Highly Selective Oxidation of Carbohydrates in an Efficient Electrochemical Energy Converter: Cogenerating Organic Electrosynthesis

Article abstract of DOI:10.1002/cssc.201501593

The selective electrochemical conversion of highly functionalized organic molecules into electricity, heat, and added-value chemicals for fine chemistry requires the development of highly selective, durable, and low-cost catalysts. Here, we propose an approach to make catalysts that can convert carbohydrates into chemicals selectively and produce electrical power and recoverable heat. A 100 % Faradaic yield was achieved for the selective oxidation of the anomeric carbon of glucose and its related carbohydrates (C1-position) without any function protection. Furthermore, the direct glucose fuel cell (DGFC) enables an open-circuit voltage of 1.1V in 0.5 m NaOH to be reached, a record. The optimized DGFC delivers an outstanding output power P_max=2mW cm^-2 with the selective conversion of 0.3 m glucose, which is of great interest for cogeneration. The purified reaction product will serve as a raw material in various industries, which thereby reduces the cost of the whole sustainable process.

Full text of DOI:10.1002/cssc.201501593

DOI: 10.1002/cssc.201501593
Full Papers
Highly Selective Oxidation of Carbohydrates in an Efficient
Electrochemical Energy Converter: Cogenerating Organic
Electrosynthesis
Yaovi Holade,[a] Karine Servat,[a] Teko W. Napporn,[a] Clµudia Morais,[a] Jean-Marc Berjeaud,[b]
and Kouakou B. Kokoh*[a]
The selective electrochemical conversion of highly functional-
ized organic molecules into electricity, heat, and added-value
chemicals for fine chemistry requires the development of
highly selective, durable, and low-cost catalysts. Here, we pro-
pose an approach to make catalysts that can convert carbohy-
drates into chemicals selectively and produce electrical power
and recoverable heat. A 100% Faradaic yield was achieved for
the selective oxidation of the anomeric carbon of glucose and
its related carbohydrates (C1-position) without any function
protection. Furthermore, the direct glucose fuel cell (DGFC) en-
ables an open-circuit voltage of 1.1 V in 0.5m NaOH to be
reached, a record. The optimized DGFC delivers an outstanding
output power Pmax =2 mWcmÀ2 with the selective conversion
of 0.3m glucose, which is of great interest for cogeneration.
The purified reaction product will serve as a raw material in
various industries, which thereby reduces the cost of the
whole sustainable process.
Introduction
Energy converters are used with the aim to supply stationary
and mobile devices with the minimum environmental
impact.[1] In this frame, the development of renewable and
low-cost energy sources is emerging to reduce our energy de-
pendence on fossil fuels.[2] Notably, carbohydrates derived
from biomass are abundant, renewable, and nontoxic organic
compounds that have a potentially high energy.[3] Indeed, car-
bohydrates (glucose, etc), cellulose, and hemicellulose repre-
sent up to 70% of biomass (an extensive and endlessly renew-
able resource).[3a,4] Heterogeneous and enzymatic catalyzes
have focused on the selective glucose oxidation to gluconic
acid, a mild organic acid that has received tremendous interest
in various fields (annual production ꢀ100000 t) as its deriva-
tives (d/g-gluconolactone, sodium and calcium salts) are used
in the food, pharmaceutical, and cosmetic industries.[5] There-
fore, the search for highly selective catalysts to oxidize the
hemiacetal function exclusively at the carbon C1-position (Fig-
ure S1a) has become a prerequisite. The effective development
of an efficient direct glucose fuel cell (DGFC), however, is ham-
pered by the design of durable alkaline anion-exchange mem-
branes (AEM) as well as the synthesis of advanced electrode
materials. In a DGFC, the fuel (fed directly into the anode with-
out any previous chemical modification) is oxidized directly on
the anode.[6] Indeed, the AEM (mostly from Solvay and Tokuya-
ma) must withstand the strong alkalinity of the solution.[6a,7]
The effectiveness of an AEM in solid alkaline membrane fuel
cells (SAMFCs), however, is linked directly to the experimental
conditions (fuel, temperature, etc),[6a,7c–e,8] and better perform-
ances for Fumatech are observed at low temperatures (such as
those used for carbohydrate-based SAMFCs: <508C).[8]
The anode catalyst must withstand the poisoning effect of
species adsorbed strongly that come from the oxidation of or-
ganics, and the cathode must be able to overcome the slug-
gishness of the oxygen reduction reaction (ORR).[9] A simplified
Pourbaix diagram of glucose in aqueous medium based on
thermodynamic data is shown in Figure 1 (left y axis).[10] This
displays the open-circuit potential (OCP) of the ORR and the
glucose oxidation reaction (GOR) with two exchanged elec-
trons (nex =2). The potentials are referenced to the standard
hydrogen electrode (SHE), whereas those scaled to the saturat-
ed calomel electrode (SCE), silver/silver chloride electrode
(SSCE), and reversible hydrogen electrode (RHE) are reported
in Figure S1. For the ORR, the OCP is ꢀ1.18 V vs. RHE in aque-
ous solutions. Recently, advances in materials science have al-
lowed the preparation of active catalysts able to deliver an
OCP>1 V vs. RHE (overpotential <200 mV) with good kinetics
at 0.8–0.95 V vs. RHE.[9b,11] However, the most challenging issue
for SAMFCs is the anodic reaction. To date, the oxidation of or-
ganic molecule at the anode, even with the most active nano-
catalyst, largely occurs with an overpotential ꢁ200 mV, and ac-
tivation is very difficult. Theoretically, at pH 14, the OCP is
À1.12 V vs. SHE or À0.294 V vs. RHE for the GOR. Unfortunate-
[a] Dr. Y. Holade, Dr. K. Servat, Dr. T. W. Napporn, Dr. C. Morais,
Prof. Dr. K. B. Kokoh
Department of Chemistry
IC2MP CNRS UMR 7285
UniversitØ de Poitiers
4 rue Michel Brunet - B27, TSA 51106, 86073 Cedex 9 (France)
E-mail: boniface.kokoh@univ-poitiers.fr
[b] Prof. Dr. J.-M. Berjeaud
EBI UMR 7267 CNRS
UniversitØ de Poitiers
1 rue Georges Bonnet, B36/37, TSA 51106, 86073 Poitiers cedex 09 (France)
Supporting Information for this article is available on the WWW under
http://dx.doi.org/10.1002/cssc.201501593.
ChemSusChem 2016, 9, 252 – 263
252
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Results and Discussion
A broad overview of this study with major goals is provided in
Scheme S1 in the Supporting Information.
Au nanomaterials characterization by TEM and X-ray photo-
electron spectroscopy and their electrocatalytic activity
Typical TEM images of the Au/C nanocatalyst are shown in
Figure 2 (panels i–iv). The total metal loading determined from
thermogravimetric analysis (TGA; Figure S2a) is 21 wt% as ex-
pected (theoretical: 20 wt%). The average particle size is (5.7Æ
0.1) nm as can be seen from the histograms presented inset in
Figure 2 (panel i). Basically, the nanocrystals are deposited ran-
domly onto the support without a well-defined orientation.
The high-resolution transmission electron microscopy (HRTEM)
micrographs shown in Figure 2 (panels v–vi) display a truncated
octahedral shape with different degrees of truncation. The
nanoparticles have facets oriented along the (111) and (200)
crystallographic planes instead of the common (111) and
(100) planes predicted by Wulff’s theorem.
Figure 1. From glucose to electrochemical energy: direct alkaline fuel cell.
Left y axis: simplified Pourbaix diagrams for glucose (blue lines) and water
(brown lines) systems. Right y axis: theoretically recoverable voltage, OCV
(green line); the dashed zone shows the possible values.
ly, Pt, which allows an OCP of ꢀ+0.05 V vs. RHE (most dehy-
drogenation catalysts), is deactivated rapidly. The compromise
relies on Pt-based electrocatalysts. Still out of control, the reac-
tion induces CÀC bond cleavage to lead to unrecoverable
products that also deactivate the catalyst. Thorough glucose
electro-oxidation involves 24 electrons per molecule and ena-
bles an OCV of 1.25 V to be harvested, which constitutes a free
energy of 2871 kJmolÀ1, that is, 4.430 kWhkgÀ1.[3a,6a,12]
We further characterized the Au/C surface state by X-ray
photoelectron spectroscopy (XPS). The high-resolution XPS
spectrum of the Au4f region is depicted in Figure 3a. The ob-
served doublets are related to spin–orbit splitting (Æ1/2). The
doublets of Au are situated at binding energies (BEs) of 83.9
(Au4f7/2) and 87.6 eV (Au4f5/2).[15] The presence of oxide AuOx
is indicated by doublets at BE=85.5 (Au4f7/2) and 89.1 eV
(Au4f5/2). Energy-dispersive X-ray (EDX) spectroscopy and XRD
(Figure S2b–c) agree completely that the oxide amount is neg-
ligible. Currently, upon exposure to ambient air, a thin protec-
tive layer safeguards the metal surface from deep oxidation
(passivation layer).
As indicated above, the last few decades of research has
dealt with nex =2 towards gluconic acid (pKa =3.76). Therefore,
as charted in Figure 1 (right y axis), the electrochemical conver-
sion of glucose as a sustainable fuel affords OCV=1.154 V for
pHꢂ3.76 and OCV=1.04+0.03”pH for pHꢁ3.76. Particular
operating points highlighted in Figure 1 (black for biofuel cells
(BFCs)[13] and red for DGFCs) arise as the two new paradigms
in the GOR to value electricity. Thus, a specific energy of
0.435 kWhkgÀ1 is expected at pH 14. Even if it represents only
ꢀ10% of the value from the total glucose oxidation, the con-
cept is better as gluconate (thus, gluconic acid) is a high
added-value product compared to carbonate from total oxida-
tion. Currently, DGFCs that use an AEM or not deliver a maxi-
mum output power density Pmax of 0.5–1.5 mWcmÀ2. Surpris-
ingly, Fujiwara et al.[14] reported a DGFC designed with an AEM
that delivers a Pmax of 20 mWcmÀ2 at 308C in 0.5m KOH with
The electrochemical behavior of Au/C in 0.1m NaOH is
shown in Figure 3b (dashed plots). Two major features during
the positive scan (peak A’: metal oxidation) and the negative
scan (peak C’: oxide reduction) typify Au in an alkaline
medium.[15–16] In the presence of 10 mm glucose, the electroca-
talytic oxidation of glucose is marked by peaks A1, A2, A3 (for-
ward), and C1 (backward). These peaks are well displayed in
Figure S3. Notably, the oxidation peak A2 is associated with
a shoulder around 0.9 V vs. RHE. This shoulder is situated in
the potential range in which the adsorption of hydroxyl spe-
cies occurs at the Au surface. Actually, glucose electro-oxida-
tion at the Au/C catalyst surface involves three different spe-
cies of Au: Au metal (peak A1 ~0.5 V vs. RHE), Au(OH)x
(peak A2 ~1.2 V vs. RHE), and AuOx (peaks A3 ~1.3 V vs. RHE and
C1 ~1.2 V vs. RHE). The phenomenon that is marked by the
peak A2 is initiated from the shoulder at around 0.9 V vs. RHE.
The main phenomena that occur at these peaks will be deter-
mined thoroughly later. The onset potential is ꢀ0.2 V vs. RHE,
which correlates roughly to 0.49 V of overpotential. However,
previous studies show that the GOR starts here at least 100 mV
earlier.[17] The current density is 285 AcmÀ2 gÀ1 (20 AgÀ1 or
2 mAcmÀ2) at 0.5 V vs. RHE and 1025 AcmÀ2 gÀ1 (i.e., 72 AgÀ1
or 6 mAcmÀ2) at the peak, that is, 1.2 V vs. RHE. Here, the cata-
lyst shows superior electrocatalytic activity because of its good
0.5m glucose that contained
a high metal loading of
3 mgmetal cmÀ2 (carbon-free PtRu (cathode) and Pt nanoparticles
(anode)). With regard to the scarcity and high price of Pt and
Ru, this kind of design requires noble metal reduction in the
electrode catalysts.[11a] No stability test was performed as the
anode material (Pt) is deactivated swiftly by reaction intermedi-
ates. Our main goal herein is to prove the feasibility of the
electrochemical cogeneration process by harvesting electrical
power as the fuel is oxidized selectively to added-value chemi-
cals. For this purpose, highly active and selective Au nanoparti-
cles were prepared as the anode catalyst from a simple
method without surfactant. For the first time, a unique final
product was determined from a versatile experimental setup
thanks to the careful and judicious coupling of electrochemis-
try to analytical chemistry.
ChemSusChem 2016, 9, 252 – 263
www.chemsuschem.org
253
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Figure 2. TEM images of gold nanoparticles dispersed on Vulcan XC 72R carbon (Au/C, 20 wt%) with different magnifications and i) the histogram of nanopar-
ticle size distribution (inset). ii) Overview (50 nm scale bar), iii– iv) near view (10 nm scale bar), and v–vi) close view that shows HRTEM micrographs of one
nanoparticle (zone axis: [101]).
dation of glucose in alkaline media. We first postulate that the
GOR might concern the aldehyde function (C1-position: nex =
2), the primary alcohol (C6-position: nex =4), or both (nex =6).
Spectroelectrochemical determination of the reaction inter-
mediates and products
Cyclic voltammetry coupled with IR spectroscopy: CV-FTIRS
(SPAIRS)
To prove our previous hypothesis (nex ꢂ6), we recorded FTIR
Figure 3. a) High-resolution XPS spectra of the Au4f core level of gold nano-
particles dispersed on Vulcan XC 72R carbon. b) CVs of a Au/C electrode ma-
spectra of glucose and different possible intermediates or
terial at 20 mVsÀ1 in a 0.1 molLÀ1 NaOH electrolytic solution at 258C in the
absence (black dashed line) and presence of 10 mmolLÀ1 glucose (solid red
line).
products in 0.1m NaOH as references for band assignment.
The chemical structures of these possible reaction intermedi-
ates or products are displayed in Figure S4, their reflectance
FTIR spectra are shown in Figure S5a, and the main characteris-
tics are summarized in Table S1. Finally, concentrated standards
were used to better discern bands as no significant effect on
their position was observed during preliminary tests (Fig-
ure S5b). Preliminary optimized GOR conditions based on glu-
cose concentration (10–100 mm) and using the state-of-the-art
Pt/C catalyst are given in Figure S6a–e. However, very low con-
centrations give a low amount of reaction products so it will
be difficult to identify them by IR spectroscopy. Therefore,
50 mm glucose was used. The single potential alteration infra-
red spectroscopy (SPAIRS) results are shown in Figure 4. The
CV shown in Figure 4a displays the forward-going process (red
kinetics compared to that in previous studies.[17] The scan rate
effect on the Au/C electrode material is reported in Figure S3.
The analysis indicates that the electrochemical processes are
mainly under the diffusion control of the reactants and/or
products with a substantial contribution from adsorption. Inde-
pendent investigation by the temperature effect has shown
that the electrochemical activation energy is less than
50 kJmolÀ1; which thereby endorses our conclusion. Before
a DGFC is designed, methods of chemical analysis will be cou-
pled with electrochemical methods for the determination of
the reaction products that result from the electrocatalytic oxi-
ChemSusChem 2016, 9, 252 – 263
www.chemsuschem.org
254
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
tion), or glucarate (both C1 and C6 oxidation). Other-
wise, neither compound resulted in CÀC bond cleav-
age (CO: n˜ =1900–2100 cmÀ1, carbonate: n˜ =
1396 cmÀ1, CO2: n˜ =2340–2350 cmÀ1) nor gluconolac-
tone (specific nC=O band at n˜ =1742 cmÀ1) are ob-
served. The first assessment indicates a selective oxi-
dation at the whole electrode potential range with-
out CÀC bond breaking and the last one highlights
a fast gluconolactone hydrolysis as it is an unavoida-
ble intermediate. Thereby, these results validate our
initial hypothesis if we assume that nex ꢂ6. Such ob-
servations indicate that our revisited bromide anion
exchange (BAE) protocol changed the GOR kinetics
drastically as both CÀC bond cleavage and glucono-
lactone were observed by many methods if metal or
enzymatic catalysts were used previously.[3a,18]
Figure 4. CV-FTIRS experiments on a Au/C electrode material at a scan rate of 1 mVsÀ1 in
0.1 molLÀ1 NaOH+50 mmolLÀ1 glucose: a) CVs recorded in the FTIRS cell and b) CV-FTIR
spectra.
arrows) and reverse scan (blue arrows). FTIR spectra, plotted
every 0.1 V, are depicted in Figure 4b for the positive scan
Chronoamperometry coupled with IR spectroscopy: CA-FTIRS
(0.1–1.4 V vs. RHE) and in Figure S6 f for the reverse scan (1.4–
0.1 V vs. RHE). Basically, for the in situ reflectance mode,
a band that faces downward indicates the formation of a com-
pound, whereas a band that faces upwards signifies its remov-
al from the window–electrode interface (an intermediate or
simple diffusion process). We underscore important current
trends that voltammograms recorded in the IR cell are typical
of those already obtained in a conventional electrochemical
cell. Thus, all processes are typical of the reactions involved.
During the positive scan, oxidation starts very early at Eꢂ0.2 V
vs. RHE (Figure 4a), which thereby corroborates the results
shown in Figure 3b. Accordingly, two bands that appear syn-
chronously at approximately n˜ =1584 and 1413 cmÀ1 can be
seen in the spectra presented in Figure 4b. As indicated in
Table S1, these bands correspond to the asymmetric stretching
vibration nas(OÀCÀO) and symmetric ns(OÀCÀO) of COOÀ (oxi-
dized carbohydrate specific functions), respectively. Glucose
consumption is indicated (through its dehydrogenated species)
by the positive bands at approximately n˜ =1200 (distortion
band) and 2500–2900 cmÀ1 (stretching band). The amount of
water in the thin electrolyte layer during the reaction is dem-
onstrated by its d(H2O) band at approximately n˜ =1660 cmÀ1.
The same bands are present in the reverse scan (Figure S6 f).
The identification of the bands during the reverse
The CA-FTIRS technique allows the online monitoring of the
electrocatalytic reaction by following the vibration bands at
a fixed electrode potential. CA-FTIRS investigations were per-
formed at various electrode potentials from 0.2 to 1.4 V vs.
RHE. Although CA results change from one potential to anoth-
er (rational trend), we were surprised to find that the bands
were unchanged. A CA plot for 0.5 V vs. RHE is shown in Fig-
ure 5a and its spectra are presented in Figure 5b. The intensity
of the CA-FTIRS bands increases over time (an increase in con-
version rate). Moreover, the spectra show the same bands as
those from CV-FTIRS. All of the bands appear during the first
minutes. The solution to the FTIRS dilemma about the bands
at n˜ =1413 and 1584 cmÀ1 consists of using complementary
techniques using chromatography and spectroscopy in
tandem.
Electrosynthesis: Carbohydrate electrolysis
After the spectro-electrochemistry experiments, electrolysis
was performed in potentiostatic mode by applying different
electrode potentials from 0.2 to 1 V vs. RHE in preliminary
tests. Chromatography analyses did not show any difference in
scan is a challenge because the species are already in
the solution (during the positive scan), so we per-
formed a fresh experiment by starting at the upper
potential (1.4 V vs. RHE) for the precise and unequiv-
ocal determination of the nature of the processes
that take place. Surprisingly, the results (not shown)
showed a similar voltammogram and the same
bands. These results, reported herein for the first
time, mean that the peak observed during the re-
verse scan comes from glucose oxidation, not from
intermediates adsorbed strongly (absent).
Furthermore, in-depth analysis of the spectra
shows a dilemma for the assignment of the bands at
n˜ =1413 and 1584 cmÀ1. Indeed, they could belong
Figure 5. CA-FTIRS experiments on a Au/C electrode material at 0.5 V vs. RHE in
0.1 molLÀ1 NaOH+50 mmolLÀ1 glucose: a) Chronoamperogram recorded in the FTIRS
to gluconate (C1 oxidation), glucuronate (C6 oxida- cell and b) CA-FTIR spectra plotted every 6 min.
ChemSusChem 2016, 9, 252 – 263
www.chemsuschem.org
255
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
tions will be undertaken to test this hypothesis. From the gluc-
onate calibration plots (Figure S8b), its concentration (c) was
evaluated to be 13 mm after 7.4 h of electrolysis. Based on
gluconate, the conversion is 65%. The mean experimental ex-
changed number of electrons (nexp) was evaluated by Equa-
tion (1).
Qox
ð1Þ
ð2Þ
ð3Þ
nexp
¼
FVDc
nexpðexperimentalÞ
nexðtheoreticalÞ
tF ¼ 100
nexFVc
Qox
t ¼ 100
Figure 6. Electrolysis of 20 mmolLÀ1 glucose in 0.1 molLÀ1 NaOH on a Au/C
electrode material at 0.8 V vs. RHE. Oxidation current Iox (left y axis, black
open circles) and quantity of electricity Qox (Right y axis, blue solid circles).
in which Qox [C] is the involved quantity of electricity, Dc
[molLÀ1]=c0Àcf, which are the initial and final states for glu-
cose, c [molLÀ1] is the concentration of gluconate (peak A), V
[L] is the volume of the electrolysis solution (V=0.043 L), F is
the Faraday constant (96485 CmolÀ1), tF is the Faradaic yield,
and t is the catalyst efficiency.
the distribution of the reaction products. Consequently, the
electrosynthesis experiment in this study concerns a process
set at 0.8 V vs. RHE. The electrolysis of 20 mm glucose results
in terms of oxidation current (Iox, left y axis) and quantity of
electricity (Qox, right y axis) are displayed in Figure 6. The pro-
gressive decrease in Iox may be attributed reasonably to glu-
cose consumption and/or to catalyst deactivation. Indeed,
strongly adsorbed oxygenated species that come from GOR on
the catalyst surface cause its deactivation even if Au is one the
most stable catalysts for GOR. After 7.5 h of electrolysis, the
produced quantity of electricity is Qox =123 C. Other carbohy-
drates were also studied; Qox =131 C for galactose (the isomer
of glucose) and Qox =87 C for lactose (a dimer of glucose and
galactose; Figure S7). More quantitative data will be extracted
after chromatographic analyses.
If we assume that the theoretical exchanged number of
electrons is nex =2 (gluconate, see above), the theoretical
quantity of electricity is Qth =2”0.02”0.043”96485=166 C
for full conversion (20 mm glucose). Consequently, the involved
quantity of electricity after 7.5 h of electrolysis represents 74,
79, and 53% for glucose, galactose, and lactose, respectively.
To obtain the time-dependent experimental number of elec-
trons [Eq. (1)] and, subsequently, the experimental Faradaic
yield [Eq. (2)], we performed the electrolysis of 50 mm glucose,
and the results are reported in Figure 7b (Dc=c, from peak A)
and Figure S9a–c. Only the intensity (Figure S9b) and the area
(Figure S9c) of peak A increase significantly over time during
electrolysis, and the contribution from this peak is very small
(<10%; Figure S9c). At the end of the electrolysis, nexp =2.05
for tF =102%. Furthermore, nexp =2.2 and tF =110% were ob-
tained for 20 mm glucose. A slight deviation of these values
from theoretical ones (nexp =2; tF =100%) was expected as the
amount of peak B was considered [Dc=c was underestimated
in Eq. (1)]. If we consider the electrolysis of 20 mm glucose
High-performance liquid ionic chromatography: Identifica-
tion of the product(s)
With the aim to determine exactly the nature of the final prod-
uct(s), we performed high-performance liquid ionic chromatog-
raphy (HPLIC). The chromatograms shown in Fig-
ure S8a indicate that the retention time (tR) is 5–6,
10, and 21 min for gluconate, glucuronate, and gluca-
rate, respectively. The qualitative HPLIC analysis of
the sample after electrolysis is depicted in Figure 7a,
which highlights the presence of two unresolved
peaks A (tR =5.5 min) and B (tR =6.1 min). As indicat-
ed, peak A belongs to gluconate. However, peak B
does not belong either to glucuronate or glucarate.
Seminal studies with bulk Au or Pt electrodes have
shown that this peak belongs to gluconate, and
other derivative compounds from CÀC bond break-
ing were also identified.[18d,19] We combined in situ
FTIRS studies and HPLIC analyses to be able to state
Figure 7. a) HPLIC chromatograms: reference (top) and analyte from the electrolysis of
20 mm glucose (bottom: diluted twice before injection). b) Experimental exchanged
number of electrons (left y axis, black) and experimental Faradaic yield (right y axis, blue)
unambiguously that
CÀC bond cleavage does not occur. It is possible that
A and B both belong to gluconate. Other investiga- based on the electrolysis of 50 mm glucose.
ChemSusChem 2016, 9, 252 – 263
www.chemsuschem.org
256
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
into 13 mm gluconate, the catalyst efficiency [Eq. (3)] is t=
88%. This value is 89% towards galactonate (14 mm) if we
consider galactose. The conversions obtained here (65–70%)
are similar to those reported for heterogeneous catalysis.[3a,5d]
NMR spectra, which offer a general fingerprint for carbon
atoms at the C1-position. To obtain good NMR references, we
first analyzed commercially available compounds: glucose (Fig-
ure S11a–c), gluconate (from sodium gluconate, Sigma–Aldrich,
Figure S12a–d), and gluconolactone (Sigma–Aldrich, Fig-
ure S13a–b). To mimic our experimental conditions in which
gluconate is firstly obtained (electrolysis), then H+ exchanged
(resin), and finally lyophilized, we used gluconate salt to gener-
ate gluconic acid (Figure S14a–b). A comparison of NMR spec-
tra (1H, 13C, 13C-DEPT135, 1H-13C) of gluconate before (Fig-
ure S12) and after lyophilization (Figure S14) indicates that the
resin enables effective proton exchange.
Analytical investigation of the final reaction product
FTIRS and liquid chromatography coupled with mass spec-
trometry
To validate previous results, analytical analysis by FTIRS and
liquid chromatography coupled with mass spectrometry (LC–
MS) was performed. Experiments were performed after lyophi-
lization in which a brown and sticky solid is obtained from glu-
cose electrolysis, typical of gluconic acid.[5b,10a] This was firstly
dissolved in 0.1m NaOH to compare its FTIR spectrum with
that of standards. Two FTIR spectra obtained under reproduci-
ble conditions are shown in Figure 8a. They confirm the IR
As gluconic acid is available in solution (49–53 wt% in H2O,
Sigma–Aldrich), we lyophilized it firstly to remove water. Im-
portantly, the 13C NMR spectrum of lyophilized gluconic acid
solution (Figure 9a) and that of gluconate after H+ exchange
and lyophilization (Figure S14a) are identical. The 13C NMR
spectrum of the glucose electrolysis sample shown in Fig-
ure 9b (the obtained 13C-DEPT135 spectrum is report-
ed in Figure S15) is the overlay of that from gluco-
nate after H+ exchange and glucose (the conversion
is not 100%). In particular, the 13C chemical shift of
C1 is 95.9 and 92.1 ppm for b-glucose and a-glucose,
respectively.
Quantification
from
Figure S11a
(1H NMR) shows that b/a=75:25 as expected.[18b] For
gluconate, gluconic acid, g-gluconolactone, and d-
gluconolactone, the 13C chemical shift for C1 is 178.7,
176.9, 175.8, and 173.6 ppm in agreement previous
reports.[20] ChemDraw simulations confirmed these
shifts. As mentioned above, FTIRS analysis of the
product (Figure 8a) did not disclose any gluconolac-
tone specific band (n˜ =1742 cmÀ1). Thus, we believe
that the lyophilization induces partial gluconic acid
cyclization into gluconolactone; a common phenom-
enon known as “lactonization”.[5b,20b,21] Then, glucono-
Figure 8. a) Two reproducible FTIR spectra of the product recorded in 0.1m NaOH. b) LC–
MS negative ionization mass spectrum (M-1) of the product.
bands of gluconate, which underpins the conclusion that gluc-
onate is the main reaction product (based on the conclusions
drawn from HPLIC). Furthermore, LC–MS enables us to distin-
guish different acids based on the mass of their pseudomolec-
ular ions [MÀH]À: m/z=209 (glucaric), 195 (gluconic), and 193
(glucuronic) in negative-ionization mode (Figure S10). The
spectrum from the LC–MS of the sample from glucose electrol-
ysis is shown in Figure 8b. The peak at m/z=195 belongs to
gluconic acid [MÀH]À, whereas those at m/z=129 and 391 are
attributed to the fragmentation and simple dimerization pro-
cesses. Spectra from galactose and lactose electrolyses are
shown in Figure S10d and S10 f, respectively. Consequently,
the electro-oxidation of carbohydrates on Au/C concerns exclu-
sively the C1-position and involves a two-electron process
(nex =2). HPLIC and LC–MS are suitable methods for the accu-
rate and effective determination of the final reaction products.
lactone is hydrolyzed swiftly into gluconate in H2O/NaOH
media (FTIRS conditions), which is contrary to that in D2O
media (NMR conditions) in which hydrolysis does not occur.
For the first time, this rational and methodological study re-
veals that the GOR at the Au/C catalyst does not affect its
carbon skeleton, and we assume that HPLIC peaks A and B
belong both to gluconate.
Scheme of the electrocatalytic oxidation reaction
On the basis of all of our results, we can, therefore, resume
glucose electro-oxidation in an alkaline medium on the Au
nanocatalyst (Scheme 1). The reaction starts at an electrode
potential E1 with glucose molecule adsorption through the b-
hydrogen atom of C1 (good spatial arrangement,[18b] 1*) and
yields intermediate 2* (positive bands at n˜ =1030, 1050, and
1080 cmÀ1).[18b] The adsorption takes place at an active site de-
noted by “S”. S can be Au metal (E<0.3 V vs. RHE), Au(OH)x
(E>0.3 V vs. RHE), or AuOx (E>1.2 V vs. RHE), which depends
on the electrode potential.[16] For E>0.3 V vs. RHE, the partici-
pation of the catalyst in the GOR through Au(OH)x and/or AuOx
leads to the well-known bifunctional mechanism. At E2 ꢁE1,
NMR spectroscopy
Qualitative NMR spectroscopy was performed to substantiate
previous results. We especially focused on 13C NMR and 13C dis-
tortionless enhancement by polarization transfer (DEPT135)
ChemSusChem 2016, 9, 252 – 263
www.chemsuschem.org
257
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Glucose alkaline fuel cell performances
For the DGFC, Au/C was used as the anode and Pt/C
as the cathode catalyst. Pt/C exhibits impressive du-
rability towards the ORR in an alkaline medium with
only 12% loss in its electrochemical active surface
after 1000 potential cycles from 0.05 to 1.1 V vs. RHE
(Figure S16). In addition, it shows superior ORR per-
formances than related carbon-supported electroca-
talysts prepared by chemical methods.[11d] A photo-
graph of the homemade DGFC is shown in Fig-
ure 10a, in which the AEM is fitted between the
cathodic (+) and anodic (À) compartments (see the
thin black layer).
We first examined whether our design enables
ionic conductivity by recording CV in the three-elec-
trode configuration (counter and working (together
with reference) electrodes on either side of the mem-
brane). The outcomes reported in Figure S17 show
the efficiency of the constructed cell. The assembled
fuel cell is shown in Figure 10b and c. The two black
brushes on both ends (Figure 10b) show connections
to the tiny reference electrodes (identical: SSCE).
Meanwhile, blue (anode) and red (cathode) brushes
show the electrode wiring, which enables us to dis-
play their OCP values on the corresponding multime-
ter screen (see video in SI). In 0.5m KOH (pH 13.7),
SSCE=+1005 mV vs. RHE. A bi-potentiostat (AUTO-
LAB PGSTAT302N, Netherlands) was connected to the
DGFC to monitor its OCV over time and then to
record its polarization curves. The OCP (left y axis)
and OCV (right y axis) measured over time (up to sev-
eral hours) are shown in Figure 10d, and a steady-
state was reached after a few minutes. Most studies
of DGFCs used linear sweep voltammetry (LSV) to
record polarization curves.[14,22] Basu and Basu report-
ed that both LSV at a scan rate of 1 mVsÀ1 and the
common potentiostatic discharge method (PSD;
Figure 9. 13C NMR of a) lyophilized gluconic acid solution and b) the sample from glucose
electrolysis after lyophilization.
the deprotonation of 2* leads to adsorbed gluconolactone (3*/
4*), which is quickly either desorbed (5*) or hydrolyzed into an
alkoxide 6* at E3 (ꢁE2) in the presence of hydroxides M(OH)x
(M=Au). This strong base undergoes self-rearrangement rapid-
steps of cell voltage (Ecell)) led to the same results.[22a] However,
the scheduled program for our measurements was PSD with
a step potential of 0.05 V between the OCV and 0.1 V after test-
ing both methods (Figure S18). We also examined the effect of
the supporting electrolyte with 0.1m glucose (Figure S19). De-
spite the very high OCV of 1.1 V in 0.5m NaOH compared to
the OCV of 0.90 V in 0.5m KOH, KOH allows us to deliver
a higher current density supported by Pmax =0.86 mWcmÀ2
compared to Pmax =0.70 mWcmÀ2 for the NaOH electrolyte. We
noted that this difference comes mostly from the cathode in
which the potential decreases more quickly around the range
of 0.9–0.7 V vs. RHE in 0.5m NaOH. This has been ascribed to
the contribution of Na+ and K+ ions on the ORR performances
as KOH leads to better kinetics in the potential range of 0.9–
0.8 V vs. RHE.[9a,23] During ORR studies, it is well known that the
KOH electrolyte enables higher current densities to be ob-
tained than NaOH. It should be indicated that an OCV of 1.1 V
is the best value so far reported for such DGFCs.[6a,14,17a,22,24]
We next assessed further features with the use of 0.5m
ly to yield adsorbed gluconate 8*/9* (n˜ =1412 and 1583 cmÀ1
)
via 7*. Desorption at E4 (ꢁE3) in the presence of hydroxides
M(OH)x gives gluconate in solution (10*) and, subsequently, re-
leases fresh active sites S upon which a new glucose molecule
can be adsorbed for further reaction. Depending on the pH
value, gluconic acid (11*) can be observed (n˜ =1720–
1780 cmÀ1), and possible direct pathways can exist from 5* to
10* and/or 11*. The kinetics of each step depends undoubted-
ly on experimental conditions (temperature, concentration,
etc.). Definitely, this reaction scheme could be more likely ex-
tended to other monosaccharides (e.g., galactose) than poly-
saccharides (e.g., lactose).
ChemSusChem 2016, 9, 252 – 263
www.chemsuschem.org
258
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Scheme 1. Proposed catalytic reaction cycle for glucose electro-oxidation in an aqueous medium on gold nanomaterials. S denotes the active site of gold
that can be Au metal (E<0.3 V vs. RHE), Au(OH)x (E>0.3 V vs. RHE), or AuOx (E>1.2 V vs. RHE). Ei (i=1–4) is the electrode potential; M(OH)x is the hydroxyl
species involved in the reaction (M=Au). Chemical equilibria indicated by green arrows between compounds 5*, 10*, and 11* depend on the pH of the
medium.
Figure 10. a) Photograph of the homemade fuel cell unit using an AEM fitted between the cathode (+) and anode (À) compartments (see the thin black
layer). b) Photograph of the fuel cell that shows the different wiring (close view). c) Photograph taken during fuel-cell operation: multimeters show the poten-
tials of the electrodes versus the reference electrode, SSCE (left for anode, white, and right for cathode, red). d) Left y axis: evolution over time of the anode
and cathode potentials versus RHE; right y axis: measured OCV. Anode: 20 wt% Au/C (0.18 mgAu cmÀ2); cathode 20 wt% Pt/C (0.17 mgPt cmÀ2); Fumatech
AEM. Anode: 0.5 molLÀ1 KOH+0.3 molLÀ1 glucose (deoxygenated by N2) and cathode: 0.5 molLÀ1 KOH+O2 for measurements in c and d.
KOH. DGFC performances at different concentrations are pre-
sented in Figure 11 a and b (Figure S20). The OCV is quite simi-
lar, 0.90 V (Æ20 mV); Pmax =0.86 (0.1m), 1.43 (0.2m), 2.02
(0.3m), and 1.52 mWcmÀ2 (0.4m). The value of Pmax of
2.02 mWcmÀ2 for 0.3m glucose obtained here for only
0.2 mgAu cmÀ2 at 258C is highly improved (at least twofold)
compared to 1.08 mWcmÀ2 (1.2 mgPtRu cmÀ2 at 308C),[25]
0.52 mWcmÀ2 (0.45 mgAuPtPd cmÀ2 at 308C),[22b] and 1.1 mWcmÀ2
(0.6 mgAu cmÀ2 at 308C);[22c] except the unpredictable value of
20 mWcmÀ2 reached with an unrealistic metal loading of
3 mgPt cmÀ2 (0.5m glucose, 308C).[14] Here the metal loading is
15-fold lower, and our Au/C catalyst is more stable than the Pt
catalyst used there.
Furthermore, with the increase of the glucose concentration,
higher current and power densities are expected because of
the presence of a greater number of reacting molecules. This
is the case for glucose concentrations up to 0.3m. However,
experimental results herein (Figure 11 a) and reported previous-
ly[8,22a,b,26] show that a too high glucose concentration is ac-
companied by a decrease in the cell performance. We believe
that there are at least two rational reasons that explain the
output power decrease for a certain limit of glucose concentra-
ChemSusChem 2016, 9, 252 – 263
www.chemsuschem.org
259
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Figure 11. a) Fuel cell polarization curves for different concentrations of glucose in terms of cell voltage (ECell: left y axis) and power density (P: right y axis).
b) Effect of glucose concentration of the OCV (right y axis) and maximum power density (Pmax: left y axis). c) Typical behavior of the anode and cathode poten-
tials recorded during the polarization curve measurements (0.3m glucose). Anode: 20 wt% Au/C (0.18 mgAu cmÀ2); cathode 20 wt% Pt/C (0.17 mgPt cmÀ2); Fu-
matech AEM. Anode: 0.5 molLÀ1 KOH+glucose (deoxygenated by N2); cathode: 0.5 molLÀ1 KOH+O2.
tion. The first is the coverage rate of the electrode surface pro-
vided by adsorbed reactive molecules to lead to a competition
between glucose molecules and hydroxyl ions. Indeed, the in-
crease of the fuel concentration results in a saturation of the
anode catalyst surface by excessively adsorbed glucose mole-
cules. As the presence of adsorbed hydroxyl (OHads) species is
crucial for glucose oxidation,[16] the difficulty in their adsorption
on the active sites may diminish the cell performance signifi-
cantly. Similarly, an increase of the electrolytic solution (NaOH
or KOH) concentration will provide more reacting OHÀ species
and enhance the kinetics of the glucose oxidation reaction
(Figure S18d). However, a too high concentration of electrolyte
(>1m) will lead to the degradation of the AEM and a decrease
of the number of sites available for substrate (glucose mole-
cules) adsorption at the anode.[22a,26c] The second reason is re-
lated to the mass transport phenomenon: remove reaction
products and route reactants to catalytic sites. It is known that
the mass-transport phenomenon is one of the most limiting
parameters for a fuel cell that uses an organic molecule as the
fuel. The principle of mass transport, also referred to as a diffu-
sion-limiting process, consists of bringing the reacting species
close enough to the surface of the catalyst and removing the
species formed at the surface into the bulk of the solution. Ini-
tially, an increase of the fuel concentration is accompanied by
an increase in cell performances (Figure 11 a). However, a too
high glucose concentration will lead to poor mass transport
that may reduce the cell performances significantly. An in-
crease of the fuel concentration increases the solution viscosity
that can limit the transport of the fuel through the solution
and catalytic layer toward the active sites at which the oxida-
tion reaction occurs. With a molecule such as glucose, which
has a low diffusivity (D=6.9”10À10 m2 sÀ1),[22a] especially in
a batch cell in which no product outlet is provided, the de-
crease in cell performance with an increase in the glucose con-
centration after a certain limit is clear.[8,22a,b,26] Even if a flow
cell is used, a decrease of the cell performances at high con-
centration is usually observed, and the validity of the two
aforementioned hypotheses has been verified for a wide range
of systems based on organic molecules (formic acid, methanol,
ethanol, glycerol, etc.).[7e,9b,18c,27]
Behavioral changes in the anode (EA) and cathode (EC) elec-
trode potentials when the fuel cell delivers current were scruti-
nized thanks to our implanted SSCE reference electrodes. The
recorded values for 0.3m glucose are depicted in Figure 11 c.
From Ecell =0.89 (OCV) to 0.1 V, EC goes from 1.03 to 0.79 V vs.
RHE (DEC =0.24 V), and EA undergoes a prominent increase
from 0.13 to 0.59 V vs. RHE (DEA =0.46 V). Such results denote
that performances are limited mostly by the anodic reaction as
expected as the designed DGFC operates under batch condi-
tions without fuel flow and the prompt removal of reaction
products from electrodes. The future design of a flow-cell
system might decrease the mass-transport limitation and lead
to a better interaction between the reagents and the electrode
surfaces.
Conclusions
In this work, we aimed towards the electrochemical valoriza-
tion of carbohydrates by using a direct alkaline fuel cell
(DGFC). We scrutinized the fundamental aspects of the anode
electrocatalytic reaction thoroughly by coupling electrochemi-
cal (cyclic voltammetry, chronoamperometry, electrolysis) to
spectroscopic (cyclic voltammetry coupled with IR spectrosco-
py, NMR), spectrometric (LC–MS), and chromatographic (high-
performance liquid ionic chromatography) methods. Au/C was
the anode electrode material, and Pt/C was used on the catho-
dic side. Both were made by the “bromide anion exchange”
approach with a good yield (>90%). Characterization (TEM, X-
ray photoelectron spectroscopy, XRD) reveals well-dispersed
and unoxidized metal nanoparticles on the support. The judi-
cious combination of (electro)analytical techniques allowed the
unambiguous determination of the reaction product over the
Au/C catalyst. Glucose was oxidized selectively without CÀC
bond cleavage to gluconate (which can be altered into glucon-
ic acid on ion-exchange resin) contrary to previous reports.
The Faradaic efficiency is nearly 100%. With the proposed reac-
tion scheme, this work is a crucial step towards the electrosyn-
thesis of organics without restrictive conditions as in organic
chemistry synthesis.
ChemSusChem 2016, 9, 252 – 263
www.chemsuschem.org
260
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
In 0.5m KOH and 0.3m glucose, the DGFC has an open-cir-
cuit voltage of 0.9 V and delivers a maximum power of
2.02 mWcmÀ2, which is at least twofold higher than the report-
ed counterparts. Our findings underline that the fuel cell per-
formances are limited by phenomena that occur at the anode,
especially mass transport. Finally, the efficiency of the electro-
chemical production of energy and chemicals was demonstrat-
ed successfully by a high-output-power DGFC in which glucose
is oxidized selectively at the anode. We are highly confident
that upcoming improvements in cell design, namely, size, flow
conditions, low surface resistance in a suitable membrane-elec-
trode-assembly, and operating temperature, will lead to cur-
rent and power increases for wide stationary applications. This
work continues the research in which efforts are devoted to
develop advanced, environmentally friendly energy converters
and emphasizes that the cogeneration of clean energy and
chemicals is utterly doable.
(Spectro)electrochemical and analytical measurements
The catalyst ink was prepared by mixing MQ water (375 mL) and
Nafion suspension (50 mL) in an ultrasonic bath (water). Then, cata-
lyst powder (4 mg) was added to obtain a homogeneous ink
(delay depends on the apparatus[30]).
CV: Electrochemical CV tests were conducted by using a conven-
tional three-electrode cell using a potentiostat EG&G PARC Model
362 (Princeton Applied Research). The reference electrode was
a RHE. The working electrode consisted of 3 mL catalyst ink depos-
ited onto a well-polished glassy carbon disk (GC: 0.071 cm2)
through an abrasive disk with alumina powders of 1, 0.3, and
0.05 mm. A slab of GC (6.48 cm2) was used as the counter electrode.
NaOH (97%) and glucose (d-(+)-glucose, 99.5%) from Sigma–Al-
drich were used as the electrolyte solution and fuel, respectively.
Spectroelectrochemical measurements: Details of in situ FTIRS
measurements by using a Bruker IFS 66v spectrometer are de-
scribed elsewhere.[31] A slab of GC and RHE served as the counter
and reference electrodes, respectively. The working electrode con-
sisted of 3 mL of catalyst ink deposited onto a GC disk (8 mm diam-
eter). The amount of Nafion in the ink was halved to reduce its in-
terference (total volume unchanged). A small amount of ink was
thus deposited to avoid reducing the IR beam with carbon black
absorption. Excellent reflectivity was obtained by pressing the
working electrode against the CaF2 window to obtain a thin layer
of electrolytic solution. Spectroelectrochemical analyses consist of
coupling either CV to FTIRS (SPAIRS) or CA to FTIRS. CV-FTIRS con-
Experimental Section
Preparation of the nanocatalysts
The nanocatalysts used for the electrochemical conversion of glu-
cose were prepared according to the BAE method[28] using carbon
black (Vulcan XC 72R, Cabot, pretreated thermally to boost the
electrocatalytic properties of the nanoparticles[29]) as the support
(thereafter denoted as Au/C) for a targeted metal loading of
20 wt%. Basically, to prepare 100 mg of catalyst, tetrachloroauri-
c(III) acid trihydrate (HAuCl4·3H2O; 40.0 mg; Sigma–Aldrich,
ꢁ99.9%) was dissolved in a reactor that contained ultrapure water
(100 mL; MQ: Milli-Q Millipore, 18.2 MW cm at 293 K) at 258C
under stirring. This was followed by the addition of KBr (17.8 mg;
Sigma–Aldrich, ꢁ99%) under vigorous stirring. Then, carbon black
(80 mg) was added under constant ultrasonic homogenization for
45 min. Afterwards, the metal salt was reduced by the addition of
sodium borohydride (15 mL, 0.1 molLÀ1 NaBH4; Sigma–Aldrich,
99%) and the mixture was stirred vigorously at 408C for 2 h. Final-
ly, the carbon-supported gold nanoparticles were collected by fil-
tration, washed several times with MQ water, and dried in an oven
at 408C for 12 h. The nanocatalyst was prepared with a synthesis
yield of 94%. The Pt/C catalyst used as cathode electrode material
was prepared in the same way using hexachloroplatinic(IV) acid
hexahydrate (H2PtCl6·6H2O; 53.1 mg; Sigma–Aldrich, ꢁ37.50% Pt
basis) as metal precursor. The synthesis yield was 91% for the Pt/C
catalyst.
sists of recording the electrode reflectivity RE at different potentials
i
Ei in steps of 0.05 V at 1 mVsÀ1, whereas CA-FTIRS concerns spec-
trum acquisition every 3 min at a set electrode potential. A setup
of the m-AUTOLAB Type III (Metrohm Autolab BV, Netherlands) po-
tentiostat was used for electrochemistry and OPUS software
(Bruker) was used for IR spectroscopy.
Electrolysis and chromatographic analyses: Electrolysis was per-
formed by using a Pyrex two-compartment cell[19a] separated with
a AEM (Fumasep FAA, Fumatech). The AEM preventing the con-
tents from mixing from the two compartments and provides the
current relay through ion exchange between the auxiliary elec-
trode compartment (GC: 1.5 cm”7.2 cm) and that of the working
electrode. The latter is a square plate (2 cm side) of Carbon Paper
(Spectracarb 2050L-1050; Fuel Cell Store, TX), similar to Toray
Carbon Paper 090. Ink (50 mL) was deposited onto each face of the
electrode. A RHE was used as the reference electrode, which was
in contact with the working electrode compartment by a Luggin
bridge. This compartment was filled with 43 mL of solution, which
was stirred slightly during electrolysis by a bar magnet located at
the bottom of the solution. After preliminary tests, electrolyzes
were finally performed with a CA program by using a potentiostat
EG&G PARC Model 362. This choice was motivated by the risk of
having contaminants in the final products, some of which come
from the catalytic surface regeneration at high potential. At the
end of the electrolysis, the collected sample was divided into two
parts. The first part was analyzed immediately by HPLIC (Dionex
system ISC 5000) in gradient elution with a conductivity detector
(CD-5000: allows the elution by conductivity strength) and an am-
perometric detector (ED-5000: allows in situ electrolysis for qualita-
tive and quantitative analyses). The HPLIC included an autosampler
(AS50 Automated Sample Injector), a sample loop (20 mL), and
a column 2”250 mm (IonPac AS15), which operated at 308C.
A constant flow of eluent (0.3 mLminÀ1 at pressure of ꢀ1500 psi)
was provided by a pump (ICS-5000 P). The used eluent was
Nanomaterials characterization
The obtained nanomaterials were characterized physicochemically
by TGA (real metal loading; by using a TA Instruments SDT Q-600
apparatus), XRD (crystallographic structure and crystallites size; by
using an EMPYREAN (PANanalytical) diffractometer in Bragg–Bren-
tano q–q configuration), TEM (morphology, particles size disper-
sion; by using a TEM/STEM JEOL 2100 UHR microscope at 200 kV),
TEM coupled to EDX spectroscopy (elemental analysis; by using
a JED Series AnalysisProgram, JEOL), and XPS (oxidation state of
the surface; by using a Kratos Axis Ultra DLD spectrometer). Most
of these techniques are described elsewhere.[29]
ChemSusChem 2016, 9, 252 – 263
www.chemsuschem.org
261
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
a 10 mmolLÀ1 NaOH solution prepared from a concentrated stock
solution (Acros Organics: 50 wt%, density=1.5). The HPLIC was
controlled by Chromeleon 6.80 software.
[7] a) Y.-J. Wang, J. Qiao, R. Baker, J. Zhang, Chem. Soc. Rev. 2013, 42, 5768–
5787; b) H. Zhang, P. K. Shen, Chem. Rev. 2012, 112, 2780–2832; c) J. R.
Varcoe, P. Atanassov, D. R. Dekel, A. M. Herring, M. A. Hickner, P. A. Kohl,
A. R. Kucernak, W. E. Mustain, K. Nijmeijer, K. Scott, T. Xu, L. Zhuang,
Energy Environ. Sci. 2014, 7, 3135–3191; d) E. H. Yu, U. Krewer, K. Scott,
Energies 2010, 3, 1499; e) R. Pathak, S. Basu, Electrochim. Acta 2013, 113,
42–53.
[8] A. Ilie, M. Simoes, S. Baranton, C. Coutanceau, S. Martemianov, J. Power
Sources 2011, 196, 4965–4971.
[9] a) A. Wieckowski, E. R. Savinova, C. G. Vayenas, Catalysis and Electrocatal-
ysis at Nanoparticle Surfaces, Marcel Dekker, Inc., New York, 2003; b) H.
Corti, E. R. Gonzalez, Direct Alcohol Fuel Cells, Springer Netherlands, Dor-
drecht, 2014.
Analytical analyses. Lyophilization, LC–MS, and NMR spectrosco-
py: The second part of the electrolysis solution was passed
through a cationic resin (DOWEX 50WX8-200 ion-exchange resin;
Sigma–Aldrich) to remove salts and convert carboxylates to their
protonated forms. The final filtrate was frozen and lyophilized
(Freeze PowerDry LL 3000; Thermo Scientific) to remove water by
sublimation. The LC–MS used an ion-exclusion column (HPX-87H,
30 cm, Aminex) in isocratic elution at 0.6 mLminÀ1 (solvent: water
with 0.2% formic acid) and a mass spectrometer Xevo Q-TOF from
Waters (software: MassLinx, Waters) in negative-ionization mode
(cone voltage: 20–40 V). NMR spectra were recorded at 297 K in
D2O solvent by using a 400 MHz (1H) or 100 MHz (13C) instrument,
employing standard software provided by the manufacturer
(Bruker). Various techniques were used to improve assignments:
[10] a) P. M. Collins, Dictionary of Carbohydrates, 2nd ed., Chapman & Hall/
CRC, London, 2006; b) R. K. Thauer, K. Jungermann, K. Decker, Bacteriol.
Rev. 1977, 41, 100–180.
[11] a) M. K. Debe, Nature 2012, 486, 43–51; b) L. Su, W. Jia, C.-M. Li, Y. Lei,
ChemSusChem 2014, 7, 361–378; c) A. Anastasopoulos, J. C. Davies, L.
Hannah, B. E. Hayden, C. E. Lee, C. Milhano, C. Mormiche, L. Offin, Chem-
SusChem 2013, 6, 1973–1982; d) Y. Holade, N. Sahin, K. Servat, T. Nap-
porn, K. Kokoh, Catalysts 2015, 5, 310–348; e) J. Greeley, I. E. L. Ste-
phens, A. S. Bondarenko, T. P. Johansson, H. A. Hansen, T. F. Jaramillo, J.
Rossmeisl, I. Chorkendorff, J. K. Nørskov, Nat. Chem. 2009, 1, 552–556;
f) J. Kibsgaard, Y. Gorlin, Z. Chen, T. F. Jaramillo, J. Am. Chem. Soc. 2012,
134, 7758–7765; g) Y. Zhang, Y.-C. Hsieh, V. Volkov, D. Su, W. An, R. Si, Y.
Zhu, P. Liu, J. X. Wang, R. R. Adzic, ACS Catal. 2014, 4, 738–742.
[12] L. Mor, Z. Rubin, P. Schechner, J. Fuel Cell Sci. Technol. 2008, 5, 014503–
014503.
1
1H, 13C, 1H-1H (COSY), H-13C (HSQC), and DETP 135 NMR.
Direct glucose alkaline fuel cell design
A homemade two-compartment Teflon cell (with the same AEM as
for electrolysis) operated under batch conditions was designed for
DGFC tests (Figure 10). The electrode was a 1 cm2 square plate
(using both sides: 2 cm2) of Carbon Paper. Catalyst ink was pre-
pared similarly to that used for CV. The anode catalyst was 20 wt%
Au/C (0.18 mgAu cmÀ2), and the cathode catalyst was 20 wt% Pt/C
(0.17 mgPt cmÀ2). “BASi Reference Electrodes, 3 cm” (RE-6 Ag/AgCl,
MF-2078) purchased from Alvatek Ltd (UK) with a flexible wire con-
nector were used to follow each electrode potential.
[13] a) E. Katz, K. MacVittie, Energy Environ. Sci. 2013, 6, 2791–2803; b) W.
Gellett, M. Kesmez, J. Schumacher, N. Akers, S. D. Minteer, Electroanalysis
2010, 22, 727–731; c) D. P. Hickey, F. Giroud, D. W. Schmidtke, D. T.
Glatzhofer, S. D. Minteer, ACS Catal. 2013, 3, 2729–2737.
[14] N. Fujiwara, S.-i. Yamazaki, Z. Siroma, T. Ioroi, H. Senoh, K. Yasuda, Elec-
trochem. Commun. 2009, 11, 390–392.
ˇ
[15] K. Juodkazis, J. Juodkazyt, V. Jasulaitien, A. Lukinskas, B. Sebeka, Electro-
chem. Commun. 2000, 2, 503–507.
ˇ ´
[16] a) M. W. Hsiao, R. R. Adzic, E. B. Yeager, J. Electrochem. Soc. 1996, 143,
759–767; b) M. Pasta, R. Ruffo, E. Falletta, C. M. Mari, C. Della Pina, Gold
Bull. 2010, 43, 57–64; c) S. HebiØ, L. Cornu, T. W. Napporn, J. Rousseau,
B. K. Kokoh, J. Phys. Chem. C 2013, 117, 9872–9880; d) S. HebiØ, Y.
Holade, K. Maximova, M. Sentis, P. Delaporte, K. B. Kokoh, T. W. Napporn,
A. V. Kabashin, ACS Catal. 2015, 5, 6489–6496.
Acknowledgements
This research was supported by the French National Research
Agency (ANR) in the framework of the “ChemBio-Energy program
2012–2015” including IC2MP and EBI. The authors are grateful to
Dr. JØrôme DØsirØ (University of Poitiers) for useful discussions
during NMR analyses.
[17] a) D. Basu, S. Basu, Int. J. Hydrogen Energy 2011, 36, 14923–14929; b) C.
Zhu, S. Guo, S. Dong, Adv. Mater. 2012, 24, 2326–2331; c) L. Yan, A.
Brouzgou, Y. Meng, M. Xiao, P. Tsiakaras, S. Song, Appl. Catal. B 2014,
150–151, 268–274.
[18] a) P. Tonda-Mikiela, T. W. Napporn, C. Morais, K. Servat, A. Chen, K. B.
Kokoh, J. Electrochem. Soc. 2012, 159, H828–H833; b) F. Largeaud, K. B.
Kokoh, B. Beden, C. Lamy, J. Electroanal. Chem. 1995, 397, 261–269;
c) Y.-L. Yang, X.-H. Liu, M.-Q. Hao, P.-P. Zhang, Int. J. Hydrogen Energy
2015, 40, 10979–10984; d) K. B. Kokoh, J. M. LØger, B. Beden, H. Huser,
C. Lamy, Electrochim. Acta 1992, 37, 1909–1918.
[19] L. H. E. Yei, B. Beden, C. Lamy, J. Electroanal. Chem. Interfacial Electro-
chem. 1988, 246, 349–362.
[20] a) V. H. Pomin in Glycosylation (Ed.: S. Petrescu), INTECH, Rijeka, 2012,
pp. 63–98; b) J.-C. Portais, P. Tavernier, I. Besson, J. Courtois, B. Courtois,
J.-N. Barbotin, FEBS Lett. 1997, 412, 485–489; c) C. S. Tsai, H. G. Ye, J. L.
Shi, Arch. Biochem. Biophys. 1995, 316, 155–162.
[21] a) Z. Zhang, P. Gibson, S. Clark, G. Tian, P. Zanonato, L. Rao, J. Solution
Chem. 2007, 36, 1187–1200; b) Y. Pocker, E. Green, J. Am. Chem. Soc.
1973, 95, 113–119.
[22] a) D. Basu, S. Basu, Electrochim. Acta 2011, 56, 6106–6113; b) D. Basu, S.
Basu, Int. J. Hydrogen Energy 2012, 37, 4678–4684; c) L. Li, K. Scott, E. H.
Yu, J. Power Sources 2013, 221, 1–5.
Keywords: carbohydrates
conversion · fuel cells · heterogeneous catalysis
·
electrochemistry
·
energy
[1] a) B. C. H. Steele, A. Heinzel, Nature 2001, 414, 345–352; b) J. Greeley,
N. M. Markovic, Energy Environ. Sci. 2012, 5, 9246–9256.
[2] a) S. Chu, A. Majumdar, Nature 2012, 488, 294–303; b) M. Bellini, M.
Bevilacqua, J. Filippi, A. Lavacchi, A. Marchionni, H. A. Miller, W. Ober-
hauser, F. Vizza, S. P. Annen, H. Grützmacher, ChemSusChem 2014, 7,
2432–2435.
[3] a) G. D. Watt, Renewable Energy 2014, 72, 99–104; b) J. McGinley, F. N.
McHale, P. Hughes, C. N. Reid, A. P. McHale, Biotechnol. Lett. 2004, 26,
1771–1776.
[4] A. M. Ruppert, K. Weinberg, R. Palkovits, Angew. Chem. Int. Ed. 2012, 51,
2564–2601; Angew. Chem. 2012, 124, 2614–2654.
[5] a) I. Dencic, J. Meuldijk, M. Croon, V. Hessel, J. Flow Chem. 2011, 1, 13–
23; b) S. Ramachandran, P. Fontanille, A. Pandey, C. Larroche, Food Tech-
´
[23] a) N. M. Markovic, P. N. Ross, Surf. Sci. Rep. 2002, 45, 117–229; b) N. M.
ˇ ´
´
nol. Biotechnol. 2006, 44, 185–195; c) M. Varnicic, T. Vidakovic-Koch, K.
Sundmacher, Electrochim. Acta 2015, 174, 480–487; d) P. Qi, S. Chen, J.
Chen, J. Zheng, X. Zheng, Y. Yuan, ACS Catal. 2015, 5, 2659–2670.
[6] a) E. H. Yu, X. Wang, U. Krewer, L. Li, K. Scott, Energy Environ. Sci. 2012,
5, 5668–5680; b) N. Fujiwara, Z. Siroma, S.-i. Yamazaki, T. Ioroi, H.
Senoh, K. Yasuda, J. Power Sources 2008, 185, 621–626.
´
Markovic, H. A. Gasteiger, P. N. Ross, J. Phys. Chem. 1996, 100, 6715–
6721; c) M. Nesselberger, S. Ashton, J. C. Meier, I. Katsounaros, K. J. J.
Mayrhofer, M. Arenz, J. Am. Chem. Soc. 2011, 133, 17428–17433.
[24] C.-C. Chen, C.-L. Lin, L.-C. Chen, J. Power Sources 2015, 287, 323–333.
[25] D. Basu, S. Basu, Electrochim. Acta 2010, 55, 5775–5779.
ChemSusChem 2016, 9, 252 – 263
www.chemsuschem.org
262
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
[26] a) R. L. Arechederra, B. L. Treu, S. D. Minteer, J. Power Sources 2007, 173,
156–161; b) Y. S. Li, T. S. Zhao, Z. X. Liang, J. Power Sources 2009, 187,
387–392; c) C. Rice, S. Ha, R. I. Masel, P. Waszczuk, A. Wieckowski, T. Bar-
nard, J. Power Sources 2002, 111, 83–89.
[27] a) C. A. Apblett, D. Ingersoll, S. Sarangapani, M. Kelly, P. Atanassov, J.
Electrochem. Soc. 2010, 157, B86–B89; b) F. M. Cuevas-MuÇiz, M. Guerra-
Balcµzar, J. P. Esquivel, N. SabatØ, L. G. Arriaga, J. Ledesma-García, J.
Power Sources 2012, 216, 297–303.
[29] Y. Holade, C. Morais, K. Servat, T. W. Napporn, K. B. Kokoh, Phys. Chem.
Chem. Phys. 2014, 16, 25609–25620.
[30] B. G. Pollet, Electrocatalysis 2014, 5, 330–343.
[31] Y. Holade, C. Morais, K. Servat, T. W. Napporn, K. B. Kokoh, ACS Catal.
2013, 3, 2403–2411.
Received: December 1, 2015
[28] Y. Holade, C. Morais, S. Arrii-Clacens, K. Servat, T. W. Napporn, K. B.
Kokoh, Electrocatalysis 2013, 4, 167–178.
Published online on January 18, 2016
ChemSusChem 2016, 9, 252 – 263
www.chemsuschem.org
263
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim