Thermal Behavior of d-Ribose Adsorbed on Silica: Effect of Inorganic Salt Coadsorption and Significance for Prebiotic Chemistry

Article abstract of DOI:10.1002/chem.201601418

Understanding ribose reactivity is a crucial step in the “RNA world” scenario because this molecule is a component of all extant nucleotides that make up RNA. In solution, ribose is unstable and susceptible to thermal destruction. We examined how ribose behaves upon thermal activation when adsorbed on silica, either alone or with the coadsorption of inorganic salts (MgCl₂, CaCl₂, SrCl₂, CuCl₂, FeCl₂, FeCl₃, ZnCl₂). A combination of¹³C NMR, in situ IR, and TGA analyses revealed a variety of phenomena. When adsorbed alone, ribose remains stable up to 150 °C, at which point ring opening is observed, together with minor oxidation to a lactone. All the metal salts studied showed specific interactions with ribose after dehydration, resulting in the formation of polydentate metal ion complexes. Anomeric equilibria were affected, generally favoring ribofuranoses. Zn²⁺stabilized ribose up to higher temperatures than bare silica (180 to 200 °C). Most other cations had an adverse effect on ribose stability, with ring opening already upon drying at 70 °C. In addition, alkaline earth cations catalyzed the dehydration of ribose to furfural and, to variable degrees, its further decarbonylation to furan. Transition-metal ions with open d-shells took part in redox reactions with ribose, either as reagents or as catalysts. These results allow the likelihood of prebiotic chemistry scenarios to be evaluated, and may also be of interest for the valorization of biomass-derived carbohydrates by heterogeneous catalysis.

Full text of DOI:10.1002/chem.201601418

DOI: 10.1002/chem.201601418
Full Paper
&
Prebiotic Chemistry
Thermal Behavior of d-Ribose Adsorbed on Silica: Effect of
Inorganic Salt Coadsorption and Significance for Prebiotic
Chemistry
Mariame Akouche,[a] Maguy Jaber,[b] Emilie-Laure Zins,[c] Marie-Christine Maurel,[d] Jean-
Francois Lambert,*[a] and Thomas Georgelin*[a]
Abstract: Understanding ribose reactivity is a crucial step in
the “RNA world” scenario because this molecule is a compo-
nent of all extant nucleotides that make up RNA. In solution,
ribose is unstable and susceptible to thermal destruction.
We examined how ribose behaves upon thermal activation
when adsorbed on silica, either alone or with the coadsorp-
tion of inorganic salts (MgCl2, CaCl2, SrCl2, CuCl2, FeCl2, FeCl3,
ZnCl2). A combination of 13C NMR, in situ IR, and TGA analy-
ses revealed a variety of phenomena. When adsorbed alone,
ribose remains stable up to 1508C, at which point ring open-
ing is observed, together with minor oxidation to a lactone.
All the metal salts studied showed specific interactions with
ribose after dehydration, resulting in the formation of poly-
dentate metal ion complexes. Anomeric equilibria were af-
fected, generally favoring ribofuranoses. Zn2+ stabilized
ribose up to higher temperatures than bare silica (180 to
2008C). Most other cations had an adverse effect on ribose
stability, with ring opening already upon drying at 708C. In
addition, alkaline earth cations catalyzed the dehydration of
ribose to furfural and, to variable degrees, its further decar-
bonylation to furan. Transition-metal ions with open d-shells
took part in redox reactions with ribose, either as reagents
or as catalysts. These results allow the likelihood of prebiotic
chemistry scenarios to be evaluated, and may also be of in-
terest for the valorization of biomass-derived carbohydrates
by heterogeneous catalysis.
Introduction
RNA is a polymer of nucleotides, themselves composed of
phosphates, nucleobases and a specific pentose carbohydrate,
namely d-ribose. Therefore, studies of ribose synthesis, of its
conditions of stability, and of its further reactions such as phos-
phorylation and glycosylation are particularly interesting to
explain the formation of RNA.
In the prebiotic scenario of the “RNA world”,[1] RNA oligomers
are considered as the first biological molecules to have
emerged. This idea is based on the fact that some RNAs are
the only biopolymers to possess both information capability
and catalytic properties, which are currently observed in sever-
al ribozymic motifs and species such as viro¨ıds. Combining
both abilities in the same molecule eschews the “chicken-and-
egg” dilemma that would be implied if the choice were
between DNA and proteins.
Ribose can be synthesized under abiotic conditions together
with other carbohydrates from the formose reaction,[2] based
on the polymerization of formaldehyde; in particular, its syn-
thesis has been reported in analogues of interstellar ices.[3]
Once ribose is formed, several pathways could lead to nucleo-
sides, and recent investigations have shown good yields and
selectivities.[4] Yet these scenarios are plagued by several prob-
lems. First, ribose is often a rather minor product of the for-
mose reaction.[5] Second, assuming it can be synthesized, the
ribose molecule is quite unstable in aqueous solution.[6] For ex-
ample, at pH 9 and 608C, its half-life is about 50 h; closer to
physiological conditions, at pH 7 and 378C, it should be
around 500 h if one extrapolates from the data of Miller and
co-workers.[6] Then, as in many prebiotic scenarios, there is a
selectivity problem. Not only does life use only d-ribose, ex-
cluding l-ribose, but among the four anomeric forms of this
molecule (cf. Scheme 1), nucleotides and RNA exclusively use
the b-d-ribofuranose form. In aqueous solutions, in contrast,
this form is not predominent, typically accounting for 12% of
all ribose in solution.[7]
[a] M. Akouche, J.-F. Lambert, T. Georgelin
Sorbonne Universitꢀs, UPMC Univ Paris 06 and CNRS UMR 7197, LRS case
courrier 178, UPMC 4 Pl. Jussieu, 75252 PARIS CEDEX 05 (France)
E-mail: thomas.georgelin@upmc.fr
jean-francois.lambert@upmc.fr
[b] M. Jaber
Sorbonne Universitꢀs, UPMC Univ Paris 06 and CNRS UMR 8220, LAMS,
case courrier 225, UPMC 4 Pl. Jussieu, 75252 Paris CEDEX 05 (France)
[c] E.-L. Zins
Sorbonne Universitꢀs, UPMC Univ Paris 06 and CNRS UMR 8233, MONARIS,
case courrier, UPMC 4 Pl. Jussieu, 75252 Paris CEDEX 05 (France)
[d] M.-C. Maurel
UMR 7205- ISyEB, CNRS-MNHN-UPMC Univ Paris 06, 75005 Paris (France)
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/chem.201601418. It contains further experimental
details (tables and figures).
Chem. Eur. J. 2016, 22, 1 – 14
1
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
drates[15] but even in this case many questions remain unan-
swered because these studies are often concerned only with
applications to the analysis of carbohydrate mixtures. Studies
of sugars in acidic solutions or brines are also relevant because
they illustrate some of the possible reactions of carbohydrates,
from ring opening to the formation of furfural by dehydration
when the activity of water is low enough. Lactones (cyclic
esters) can also be formed but they require an oxidative step:
the sugar ring is opened and isomerized to the enediol form,
and the latter is oxidized to a linear carboxylic acid, which pro-
duces the lactone by internal condensation with recyclization.
In the present study we evaluate the stability of ribose
under thermal activation and analyze the chemical behavior of
ribose on silica surfaces, both alone and in the presence of in-
organic salts. In a prebiotic setting, the deposition of ribose
from an aquifer, presumably upon drying, could have been ac-
companied by the co-deposition of minerals from sea water.
We chose not to attempt to reproduce a realistic composition
of primordial seawater, but to focus on simple systems consist-
Scheme 1. Ribose polymorphs.
Recent studies have attempted to solve the stability and the
selectivity problems by invoking the formation of borate or sili-
cate complexes in solution.[2,8] These complexes result from
the formation of two covalent bonds between the oxygen
atoms of two -OH groups of the ribose molecule and the
boron (resp. silicon) atom. It has been observed that such
a complexation stabilizes ribose, and furthermore, preferably
its furanose forms. Theoretical results have shown that indeed
silicate/ribose complexes should be formed exclusively with
the furanose form, because its HO-C-C-OH dihedral angle is
sufficiently small to allow the formation of a planar five-
membered ring.[8c]
ing of ribose and the chloride salt of Mg2+, Ca2+, Sr2+, Fe2+
,
Fe3+, Cu2+, and Zn2+. Mg2+ is well-known for its role in the co-
ordination chemistry of nucleotides. Compared with the other
alkaline earth cations (Ca2+ and Sr2+), Mg2+ can provide infor-
mation on the role of Lewis acidity, which increases when
going up the alkaline earth column in the periodic table. Fur-
thermore, Mg2+ and Ca2+ easily leach from meteoritic miner-
als.[16] The other four elements are transition metals that could
have leached from the prebiotically important sulfide miner-
als.[17] Fe2+, Fe3+, and Cu2+ have open (nd) subshells, allowing
them to take part in redox reactions.
The silicate or borate scenarios have shown the potential of
considering inorganic/organic interactions to study the origins
of life. They deal with interactions in solution; another direc-
tion of research is the interaction of small biological molecules
with surfaces of inorganic materials, i.e., interfacial chemistry. It
was first proposed in 1951[9] that mineral surfaces could have
played a role in the emergence of life. Later, experimental
studies showed that, for example, silica and clay surfaces can
indeed induce prebiotically interesting reactions such as the
formation of peptide bonds between amino acids,[10] which is Materials and Methods
related to the capacity for protein adsorption,[11] or the oligo-
Adsorption procedures
merization of nucleotides.[12] Our team has demonstrated the
ability of silica and other surfaces to promote the oligomeriza-
tion of amino acids to peptides or of monophosphate to poly-
phosphates.[13] We have also recently reported that the silica
surface stabilizes d-ribose upon drying, and causes a small but
significant selection of the furanose form.[14]
Ribose adsorption was carried out by impregnation methods.
In this procedure, 300 mg of amorphous silica (380 m2 gÀ1) was
wetted with 4 mL of a ribose solution. The ribose concentra-
tions were chosen so as to provide 5.0, 10.0, and 15.0
weight% ribose loading with respect to the silica support. The
pastes obtained after wetting of the silica were stirred at room
temperature for 3 h and then dried overnight at 708C in an
oven under room humidity. Samples obtained after drying will
be denoted as 5% ribose/SiO2, etc. “ribose/SiO2”, with the per-
centage value omitted, will refer to the “basis” sample, that is,
to the 10% loading.
The present work intends to expand these initial results by
focusing on the thermal reactivity of ribose adsorbed on amor-
phous silica (SiO2). Silica represents a realistic model of mineral
phases existing on the primitive earth such as cherts; it is also
a desirable support from the point of view of the ease of char-
acterization by such techniques as IR, NMR, and TGA.[13h] Our
goal is to study ribose reactions of prebiotic interest such as
glycosylation or phosphorylation on the surface of silica. The
study of the thermal reactivity of ribose is a prerequisite for
such studies because the reagents should not be destroyed
under the conditions tested.
Experiments were also carried out in which ribose was coad-
sorbed with divalent metal salts. 10.0 weight% of ribose per
silica weight were coadsorbed with metal chloride, using
ribose/metal molar ratios of either 1:1 or 2:1. The samples
were equilibrated and dried in the same way as for the
ribose/SiO2 series. Dried samples will be called ribose-metal
(x:1)/SiO2; when the molar ratio is omitted, it is understood to
be equal to (1:1).
To our knowledge, there is no information about the thermal
reactivity of pentoses adsorbed on mineral surfaces. There
exist some studies on the thermal reactivity of bulk carbohy-
&
&
Chem. Eur. J. 2016, 22, 1 – 14
www.chemeurj.org
2
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
Analytical techniques
X-ray powder diffraction (XRD) was carried out with a Bruker
D8 Avance diffractometer using CuKa radiation (wavelength
l=1.5404 ꢁ). XRD patterns were recorded between 38 and 708
with a step size of 0.058.
TGA of the samples was carried out with a TA Instruments
Waters LLC, with a SDT Q600 analyzer, using a heating rate b=
58CminÀ1 under dry air flow (100 mLminÀ1). For silica-support-
ed samples, masses were normalized to the residual mass at
8008C, which corresponds to anhydrous silica without organic
matter. All the weight losses assigned to transformations of
supported organic molecules were quantified between limits
determined from the DTG traces, and corrected for the weight
loss of the bare support in the same temperature range.
Infrared (IR) spectra of solid samples were recorded in the
transmission mode on self-supported pellets in a cell fitted
with KBr windows. Samples may have two positions in the cell;
one position is in the oven, allowing in situ thermal treatment
under vacuum or various atmospheres, and the second posi-
tion allows room-temperature recording of spectra without re-
exposure to air. Fourier-transform (FT) IR spectra were recorded
with a Bruker-Vector 22 FTIR spectrometer equipped with
a deuterated triglycine sulfate (DTGS) detector, having a
nominal resolution of 4 cmÀ1, by adding 128 scans.
Figure 1. DTG traces of the silica support, of bulk ribose, and of samples 5%
ribose/SiO2, 10% ribose/SiO2, and 15% ribose/SiO2.
ing in quantitative combustion (Figure 1). No significant
weight loss was observed before 1958C, but already after melt-
ing, liquid ribose shows signs of transformation such as brown-
ing and a typical lactone odor. In addition, as the following
paragraph shows, tautomerization reactions are suggested by
IR, which cannot be detected by TG because they do not
cause any weight loss. This behavior is compatible with the
rather limited information available in the literature on the TG
of pentoses.[18] The endothermic event at 1958C could be due
to a dehydration step (the corresponding weight loss amounts
to between one and two water molecules per ribose unit). It
has been claimed to be indicative of carbohydrate polymeri-
zation by condensation, but the matter has not been investi-
gated in detail. The material collected after this step was black
and vitreous. Full polymerization to infinite chains would result
in the elimination of only one water molecule per ribose, so
the larger values measured indicate that modifications of the
ribose skeleton also occur in the 1958C event.
Attenuated total reflectance Fourier-transform infrared spec-
troscopy (ATR-FTIR) was used complementarily to investigate
the effect of the temperature on bulk ribose. For these experi-
ments, the samples were heated inside a heat chamber at am-
bient pressure. For each sample, 40 scans were added and the
spectra were registered in the 400–4000 cmÀ1 spectral region
with a nominal resolution of 1 cmÀ1. The ATR-FTIR approach
provides spectral information while minimizing sample
preparation.
Solid-state 13C cross-polarization magic angle spinning (CP-
MAS) NMR spectra were recorded at room temperature with
a Bruker Avance 500 spectrometer with a field of 11.7 T
equipped with a 4 mm MAS probe with a spinning rate of
10 kHz. The pulse angle was 308, decoupling was applied at 50
KHz, and the recycle time was 10 s for bulk ribose and 1 s for
adsorbed ribose. The contact time was 1000 ms (other values
were also tested in one instance, see below).
FTIR spectroscopy
The IR spectra of bulk d-ribose were recorded at increasing
temperatures, below and above the melting point (Figure 2 A:
rapid heating, and 2B: slow heating). At room temperature,
the spectrum presents vibrational bands characteristic of cyclic
ribose. Table SI-1 shows that our data are in good agreement
with those previously published by Mathlouthi and Seuvre.[19]
Some of the proposed assignments are open to question; in
particular, these authors have claimed to observe vibrations at-
Liquid-state 13C NMR spectra were recorded with a Bruker
Avance III 500 spectrometer, using a 1D sequence with power-
gated decoupling, a 308 flip angle and a recycle time of 2 s.
Results and Discussion
Thermal reactivity of bulk ribose
TGA
tributable to both the pyranose form (1116 cmÀ1, 890 cmÀ1
)
and the furanose form (1160 cmÀ1, 910 cmÀ1), which seems at
odds with solid-state NMR results (see below). In fact, we car-
ried out numerical calculations suggesting that it is extremely
difficult to discriminate the different anomers based on their
vibrational frequencies. These data will be the object of a later
publication. For the time being, we may underline that the
most intense vibrations of d-ribose fall between 1200 and
Figure 1 shows the DTG signals of the various samples heated
under dry air flow. Bulk d-ribose shows an endothermic event
without weight change at 908C, due to ribose melting,[15a] fol-
lowed by four weight loss events at 1958C (endothermic,
À20% of starting mass), 2808C and 3208C (both exothermic,
À51%) and 5108C (strongly exothermic, À29%) finally result-
Chem. Eur. J. 2016, 22, 1 – 14
www.chemeurj.org
3
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
Scheme 2. Thermally-induced ring opening (A) and enolization (B) equilibria
of d-ribose (with the b-furanose polymorph chosen as the starting point).
Transformation of the enolic form into d-ribulose (C) is also illustrated.
Furthermore, the shapes of the bands between 1500 and
500 cmÀ1 are strongly affected by heating: below 858C, well-
resolved bands are observed whereas above the melting point,
they are significantly broadened and an envelope is observed.
This suggests a degree of chemical heterogeneity in the liquid.
In the second series of experiments in which a more gentle
heating was applied to 5.0 mg of d-ribose (Figure 2B), a similar
trend was observed, but the linear d-ribose and the enediol
species only appeared at higher temperatures (around 1508C
instead of 908C). Furthermore, the amounts of these two spe-
cies appear to be more limited. We believe that polymerization
is in competition with the ring-opening process, and is more
favored under gentle heating, that is, in the second heating
procedure for IR, but also in the TGA experiments.
Figure 2. Influence of heating under air on IR-ATR of bulk d-ribose. A) rapid
heating of 0.1 mg of d-ribose; B) progressive heating of 5 mg of d-ribose.
900 cmÀ1, which will later preclude their observation in ribose/
SiO2 samples because of overlap with the lattice vibrations of
the SiO2 support. At room temperature, no carbonyl signal was
present, meaning there was no linear ribose (or products of
further degradation, cf. infra).
Thermal reactivity of adsorbed ribose
X-Ray diffraction (XRD) and TGA
We estimate that a physical monolayer of ribose molecules on
a surface would correspond to 1.6 molecules per nm2. This
Figure can be compared to the surface densities of ribose in
our samples, namely 0.5, 1.1, and 1.5 molecules per nm2 for 5,
10, and 15% ribose/SiO2 respectively, suggesting submonolay-
er coverage (but close to the monolayer for 15% ribose/SiO2).
Indeed no trace of bulk ribose was detected by XRD in any of
these samples.
Two series of spectroscopic experiments were carried out to
investigate the thermal processes occurring in this crystalline
sample. In the first series of experiments, 0.1 mg of d-ribose
was subjected to rapid heating by using a heatgun. In the
second series of experiments, 5.0 mg of d-ribose was
progressively heated inside a heating chamber.
For quick heating, as long as the sample remained below
the melting point, no structural evolution was observed. More
specifically, no vibrational bands were observed in the carbonyl
region. Above the melting point (908C), two vibrational bands
appeared at 1722 cmÀ1 (nC=O) and 1630 cmÀ1 (dOH; Fig-
ure 2 A). They are respectively diagnostic of linear d-ribose and
the enediol species.[20] This means that in bulk liquid ribose,
thermal activation has caused the opening of the ribose ring
and that a keto-enol isomerization has taken place (Scheme 2),
similar to the reactivity of ribose in solution.
TGA of the silica support (Figure 1) shows that it undergoes
a strongly endothermic loss of physisorbed water up to 1008C,
followed by a continuous, low-intensity weight loss between
100 and 6008C, without clearly defined maxima, that can be
assigned to the condensation of surface silanols.[21]
Figure 1 also shows the differential thermograms (DTG) of
5.0, 10.0, and 15.0% ribose/SiO2. As on the raw silica support,
desorption of physisorbed water is observed between 20 and
110 8C. Two broad events at 170 and 3088C are present only in
&
&
Chem. Eur. J. 2016, 22, 1 – 14
www.chemeurj.org
4
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
Table 1. Weight losses (in weight% of the silica support) corresponding
to the two DTG peaks attributable to supported ribose.
Sample
Weight loss [%]
2nd peak
1st peak
total
5% ribose/SiO2
10% ribose/SiO2
15% ribose/SiO2
1.1
2.5
2.6
3.8
7.3
10.5
4.9
9.8
13.1
the ribose-containing samples; as shown in Table 1, the total
weight loss for these two events (corrected for the contribu-
tion of the support) is close to the nominal ribose loading, and
therefore most of the supported ribose has been eliminated
by 4508C.
The event at 1708C is endothermic. The associated weight
loss amounts to one to two molecules of water per ribose
unit—in fact, close to two molecules for 5% and 10% ribose/
SiO2. The event at 3088C is exothermic, probably correspond-
ing to combustion of the remaining organic matter.
FTIR analyses were carried out on supported ribose samples.
The spectra contain less usable information than those of bulk
ribose because of the predominance of vibrational bands of
silica. Only the region between 1900 and 1350 cmÀ1 can be ob-
served without interference from the silica bands. Our previous
work[14] showed that ribose is adsorbed on silica under its
cyclic form and no trace of degraded products is observable at
room temperature, as confirmed by the results of 13C NMR
spectroscopic analysis. All the IR bands recorded at room tem-
perature may be assigned to cyclic ribose and silica nano-
particles.
Figure 3. IR spectra of 10% ribose/SiO2 activated under vacuum up to
2508C.
The IR spectrum of 10% ribose/SiO2 is shown in Figure 3 (it
may be compared with the spectra of 5 and 15% ribose/SiO2
given in Figure SI-1). At room temperature, the observable
bands (mostly the dHCH at 1455 and 1410 cmÀ1) could be as-
cribed to adsorbed cyclic ribose, with no indication of degra-
dation. Activation at temperatures lower than 1508C did not
result in any significant modification of the spectrum. From
150 to 1808C, a vibrational band grew in the nC region, and
=
O
as in the case of bulk ribose (cf. supra), it can probably be as-
signed to linear ribose; it seemed to shift as a function of tem-
perature, because it was first centered at 1723 cmÀ1 and
moved to 1728 cmÀ1 after activation at the final temperature
of 2508C. The shift of this band with respect to bulk ribose
could be due to the isomerization to linear ribulose, which has
been reported for ribose submitted to thermal activation in so-
lution.[20] The intermediate in the formation of this compound
is the enediol form of ribose (cf. Scheme 2), which should have
a characteristic vibration in the 1630–1650 cmÀ1 range. No new
signal was clearly observed in this region, but a small contribu-
tion would probably not be distinguished from the residual
dHOH band of H2O, and therefore we cannot exclude the
presence of this intermediate.
Scheme 3. A three-step mechanism leading to the formation of ribonolac-
tone from d-ribose (see text).
reports,[22] a three-steps mechanism is involved as illustrated in
Scheme 3: A) formation of linear ribose (already illustrated in
Scheme 2); B) oxidation of the carbonyl group to ribonic acid;
and C) dehydration to ribonolactone.
To confirm the hypothesis that an oxidative process is taking
place as postulated in Step B, thermal activation of ribose/SiO2
was carried out in O2 flow and the vibrational bands that we
had assigned to linear ribose and to ribonolactone were quan-
tified. The band of ribonolactone was five times more intense
at 2508C under O2 flow than under (nominal) vacuum. At the
same time, the proportion of linear ribose was twice as high
under O2 flow as under vacuum: the opening of the ribose
ring seems also to be facilitated under O2. The occurrence of
an oxidative reaction under nominal vacuum remains surpris-
From 180 to 2508C, a second vibrational band appears at
1775 cmÀ1, which could be attributed to a cyclic ester such as
that present in ribonolactone.[22] The formation of ribonolac-
tone from ribose is an oxidative process. According to previous
Chem. Eur. J. 2016, 22, 1 – 14
www.chemeurj.org
5
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
ing because our systems are not supposed to contain
oxidants. The most logical candidate would be resid-
ual O2 in the vacuum ramp. Attempts to duplicate
our results revealed that the intensity of the band as-
signed to ribonolactone was indeed strongly depen-
dent on the experimental setting.
Table 2. Weight losses (in weight% of the silica support) corresponding to DTG peaks
in ribose-metal/SiO2 samples.
Sample
Peak Temp.
Weight
losses [%]
Thermal effect
(DTA)
Total weight loss
all ribose peaks [%]
Tmax [8C]
ribose/silice
38
170
304
65
165
~370
57
172
~320
53
172
~315
42
200
39
199
43
2.0
2.5
7.3
14.0
5.5
6.8
6.4
3.4
4.5
3.3
4.4
5.2
3.9
10.6
3.0
9.6
1.8
2.6
6.8
4.1
5.0
6.1
endothermic
endothermic
exothermic
endothermic
endothermic
(weak)
endothermic
endothermic
exothermic
endothermic
endothermic
(weak)
endothermic
exothermic
endothermic
exothermic
endothermic
?
9.8
12.3
7.9
With respect to the effect of ribose loading, the
amount of linear ribose formed at a given tempera-
ture increases faster than the initial amount of ribose
(Figure SI-2). This could hint at a chemical heteroge-
neity of the ribose/surface interaction. As for the
amount of ribonolactone, it does not show a clear
trend as a function of ribose coverage.
ribose-Mg/SiO2
ribose-Ca/SiO2
ribose-Sr/SiO2
9.6
Effect of the coadsorption of divalent cation salts
on the reactivity of adsorbed ribose
ribose-Fe 2+/SiO2
ribose-Fe 3+/SiO2
ribose-Cu/SiO2
10.6
9.6
XRD and TGA
In ribose-Sr/SiO2, XRD analysis revealed the presence
of bulk SrCl2. No bulk salt phases were observed for
the other ribose-metal/SiO2 samples. This probably
means that metal cations (and chloride anions as
well!) are well dispersed on the silica surface (see
Figure SI-3 for X-ray diffractograms of the various
samples).
144
257
47
183
320
9.4
(weak)
ribose-Zn/SiO2
endothermic
endothermic
exothermic
11.1
The first peak under 1008C is endothermic, the second be-
tween 150 and 2008C is generally also endothermic (the direc-
tion of the effect is unclear in the case of ribose-Cu/SiO2), and
the third peak is exothermic when the thermal effect can be
ascertained. Furthermore, the sum of the weight losses for the
second and third peak is close to 10% for a majority of the
samples (10Æ0.6%), that is, close to the theoretical ribose
loading, and it seems logical therefore to assign both of them
to the elimination of ribose; note however that significantly
higher values are measured for ribose-Zn/SiO2 (11.1%) and
ribose-Mg/SiO2 (12.3%), and a significantly lower value for
ribose-Cu/SiO2 (7.9%).
DTG traces of the various ribose-metal/SiO2 samples are pre-
sented in Figure 4 and the weight losses corresponding to the
identifiable thermal events can be found in Table 2.
Most of the ribose-metal/SiO2 samples, except ribose-Fex+
/SiO2, have similar TGA profiles, with three peaks in the DTG.
The first peak below 1008C can be attributed to the loss of
weakly held water, which may be adsorbed on the free surface,
interacting with ribose or bound to the cations (in their hydra-
tion sphere). It is also present in ribose-free metal salt/SiO2
samples that were investigated separately (see Figure SI-4). In
both Mg/SiO2 and Cu/SiO2, the shape of the peak shows
a structure that probably indicates well-defined transitions be-
tween [M(H2O)6]2+ complexes of well-defined stoichiometry.
Generally speaking, it is difficult to discriminate between
metal-bound and surface-bound water. However, the alkaline
earth/SiO2 series shows a strong variation in the amount of
weight loss that follows the trend in hydrophilicity, with
ribose-Mg/SiO2 (17%)> ribose-Ca/SiO2 (7.2%)> ribose-Sr/SiO2
(6.45%), whereas transition-metal/SiO2 samples contain lower
amounts of water (5.0 to 5.5%).
Among the ribose-metal/SiO2 samples, weight losses are
also strongly variable. In ribose-alkaline earth/SiO2 samples,
they are very similar to the ribose-free samples, suggesting
that hydration water mostly interacts with the metal cations
and that this interaction is not disturbed by the presence of
Figure 4. DTG traces of ribose-metal/SiO2 samples
&
&
Chem. Eur. J. 2016, 22, 1 – 14
www.chemeurj.org
6
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
ribose. On the other hand, in ribose-transition metal/SiO2
samples, the presence of ribose does cause some changes in
hydration, with ribose-Cu/SiO2 being the least hydrated.
The second, endothermic peak in DTG traces is reminiscent
of the peak that was observed at 1708C in ribose/SiO2, with
a 2.5% weight loss. Note first that the metal salts themselves
do not give rise to a TG event in this temperature range (Fig-
ure SI-4; CuCl2/SiO2 is an exception, with a small peak at 2878C
that may be due to a redox between chloride ions and Cu2+).
In the presence of alkaline earth metals, the second peak is
also located at 1708C with a larger weight loss that depends
on the nature of the metal cation: about 3.4, 5.5, and 4.4%, re-
spectively, in the presence of calcium, magnesium, and stronti-
um. Transition metals induce different changes. In the pres-
ence of copper, this peak shifts to a lower temperature (1448C)
with a 2.5% weight loss. In contrast, in the presence of zinc,
this peak shifts to a higher temperature (1858C) with a 4.1%
weight loss. Some of the weight loss values seem too high to
be assigned only to ribose dehydration reactions, because
even extensive dehydration to furfural (see IR results and dis-
cussion) would result in the loss of only three water molecules
per ribose; that is, a loss of 3.6% on the basis of a 10% ribose
loading. Thus, some metal ions promote extensive ribose
degradation at low temperatures.
The highest temperature peak observed in the thermograms
shows an exothermic event that can be attributed to the oxi-
dation of residual organic matter. In comparison to ribose/SiO2
(where Tmax =3048C), this peak is considerably broadened and
shifted to higher temperatures in ribose-alkaline earth
metals/SiO2 systems (Tmax between 315 and 3708C) and ribose-
Zn/SiO2 (Tmax about 3208). In contrast, the combustion of
organic matter occurs at lower temperatures for ribose-Cu/SiO2
(Tmax =2578C) and even more so for ribose-Fex+/SiO2, for which
the combustion event cannot be separated from the second
peak. In the latter two cases, the low-temperature combustion
event suggests that the supported metal ion is intimately in-
volved in the oxidation event.
FTIR spectroscopy
Room temperature spectra: The metal ions we investigated
show different effects on the IR spectrum, both at room
temperature and upon thermal activation. The spectra of the
different samples are presented in Figure 5 and Figure 6.
Before thermal activation, differences with ribose/SiO2 are
already apparent (compare lowest spectra in Figure 5 and
Figure 6 with Figure 3). The dHOH band of water around
1630 cmÀ1 is a good indicator of the hydration state of the
material. It is more intense in all ribose-metal/SiO2 than in
metal-free samples, and for the alkaline earth cations its inten-
sity follows the order of hydrophilicity (Mg2+ @Ca2+ >Sr2+), in-
dicating that a lot of water is bound to the cations at room
temperature. For the transition-metal cations, the order of in-
tensity is (Fex+ @Cu2+ =Zn2+); however, the maximum intensi-
ty for the ribose-Fe/SiO2 samples is artificially inflated by the
presence of another band in close vicinity.
Figure 5. IR spectra as a function of activation temperature of ribose-Mg/
SiO2 (A), ribose-Ca/SiO2 (B), and ribose-Sr/SiO2 (C), with (1:1) molar equiva-
lent.
As regards the dHCH vibrations, in ribose-transition metal/
SiO2, the band at 1452 cmÀ1 is unaffected by the presence of
the cations, whereas the second band shows significant batho-
chromic (red) shifts. We evaluate them at À20, À18, and
À13 cmÀ1, respectively, in ribose-Zn/SiO2, ribose-Fex+/SiO2, and
ribose-Cu/SiO2, for a ribose/metal molar ratio of 1:1, but it is
less important in the ribose-metal (2:1)/SiO2 series (ca. 8 cmÀ1).
This indicates a direct interaction between metals and sugar,
probably involving the C5 carbon. The effect of alkaline earth
cations on this band are more complex: it appears split into
two components, one blueshifted and the second redshifted,
Chem. Eur. J. 2016, 22, 1 – 14
www.chemeurj.org
7
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
The shoulder at 1661–1662 cmÀ1 in ribose-Fex+/SiO2 and
1658 cmÀ1 in ribose-Cu/SiO2 is probably characteristic of the
enediol. The position of this band would then be significantly
blueshifted with respect to the same species in bulk ribose,
and this might be the effect of complexation to the metal
ions. Enediols, like aromatic diols such as catechol or dopa-
mine, are known to be good complexing agents,[23] including
for Fex+ centers. The complexes of enediol with iron and
copper ions could be stabilized by crystal field stabilization
energy (CFSE) effects, causing a shift of the ribose isomeriza-
tion equilibria towards this form.
With regard to the alkaline-earth-containing samples, a small
band is observed at 1702 cmÀ1 for ribose-Mg/SiO2 and 1710–
1715 cmÀ1 for ribose-Ca/SiO2, but it is not present for ribose-
Sr/SiO2. An assignment to complexed linear ribose may also be
proposed here on the basis of parsimony, but the carbonyl
shift would be more important in ribose-Mg/SiO2 than in the
other samples, which is somewhat surprising.
Effect of thermal activation on spectra: We shall now consider
the evolution of IR spectra upon thermal activation. In ribose-
Mg (1:1)/SiO2, no changes are observed from room tempera-
ture to 1008C, except for a gradual decrease of the band at
1630 cmÀ1 indicative of dehydration. At 1508C, a conspicuous
new band appears at 1533 cmÀ1 (nC
region), which is
=
C
a region in which no signal was detected for ribose/SiO2 irre-
spective of the temperature. At the same time, an absorption
in the 1400–1500 cmÀ1 region was also modified with the ap-
pearance of a new maximum at 1415 cmÀ1, and the carbonyl
band was redshifted by À4 cmÀ1; it seems to disappear above
1808C. These observations suggest that ribose has been
strongly dehydrated, with the concomitant production of C=C
double bonds. One logical outcome of this dehydration reac-
tion would be the formation of furfural (cf. Scheme 4), but this
Scheme 4. Furfural and furan, compared with d-ribose.
Figure 6. IR spectra as a function of activation temperature of ribose-Zn/SiO2
(A), ribose-Fe/SiO2 (B), and ribose-Cu/SiO2 (C), with (1:1) molar equivalent.
in ribose-Mg/SiO2 and ribose-Sr/SiO2, whereas only the blue-
shifted component appears in ribose-Ca/SiO2.
product should have a very strong nC band at 1714 cmÀ1 be-
=
O
New bands are also apparent in the carbonyl stretching
cause of its aldehyde function, which is not observed here. The
product that forms must therefore have C=C double bonds
and no aldehyde function. This is the case for furan, C4H4O,
and the subtraction spectrum is indeed rather compatible with
that of furan.[24] The evolution of ribose-Mg (2:1)/SiO2 is almost
indiscernible from that of ribose-Mg (1:1)/SiO2.
(nC O) region in ribose-Fex+/SiO2 and ribose-Cu/SiO2, for both
=
molar ratios, but not in ribose-Zn/SiO2. The first region, at
1714 cmÀ1, appears for ribose-Cu/SiO2 and ribose-Fe2+/SiO2
(but not ribose-Fe3+/SiO2). The similarity with the aldehyde
band of furfural is probably coincidental because the other
bands of this molecule are not present. We tentatively attri-
bute the band at 1714 cmÀ1 to a linear ribose; however, given
that its wavenumber is significantly lower than in free linear
ribose, we suggest that it corresponds to linear ribose
complexed by iron(II).
For ribose-Ca/SiO2 and ribose-Sr/SiO2, more complex
changes are observed, with no less than six new bands appear-
ing at 1508C and above (1764, 1714–1718, 1580, 1538, 1475,
and 1411 cmÀ1). They correspond to several different products
because their relative intensities change with the temperature.
&
&
Chem. Eur. J. 2016, 22, 1 – 14
www.chemeurj.org
8
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
One of the products could be ribonolactone (nC
at
central to the main question motivating the present study,
namely, the stability of ribose.
=
O
1764 cmÀ1), which remains a minor species. The two bands in
the 1500–1600 cmÀ1 range can be assigned, as above, to nC
=
C
bands, possibly of different products. They appear at the same
time as the nC at 1714–1718 cmÀ1 and together, these obser-
=
O
vations could indicate the formation of furfural. In comparison
to ribose-Mg/SiO2, we propose that in ribose-Ca/SiO2 and
ribose-Sr/SiO2 both furfural (completely dehydrated product)
and furan (dehydrated and decarboxylated) are formed.
Heating of ribose-Zn/SiO2 to 1508C does not cause any
changes in the spectrum. The first modifications are only ap-
parent at 1808 and become conspicuous at 2008C. They are
very similar to what was observed at lower temperatures in
ribose-Mg/SiO2 and ribose probably undergoes the same trans-
formations here, only at a significantly higher temperature.
Thus, the presence of Zn2+ in a 1:1 ratio causes a thermal sta-
bilization of ribose as compared with ribose/SiO2, and, actually,
also to the other ribose-metal/SiO2. In contrast, the bands that
we have previously assigned to linear ribose (1730 cmÀ1) and
ribonolactone (1780 cmÀ1) are observed in ribose-Zn (2:1)/SiO2
already at 1508C. We hypothesize that Zn2+ is able to form
a complex with cyclic ribose that results in the stabilization of
the latter molecule with respect to ring opening. If the stoichi-
ometry of this complex is 1:1, then all ribose molecules may
be complexed, and thus stabilized, in ribose-Zn (1:1)/SiO2,
whereas half of them would remain uncomplexed in ribose-Zn
(2:1)/SiO2, and therefore follow the same transformation as
without zinc.
NMR spectroscopy
The 13C NMR spectrum of ribose in water solution has been
reported previously and can be used as a basis for peak ass-
ignments in other samples. The spectrum we obtained (Figure
SI-5) was identical to that published by Ortiz et al.[27] In general,
four signals can be observed for each carbon, corresponding
to the four anomers (although there is some overlap); the ap
and bp forms are more difficult to discriminate than the others.
The C1 peaks are particularly well resolved and can be used to
quantify the anomeric ratios; it can also be verified that very
similar ratios are obtained on the basis of the C4 peaks.
We have also recorded liquid-state 13C NMR spectra of solu-
tions containing ribose and metal chlorides (1:1 ratio) to assess
the complexing power of these salts (Figure SI-6, and Table SI-
2). The addition of MgCl2, CaCl2, or ZnCl2 did not have much
effect on the chemical shifts of the various peaks, with dis-
placements in general smaller than 0.1 ppm (except for the C5
peaks for which +0.2 and +0.35 ppm shifts were observed in
two cases). The intensity ratios were not significantly altered
either. In contrast, the addition of SrCl2 caused shifts up to
1.0 ppm, making the assignments of some peaks difficult, and
the contribution of the af anomeric form was significantly ex-
alted. In summary, the modifications of the 13C NMR spectra
are very minor and do not indicate coordination of ribose to
Mg2+, Ca2+, or Zn2+ in aqueous solution (nor any strong bind-
ing of ribose to ClÀ); at most, the results indicate the possible
formation of weakly H-bonded adducts. Sr2+ is a possible ex-
ception. These conclusions are significantly different from
those of Ortiz et al.[27] who observed more significant peak
shifts (a few tenths to 2.0 ppm) for ribose in aqueous solution
when KCl and NaCl were added to the medium; it should be
noted, however, that these authors used much higher salt
concentrations.
Copper and iron induce very different thermal reactivities
compared with zinc and alkaline earths. Below 1508C, linear
ribose (1720 cmÀ1) and enediol (1658 cmÀ1) are present, the
latter in large quantities, in ribose-Cu/SiO2 and ribose-Fex+
/SiO2.
In ribose-Cu/SiO2, at 1508C, a very strong band grows in the
nC region at 1731 cmÀ1 and the enediol band disappears. No
=
O
bands are apparent in the nC region (this is in exact opposi-
=
C
tion to ribose-Mg/SiO2). Several assignments could be pro-
posed for this nC band. It might be due to one or several al-
=
O
dehydes, including glycolaldehyde,[25] formed by catalytic
retro-aldol fragmentation of ribose.[26] Or it might be due to
a carboxylic acid formed through the oxidation of ribose by
Cu2+, which would then be acting as a stoichiometric reagent.
Given that the amount of the corresponding product, as indi-
cated by the relative band intensity, is about twice lower in
ribose-Cu (2:1)/SiO2 than in ribose-Cu (1:1)/SiO2, we would
favor the second interpretation, but clearly more work ought
to be done to clarify this point. Local browning of the sample
was observed, possibly due to the products of copper
reduction.
In ribose-Fe/SiO2 samples the reactivity seems different
again. The complexed enediol form is preserved at high tem-
peratures, whereas a new, sharp band appears at 1558 cmÀ1
and shoulders at 1510 and 1490 cmÀ1 at higher temperatures.
Thus, here, dehydration reactions must be considered again.
We will not endeavor to precisely identify the reaction prod-
uct(s) formed in these samples because their nature is not
Figure 7. 13C NMR spectra of bulk ribose, ribose/SiO2, and ribose-diamagnetic
Metal/SiO2 systems
Chem. Eur. J. 2016, 22, 1 – 14
www.chemeurj.org
9
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
The 13C CP-MAS NMR spectra of ribose/SiO2, ribose-Mg/SiO2,
ribose-Ca/SiO2, and ribose-Zn/SiO2 are presented in Figure 7
and compared with that of bulk ribose.
Ca/SiO2, both anomeric forms of ribofuranose increase, but the
af to a larger extent than the bf, leading to a b/a ratio of 0.8
and 0.7, respectively, as compared with 1.6 in ribose/SiO2. In
contrast, the opposite trend is observed in ribose-Zn/SiO2: the
b/a ratio actually increases to 3.6.
Ribose interaction with the silica surface in the absence of
cations has been discussed in a previous paper.[14] In solution,
the pyranose forms are predominant (82% of the total, for
18% of furanose forms). After adsorption on the silica surface,
the NMR peaks are much broader, which is typical of solid-
state NMR analysis, for where the molecules have a low mobili-
ty. NMR peak positions remain compatible with the chemical
shifts of ribose carbon signals. Thus, only ribose is present on
the silica surface, in its four anomeric forms; yet adsorption
somewhat modifies the anomeric equilibria, as the furanose
forms increase to about 25%, specifically due to promotion of
the af form.
General Discussion
Some articles concerning amino acids have drawn attention to
the prebiotic potential of bulk organic molecules,[28] and there-
fore the study of bulk ribose is relevant. Our initial study of the
reactivity of bulk ribose can be used as a benchmark against
which to evaluate the stability of ribose molecules in other en-
vironments. Melting at 908C already induces chemical hetero-
geneity, as indicated by the IR study and by simple visual in-
spection. The inception of some chemical transformations de-
pends on the heating procedure. One particular transformation
that is easy to demonstrate spectroscopically is ring opening
to linear ribose, followed by isomerization to enediol. Flash
heating induces ring opening already at 908C, and even upon
gentle heating these reactions clearly appear at 1508C. It can
be said with confidence that bulk ribose is no longer stable at
1508C.
When ribose is deposited on SiO2, it remains in the form of
a cyclic monomer. In particular, there are no observations that
might indicate polymerization, probably due to the restriction
in molecular mobility upon adsorption that prevents the mon-
omers from encountering each other. The only effect of ad-
sorption on ribose speciation is modification of the anomeric
equilibria, as discussed below. At 1508C and above, ribose un-
dergoes ring opening; between 180 and 2008C, oxidation
must occur because a minor amount of ribonolactone is
formed. The most likely oxidant is adventitious dioxygen re-
maining in the vacuum ramp (and if it is the case, the corre-
sponding reaction would not be of prebiotic significance be-
cause the primordial atmosphere of the Earth was anoxic). An-
other possibility would be adsorbed water; although normally
a weak oxidant, its reaction could be enhanced by removal of
its reduction product, that is, H2, under vacuum. However, we
did not observe any H2 formation, and, moreover, the remain-
ing amount of adsorbed water at 1808C is certainly quite
small; therefore we do not favor this possibility. At 2508C, not
much organic matter remains on the surface. Also interesting
is what is not observed on silica: there is no evidence of the
formation of unsaturated compounds with C=C double bonds
that would be formed by dehydration of ribose. Now the silica
surface is moderately acidic (with about 4.5 Si-OH surface
groups per nm2, of varying Brçnsted acidity). Under anhydrous
conditions and with the possibility of acid catalysis, ribose
could evolve all the way to furfural (a reaction, accompanied
by the loss of three water molecules per ribose, that was evi-
denced as early as 1943[29]), or stop at an intermediate dehy-
dration step. In fact, investigations on the dehydration of the
closely related xylose molecule have shown that the Brçnsted
acidic centers of silica were much less active than the Lewis
acidic centers of silico-aluminas.[30] Nevertheless, the well-
defined thermal event at 1508C in TG suggests that ribose
Likewise, for the ribose-metal/SiO2 samples, all individually
resolved peaks may be assigned to ribose carbon atoms, and
the remaining part of the spectrum is compatible with the en-
velope of overlapping ribose peaks. Table SI1 summarizes the
chemical shifts of the various samples, at least those that
could be determined precisely. Important shifts are observed
in some cases with respect to ribose in solution, especially for
ribose-Mg/SiO2 for which the peaks corresponding to C1 are
shifted between À3.6 and +7.2 ppm, whereas those of C4 and
C5 are affected to a lesser extent; all the enantiomers are
affected. In ribose-Ca/SiO2, in contrast, the peaks of C4 and C5
undergo more significant shifts than those of C1.
In ribose-Zn/SiO2, among the signals that can be clearly dis-
criminated, the C1 peaks of the two furanose isomers are
strongly affected, their C2 and C4 peaks much less so. The C1
peaks of the pyranose forms are also shifted, but the effect is
three times smaller than for the furanoses, leading us to
believe that the latter are more strongly complexed.
Quantification of the anomeric forms was carried out on the
basis of the C1 peaks, which are the best resolved; the results
are reported in Table 3. In some samples the C4 peaks were
sufficiently well-resolved to provide an independent estimate
of the enantiomeric ratios, and these estimates were not
significantly different from the C1-based ratios.
Coadsorption with metallic salts promotes the ribofuranose
forms over the ribopyranoses, as did the adsorption on bare
silica, but to a larger extent. The proportion of ribofuranoses
increases from 24% in ribose/SiO2 to 34, 51, and 64%, respec-
tively, in ribose-Mg/SiO2, ribose-Ca/SiO2, and ribose-Zn/SiO2.
The b/a ratios are also affected. In ribose-Mg/SiO2 and ribose-
Table 3. Anomeric ratios calculated on the basis of the intensity of the
C1 peaks.
Sample
ap
bp
Total
af
bf
Total
[%]
[%]
pyranose
[%]
[%]
furanose
Ribose solution
Solid ribose
ribose/SiO2
ribose-Mg/SiO2
ribose-Ca/SiO2
ribose-Zn/SiO2
20
n.d.
29
18
23
27
62
n.d.
47
48
26
9
82
100
76
66
49
6
–
15
18
29
14
12
–
9
16
22
50
18
0
24
34
51
64
36
&
&
Chem. Eur. J. 2016, 22, 1 – 14
www.chemeurj.org
10
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
dehydration does occur in ribose/SiO2. It is possible that the
dehydration products do not stay on the surface and are
quickly desorbed under the conditions of IR analysis (vacuum),
whereas they burn in the air flow of the TG analysis.
formation of only furanose isomers. Nevertheless, the use of
borate has been disputed. The emergence of life probably oc-
curred around 3.7 Ga, but the oldest boron-containing mineral
is probably only 3.6 Ga old.[34] Moreover, borates were probably
not in sufficiently high concentration to be used in chemical
reactions or to precipitate to allow prebiotic reactions at the
interface.
In summary, cyclic ribose adsorbed on silica is stable up to
about 1508C; this sets an upper limit on the conditions under
which prebiotic scenarios involving this molecule could be
invoked.
The question of ribose anomeric speciation goes beyond fur-
anose vs. pyranose. Today’s RNA is more specifically constitut-
ed of the b-d-ribofuranose isomer and therefore the a vs. b dis-
tribution of ribose is also relevant. Recently, Singh et al.[35] stud-
ied the glycosylation of adenine in water/acetonitrile mixtures.
They observed the preferential formation of a b-nucleotide
and tried to rationalize this observation in a mechanism involv-
ing a nucleophilic attack of the adenine base on ribose with in-
version of configuration, making the hypothesis that the ap-
proach of the nucleophilic base was favored by a specific pat-
tern of H-bonding with the af ribose anomers. The details of
the mechanism are likely to be different on the silica surface,
and insufficient molecular information is available to speculate
on this, but the special behavior of Zn chloride, which favors
the b anomer, would be worth investigating in glycosylation
experiments.
Divalent metal salts can definitely interact with ribose in
a specific way. The interaction is minimal in aqueous solution,
at least for the concentration we investigated: according to
liquid-state 13C NMR spectroscopic analysis, only very weak,
outer-sphere interactions are established between ribose and
metal salts (with the possible, and quite unexpected, exception
of Sr2+). The metal cations must be present as aqua com-
plexes, and remain so as long as the water activity is high
enough. As soon as drying removes water ligands, these cat-
ions establish rather strong interactions with ribose and may
already cause significant transformations at the low tempera-
ture of 708C. It is all but certain that every cation we tested is
able to form specific complexes with ribose upon co-adsorp-
tion on silica. This is indicated by IR spectroscopic analysis, for
which the dHCH shifts betray a strong interaction of the C5
carbon with the metal through its C-O-H moiety. IR analysis
cannot be used to determine whether the other carbon atoms
are involved in the interaction (at least not in the wavenumber
range accessible under our experimental conditions), but solid-
state 13C NMR spectroscopic analysis clearly shows strong shifts
of at least some of the other carbon signals, and we may con-
clude that ribose acts as a polydentate ligand. Establishing the
precise stoichiometry and molecular structure of the resulting
complexes goes beyond the aims of the present article and
would require much additional work. In the solid state, some
cations such as Ca2+ form 1-D coordination polymers with
pentoses,[31] but in view of the submonolayer density of ribose
in our samples this does not seem likely, and the complexes
formed on silica may have original structures.
Co-adsorption of inorganic salts causes striking modifica-
tions of the thermal reactivity of adsorbed ribose. We do not
believe that interaction with the anion is the main factor, given
that we used the same compensating anion in all samples, and
observed different behaviors depending on the nature of the
cation. Already after drying at 708C ring opening is observed
with Mg2+ and Ca2+. This isomerization to linear ribose could
be catalyzed by coordination to the alkali metal cation. Beyond
the reaction kinetics, the equilibrium (if it is indeed attained)
does not favor the linear form, which remains a minority spe-
cies and could not be separately observed by NMR spectro-
scopic analysis are. Zn2+ and Sr2+ do not cause ring opening
(see below for further discussion). Cu2+ and Fex+ cause a more
extensive transformation to the enediol form, probably
because of their capacity to make coordination compounds
with the latter.
The enantiomeric distribution of ribose in our samples is
also interesting. In a previous communication, we noted that
the furanose isomers were more prominent in ribose/SiO2 than
in water solution.[14] The tendency toward furanose stabiliza-
tion is stronger in ribose-Mg/SiO2 and ribose-Ca/SiO2 (for
which they respectively account for 34 and 51% of total
ribose, compared with 24%). It is even more pronounced in
ribose-Zn/SiO2, with 64% of furanose isomers, but this sample
is unique in favoring the b over the a isomer, whereas the op-
posite effect is observed with the other cations. The predomi-
nance of furanose is interesting because RNA exclusively uses
a furanose form of ribose, which is not the predominant
isomer in pure water solution.
Thermal activation under vacuum results, in most cases, in
rather abrupt modifications of the IR spectrum at about 1508C,
as on bare silica. With alkaline earth metals, unsaturated prod-
ucts resulting from ribose dehydration appear. For ribose-Mg/
SiO2, only weak carbonyl signals are observed, suggesting that
furfural, the end product of complete ribose dehydration (with
a loss of three H2O per ribose), is not predominant. A possible
explanation would be that carbon centers in the furanose
cycle are dehydrated first (with a loss of up to two H2O per
ribose), while keeping the -CH2OH group intact. However, this
does not correspond with known dehydration mechanisms,[36]
and besides TG shows that this sample loses more than two
H2O per ribose, not less, in the dehydration peak at 1658C. We
have mentioned that IR data are compatible with transforma-
tion of ribose into furan; the calculated weight loss would be
5.5% (reported to the mass of SiO2), in keeping with its
quantification in TG. Furan can be obtained from furfural by
The evolution of the anomeric ratio has been studied in the
presence of borate[8a,32] and silicate[8c,33] anions. In both cases,
the use of borate or silicate during the formose reaction stabi-
lized ribose, and, more specifically, its furanose forms. Ribose
speciation evolved even more by association of borate and
alkali or alkaline earth cations.[32a] The best result obtained was
for the association of potassium and borate, which led to the
Chem. Eur. J. 2016, 22, 1 – 14
www.chemeurj.org
11
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
decarbonylation, although the catalysis of this transformation
usually involves transition metals.
troscopic signature. Under TG conditions, i.e., under air flow,
the reactivity seems noticeably different from vacuum condi-
tions, in contrast to what is observed for the other systems: all
the adsorbed ribose is removed in an exothermal reaction
taking place at 2008C, meaning that the Fex+ ions catalyze the
combustion of ribose by dioxygen in the air. From a prebiotic
point of view, this is less relevant than the reactivity under
vacuum.
In contrast, for ribose-Ca/SiO2 and ribose-Sr/SiO2, a strong
carbonyl peak appears at 1714–1716 cmÀ1 at the same time as
the C=C bands, consistent with the formation of furfural. The
latter molecule would coexist with furan, because the experi-
mental weight losses fall between the theoretical values corre-
sponding to furfural formation (À3.2% expected weight loss)
and furan formation (À5.5%), while a minor amount of ribono-
lactone would be formed in a side reaction. The formation of
furfural deserves some comments. Obtaining furfural from d-
xylose (a pentose rather similar to d-ribose, itself derived from
the hydrolysis of hemicellulose) is a challenge in green chemis-
try. This reaction is usually carried out in slurries, and the effect
of metal salt additives has been studied in some depth. An un-
desirable side reaction is the formation of polymeric humins
from furfural condensation.[37] In our samples, when furfural is
formed, it does not polymerize because of its limited mobility
on the silica surface, and therefore catalytic systems based on
CaCl2/SiO2 or SrCl2/SiO2 could perhaps be applied for the indus-
trial synthesis of furfural. At any rate, mechanistic insights
gained on xylose transformation may also prove useful to un-
derstand ribose transformation,[37–38] even though the experi-
ments are carried out in quite different environments (concen-
trated slurries instead of solid-vacuum interface). In particular,
Enslow et al.[37] have proposed that ribose complexation
through its C-OH groups can weaken the CÀO bonds and thus
facilitate dehydration, in a type of Lewis acid catalysis.
Conclusion
With respect to bulk ribose, and even more to ribose in aque-
ous solutions, adsorption on silica results in stabilization of the
cyclic molecules up to about 1508C, thus extending the useful
“temperature window” for prebiotic reactions by more than
508C; the anomeric speciation is only slightly modified with re-
spect to the solution. At 1508C, ring opening starts to occur,
and this is a doorway to pentose isomerization and loss of
chemical specificity.
The alkaline earth and transition metal salts investigated
here (MgCl2, CaCl2, SrCl2, CuCl2, FeCl2, FeCl3, ZnCl2) show little
tendency to interact with ribose in solution. However, when
they are deposited together with ribose on the silica surface
and submitted to drying, they form coordination complexes in
which ribose acts as a multidentate ligand. The detailed
structures of these complexes have not yet been resolved.
Complexation influences the anomeric ratios, with the most
remarkable effect being observed for Zn2+, which favors the
furanosic forms and especially the bf isomer.
Ribose-Zn/SiO2 undergoes the same transformations as
ribose-Mg/SiO2, but at a considerably higher temperature:
cyclic ribose is intact at 1508C, it starts transforming into dehy-
dration products at 1808C, and is extensively transformed,
probably into furan, only at 2008C (this is compatible with Tmax
and weight loss observed in TG). Thus, from the point of view
of ribose stabilization that was the driving force of the present
study, co-adsorption with Zn2+ on silica is the most efficient
treatment. It is crucial that Zn2+ prevents ring opening to
linear ribose, because the formation of the latter intermediate,
followed by isomerization to enediol, can easily result in fur-
ther reactions, for example, to ribulose, arabinose, and thus
presumably to loss of (pre)biological selectivity.
The effects of inorganic salts on the thermal reactivity of
ribose are very diverse depending on the chemical nature of
the cations. Alkaline earth cations do not extend the tempera-
ture stability of ribose with respect to the salt-free system,
and, in fact, they cause a minor amount of ring opening al-
ready at low temperature. At 1508C and above, they catalyze
ribose dehydration leading to furfural and also, to variable ex-
tents, decarbonylation of the latter to furan. These reactions
probably follow a Lewis acidic catalytic mechanism, and might
be useful for the valorization of biomass-derived products.
Copper and iron ions cause a more extensive ring opening
and isomerization at low temperature (to the enediol form). At
higher temperatures, they initiate complicated reactions in
which their ability to take part in redox reactions, either as re-
agents or as catalysts, plays a role together with their Lewis
acid catalytic properties. They certainly do not look promising
to stabilize ribose at high temperatures.
In the presence of copper (Cu2+), ribose transformation fol-
lows a different path. As mentioned above, the nC indicative
=
C
of dehydration is not observed but a strong nC O band appears.
=
The weight loss is smaller than for all other cations tested and
occurs at a lower temperature. The reaction that occurs could
be a partial oxidation of the ribose skeleton. Although a minor
amount of ribose can be oxidized by residual O2 in the
vacuum ramp, the phenomenon that occurs here is more ex-
tensive and it is possible that Cu2+ itself acts as an oxidizing
agent, thus playing the role of a reagent and not of a catalyst.
Phenomena observed with iron (Fe2+ and Fe3+) are com-
plex. As mentioned, they involve ring opening at low tempera-
tures and probably dehydration reactions at 1508C and higher
under vacuum, without the formation of an aldehyde function.
In this respect, zinc appears unique because it preserves
cyclic ribose at least up to 1808C. Although the present work
may have applications other than in prebiotic chemistry, from
the latter point of view, it will be interesting to study the reac-
tivity of ribose-Zn/SiO2 systems in further reactions that might
lead to RNA, namely glycosylation and phosphorylation
processes.
This reactivity is grossly similar to that observed with Mg2+
yet the product formed does not have exactly the same spec-
,
&
&
Chem. Eur. J. 2016, 22, 1 – 14
www.chemeurj.org
12
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
[14] T. Georgelin, M. Jaber, F. Fournier, G. Laurent, F. Costa-Torro, M.-C.
Maurel, J.-F. Lambert, Carbohydr. Res. 2015, 402, 241–244.
[15] a) A. Raemy, T. F. Schweizer, J. Therm. Anal. 1983, 28, 95–108; b) P. Tom-
asik, M. Palasinski, S. Wiejak, Adv. Carbohydr. Chem. Biochem. 1989, 47,
203–278; c) Y. Roos, Carbohydr. Res. 1993, 238, 39–48; d) M. Hurtta, I.
Pitkꢂnen, J. Knuutinen, Carbohydr. Res. 2004, 339, 2267–2273; e) S. T.
Beckett, M. G. Francesconi, P. M. Geary, G. Mackenzie, A. P. E. Maulny,
Carbohydr. Res. 2006, 341, 2591–2599; f) A. Magon, M. Pyda, Carbohydr.
Res. 2011, 346, 2558–2566; g) J. W. Lee, L. C. Thomas, S. J. Schmidt, J.
Agric. Food Chem. 2011, 59, 684–701.
Acknowledgements
Mariame Akouche was supported by a Ph.D. grant from the
Ile-de-France region, under the DIM-ACAV program. The
SESAME program of the French Ile de France Region is
acknowledged for financial support for the purchase of the
500 MHz NMR spectrometer.
[16] M. R. M. Izawa, H. W. Nesbitt, N. D. MacRae, E. L. Hoffman, Earth Planet.
Sci. Lett. 2010, 298, 443–449.
[17] a) G. Wꢂchtershꢂuser, System. AppI. MicrobioI. 1988, 10, 207–210;
b) A. Y. Mulkidjanian, Biol. Direct 2009, 4, 26.
Keywords: heterogeneous catalysis · RNA · silicon · surface
analysis · surface chemistry
[18] a) M. C. Ramos-Sanchez, F. J. Rey, F. J. M.-G. M. L. Rodriguez-Mendez, J.
Martin-Gil, Thermochim. Acta 1988, 134, 55–60; b) U. Rꢂisꢂnen, I. Pitkꢂ-
nen, H. Halttunen, M. Hurtta, J. Therm. Anal. Calorim. 2003, 72, 481–
488; c) M. Lappalainen, I. Pitkꢂnen, H. Heikkilꢂ, J. Nurmi, J. Therm. Anal.
Calorim. 2006, 84, 367–376.
[19] M. Mathlouthi, A.-M. Seuvre, J. L. Koenig, Carbohydr. Res. 1983, 122, 31–
47.
[20] V. A. Yaylayan, A. A. Ismail, Carbohydr. Res. 1992, 11, 149–158.
[21] L. T. Zhuravlev, Colloids Surf. A 2000, 173, 1–38.
[22] R. G. Compton, C. H. Bamford, C. F. H. Tipper, Ester Formation and Hy-
drolysis and Related Reactions, Elsevier, Amsterdam, 1972.
[23] T. Rajh, L. X. Chen, K. Lukas, T. Liu, M. C. Thurnauer, D. M. Tiede, J. Phys.
Chem. B 2002, 106, 10543–10552.
[24] J. S. Kwiatkowski, J. Leszczynski, I. Teca, J. Mol. Struct. 1997, 436–437,
451–480.
[1] W. Gilbert, Nature 1986, 319, 618.
[2] S. A. Benner, H.-J. Kim, M. A. Carrigan, Acc. Chem. Res. 2012, 45, 2025–
2034.
[3] C. Meinert, I. Myrgorodska, P. de Marcellus, T. Buhse, L. Nahon, S. V. Hoff-
mann, L. Le Sergeant d’Hendecourt, U. J. Meierhenrich, Science 2016,
352, 208–212.
[4] S. Becker, I. Thoma, A. Deutsch, T. Gehrke, P. Mayer, H. Zipse, T. Carell,
Science 2016, 352, 833–836.
[5] R. Shapiro, Orig. Life Evol. Biosph. 1988, 18, 71–85.
[6] R. Larralde, M. P. Robertson, S. L. Miller, Proc. Natl. Acad. Sci. USA 1995,
92, 8158–8160.
[7] D. Sisak, L. B. McCusker, G. Zandomeneghi, B. H. Meier, D. Blꢂser, R.
Boese, W. B. Schweizer, R. Gilmour, J. D. Dunitz, Angew. Chem. 2010, 49,
4503–4505.
[8] a) A. Ricardo, M. A. Carrigan, A. N. Olcott, S. A. Benner, Science 2003, 63,
96–96; b) H.-J. Kim, A. Ricardo, H. I. Illangkoon, M. J. Kim, M. A. Carrigan,
F. Frye, S. A. Benner, J. Am. Chem. Soc. 2011, 133, 9457–9468; c) J. B.
Lambert, G. Lu, S. R. Singer, V. M. Kolb, J. Am. Chem. Soc. 2004, 126,
9611–9625; d) A. Vꢃzquez-Mayagoitia, S. R. Horton, B. G. Sumpter, J.
Sponer, J. E. Sponer, M. Fuentes-Cabrera, Astrobiology 2011, 11, 115–
121.
[25] T. J. Johnson, R. L. Sams, L. T. M. Profeta, S. K. Akagi, I. R. Burling, R. J. Yo-
kelson, S. D. Williams, J. Phys. Chem. A 2013, 117, 4096–4107.
[26] M. S. Holm, Y. J. Pagꢃn-Torres, S. Saravanamurugan, A. Riisager, J. A. Du-
mesic, E. Taarning, Green Chem. 2012, 14, 702–706.
[27] P. Ortiz, J. Fernꢃndez-Bertrꢃn, E. Reguerac, Spectrochim. Acta Part A
2005, 61, 1977–1983.
[28] a) M. Rodriguez-Garcia, A. J. Surman, G. J. T. Cooper, I. Suꢃrez-Marina, Z.
Hosni, M. P. Lee, L. Cronin, Nature Com. 2015, 6, 8385; b) V. M. Kolb, Int.
J. Astrobiol. 2012, 11, 43–50.
[29] R. C. Hockett, A. Guttag, M. E. Smith, J. Am. Chem. Soc. 1943, 65, 1–3.
[30] S. J. You, Y. T. Kim, E. D. Park, React. Kinet. Mech. Cat. 2014, 111, 521–
534.
[9] D. Bernal, The Physical Basis of Life, Routledge and Kegan Paul, London,
1951.
[10] J.-F. Lambert, Origins Life Evol. Biospheres 2008, 38, 211–242.
[11] a) S. Balme, R. Guegan, J.-M. Janot, M. Lepoitevin, M. Jaber, P. Dejardin,
X. Bourrat, M. Motelica-Heino, Soft Mater. 2013, 9, 3188–3196; b) M.
Lepoitevin, M. Jaber, R. Guegan, J. M. Janot, P. Dejardin, F. Henn, S.
Balme, Appl. Clay Sci. 2014, 95, 396–402.
[31] C. B. Black, H.-W. Huang, J. A. Cowan, Coord. Chem. Rev. 1994, 135/136,
165–202.
[12] J. P. Ferris, A. R. J. Hill, R. Liu, L. E. Orgel, Nature 1996, 381, 59–61.
[13] a) M. Meng, L. Stievano, J.-F. Lambert, Langmuir 2004, 20, 914–923;
b) L. Stievano, L. Piao, I. Lopes, M. Meng, D. Costa, J.-F. Lambert, Eur. J.
Mineral. 2007, 19, 321–331; c) I. Lopes, L. Piao, L. Stievano, J.-F. Lam-
bert, J. Phys. Chem. C 2009, 113, 18163–18172; d) J.-F. Lambert, L. Stie-
vano, I. Lopes, M. Gharsallah, L. Piao, Planet. Space Sci. 2009, 57, 460–
467; e) M. Bouchoucha, M. Jaber, T. Onfroy, J.-F. Lambert, B. Xue, J. Phys.
Chem. C 2011, 115, 21813–21825; f) M. Jaber, J. Spadavecchia, H. Bazzi,
T. Georgelin, F. Costa-Torro, J.-F. o. Lambert, Amino Acids 2013, 45, 403–
406; g) J.-F. Lambert, M. Jaber, T. Georgelin, L. Stievano, Phys. Chem.
Chem. Phys. 2013, 15, 13371–13380; h) T. Georgelin, M. Jaber, H. Bazzi,
J.-F. Lambert, Origins Life Evol. Biospheres 2013, 43, 429–443; i) T. Geor-
gelin, M. Jaber, T. Onfroy, A.-A. Hargrove, F. Costa-Torro, J.-F. o. Lambert,
J. Phys. Chem. C 2013, 117, 12579–12590; j) M. Jaber, T. Georgelin, H.
Bazzi, F. Costa-Torro, J.-F. Lambert, G. Bolbach, G. Clodic, J. Phys. Chem.
C 2014, 118, 25447–25455.
[32] a) A. F. Amaral, M. Matild Marques, J. L. da Silva, J. J. R. Frafflsto da Silva,
New J. Chem. 2008, 32, 2043–2049; b) Y. Furukawa, M. Horiuchi, T. Kake-
gawa, Origins Life Evol. Biospheres 2013, 43, 353–361.
[33] J. B. Lambert, S. A. Gurusamy-Thangavelu, K. Ma, Science 2010, 327,
984–986.
[34] R. M. Hazen, A. Bekker, D. L. Bish, W. Bleeker, R. T. Downs, J. Farquhar,
J. M. Ferry, E. S. Grew, A. H. Knoll, D. Papineau, J. P. Ralph, D. Sverjensky,
J. W. Valley, Am. Mineral. 2011, 96, 953–963.
[35] P. Singh, A. Singh, J. Kaura, W. Holzer, RSC Adv. 2014, 4, 3158–3161.
[36] M. Wang, C. Liu, Q. B. Li, X. X. Xu, J. Mol. Model. 2015, 21, 296.
[37] K. R. Enslow, A. T. Bell, ChemCatChem 2015, 7, 479–489.
[38] B. Danon, G. Marcotullio, W. de Jong, Green Chem. 2014, 16, 39–54.
Received: March 25, 2016
Published online on && &&, 0000
Chem. Eur. J. 2016, 22, 1 – 14
www.chemeurj.org
13
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
FULL PAPER
On silica chemistry: Ribose reactivity
upon thermal activation when adsorbed
on silica, either alone or with the
coadsorption of inorganic salts, in the
“RNA world” scenario is examined (see
scheme). These results allow the likeli-
hood of prebiotic chemistry scenarios to
be evaluated, and may also be of inter-
est for the valorization of biomass-de-
rived carbohydrates by heterogeneous
catalysis.
&
Prebiotic Chemistry
M. Akouche, M. Jaber, E.-L. Zins,
M.-C. Maurel, J.-F. Lambert,* T. Georgelin*
&& – &&
Thermal Behavior of d-Ribose
Adsorbed on Silica: Effect of Inorganic
Salt Coadsorption and Significance for
Prebiotic Chemistry
&
&
Chem. Eur. J. 2016, 22, 1 – 14
www.chemeurj.org
14
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!