A solid-state pH sensor for nonaqueous media including ionic liquids

Article abstract of DOI:10.1021/ac303354q

We describe a solid state electrode structure based on a biologically derived proton-active redox center, riboflavin (RFN). The redox reaction of RFN is a pH-dependent process that requires no water. The electrode was fabricated using our previously described 'stuffing' method to entrap RFN into vapor phase polymerized poly(3,4-ethylenedioxythiophene). The electrode is shown to be capable of measuring the proton activity in the form of an effective pH over a range of different water contents including nonaqueous systems and ionic liquids (ILs). This demonstrates that the entrapment of the redox center facilitates direct electron communication with the polymer. This work provides a miniaturizable system to determine pH (effective) in nonaqueous systems as well as in ionic liquids. The ability to measure pH (effective) is an important step toward the ability to customize ILs with suitable pH (effective) for catalytic reactions and biotechnology applications such as protein preservation.

Full text of DOI:10.1021/ac303354q

Technical Note
pubs.acs.org/ac
A Solid-State pH Sensor for Nonaqueous Media Including Ionic
Liquids
Brianna C. Thompson,* Orawan Winther-Jensen, Bjorn Winther-Jensen,§
,
†,¶
†,‡
,†,‡
and Douglas R. MacFarlane*
†
School of Chemistry, Monash University, Clayton, VIC 3800, Australia
‡
ARC Centre of Excellence for Electromaterials Science, Monash University, Clayton, VIC 3800, Australia
§Department of Materials Engineering, Monash University, Clayton, VIC 3800, Australia
*
S Supporting Information
ABSTRACT: We describe a solid state electrode structure based
on a biologically derived proton-active redox center, riboflavin
(
RFN). The redox reaction of RFN is a pH-dependent process that
requires no water. The electrode was fabricated using our
previously described ‘stuffing’ method to entrap RFN into vapor
phase polymerized poly(3,4-ethylenedioxythiophene). The elec-
trode is shown to be capable of measuring the proton activity in the
form of an effective pH over a range of different water contents
including nonaqueous systems and ionic liquids (ILs). This
demonstrates that the entrapment of the redox center facilitates
direct electron communication with the polymer. This work
provides a miniaturizable system to determine pH (effective) in
nonaqueous systems as well as in ionic liquids. The ability to measure pH (effective) is an important step toward the ability to
customize ILs with suitable pH (effective) for catalytic reactions and biotechnology applications such as protein preservation.
urrently, measurement of proton activity (PA) (and
thereby pH) in nonaqueous systems is hampered by lack
that a means of directly measuring and adjusting PA in such
hydrated IL mixtures is a vital issue.
C
of adequate and easily available measurement systems. Use of
conventional glass-electrode pH probes, without considering
the junction between the media and the water-based reference
solution in the pH electrode, often produces unreliable or
questionable measurements. In recent years, research in the
field of ionic liquids (ILs) has been venturing further into
biotechnological applications and a measure of the effective PA
of the liquid is an important property in this context. This is
particularly true for an understanding of phenomena such as
the stabilization of proteins by some ILs, as well as the use of
In previous studies, the PA of ILs have been estimated by
1
0−13
potentiometric titration,
by using solvatochromic indica-
1
4−19
tors (often using the Hammett method),
of tagged solvent molecules added to the IL or by observation
of proton chemical shift measurements in protic ILs (PILs)
by NMR studies
20
6
using NMR. Byrne and Angell showed that the increase in the
δ(N−H) shift of the PIL indicated an increase in the basic
6
character of the IL. Electrochemical methods have emerged
recently; Bautista-Martinez et al. established an electrochemical
method based on measuring the potential difference between
the onset of the proton reduction and hydrogen oxidation
1
−4
ILs more generally as solvents for pH-sensitive catalysis.
An
2
1
ability to measure PA in ILs would not only help to control/
study reactions where the PA is crucial, but also open up the
route to tailor ILs with suitable PA to preserve therapeutic
reactions to measure PA. Barhdadi et al. used two
electrochemical approaches to determine pK of N-bases in
a
10
several aprotic ILs. Kanzaki and co-workers studied acid−base
properties of N-methylimidazolium based PILs using potentio-
metric and calorimetric titrations. They established correlation
5
,6
protein formulations. Deviation from the protein preferred
PA can affect its folding energy and stability, causing undesired
5
behavior, such as aggregation and fibrillization. Angell et al.
between pK
s
(the acid−base equilibrium constant) and pK and
a
have shown that varying PA can radically change normal folding
found that proton-donating ability of acids is different in the
6
22
processes of proteins. Different proteins have different
PILs. However, these methods are usually based on the use of
7
preferred PA ranges, e.g., lysozyme was stabilized in IL with
a bubbling H electrode which limits their practical use. The
2
10
5
dynamic determination of pK based on cyclic voltammetry
the PA around 4−6. Recent work involving an IL stabilized
a
form of the protein interleukin II, which is of interest in skin
8
cancer treatment, showed that it was necessary to control the
Received: November 19, 2012
Accepted: March 5, 2013
PA in a narrow range in order to fully stabilize this protein. A
9
buffer IL proved to be the means of doing so, demonstrating
©
XXXX American Chemical Society
A
dx.doi.org/10.1021/ac303354q | Anal. Chem. XXXX, XXX, XXX−XXX

Analytical Chemistry
Technical Note
has only been shown thus far to be applicable for measuring
times with water. This led to removal of the excess reduced iron
from the film, collapse of the expanded structure of the polymer
network and trapping of the RFN inside the film. The
composite material was dried at room temperature overnight
before use. Typically, the required area of the working electrode
was only ∼2 × 3 mm.
pK of acid−base systems dissolved in large amount of neutral
a
aprotic IL, not for measuring the PA of the IL itself.
In this paper, we describe the fabrication of a miniaturizable
solid-state electrode structure which is PA sensitive in
nonaqueous liquids. By calibrating the sensor in aqueous buffer
systems, it is possible to estimate equivalent PA in nonaqueous
media and thereby obtain a measure of the apparent or effective
pH (termed “pH (effective)”) in these nonaqueous media.
Given that a standard state for the proton has yet to be
established in these media, this measurement cannot represent
an absolute activity; nonetheless it provides an accessible
method of comparing effective PA across a range of media. The
sensor is based on the biologically derived, proton-active redox
center, riboflavin (RFN). RFN is an essential water-soluble
PA Sensing. The PEDOT/RFN composite electrode was
used as the working electrode in a 3-electrode electrochemical
cell; this allows PA sensing of the electrolyte solution. The
voltage shift of the redox couple of the entrapped RFN was
observed using slow scan rate cyclic voltammetry. Platinum
wire and platinum/cobaltocene were used as counter and
pseudoreference electrode, respectively. As comparisons of
peak potentials were to be made across different solvent
systems, an internal standard redox couple was added into the
2
7,28
vitamin (vitamin B ), which forms the electron shuttling center
electrolyte solution prior to measurement.
As the solubility
2
of flavin adenine dinucleotide (FAD), a coenzyme important in
many enzymatic reactions. The oxidation and reduction of RFN
of the commonly used ferrocene was insufficient in aqueous
solutions, cobaltocene was used as reference redox couple in
these solutions. The choice of cobaltocene or ferrocene as an
internal reference in this work is designed to remove as far as
possible any effect of the changing solvent environment on the
redox potential of the reference system. Ten millimoles of
cobaltocene (Sigma-Aldrich) was added to 1 mL of the
electrolyte solution to be tested, and gently heated to 40 °C for
up to an hour to assist dissolution.
(
see Scheme 1) is a PA-dependent process that requires no
Scheme 1. Reduction and Oxidation of RFN Shows a
Proton-Dependent Process Involving Two Electrons
Nitrogen gas was bubbled into the electrolyte for at least 30
min before CV scans were started. Scans were also performed
in electrolyte systems without cobaltocene. At least 3 CV scans
at 10 mV/s were made in each electrolyte, with the voltage
range between −1.5 and −0.2 V vs platinum. The peak
positions of the RFN and cobaltocene reduction and oxidation
peaks were measured, and these peak positions were used to
provide an estimate of the pH of the electrolyte solution
(Figure 1).
2
3,24
water.
The reduction and oxidation of the RFN complex
can be followed electrochemically. In this work, a PA sensitive
electrode was constructed by incorporation of RFN into a
conducting polymer, poly(3,4-ethylenedioxythiophene)
pH Calculations. CV’s from various pH buffers described in
section Buffer Preparation, using PEDOT/RFN as the working
electrode, were used to construct a calibration curve. The peak
potentials (E , E ) or E obtained from the mid point
(
PEDOT), using the stuffing method previously shown to
ox
red
0
successfully establish the direct communication between the
between Eox and Ered of RFN were determined relative to the
peak potential of cobaltocene, and plotted against the pH of the
solution. Linear regressions of the data were performed to
determine formulas and the standard error for estimations of
pH.
2
5,26
polymer and a biomolecule.
After calibration in various
buffer-systems, the sensor capability is then demonstrated in a
nonaqueous acid−base titration and to determine pH
(
effective) of a protic IL of particular interest in biotechno-
logical applications, choline dihydrogenphosphate (choline
dhp).
To determine pH (effective) of the hydrated IL choline
dihydrogen phosphate (choline dhp), the peak positions of
RFN relative to the standard cobaltocene peaks were
determined by CV scans of PEDOT/RFN in the mixtures of
10%, 20%, 40%, 60% and 80% (w/w) choline dhp in distilled
METHODS
■
Buffer Preparation. To test the performance of the
PEDOT/RFN electrode as a pH (effective) sensor in solutions
with a range of ions, aqueous buffers were prepared across a pH
range of 3−9. Details of the buffers used are as follows;
Potassium hydrogen phthalate buffers (C H KO ): pH 3.05,
water. The difference in E voltages (RFN vs cobaltocene) and
0
the formulas displayed in Figure 2 were used to determine the
pH (effective) of the mixtures.
Acid−Base Titrations in Nonaqueous Media. PEDOT/
RFN electrodes were used to monitor the pH (effective)
change during the titration of triflic acid (Sigma) with n-
butylamine (Sigma). Triflic acid forms a solid salt at the end
point with butylamine; therefore, anhydrous propylene
carbonate was used as a medium for the titration. One milliliter
of triflic acid was mixed with 2 mL f propylene carbonate. Seven
millimolar tetrabutylammonium hexafluorophosphate and 8
mM ferrocene were added as supporting electrolyte and as the
internal standard, respectively. Ferrocene was used instead of
cobaltocene in this experiment due to significant proton
reduction at potentials higher than the redox reaction of
8
5
4
pH 4.00, pH 5.06. Sodium phosphate buffers (Na HPO /
2
4
NaH PO ): pH 4.4, pH 6.12, pH 7.00, pH 7.91, pH 8.55. The
2
4
pH of these buffers was measured using a Mettler Toledo pH
meter.
PEDOT-RFN Electrode Preparation. The method of
trapping the RFN in the network is based on the procedure
25
26
described by Winther-Jensen et al. and Thompson et al.
Vapor phase polymerized PEDOT film was prepared as
2
5,26
previous reported
(see detail in Supporting Information).
The unwashed film was soaked in a solution consisting of a
saturated aqueous RFN solution (1.5 g/L) and washed several
B
dx.doi.org/10.1021/ac303354q | Anal. Chem. XXXX, XXX, XXX−XXX

Analytical Chemistry
Technical Note
cobaltocene. CV scans in the range of +0.5 to −1 V (vs
platinum) were performed. The pH (effective) was determined
from the shift of the RFN reduction peak after conversion to
the cobaltocene potential scale based on the Ferrocene/
shifts linearly with pH, and can be used to yield pH estimates in
a small volume of electrolyte. A slope of 0.059 V per decade
would be expected for a 1 e /1 H or 2 e /2 H process and
−
+
−
+
the observed slope in the pH range 3−9 is close to this value.
2
+
30
Ferricinium − Co(C H ) /Co(C H ) difference being
However, as shown recently, the process is somewhat more
5
5
5
5 2
2
9
−
1.337 V. The reduction peak potentials were used (see
complex and involves several intermediate steps, the result of
which could produce a lower slope overall.
Supporting Information) in this case due to poor resolution of
the oxidation peak potentials in highly acidic and highly basic
conditions.
pH (Effective) Estimates in a Nonaqueous Acid−Base
Titration. To demonstrate the use of the PEDOT-RFN
electrode in a nonaqueous context, the pH (effective) was
monitored during addition of butylamine to triflic acid as
concentrated solutions (33% v/v) in propylene carbonate.
Protic ionic liquids based on the triflate anion have been
explored extensively by Watanabe’s group in recent years and
their potentiometric measurements provide some insight into
the effective proton activity in these media under various
Safety Considerations. Triflic acid fumes when expose to
moisture in the air and has to be handled in a fume hood.
Acid−base titration of triflic acid and butylamine has to be
carried out in a fume hood as the reaction can be vigorous.
Personal protective equipment (gloves, apron and goggles) has
to be worn.
31
RESULTS AND DISCUSSION
conditions of temperature and concentration. The titration
began from pure triflic acid and butylamine was gradually
added. A CV was measured after each addition of butylamine
and the reduction potential differences between RFN and
ferrocene were obtained. The pH (effective) estimates were
calculated as described in the Methods section and the resulting
titration curve is shown in Figure 3. Triflic acid is a strong acid
■
Aqueous Buffer Measurements. The PEDOT-RFN
electrode was first tested in aqueous buffers to confirm the
linearity of the RFN peak potential shifts across a range of pH
ranges between 3 and 9 with a variety of buffer compositions.
Typical cyclic voltammograms (CVs) are shown in Figure 1.
Figure 1. CVs of PEDOT/RFN in aqueous pH 7.0 phosphate buffer
with a cobaltocene (Cc) internal reference. The peaks used in the pH
calculations are labeled.
Figure 3. pH (effective) estimate during an acid−base titration of
triflic acid with butylamine as concentrated solutions in propylene
carbonate. The pH (effective) error estimates from the calibration
curve (±0.02) are within the size of the data points.
The RFN peak positions relative to the cobaltocene peak
shifted linearly with pH, with higher voltages observed in
solutions with lower pHs, as shown in Figure 2. Beyond the pH
and thereby understood to be an active source of protons. At
the beginning of the titration, a strong proton reduction wave
appeared and, only after a small amount of butylamine
(
Xbutylamine = 0.008) had been added, was the riboflavin
reduction peak distinguishable from the proton reduction
wave. Upon increasing the amount of butylamine, the pH
(
3
effective) increased and a clear step was observed between pH
and pH 8. From an estimate of the point of maximum slope
in Figure 3, the end point was estimated to be pH (effective)
6
0
0
.4 where the mole fraction of butylamine (Xbutylamine) was
.511 ± 0.006. The end point is quite clearly not Xbutylamine
.5 from Figure 3 and this is thought to be the result of small
=
Figure 2. E0 (RFN) in the PEDOT/RFN electrode versus a
cobaltocene (Cc) reference redox pair in a range of buffer solutions.
amounts of impurities in the butylamine (including water and
CO2 absorbed during the procedure). This illustrates the
usefulness of the electrode in revealing such impurity issues. In
the context of protic ILs, which are often prepared in a similar
fashion, a 1:1 mol ratio used to produce such an ionic liquid
would actually be slightly acid-rich, potentially producing a
distinctly acidic IL.
range 3−9, the peak positions become less responsive to pH.
2
The R value of the linear fit in Figure 2 is 0.99, indicating good
linearity of the data. The calculated standard errors associated
with estimates of pH based on the linear regression were ±0.02.
Thus, these studies in aqueous solution demonstrate that the
PEDOT-RFN electrode is capable of producing a signal that
pH (Effective) Estimates in a Hydrated IL across a
Range of Water Activities. The PEDOT/RFN electrode
C
dx.doi.org/10.1021/ac303354q | Anal. Chem. XXXX, XXX, XXX−XXX

Analytical Chemistry
Technical Note
material was used in choline dhp with a range of water contents
to provide an insight as to how pH (effective) is affected by the
water activity. Choline dhp is typical of the hydrated ILs used in
biological applications. The estimates of pH (effective) based
choline dhp water mixtures and in a nonaqueous acid−base
titration. The general approach we describe here of the use of a
conducting polymer to host a biomolecule/redox center as a
chemical sensing species has wider applicability to biotechno-
logical applications, particularly for applications involving
proteins or biochemical processes, for example, biosensors
and biocatalytic electrodes.
on shifts in E are shown in Figure 4 below.
0
ASSOCIATED CONTENT
Supporting Information
■
*
S
Vapor phase polymerized (VPP) PEDOT film preparation and
Standard curve for Ered of RFN vs cobaltocene. This material is
available free of charge via the Internet at http://pubs.acs.org.
AUTHOR INFORMATION
Corresponding Author
D.R.M.: address, School of Chemistry Monash University,
■
*
Clayton, VIC 3800 Australia; e-mail, douglas.macfarlane@
monash.edu; fax, +61 3 9905 4597; tel, 61 3 9905 4540. B.C.T.:
e-mail, brianna@uow.edu.au; fax, +61 2 4298 1477; tel, 61 2
Figure 4. pH (effective) estimates for choline dhp across a range of
water contents.
4
298 1905.
These results clearly indicate a somewhat unexpected set of
trends in pH (effective) as water content is varied in the choline
dhp-water system. Beginning with the dilute aqueous solution,
the pH (effective) initially drops below that of the dilute
aqueous value (∼5.5) with increasing choline dhp concen-
tration. This reflects the impact of the increasing dhp content
on ion activity in an electrolyte solution such as this. At higher
salt concentrations, the trend reverses and the pH (effective)
begins to rise with increasing salt content due to the impact of
Present Address
B.C.T.: Intelligent Polymer Research Institute, University of
Wollongong, AIIM Facility, Innovation Campus, University of
Wollongong, NSW 2522, Australia.
¶
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
D.R.M., B.W.-J. and O.W.-J. gratefully acknowledge the
Australian Research Council for fellowships.
the decreasing water activity on the dhp dissociation reaction
+
(
eq 1); the free energy of formation of the H O species
3
represents an important contributing factor in the overall free
energy change in this process.
REFERENCES
■
H PO + H O = HPO4 + H O+
−
2−
(1) Adams, C. J.; Earle, M. J.; Roberts, G.; Seddon, K. R. Chem.
Commun. 1998, 2097−2098.
2) Adams, C. J.; Earle, M. J.; Seddon, K. R. Green Chem. 2000, 2,
2
4
2
3
(1)
(
Solvation of the dhp anions by one another in complex-ion
32
21−23.
species of the type discussed by Johannsson et al. may also be
a factor in this increasing pH (effective) as this type of solvation
stabilizes the labile proton in the complex. Further studies to
confirm this behavior will be reported elsewhere.
(
3) Forsyth, S. A.; MacFarlane, D. R.; Thomson, R. J.; Von Itzstein,
M. Chem. Commun. 2002, 714−715.
(
(
4) Welton, T. Coord. Chem. Rev. 2004, 248, 2459−2477.
5) Angell, C. A.; Byrne, N.; Belieres, J. P. Acc. Chem. Res. 2007, 40,
This work has demonstrated that electrodes constructed
from PEDOT combined with the PA-sensitive biological redox
mediator provide a simple way to electrochemically estimate
the pH of very low water content media. By being able to
measure the pH (effective) of media across a range of water
activities and neat acid/base compositions of the type used in
protic ionic liquids, a better understanding and control of
proton activity can be realized. The reference redox couple
1228−1236.
(6) Byrne, N.; Angell, C. A. J. Mol. Biol. 2008, 378, 707−714.
(7) Byrne, N.; Wang, L. M.; Belieres, J. P.; Angell, C. A. Chem.
Commun. 2007, 2714−2716.
(8) Foureau, D. M.; Jones, C. P.; Weaverz, K.; Vrikkis, R. M.;
MacFarlane, D. R.; Mckillop, I. H.; Salow, J. C.; Elliott, G. D. J.
Immunother. 2010, 33, 896.
(
9) MacFarlane, D. R.; Vijayaraghavan, R.; Ha, H. N.; Izgorodin, A.;
Weaver, K. D.; Elliott, G. D. Chem. Commun. 2010, 46, 7703−7705.
10) Barhdadi, R.; Troupel, M.; Comminges, C.; Laurent, M.;
(
ferrocene or cobaltocene) has been dissolved in the mixture in
(
this work; however, it could equally be isolated in a separate
fritted reference electrode assembly as is common in ion
selective electrodes.
Doherty, A. J. Phys. Chem. B 2012, 116, 277−282.
(11) Geng, W. G.; Li, X. H.; Wang, L. F.; Duan, H. L.; Pan, W. P.
Acta Phys.-Chim. Sin. 2006, 22, 230−233.
(12) Kanzaki, R.; Uchida, K.; Song, X.; Umebayashi, Y.; Ishiguro, S. I.
CONCLUSION
In summary, we have demonstrated a system to measure pH
effective) trends in aqueous and nonaqueous solutions
covering pH 3−9 using a bioderived redox couple incorporated
in PEDOT. The system is suitable for miniaturization as only a
few square millimeters of electrode area is needed and there is
Anal. Sci. 2008, 24, 1347−1349.
13) Malham, I. B.; Letellier, P.; Turmine, M. Talanta 2008, 77, 48−
2.
14) Thomazeau, C.; Olivier-Bourbigou, H.; Magna, L.; Luts, S.;
Gilbert, B. J. Am. Chem. Soc. 2003, 125, 5264−5265.
■
(
(
5
(
(15) Chu, Y.; Deng, H.; Cheng, J. P. J. Org. Chem. 2007, 72, 7790−
7
793.
no need for H bubbling reference electrode. The established
(16) Xing, H. B.; Wang, T.; Zhou, Z. H.; Dai, Y. Y. J. Mol. Catal. A-
2
calibration curve was then used to estimate pH (effective) in
Chem. 2007, 264, 53−59.
D
dx.doi.org/10.1021/ac303354q | Anal. Chem. XXXX, XXX, XXX−XXX

Analytical Chemistry
Technical Note
(
17) D’Anna, F.; La Marca, S.; Noto, R. J. Org. Chem. 2009, 74,
952−1956.
18) Robert, T.; Magna, L.; Olivier-Bourbigou, H.; Gilbert, B. J.
Electrochem. Soc. 2009, 156, F115−F121.
19) Zhao, Y. W.; Long, J. X.; Deng, F. G.; Liu, X. F.; Li, Z.; Xia, C.
G.; Peng, J. J. Catal. Commun. 2009, 10, 732−736.
20) Angueira, E. J.; White, M. G. J. Mol. Catal. A: Chem. 2007, 277,
64−170.
21) Bautista-Martinez, J. A.; Tang, L.; Belieres, J. P.; Zeller, R.;
Angell, C. A.; Friesen, C. J. Phys. Chem. C 2009, 113, 12586−12593.
22) Kanzaki, R.; Doi, H.; Song, X.; Hara, S.; Ishiguro, S. I.;
Umebayashi, Y. J. Phys. Chem. B 2012, 116, 14146−14152.
23) Sawyer, D. T.; Gerber, J. N.; Amos, L. W.; De Hayes, L. J. J. Less-
Common Met. 1974, 36, 487−499.
24) Tatwawadi, S. V.; Santhanam, K. S. V.; Bard, A. J. J. Electroanal.
Chem. Interfacial Electrochem. 1968, 17, 411−420.
25) Thompson, B. C.; Winther-Jensen, O.; Vongsvivut, J.; Winther-
Jensen, B.; MacFarlane, D. R. Macromol. Rapid Commun. 2010, 31,
293−1297.
26) Winther-Jensen, B.; Chen, J.; West, K.; Wallace, G. Polymer
005, 46, 4664−4669.
27) Hultgren, V. M.; Mariotti, A. W. A.; Bond, A. M.; Wedd, A. G.
Anal. Chem. 2002, 74, 3151−3156.
28) Silvester, D. S.; Ward, K. R.; Aldous, L.; Hardacre, C.; Compton,
R. G. J. Electroanal. Chem. 2008, 618, 53−60.
29) De Vreese, P.; Haerens, K.; Matthijs, E.; Binnemans, K.
Electrochim. Acta 2012, 76, 242−248.
30) Tan, S. L. J.; Webster, R. D. J. Am. Chem. Soc. 2012, 134, 5954−
964.
31) Lee, S. Y.; Ogawa, A.; Kanno, M.; Nakamoto, H.; Yasuda, T.;
Watanabe, M. J. Am. Chem. Soc. 2010, 132, 9764−9773.
32) Johansson, K. M.; Izgorodina, E. I.; Forsyth, M.; MacFarlane, D.
R.; Seddon, K. R. Phys. Chem. Chem. Phys. 2008, 10, 2972−2978.
1
(
(
(
1
(
(
(
(
(
1
(
2
(
(
(
(
5
(
(
E
dx.doi.org/10.1021/ac303354q | Anal. Chem. XXXX, XXX, XXX−XXX