of formaldehyde and KOH.
Catalysts 8.4%Pt-l.2%Bi/Qoi and 7.4%Pt-2.9%Bi/Qoi were thus obtained.
The concentration of metals in the catalysts were measured by atomic absorption
spectroscopy after dissolving the solids, and the size of the metal particles were
measured by high resolution transmission electron microscopy (TEM) with a JEOL
100CX microscope. TEM views were taken on ultramicrotome sections of the
catalysts cut with a diamond knife after embedding in epoxy resin. The composition
of individual metal particles was determined with a field-emission gun scanning
transmission electron microscope (STEM) VG HB 501 and energy dispersive Xray (EDX) analyzer at 1.5 nm spatial resolution on thin sections.

R. Garcia et aL /Applied Catalysis A: General 127 (1995) 165-176

169

300 , , .

10

100

i
0

time (h)

Fig. 2. Effect of pH on the activity of 5%Pd/C catalyst for oxidation of glycerol. Consumption of glycerol (a)
and oxygen pressure in the solution (b) vs. time.

3. Results and discussion

3.1. Selective oxidation to glyceric acid on palladium catalysts


The conversion of glycerol on the 5 wt.-% P d / C catalyst was studied as a function
of time and pH. Fig. 2a shows that the higher the pH is, the higher is the reaction
rate. At pH 11, the rate is so high that the pressure of oxygen in the liquid phase is
zero until the end of the reaction (Fig. 2b) which indicates that the kinetics are
limited by the rate of diffusion of oxygen from the gas phase to the liquid phase.
The initial reaction rate, measured at pH 9, where there is little or no external mass
transfer, was 0.1 mol h - t m m o l ~ ) (Table 1).
The selectivity to GLYA increased with pH. Thus, at pH 7, 9 and 11, the
maximum selectivities obtained were 30, 55 and 77%, respectively. Figs. 3a and
3b give the product distributions measured in a reaction conducted at pH 11 as a

R. Gareia et al./Applied Catalysis A: General 127 (1995) 165-176

170

Table 1
Reaction data for glycerol oxidation
Catalyst

Preparation
mode"

Bi/Pt
mol
ratio

Initial rates of reaction [retool h 1


mmol t (pd or Pt) ] at different pH

Initial
selectivity
to DHA

Maximum yield
(C~o)b

(%)
2
5%Pd/C
5%Pt/C
PtBi/C
PtBi/Cex
PtBi/Ccx
PtBi/C~
PtBi/C~
PtBi/Cco.
PtBi/C,,o~

(1)
( 1)
(2)
(2)
(3)
(3)

0.95
0.15
0.47
0.19
0.48
0.13
0.37

107
112
107
66
87
100
38
52

11

26
174
177

58
375
203

96
300
277

109
347

DHA
I(Y
250
77
40
80
70
80
70
80

GLYA

8( 100)" 70(90) c
12(65) d 55(90) d
30(40) ~
20(70)
30(60)
22(80)
28(60)
34(70)
37(75)

a Preparation mode, ( 1) loading of Pt by ion exchange with Pt(NH3) 3+, reduction by formaldehyde, loading of
bismuth by reduction of BiO +, (2) loading of Pt by impregnation with PtCI6 , reduction by formaldehyde,
loading of bismuth by reduction of BiO +, (3) coimpregnation of Pt and Bi, reduction by formaldehyde.
b Maximum yield at conversion given in parentheses.
CAtpH 11.
dAt pH 7.
c At pH2.

function of time and conversion, respectively. The maximum yield in GLYA (70%)
was obtained at ca. 90-100% conversion; the selectivities to the other reaction
products [DHA, tartronic (COOH-CHOH-COOH) and oxalic (COOH-COOH)
acids] were always smaller than 10% provided glycerol was present in the reaction
medium. Following 100% conversion (Fig. 3a), GLYA concentration decreased
at the expense of oxalic acid. Formic acid was probably formed transiently but
should be rapidly oxidized to carbon dioxide because Pd/C catalysts are active for
this reaction [ 16].
These results are in agreement with the present literature data on the oxidation
of aqueous solutions of alcohols on platinum metals and one may draw similar
conclusions: (i) the rate of oxidation of a primary alcohol function is higher than
that of a secondary alcohol [6,7]. (ii) the oxidation stops at the aldehyde stage
only in those instances where the primary alcohol function is located next to an
aromatic ring or a C = C double bond [ 7]. (iii) a basic pH increases the oxidation
rate because it favours the deprotonation of the hydroxyl group (first step of the
reaction) [6] or because it favours the desorption of the acid from the metal [ 17]
(last step of the reaction). At higher contact time, GLYA was oxidized to oxalic
rather than tartronic acid.
Palladium-bismuth catalysts were also used, but the deposition of bismuth on
the 5 wt.-% Pd/C catalyst decreased the rate of glycerol oxidation without improving the selectivity to GLYA or to DHA.

R. Garcia et al. / Applied Catalysis A." General 127 (1995) 165-176

300

k
I~.
\

I
~2001

~
~

~
/

\+/

GLYA

"

DHA
Tartronicacid
Oxalicacid

_"

171

time (h)

80"
b

GLYA
* DHA
M Tartronieacid
o

60"

Ox~ '

_~ 40'~.

20"

-.

20

'~

40

60

80

100

glycerol conversion (%)


Fig. 3. P r o d u c t d i s t r i b u t i o n o b t a i n e d at p H 11 on 5 % P d / C as a f u n c t i o n o f t i m e ( c u r v e a) a n d g l y c e r o l c o n v e r s i o n
(curve b).

3.2. Reaction data on unpromoted platinum catalyst


The initial rates of glycerol oxidation on 5 wt.-% Pt/C given in Table 1 are pH
dependent. The highest rate (375 mmol h - 1 m m o l ~ ) ) at neutral pH was six times
higher than on palladium. Higher oxidation rates on platinum than on palladium
were also observed in the case of glyoxal [ 12] and methanol [ 18 ] oxidations. They
were attributed to the lower redox potential or work function of palladium which
leads to a higher oxygen coverage of the surface, thus decreasing the probability
of adsorption of the organic substrate; metals with lower redox potentials such as
ruthenium, were almost inactive due to a complete oxygen poisoning of their surface
[ 12]. A similar interpretation in terms of over-oxidation of the surface may be
given to account for the lower activity of palladium in glycerol oxidation.

172

R. Garcia et aL / Applied Catalysis A." General 127 (1995) 165-176


50

a
40

< 30
1'a
-o 20
.~,

10"

0
0

20

40
glycerol

60

80

100

80

O0

(%)

conversion

50

40

< 3O
>.
rl

i2o
10'

0
-

2'0

40
glycerol

6'0

conversion

(%)

Fig. 4. Influence of the pH on the selectivity of glycerol oxidation on a commercial PtBi/C catalyst. DHA yield
( a ) and HPYA yield ( b ) vs. conversion.

The selectivity of the Pt/C catalyst was different from that of Pd/C. GLYA was
still the main product but its yield was lower (maximum yield 55% at 90% conversion) while a slightly higher yield in DHA was obtained (maximum yield 12%
compared to 8% on palladium).

3.3. Selective oxidation to dihydroxyacetone on platinum-bismuth catalysts


According to literature data [ 6,11 ], the selectivity towards the oxidation of the
secondary alcohol function can be greatly improved by associating platinum with
p-electron metals such as bismuth. A commercial 5 wt.-% Pt-5 wt.-% Bi/C catalyst
was used to determine the optimum reaction conditions and particularly the pH of
the reaction medium, to give the highest DHA yield. Figs. 4a and 4b show that at

R. Garcia et al. /Applied Catalysis A: General 127 (1995) 165-176

173

the lower pH the oxidation of glycerol gives mainly dihydroxyacetone (50% selectivity at 60% conversion). The reaction at pH = 2 was carried out without any
addition of alkali; the initial pH of ca. 4 dropped rapidly to pH 2-1.5. At higher
contact time, the primary alcohol function was slowly oxidized to give HPYA
(23 % selectivity at 100% conversion). Subsequent experiments were conducted at
pH 2.
Catalysts prepared by method ( 1 ) (platinum loading by ion exchange and bismuth loading by redox reaction of BiO + ions on hydrogen-covered platinum particles) were characterized by high resolution TEM and by STEM-EDX analysis at
1.5 nm spatial resolution on thin sections of the catalyst grains cut with an ultramicrotome. As previously observed [ 12,13], the ion-exchange method leads to
platinum particles in the size range 1-2 nm, uniformly distributed throughout the
active charcoal grains. STEM-EDX analyses of individual particles in catalysts
prepared by ion exchange and moderately loaded with bismuth (Bi/Pt molar
ratio = 0.15) showed, as previously in the case of PdBi/C catalysts [ 14,15 ], that
bismuth was very uniformly deposited on the platinum particles. Analysis of bismuth on the catalysts and in the solutions after reaction showed that there was no
bismuth leakage from catalyst PtBi/C~x (Bi/Pt molar ratio = 0.15) nor from catalyst PtBi/C~x (Bi/Pt molar ratio = 0.47) during reaction.
The product distribution obtained with catalyst PtBi/C~x (Bi/Pt = 0.47) is given
as a function of time and conversion in Figs. 5a and 5b, respectively. The initial
selectivity in DHA was 80%, but it gradually decreased so that a maximum yield
of 30% DHA was attained at 60% conversion (Fig. 5b). From the data given in
Fig. 5a, DHA appears to be rather stable and does not transform into HPYA. Indeed,
DHA molecules, once formed, have little chance to be further oxidized because
carboxylic acids (GLYA and HPYA) formed concurrently can bond strongly to
the metal surface and prevent DHA adsorption. The ketoacid seems to be issued
directly from glycerol with an initial selectivity close to 20%, which then decreased,
whilst the HPYA yield grew steadily even at high conversion (in contrast with
DHA).
PtBi/Ci catalysts obtained by preparation method (2) (loading of platinum by
impregnation, formaldehyde reduction and loading of bismuth by surface redox
reaction of BiO ions on hydrogen-covered platinum particles) were characterized
by TEM. The particle sizes are slightly larger (size range 1--4 nm) than for the
catalysts prepared by method ( 1 ) and the particle distribution in the active charcoal
grains is less uniform. However, Table 1 shows that the selectivity data are almost
similar to those obtained with the previous catalysts, i.e. the initial selectivity to
DHA was 80% and the maximum DHA yield was ca. 30% at 60% conversion. The
B i/Pt molar ratios (0.19 and 0.48) did not change significantly following reaction
which confirmed that bismuth was not leached away during reaction.
Catalysts PtBi/Ccoi, prepared by method (3) (coimpregnation of platinum and
bismuth, formaldehyde reduction) gave higher yields in DHA (Fig. 6 and Table
1 ). The difference in yield between the three preparation methods was more pro-

174

R. Garcia et al./Applied Catalysis A." General 127 (1995) 165 176


300
o

GLYAD +GLYA

DHA

HPYA

=o 200

ffl

E 100

a
i

time (h)
50'

40'

GLYAD + GLYA

DHA

HPYA

30
"O

~,

20

10

0
20

40

60

80

00

glycerol conversion (%)

Fig. 5. Product distribution obtained at pH 2 on a PtBi/Cex catalyst as a function of time (curve a) and glycerol
conversion (curve b); (GLYAD = glyceraldehyde).

nounced at lower bismuth concentrations. Thus, for a molar ratio Bi/Pt = 0.13, the
initial selectivity to DHA was 70%; the maximum DHA yield was 34% at 70%
conversion compared to ca. 21% on the two previous catalyst types. At higher
bismuth concentration (Bi/Pt molar ratio=0.37), a 37% yield in DHA was
obtained at 75% conversion. No bismuth leakage into the solution was detected. A
definitive interpretation of the improved selectivity of this catalyst must await a
detailed characterization of the structure of the bimetallic catalyst.

4. Conclusion
This preliminary study shows that, depending upon the reaction conditions and
upon the nature of the catalyst, it is possible to orientate the selectivity of glycerol

R. Garcia et al. /Applied Catalysis A: General 127 (1995) 165-176

175

40
a

< 30
"r

Q 20

"o

--__0>,10
'-

Bi/Pt = 0.15
Bi/Pt = 0.47

b
30
20

_ D ~

10

19

l"

t=o48

1:
O ~ r

0.a7
.

20
40
60
80
glycerol conversion (%)

100

Fig. 6. DHA yield vs. conversion at different Bi ]oadings (Bi/Pt molar ratios) for various catalysts: (a) PtBi/
Cex, (b) PtBi/C~, (c) PtBi/Coo~.

oxidation towards the oxidation of the primary or secondary alcohol function. These
results are encouraging because although the reaction conditions and catalysts have
not yet been fully optimized, the following points are highlighted: (i) This is, to
our knowledge, the first report to indicate that a high selectivity to glyceric acid
(77% at 90% conversion) can be obtained by air oxidation of glycerol solutions
on palladium catalysts. This one-pot, clean catalytic oxidation could be used in
place of the present mineral acid oxidation to produce glyceric acid at a much lower
cost particularly if a cheaper glycerol, from triglyceride transesterification, was
available on the market. (ii) The previous results of Kimura et al, [ 10] on glycerol
oxidation using platinum-bismuth catalysts in a batch reactor were confirmed and
greatly improved since a 37% yield in dihydroxyacetone at 70% conversion was
obtained instead of 20% at 30% conversion.

References
[ 1 ] J. Chowdhury and K. Fouhy, Chem. Eng., (Feb. 1993) 35.
[2] K. Gottlieb, H. Neitsch and R. Wessendorf, Chem.-Ing.-Tech., 66 (1994) 64.
[3] K. Miltenberger, in B. Elvers, S. Hawkins, M. Ravenscroft, J.F. Rounsaville and G. Schulz (Editors),
Ullmann's Encyclopedia of Industrial Chemistry, Vol. A 13, VCH, Weinheim, 1989, p. 507.
[4] Merck catalogue, MS Info 92-2.
[5] H. Kimura and K. Tsuto, J. Am. Oil Chem. Soc., 70 (1993) 1027.

176
[6]
[7]
[8]
[9]
[ 10]
[11]
[ 12]
[ 13]
[14]
[ 15]
[ 16]
[ 17]
[ 18]

R. Garcia et al. / Applied Catalysis A : General 127 (1995) 1 6 ~ 1 7 6

P. Vinke, D. de Wit, A.T.J.W. de Goede and H. van Bekkum, Stud. Surf. Sci. Catal., 72 (1992) l.
T. Mallat and A. Baiker, Catal. Today, 19 (1994) 247.
P.C.C. Smits, B.F.M. Kuster, K. van der Wiele and H.S. van der Baan, Carbohydr. Res., 153 (1986) 227.
P.C.C. Smits, B.F.M. Kuster, K. van der Wiele and H.S. van der Baan, Appl. Catal., 33 (1987) 83.
H. Kimura, K. Tsuto, T. Wakisaka, Y. Kazumi and Y. Inaya, Appl. Catal. A, 96 (1993) 217.
H. Kimura, Appl. Catal. A, 105 (1993) 147.
P. Gallezot, R. De Mdsanstoume, Y. Christidis, G. Mattioda and A. Schouteeten, J. Catal., 133 (1992) 47.
P. Gallezot, F. Fache, R. de M6sanstoume, Y. Christidis, G. Mattioda, A. Schouteeten, Stud. Surf. Sci. Catal.,
75 (1993) 195-204.
M. Besson, P. Gallezot, F. Lahmer, G. Fl~che and P. Fuertes, in J.R. Kosak and T.A. Johnson (Editors),
Catalysis of Organic Reactions, Marcel Dekker, New York, 1994, p. 169.
M. Besson, F. Lahmer, P. Gallezot, P. Fuertes and G. Fl~che, J. Catal., 152 (1995) 116.
B. Claudel, M. Nueilati and J. Andrieu, Appl. Catal., 11 (1984) 217.
A. Abbadi, M. Makkee, W. Visscher J.A.R. van Veen and H. van Bekkum, J. Carbohydr. Chem., 12 (1993)
573.
H.E. van Dam, k J . Wisse and H. van Bekkum, Appl. Catal., 61 (1990) 187.

Vous aimerez peut-être aussi