A Green-LED Driven Source of Hydrated Electrons Characterized from Microseconds to Hours and Applied to Cross-Couplings

Article abstract of DOI:10.1002/chem.201800626

We present a novel photoredox catalytic system that delivers synthetically usable concentrations of hydrated electrons when illuminated with a green light-emitting diode (LED). The catalyst is a ruthenium complex protected by an anionic micelle, and the urate dianion serves as a sacrificial donor confined to the aqueous bulk. By virtue of its chemical properties, this donor not only suppresses charge recombination that would limit the electron yield, but also enables this system to perform cross-couplings through the action of hydrated electrons, the first examples of which are reported here. We have investigated the kinetics of all the steps involving the electron and its direct precursor in a comparative study by means of laser flash photolysis and by monitoring product formation during LED photolysis. Despite the differences in timescales, each approach on its own already gives a complete picture of the reaction over a temporal range spanning ten orders of magnitude. Discrepancies between the kinetic parameters obtained with the two complementary techniques can be rationalized with the slow secondary chemistry of the system; they reveal that the product-based method provides a more accurate description because it also responds to the changes of the system composition during a synthesis; hence, they demonstrate that in complex systems the timescale of the experimental observation should be matched to that of the actual application.

Full text of DOI:10.1002/chem.201800626

A Journal of
Accepted Article
Title: A green-LED driven source of hydrated electrons characterized
from microseconds to hours and applied to cross couplings
Authors: Robert Naumann and Martin Goez
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: Chem. Eur. J. 10.1002/chem.201800626
Link to VoR: http://dx.doi.org/10.1002/chem.201800626
Supported by

10.1002/chem.201800626
Chemistry - A European Journal
A green-LED driven source of hydrated electrons characterized from
microseconds to hours and applied to cross couplings†
By Robert Naumann and Martin Goez∗
Abstract: We present a novel photoredox catalytic system that delivers synthetically
useable concentrations of hydrated electrons when illuminated with a green light-emitting
diode (LED). The catalyst is a ruthenium complex protected by an anionic micelle, and
the urate dianion serves as sacrificial donor confined to the aqueous bulk. Through its
chemical properties, that donor not only suppresses charge recombination that would limit
the electron yield but also enables this system to perform cross couplings via hydrated
electrons, for which we report the first examples. We have investigated the kinetics of all the
steps involving the electron and its direct precursor in a comparative study by using laser
flash photolysis and by monitoring product formation during LED photolysis. Despite the
differences in timescales, each approach on its own already gives a complete picture of
the reaction over a temporal range ten orders of magnitude wide. Noticeable discrepancies
between the kinetic parameters obtained with the two complementary techniques can be
rationalized with the slow secondary chemistry of the system; they reveal that the product-
based method provides a more accurate description because it responds also to the changes
of the system composition during a synthesis; hence, they demonstrate that in complex
systems the timescale of the experimental observation should be matched to that of the
actual application.
1 Introduction
•−
Photoredox catalysis with the hydrated electron
e
as a relay unleashes a reductive power
aq
stronger than that of most excited states, yet conveys it through an intrinsically much longer-
•−
lived intermediate (standard potential of
e
,−
2.9 V; unquenched lifetime, typically 1–2
µ
s)1.
aq
•−
As specific ”green” advantages, the attack of
the solvent is water.
e
on a substrate affords no by-product, and
aq
•−
We recently found a sustainable access to
e
with visible light by the two-photon ioniza-
aq
tion of [Ru(bpy)3]2+ as the catalyst with ascorbate Asc2− as a sacrificial donor.2 When we
raised the Asc2− concentration to a few tens of mM, this system afforded laboratory-scale
•−
concentrations of
e
with single flashes from a green solid-state laser;3 and when we addi-
aq
tionally shielded the catalyst by an anionic micelle (of sodium dodecyl sulfate SDS), a green
light-emitting diode (LED) sufficed for the same purpose,4 which now allows employing the
•−
super-reductant
e
as a routine reagent, without special instrumentation, and without the
aq
safety hazards incurred by lasers or radiolysis.
This work aims at enlarging the scope of the [Ru(bpy)3]2+ electron source to encompass
•−
also
e
-induced cross couplings, an objective intrinsically incompatible with ascorbate as
aq
∗Robert Naumann, Prof. Dr. Martin Goez,
Martin-Luther-Universität Halle-Wittenberg, Institut für Chemie, Kurt-Mothes-Str. 2, D-06120 Halle (Saale),
Germany.
E-mail: martin.goez@chemie.uni-halle.de
†Supporting Information available: NMR spectra of additional cross couplings. See DOI: ...
1
This article is protected by copyright. All rights reserved.

10.1002/chem.201800626
Chemistry - A European Journal
the sacrificial component: the monoanion HAsc− has a pKa of 11.74;5 hence, even at the
strongly basic pH of our experiments, 12.7, the potent hydrogen donor HAsc− is still present
in mM concentrations, and thus efficiently scavenges the intermediate substrate-derived
radicals before they can undergo cross coupling. To avoid this scenario, we chose the urate
dianion Ur2− (for the structural formula, see Figure 1b below) as a replacement. Urate
is a classical water-soluble antioxidant with a lower acidity of its monoanion (pKa, 9.8)6
compared to HAsc−. At pH 12.7, therefore, it exists practically quantitatively as Ur2−, which
is a good electron donor but cannot directly transfer a hydrogen atom.7
An equally strong motive for selecting Ur2− as the sacrificial agent was an envisaged
mechanistic advantage for the catalytic cycle. As discussed in detail in Section 3.1, the
•−
cornerstone for successful
e
production by LED illumination is a sufficiently slow recombi-
aq
nation between the reduced catalyst and the oxidized electron donor, the former residing
inside the micelle and the latter in the aqueous bulk. We expected Ur2− to be favourable in
this respect because its oxidized form Ur•− possesses a pKa of 9.56 and should thus be
deprotonated quasi-instantaneously at pH 12.7 to give the dianionic radical Ur•2−, which
experiences a stronger Coulombic repulsion by the anionic micelle than does the singly
charged Asc•−; and a further deprotonation of Ur•2−
more. These electrostatic effects should thus decelerate the recombination.
(
pKa ≈ 13)8 increases this effect even
As additional benefits, Ur2− is similarly inexpensive and equally bioavailable as is Asc2−
but exhibits a much lower susceptibility to oxygen; and it remains completely transparent
above 450 nm throughout the reaction whereas Asc2− has a propensity to turn brown, which
reduces the amount of light absorbed productively and heats up the samples at larger
turnover.
Before furnishing•p−roof of principle that the [Ru(bpy)3]2+
/
Ur2− system can effect cross
couplings through eaq , we present a comprehensive mechanistic and kinetic investigation
•−
of the reaction steps in our catalytic electron source up to and including the capture of
e
aq
by a substrate. Leading photochemists have recently criticised a tendency to neglect such
issues in studies on systems for photoredox catalysis, in particular with respect to the fast
kinetics of the elementary processes;9–11 on the other hand, many synthetic laboratories
are not equipped for performing time-resolved photochemical experiments on the required
timescales. In an attempt to help resolve this controversy, we compare the kinetics obtained
by nanosecond laser flash photolysis and by monitoring the evolution of the reactant and
product concentrations during preparative illumination for hours. As will emerge, the outcome
is not always identical because the experiments on the short timescales do not capture
the changes of the system during the course of a synthesis; but our results will show how
the kinetic parameters relevant for an actual application can be obtained by experiments
accessible to every laboratory.
2 Experimental
All chemicals were obtained in the highest available purity and used as received.
Nanosecond laser flash photolysis was performed with a setup described elsewhere.3 Its
relevant features include high-intensity excitation (pulse width, 5 ns) with collimated beams
sent through optically thin solutions, such that absolute concentrations of transients can be
2
This article is protected by copyright. All rights reserved.

10.1002/chem.201800626
Chemistry - A European Journal
reliably measured; detection by absorption or luminescence; and a flow-through system to
ensure that each laser flash hits fresh solution.
For the LED experiments, 3.8 ml of the degassed solutions were irradiated in a fluo-
rescence cuvette with a Prizmatix UHP-T-LED-520 (maximum optical power, 1.8 W, with
reproducible variation by the current source; emission maximum and full width at half maxi-
mum, 520 nm and 37 nm, closely matching the absorption maximum of OER in the visible;
beam collimated to a diameter of 8 mm) under continuous stirring. The partial illumination
of the sample and the volume change during the experiments were taken into account by
correcting the illumination times as derived previously.12
After predetermined times, the LED was blocked for about 1 min while aliquots of 0.2 ml
and 40 µ
l were removed. The concentrations of Cl− and of remaining [Ru(bpy)3]2+ were
determined in the former aliquot (by using an ion-sensitive electrode and the luminescence,
both with the procedures described before)4, those of Ur2− by diluting the latter aliquot with
20 ml of the basic SDS medium (50 ml, pH 12.7) and recording an absorption spectrum (see,
Figure 4a).
NMR product analysis at illumination endpoints was performed on an Agilent Technolo-
gies 500 MHz DD2 spectrometer and involved the same procedure as before (PRESAT
pulse sequence for suppressing the intense water signal, addition of 10 % v/v of D2O for
shimming and locking, acidification to pH 1 in the case of ClAc, SDS as reference for the
concentrations).3,4
3 Results and discussion
3.1 Mechanism and kinetics
We have already reported on the cyclic photoionization of [Ru(bpy)3]2+ with ascorbate Asc2−
as the sacrificial donor.2–4 Because the light is only absorbed by the complex, a replacement
of Asc2− by urate Ur2− leaves unchanged the two photoprocesses involved but modifies
the intervening and subsequent thermal reactions. Figure 1a sums up the mechanism, and
Figure 1b displays the formulas of the two key compounds [Ru(bpy)3]2+ and Ur2−
.
The active species of [Ru(bpy)3]2+ is its one-electron reduced form OER. One green
•−
photon liberates the hydrated electron
e
from OER with an intensity-proportional rate
aq
constant κionI , concomitantly recovering the complex in its ground state GS, another green
photon converts GS into the metal-to-ligand charge-transfer excited state 3MLCT with rate
constant κexcI. Electron-transfer quenching of 3MLCT by the sacrificial donor (rate constant,
kq; efficiency, η; unquenched lifetime, τ0) connects the two ends of this chain in the desired
way for a catalytic cycle, whereas recombination of OER and the donor radical (rate constant,
krec) competes with the electron ejection and acts as an unwanted shunt to GS.
Evidently, the ability of this mechanism to operate under the low photon flux I of an LED
pivots on suppressing that recombination as much as possible. At first glance, this suggests
an uphill struggle since the oxidation necessarily turns the sacrificial donor into a species
with a charge more positive by one unit, hence attracted more strongly or repelled less
weakly by the anionic micelle than was the donor. However, one can elegantly circumvent
3
This article is protected by copyright. All rights reserved.

10.1002/chem.201800626
Chemistry - A European Journal
•2− + H+
Ur2−
O
N
(a)
(b)
Ur
H
N
kq, η
HN
−O
Ur2−
O−
krec
P
N
3MLCT
OER
ks
•−
N
N
N
S
1/τ0
[Ru(bpy)3]2+
Ru2+
e
aq
κion
I
N
N
GS
N
1/τe
κexc
I
GS
•−
Figure 1: Mechanism of the cyclic source of hydrated electrons e . Graph (a), cartoon representation of the reaction scheme. Only
aq
the catalyst [Ru(bpy)3]2+ in its three forms (ground state GS, metal-to-ligand charge-transfer excited state•−MLCT, and one-electron
3
reduced form OER) is solubilized by the SDS micelle; the quencher Ur2−, its resulting radical Ur•2−, and eaq reside exclusively in the
•−
aqueous phase. The photons from the green LED are absorbed by GS and OER. After ejection, e
reacts with a hydrophilic substrate
aq
S, ultimately yielding a product P. All kinetic parameters needed for analysis with Equation 1, below, are given at the reaction arrows (in
green for the light-driven steps, the rate constants of which are proportional to the LED intensity I). Graph (b), structural formulas of the
catalyst [Ru(bpy)3]2+ in its ground state GS and of the quencher, the urate dianion Ur2−
.
this problem by employing Ur2− because the radical produced by its oxidation possesses
pKa of 9.5,6 so immediately loses a proton in sufficiently basic medium (pH in all our
experiments, 12.7) to give the dianionic intermediate Ur•2−
A first prognosis of whether a combination of [Ru(bpy)3]2+ with a particular quencher might
qualify as an electron source should be based on the kinetic parameters kq, τ0, and krec
a
.
η
,
of the two dark reactions leading to and from OER. Figure 2 collects the results obtained by
laser flash photolysis for the system of this work.
(a)
(b)
1.0
0.5
0
0.6
0.3
0
0.8
with Asc•−
10
8
6
4
2
0
with Ur•2−
with Ur•2−
0
3
t
6
/ ms
9
0
0
50
[Ur2−] / mM
with Asc•−
0
2
4
0
3
6
9
t / µs
t / ms
Figure 2: OER formation and decay in the system of this work (in the case of the decay also compared with Asc2− as the quencher),
as studied by laser flash photolysis. Common experimental conditions, 50µM [Ru(bpy)3]2+ in 50 mM aqueous SDS at pH 12.7; ionic
strength kept at the value with the respective maximum quencher concentration by adding Na2SO4; excitation with a 532 nm flash
of intensity 458 mJ cm−2. Graph (a), main plot; quenching of the 3MLCT luminescence by Ur2− (concentration in mM / colour code,
0 / black, 10 / violet, 20 / green, 30 / teal, 40 / red); traces normalized to the maximum amplitude without quencher and overlaid with best-
fit functions (dashed lines) exp(−t/τ); resulting unquenched lifetime τ0, 780 ns. Inset, corresponding Stern–Volmer plots for Ur2− using
the same colour coding as in the main plot, slope of the regression line, 16.7 M−1. Graph (b), OER decay in experiments with identical
initial radical formation; quencher concentrations / colour code, 40 mM Ur2− / orange, 18 mM Asc2− / brown; main plot, experimental
traces; inset, linearization for a second-order process, overlaid regression lines corresponding to rate constants of 8.0 × 107
M
−1s−1
(Asc•−) and 3.8 × 107
M
−1s−1 (Ur•2−).
The decays of 3MLCT in the presence of variable Ur2− concentrations all start from the
same initial value and obey clear first-order laws with rate constants exhibiting Stern–Volmer
4
This article is protected by copyright. All rights reserved.

10.1002/chem.201800626
Chemistry - A European Journal
behaviour (Figure 2a). These observations establish that Ur2− quenches 3MLCT purely
dynamically and across the micelle-water interface in the same way as does Asc2−, but
with an almost three times smaller rate constant (2.1
×
107 M−1s−1) owing to its more
positive standard potential (0.26 V)6. The good solubility in water (saturation concentration
of disodium urate, about 80 mM; of dilithium urate, several hundreds of mM) facilitates
compensating for this deceleration by raising the quencher concentration. We measured
an efficiency
value with Asc2− and compares very favourably with phenolates, anilines, thiocarbamate
or cysteine, none of which we found to reach a higher than 0.2. Owing to the long
unquenched lifetime of 3MLCT s), our standard concentration of Ur2− (40 mM)
τ0, 0.78
η
of 0.48 for OER formation with Ur2−, which is almost three quarters of the
η
(
µ
thus suffices for producing one molecule of OER for every fifth photon.
The recombination of OER and the quencher-derived radical plays an even more important
role than does the 3MLCT quenching because it determines the storage duration and
hence the probability of absorbing the ionizing photon. Decisive are not only a low rate
constant but also a reaction order higher than first order.4 Figure 2b juxtaposes decay
traces of OER accessed through 3MLCT quenching by Ur2− and Asc2−, where the quencher
concentrations were chosen such as to produce identical initial amounts in both systems.
Direct observation at a wavelength practically coinciding with the OER absorption maximum
is possible because all quencher-derived species are transparent in this spectral range4,6
while 3MLCT and GS are blanked through having an isosbestic point. As is immediately
perceived, OER recombines significantly more slowly with Ur•2− than with Asc•−, which
proves our strategy of compensating the more positive charge on the quencher radical by
a subsequent deprotonation to be workable. What is more, the inset of Figure 2b reveals
systematic and progressive deviations of the OER/Ur•2− recombination from second-order
kinetics at longer times such that the OER decay becomes increasingly slower and has
almost come to a halt at the end of the observation period. This effect is seen to be absent
on the same (10 ms) timescale when Asc2− is the sacrificial donor, and we ascribe it to a
faster removal of Ur•2− from the system by disproportionation than in the case of Asc•−
.
These encouraging results prompted us to investigate the performance of this system
with the chloroacetate assay. This assay is specific through neither responding to reductants
•−
weaker than
e
nor poisoning catalysts solubilized in SDS micelles; moreover, it directly
aq
corresponds to an environmentally relevant application. The reaction scheme is displayed in
•−
Figure 3a. The highly toxic chloroacetate ClAc captures e with a rate constant in excess
aq
of 109 M−1s−1, the exact value depending on the ionic strength of the solution. The electron
attachment is dissociative and results in the harmless chloride ion and the carboxymethyl
radical, which finally stabilizes to give the equally innocuous acetate Ac. As illustrated by the
NMR spectra shown in the Figure, exhaustive illumination of 1 mole percent of the catalyst
destroyed most (
>
95 %) of the substrate ClAc with practically complete recovery as Ac; the
1 %). However,
amount of the radical combination product succinic acid was negligible (
<
considerably more information than from the system composition at the end point is obtained
from the development of the concentrations during the illumination period. Figures 3b and 4a
collect this for the catalyst and Cl−, as well as for the sacrificial donor. The analysis involves
discontinuous operation: the illumination of the samples is interrupted at certain times to
remove small aliquots, and the associated reduction of the sample volumes is taken into
account by correcting the duration of the subsequent illumination.
5
This article is protected by copyright. All rights reserved.

10.1002/chem.201800626
Chemistry - A European Journal
(a)
(b)
1
1.0
1
0
0.8
0.6
0.4
0.2
0
0
•−
+ e
aq
ClAc
Cl−+ Ac•
Ac
0
2
1
P
/ W
0
4.2
4.1
4.0
2.0
1.9
0
2
4
6
8
10
δ / ppm
t / h
Figure 3: Characterizing the catalytic electron source with the chloroacetate assay. System composition, 0.1 mM [Ru(bpy)3]2+ and
40 mM Ur2− in 50 mM aqueous SDS at pH 12.7; 10 mM chloroacetate ClAc. Graph (a), NMR spectra with identical vertical scales before
(bottom trace) and after (top trace) illumination for 10 h with a green LED (520 nm). The toxic ClAc (red singlet, 4.12 ppm) is converted
near-quatitatively into the harmless acetate Ac (green singlet, 1.94 ppm) as shown by the reaction scheme between the traces. Graph (b),
kinetics during the experiment of graph (a). Main plot, relative concentrations of liberated chloride ions (cyan filled circles and solid fit
curve 0.954{1 − exp[−t/(2.08 h)]}) and of remaining catalyst (magenta open circles and dashed fit curve exp[−t/(5.79 h)]); inset,
relative initial slope of Cl− production as function of the LED radiative power P, with fit curve [P/(1.62 W)] 3/2
.
The catalyst concentration can be monitored conveniently and selectively through the
3MLCT luminescence, after dilution and acidification to eliminate quenching by the sacrificial
donor or its oxidation products; and the Cl− concentration can be determined without inter-
ference from other species by using an ion-sensitive electrode. As is seen in Figure 3b, the
time dependences of the [Ru(bpy)3]2+ and Cl− concentrations are excellently representable
by monoexponential functions; but the observation that this is only possible individually, with
best-fit rate constants differing by a factor of nearly three, already exposes this simplicity as
deceptive and rules out any direct interpretation of the fit parameters. Because the evolution
of the system comprises negative feedback between the concentrations of the key species,4
and because the secondary chemistry of the sacrificial donor is only rudimentarily known,
the initial slopes are the only observables on which conclusions should be based; and even
in this temporal range, the system composition is not completely certain, as will emerge
below.
With the kinetic parameters of Figure 1a, the initial rate of Cl− formation can be expressed
as the product of three terms p, q, and r,4 each of which describes the dependence on
an easily varied quantity (p, light intensity I; q, quencher concentration [Q], Q = Ur2− or
Asc2− this work; r, substrate concentration [S], S = ClAc with the chloroacetate assay). The
abbreviation Cat stands for [Ru(bpy)3]2+, and all concentrations on the r.h.s. of Equation 1
are weight-in concentrations.
ꢀ
ꢁ
d[Cl−]
dt
=
=
p × q × r
(1)
t=0
s
!
ꢂ
ꢃ
ꢂ
ꢃ−1
× 1 + 1/(kqτ0[Q]) −1/2 × 1 + 1/(ksτe[S])
κexc[Cat] η
I 3/2κion
krec
•−
The inset of Figure 3b demonstrates that the
e
yield is indeed proportional to I 3/2
,
aq
because the intensity I is proportional to the LED radiative power P in turn. This characteristic
6
This article is protected by copyright. All rights reserved.

10.1002/chem.201800626
Chemistry - A European Journal
dependence identifies the mechanism as being cyclic and featuring the monophotonic
ionization of a catalyst form that is itself produced by a monophotonic reaction and reverts
to the starting state by a second-order dark process. The term p also provides a criterion
for the selection of the LED: κexc and κion are proportional to the extinction coefficients of
OER and GS at the LED wavelength,13 but their different exponents make it much more
important to match the LED emission to the spectrum of the direct electron precursor OER
and this is intensified by the fact that κexc and κion are also proportional to the quantum yield
of the respective process,13 which is unity for the primary excitation but only a few percent
for electron ejection.
;
Figure 4 focuses on the influence of the quencher. To quantify the amount of remaining
Ur2−, we had to use an approach different from that in our earlier study with Asc2− as
the sacrificial donor, because the Folin–Ciocalteu reagent employed there also responds
to the oxidation products of Ur2−. Instead, the intense absorption of Ur2− in the UV-B
(
λmax = 295 nm, εmax = 13200 M−1cm−1; see, main plot of Figure 4a) turned out to be well
suited for the purpose. Interference from the catalyst-based species also absorbing there
plays no role on account of the large disparity of concentrations; and scaling each spectrum
to maximum reveals that the band shape at λ ≥ λmax remains invariant during the reaction,
thus ruling out a superposition of absorptions of different species in this range.
(a)
(b)
8
6
4
2
0
1.0
0.8
0.6
0.4
0.2
0
15
10
5
Q = Ur2−
Q = Asc2−
0
0
5
10
[Cl−] / mM
250
300
350
400
0
20
40
60
λ / nm
[Q] / mM
Figure 4: Consumption during, and effect of the quencher on, Cl− production with the ClAc assay. Graph (a), development of [Ur2−
]
during the experiment of Figure 3; main plot, UV spectra recorded at different illumination times; inset, Ur2− decrease (obtained by
integrating the spectra of the main plot between 295 nm and 315 nm) as function of Cl− increase with regression line of slope 1.2, same
colour code for the data points as for the Ur2− bands as in the main plot. Graph (b), initial slope of chloride formation in experiments
as in Figure 3 but with 50 mM ClAc, x mM of quencher Q and (50 − x) mM Na2SO4 for keeping the ionic strength constant. The data
for Asc2− (brown, diamonds, solid fit curve) can be represented by q (Equation 1) with kqτ0 = 37.7 M−1, which is the Stern–Volmer
constant obtained from laser flash photolysis at this ionic strength, whereas the data for Ur2− (orange, circles) differ widely from the
corresponding curve (dashed; kqτ0 = 16.7 M−1 from Figure 2a) and can only be accommodated with this Stern–Volmer constant when
the exponent in q is set to −0.933 (orange solid curve).
The conversion of ClAc into Ac theoretically requires two molecules of the sacrificial
•−
donor, the first serving as an electron donor for generating the
e
precursor OER, and
aq
the second as a hydrogen donor for intercepting the carboxymethyl radical (compare, the
scheme in Figure 3a). Both with Asc2− and with Ur2−, donor consumption is indeed found to
be proportional to substrate consumption (as measured by chloride formation) during most
of the reaction, up to 95 % and 80 % turnover; but the constant of proportionality amounts to
only 1.6 in the case of Asc2− and to only 1.2 in the case of Ur2− (see, the inset of Figure 4a).
We ascribe the deviations to the disproportionation of two donor-based radical, which
7
This article is protected by copyright. All rights reserved.

10.1002/chem.201800626
Chemistry - A European Journal
recovers one molecule of the donor. For Asc•−, this reaction has been investigated in some
detail,5 whereas for Ur•2− it is known but has largely remained unexplored.14 Depending on
what fraction of the radicals undergoes disproportionation, any constant of proportionality
between 1 and 2 can be rationalized, and the experimental observation puts this fraction at
40 % for Asc•− and at 80 % for Ur•2−. A larger amount of Ur2− recycling is also consistent
with Figure 2b, which revealed that the recombination of OER with Ur•2− — as opposed to
that with Asc•− — started noticeably to deviate from a second-order process already after
4 ms, as it must when Ur•2− is consumed by another process.
This discrepancy between the short-time kinetics with Asc2− and Ur2− in Figure 2b is
reflected by the long-time behaviour observed through the dependence of the initial Cl−
formation rate on the quencher concentration in Figure 4b. In these experiments, the ClAc
concentration was chosen so high that the deviations of the term r (Equation 1; see the
discussion below) from unity are insignificant throughout the range of [Q]. The data for Asc2−
with its regular second-order OER decay are perfectly accommodated by the term q with
the parameters extracted from measurements as in Figure 2, in other words, the laser-flash
photolysis results are sufficient for anticipating the outcome under preparative photolysis. But
the same dependence on [Ur2−] is not only impossible to fit with the corresponding kinetic
parameters from the short-time experiments but also generally impossible to fit with the
inverse square-root law for q when the value of kqτ0 is positive, as it must be; the only way to
reproduce the data with the known Stern–Volmer parameter is by increasing the exponent to
almost unity. Hence, in this system a complex secondary chemistry renders flash photolysis
useless for making quantitative predictions for LED photolysis; but experiments on short
•−
timescales still retain their value for establishing whether or not
principle.
e
formation is feasible in
aq
Figure 5 pinpoints a related divergence with respect to the electron capture by the
•−
substrate (the term r in Equation 1). This process has to compete with the
e
decay
aq
through all other channels, amongst which the reaction with the sacrificial donor plays
a major rol•e−on account of its high concentration even•t−hough donors are intrinsically
inefficient eaq scavengers. As can be seen in Figure 5a, eaq generated by laser flashes on
the [Ru(bpy)3]2+ Ur2− system in the absence of ClAc vanishes through a pseudo first-order
/
process exhibiting Stern–Volmer behaviour when the Ur2− concentration is varied. The
•−
relatively low rate constant of 9.0
×
107 M−1s−1 and long unquenched
e
lifetime of 0.81
µ
s
aq
obtained from these experiments are in line with expectation, and predict the much more
•−
efficient e scavenger ClAc (rate constant kS of electron capture, about 1.8 × 109 M−1s−1
aq
at the high ionic strength of these systems) to compete successfully.
As Figure 5b demonstrates, this is indeed the case. However, keeping τe in the function r
fixed at the values from laser flash photolysis gave poor fits of the dependences on the
substrate concentration, with deviations that were already noticeable in the case of Asc2−
and clearly inacceptable in that of Ur2−. Good representations were only obtained when τe
was treated as an adjustable quantity, and the fits converged on values of τe that were shorter
by factors of 1.8 and 2.8 than those extracted from the measurements on µs timescales.
We ascribe the contradictions between these observations on short and long timescales
to the same cause as those between Figures 2b and 4b: soon after the start of the
illumination — with Ur2− already within ms, as evidenced by the inset of Figure 2b — the
secondary chemistry of the donor radicals sets in, so the LED experiments experience the
8
This article is protected by copyright. All rights reserved.

10.1002/chem.201800626
Chemistry - A European Journal
(a)
(b)
5
8
6
4
2
0
6
3
0
4
3
2
1
0
Q
= Ur2−
Q = Asc2−
0
30
60
[Ur2−] / mM
0
0.5
1.0
1.5
2.0
0
20
40
60
t / µs
[ClAc] / mM
•−
Figure 5: Capture of e
investigated by experiments on different timescales. Common conditions, 0.1 mM [Ru(bpy)3]2+ in 50 mM
aq
aqueous SDS at pH 12.7, ionic strength in each series of measurements kept constant by adding the required amount of Na2SO4.
•−
Graph (a): main plot, e
trace in the presence of 40 mM Ur2−, ea•q− generation by laser flash photolysis (532 nm) with the sequence
aq
flash (412 mJ cm−2) — delay (5 µs) — flash (664 mJ cm−2) at 0 µs, overlaid with dashed fit curve a exp(−t/τe) with best-fit value 207 ns
for τe; inset, Stern–Volmer plot based on τe giving an unquenched e•aq− lifetime (from the intercept) of 805 ns and a rate constant for e
•−
aq
scavenging by Ur2− of 9.0 × 107
M
−1s−1. Graph (b), initial slope of Cl− formation from variable amounts of ClAc, concentration of the
sacrificial donor Q fixed at 75 mM (Asc2−, brown, diamonds) or 40 mM (Ur2−, orange, filled circles). The solid lines are fit curves of r
(Equation 1) with adjustable amplitude and kSτe (best-fit values / independently determined kS at the pertaining ionic strength / resulting
τe, 201 M−1 / 1.9 × 109
M
−1s−1 / 106 ns for Asc2− and 124 M−1 / 1.7 × 109
M
−1s−1 / 73 ns for Ur2−). The dashed lines show the
best-fit curves when τe is kept fixed at the respective value from laser flash photolysis (Asc2−, 186 ns; Ur2−, 207 ns).
additional influence of species and reactions that did not exist in most laser flash photolysis
traces, which are normally limited to some ten ms to avoid distortions by convection. This
emphasizes the necessity of characterizing the catalytic systems on a timescale comparable
to that of their application. Nevertheless, laser flash photolysis is still extremely important
because it isolates, and is thus instrumental in better understanding, the ”early” reactions. In
particular, it will spot a sacrificial donor that does not quench 3MLCT rapidly and productively
enough, or one whose radical recombines too fast with OER; hence, it•w−ill help avoid
fruitless optimization attempts of systems that are intrinsically useless as eaq sources.
3.2 Application to cross couplings
The preceding Section has shown that the [Ru(bpy)3]2+
/
Ur2− / SDS electron source is
•−
capable of•d−elivering mM concentrations of
e
upon illumination with a green LED. The
aq
attack of eaq on chlorinated and even fluorinated aliphatic or aromatic substrates effects
ejection of the halide anion; the remaining carbon-centered radical abstracts a hydrogen
atom when a suitable donor is present.3,4 Although such an exchange of halogen against
hydrogen is of general importance for syntheses and environmental remediation,15–19 the
use of the intermediate radicals as building blocks in cross-coupling reactions is even more
interesting.20–24
What would thwart such cross couplings obviously is the competitive transfer of a hydrogen
atom, and in this regard Ur2− as sacrificial species brings a decisive advantage over
Asc2− because its second pKa is lower by two units, so at the pH of our experiments the
concentration of the hydrogen-donating monoanion is a hundred times smaller than in the
case of Asc2−
.
Especially promising as coupling reagents are N-alkylated pyrroles because they are
9
This article is protected by copyright. All rights reserved.

10.1002/chem.201800626
Chemistry - A European Journal
very reactive towards radicals and exclusively afford 2-substituted products; hence, even
at moderate concentrations they allow an effective and selective radical usage.25–30 In our
strongly basic aqueous solutions, N-methyl-2-pyrrolecarboxylic acid (NMPCA) is particularly
•−
well suited for intercepting the
e
-generated radicals, as it combines this reactivity with an
aq
excellent solubility while the deprotonated carboxylate substituent increases its hydrophilic
such as to safely avoid its entry into the micelles where it might interfere with the electron
source.
Figure 6a demonstrates the successful cross coupling of NMPCA with ClAc as the radical
precursor to give 5-(2-carboxymethyl)-N-methylpyrrole-2-carboxylate
1 (for the formulas
of all cross-coupling products, see Scheme 1) in moderate yield (41 %), with the system
composition in the Figure — as for all the examples that follow — already representing an
optimum. Together with a substantial amount of acetate formation (26 %), which reflects
that even a weak hydrogen donor or one present in low concentration can compete with
the slow scavenging of the carboxymethyl radical by aromatic systems, this is identical with
the ClAc consumption (67 %). The incomplete consumption is a direct consequence of
the known trade-off between ClAc decomposition and turnover number.3 Increasing the
NMPCA concentration does not improve the cross-coupling yield because we found that
•−
this slightly activated heteroarene can also scavenge
e
to give a dihydropyrrole; this
aq
•−
reaction diminishes the fraction of
e
forming carboxymethyl radicals. Furthermore, the
aq
addition of NMPCA raises the ionic strength of the soluti•o−n, which increases krec through
the Brønsted–Bjerrum relationship,31 so diminishes the eaq generation rate.
(a)
(b)
1
1
Me′′
Me′
Me
d
c
a
b
Ac
a
b
d
D
c
D
x 5
0
1
0
1
S
S
S
c
′
c
c
a
a
b
c
Me
O
O
d
d
Me′′
O
−O
N
−O
O−
b Me
d
O
b
O−
a
d
0
0
6.6
5.9
3.9
3.6
3.5
1.8
7.8
7.6
7.4
7.2
7.0 3.5 2.2 1.8
δ / ppm
δ / ppm
•−
•−
Figure 6: Cross couplings induced by e . Common experimental parameters of the e
source, 100 µM [Ru(bpy)3]2+, 40 mM Ur2−
,
aq
aq
pH 12.7, 50 mM SDS. Shown are the 1H-NMR spectra, with identical vertical scales before (bottom trace) and after (top trace) illumination
for 14 h with a green LED, as well as the structural formula of the respective coupling product. Colour code in both spectra and formulas,
•−
orange (moiety of the e -generated radical source) and cyan (radical-scavenger moiety); in the spectra only, black (radical source S
aq
and product derived from S without participation of the radical scavenger). Graph (a), 10 mM ClAc (3.95 ppm, s; hydrogen-abstraction
product Ac, 1.81 ppm, s) and 100 mM NMPCA. Cross-coupling product 5-(2-carboxymethyl)-N-methylpyrrole-2-carboxylate; a, 6.60 ppm,
d; b, 5.88 ppm, d; c, 3.62 ppm, s; d, 3.43 ppm, s. The prominent gray signal is due to SDS (3.92 ppm, t). Graph (b), 5 mM ClBz (7.71 ppm,
d, and 7.35 ppm, d; dehalogenation and dimerization products D, 7.76, 7.44, and 7.37 ppm, all m) and 100 mM MesAc. Cross-coupling
product 3-(4-carboxyphenyl)-mesitylacetate; a, 7.85 ppm, d; b, 7.12 ppm, d; c, 6.96, s; d, 3.51 ppm, s; Me00, 2.18 ppm, s; Me, 1.84 ppm, s;
Me0, 1.80 ppm, s.
In line with this reasoning, using a much more reactive phenyl
σ
radical in•p−lace of the carb-
oxymethyl radical promotes the cross coupling. The reducing power of eaq easily cleaves
aromatic carbon–chlorine bonds,3,4,32,33 so allowed replacing ClAc with 4-chlorobenzoate
ClBz, which at halved concentrations of both the radical souce and the heteroarene raised
10
This article is protected by copyright. All rights reserved.

10.1002/chem.201800626
Chemistry - A European Journal
the yield of the cross-coupling product 5-(4-carboxyphenyl)-N-methylpyrrole-2-carboxylate
to 60 %. (For the NMR spectra, see Supporting Figure 1a.) The by-products, mainly
4
benzoate, accounted for a further 25 %. The presence of the coupling reagent is very
beneficial in this system, as in its absence the phenyl radicals also attack the catalyst, which
decreases the ClBz turnover from 85 % to 37 % under otherwise identical conditions. The
contrast to ClAc, where no such catalyst poisoning is observed, conspicuously demonstrates
the reactivity difference between the phenyl and carboxymethyl radicals.
Omitting the carboxylate group from the pyrrole should increase the cross-coupling
•−
efficiency through lowering the propensity of the heteroarene to scavenge
e
parasitically
aq
and through avoiding an increase of the ionic strength. These motivations and the good
solubility of the unsubstituted N-methyl pyrrole NMP in water prompted us employ NMP as
coupling reagent for the same phenyl radical as above. Contrary to expectation, however,
the NMR spectra collected in Supporting Figure 1b revealed a decrease in the yield of
the cross-coupling product 2-(4-carboxyphenyl)-N-methylpyrrole
3 to 30 %, which is not
due to increased by-product formation but results•f−rom a reduction of the ClBz turnover to
•−
56 %. Evidently, NMP either interferes with the eaq generator itself or intercepts
e
in a
aq
nonproductive parasitic process.
•−
A hydrophilic coupling reagent with very little tendency to scavenge
e
is mesitylacetate
aq
MesAc, owing to the isolation of the carboxylate function from the ring by a saturated
carbon.3 Although the considerable steric hindrance by its substituents must decelerate the
attack by phenyl radicals, the cross-coupling product 3-(4-carboxyphenyl)-mesitylacetate
2
is still formed with a yield of 30 %, as the NMR spectrum of Figure 6b demonstrates, but
the correspondingly longer life of the radicals results in faster catalyst poisoning, which
manifests itself by a ClBz turnover of only 58 %.
•−
Scheme 1 summarizes the
e
-induced cross couplings described in this Section. General
aq
regularities that allow an optimization are as follows. As expected,4 the catalyst concentration
exerts no influence on the product distribution but determines the amount of
•−
e
formed
aq
and the turnover rate. This is valuable with respect to synthetic output but less so from the
p•o−int of view of efficient catalyst use. Increasing the substrate concentration maximizes
eaq utilization but reduces the percentage of substrate conversion;3 hence, although the
cross-coupling product is formed in a larger absolute amount, its yield with respect to the
substrate decreases. This effect can be countered by adding the substrate in several smaller
portions distributed over the reaction duration. The concentration of the trapping reagent
is subject to a trade-off: higher concentrations accelerate the coupling but increase the
•−
amount of parasitic
opposite ways.
e
scavenging; both factors obviously influence the product yield in
aq
The Ur2− concentration is the only one that significantly affects the partitioning of the
substrate radicals between hydrogen addition and coupling with the trapping reagent. The
latter reaction channel is favoured by lowering the Ur2− concentration, and this dependence
becomes more prominent for a sterically hindered trapping reagent such as MesAc. When
the electron generator is fuelled by ascorbate instead of urate, none of the reactions of
Scheme 1 give an appreciable yield; instead, only hydrogenation products are observed.
These observations clearly identify the concentration, and above all hydrogen-donating
ability, of the sacrificial donor as the controlling factor, thus confirm the starting hypothesis
of this work.
11
This article is protected by copyright. All rights reserved.

10.1002/chem.201800626
Chemistry - A European Journal
[Ru(bpy)3]2+, SDS, Ur2−
Substrate + Trapping reagent
Product
520 nm, 14 h, H2O, pH 12.7
O
O
1
2
N
−O
(41 %)
(30 %)
O
O−
−O
O
O−
O
O
O
3
4
O−
N
N
(30 %)
(60 %)
−O
−O
•−
Scheme 1: Cross couplings induced through LED-generated e
.
aq
4 Conclusions
With regard to mechanism and kinetics, our results have provided clear evidence that
laser flash photolysis with detection on short timescales does not always provide sufficient
information on the inner workings of a photoredox catalytic cycle. Such experiments are by
necessity restricted either to the pristine starting state of the system (e.g., for investigating
the processes surrounding the quenching step) or to an assumed and idea•li−zed intermediate
state (e.g., when studying the competition between the natural decay of eaq and its capture
by a substrate). Not accessible to them are the much slower changes during continuous
illumination for periods at least a billion times longer, in particular the variation of the system
composition by the secondary chemistry of the sacrificial donor; but these changes can
significantly influence the fast reactions through the presence of newly formed species.
Analyzing the initial slopes of product formation and catalyst decay during low-intensity
preparative photolysis matches the observation window much more closely to the time frame
of the actual benchtop procedure.
•−
In addition, this work has demonstrated for the first time that
e
can also be enlisted
aq
•−
as a reagent for achieving cross-couplings. With its impressive reducing power,
e
is
aq
not deterred by aliphatic or aromatic carbon–chlorine bonds,4 so extends the palette of
substrates that can be employed. Yet, our electron generator itself is driven by a low-power
green LED; and it is sustainable because it only consumes the ubiquitous antioxidant urate.
The present study has successfully validated our hypothesis that the much lower hydrogen-
donating ability of the urate dianion compared to the ascorbate dianion is instrumental in
suppressing the substitution of halogen by hydrogen in favour of the desired cross couplings.
It has provided proof of principle by focussing on the role of the sacrificial donor, but we
envisage that tuning the photocatalyst will further optimize the yields of the cross-coupling
products.
12
This article is protected by copyright. All rights reserved.

10.1002/chem.201800626
Chemistry - A European Journal
References
[1] G. V. Buxton, C. L. Greenstock, W. P. Heiman, A. B. Ross, J. Phys. Chem. Ref. Data
1988, 17, 513–886.
[2] M. Goez, C. Kerzig, R. Naumann, Angew. Chem. Int. Ed. 2014, 53, 9914–9916.
[3] R. Naumann, C. Kerzig, M. Goez, Chem. Sci. 2017, 8, 7510–7520.
[4] R. Naumann, F. Lehmann, M. Goez, Angew. Chem. Int. Ed. 2018, 57, 1078–1081.
[5] M. B. Davies, J. Austin, D. A. Partridge, Vitamin
C: Its Chemistry and Biochemistry,
The Royal Society of Chemistry, Cambridge, 1st ed., 1991.
[6] M. G. Simic, S. V. Jovanovic, J. Am. Chem. Soc. 1989, 111, 5778–5782.
[7] L. Muñoz-Rugeles, A. Galano, J. R. Alvarez-Idaboy, Phys. Chem. Chem. Phys. 2017
,
15296–15309.
[8] J. P. Telo, Org. Biomol. Chem. 2003, 1, 588–592.
[9] S. P. Pitre, C. D. McTiernan, J. C. Scaiano, Acc. Chem. Res. 2016, 49, 1320–1330.
[10] R. Martinez-Haya, M. A. Miranda, M. L. Marin, Eur. J. Org. Chem. 2017, 2164–2169.
[11] M. Marchini, G. Bergamini, P. G. Cozzi, P. Ceroni, V. Balzani, Angew. Chem. Int. Ed.
2017, 56, 2–4.
[12] T. Kohlmann, R. Naumann, C. Kerzig, M. Goez, Photochem. Photobiol. Sci. 2017, 16,
1613–1622.
[13] M. Goez, V. Zubarev, Chem. Phys. 2000, 256, 107–116.
[14] G. L. Squadrito, R. Cueto, A. E. Splenser, A. Valavanidis, H. Zhang, R. M. Uppu, W. A.
Pryor, Arch. Biochem. Biophys. 2000, 376, 333–337.
[15] L. Huang, W. B. Dong, H. Q. Hou, Chem. Phys. Lett. 2007, 436, 124–128.
[16] H. Park, C. D. Vecitis, J. Cheng, W. Choi, B. T. Mader, M. R. Hoffmann, J. Phys. Chem.
A 2009, 113, 690–696.
[17] X. Li, J. Ma, G. Liu, J. Fang, S. Yue, Y. Guan, L. Chen, X. Liu, Environ. Sci. Technol.
2012, 46, 7342–7349.
[18] Y. Peng, S. He, J. Wang, W. Gong, Radiat. Phys. Chem. 2012, 81, 1629–1633.
[19] Z. Song, H. Tang, N. Wang, L. Zhu, J. Hazard. Mater. 2013, 262, 332–338.
[20] J. M. R. Narayanam, C. R. J. Stephenson, Chem. Soc. Rev. 2011, 40, 102–113.
[21] C. K. Prier, D. A. Rankic, D. W. C. MacMillan, Chem. Rev. 2013, 113, 5322–5363.
[22] Chemical Photocatalysis, (Ed.: B. König), DeGruyter, Berlin, 2013.
13
This article is protected by copyright. All rights reserved.

10.1002/chem.201800626
Chemistry - A European Journal
[23] M. Reckenthäler, A. G. Griesbeck, Adv. Synth. Catal. 2013, 355, 2727–2744.
[24] N. A. Romero, D. A. Nicewicz, Chem. Rev. 2016, 116, 10075–10166.
[25] Y. M. Osornio, R. Cruz-Almanza, V. Jiménez-Montaño, L. D. Miranda, Chem. Commun.
2003, 2316–2317.
[26] O. Guadarrama-Morales, F. M
4518.
éndez, L. D. Miranda, Tetrahedron Lett. 2007, 48, 4515–
[27] C. Iuga, S. O. Uribe, L. D. Miranda, A. Vivier-Bunge, Int. J. Quantum Chem. 2010, 110,
697–705.
[28] L. Marzo, I. Ghosh, F. Esteban, B. König, ACS Catal. 2016, 6, 6780–6784.
[29] J. I. Bardagi, I. Ghosh, M. Schmalzbauer, T. Ghosh, B. König, Eur. J. Org. Chem. 2018
,
34–40.
[30] M. Neumeier, D. Sampedro, M. Májek, V. A. de la Peña O’Shea, A. Jacobi von Wangelin,
R. Pérez-Ruiz, Chem. Eur. J. 2018, 24, 105–108.
[31] A. A. Frost, R. G. Pearson, Kinetics and Mechanism, John Wiley and Sons, New York,
2nd ed., 1961.
[32] V. V. Konovalov, S. S. Laev, I. V. Beregovaya, L. N. Shchegoleva, V. D. Shteingarts, Y. D.
Tsvetkov, I. Bilkis, J. Phys. Chem. A. 2000, 104, 352–361.
[33] R. Zona, S. Solar, N. Getoff, K. Sehested, J. Holcman, Radiat. Phys. Chem. 2008, 77,
162–168.
14
This article is protected by copyright. All rights reserved.

10.1002/chem.201800626
Chemistry - A European Journal
Table of contents entry
Ur2−
Ur•2− + H+
1
Ur the one... from microseconds to hours. A photo-
RT
MLCT OER
T
X−
RX
redox catalytic system driven by a green LED, con-
suming only the bioavailable urate dianion Ur2− and
•−
e
aq
0
GS
0
2
t / µs
•−
producing hydrated electrons
e
is characterized over
10
aq
10
timescales spanning ten orders of magnitude, and is
employed to achieve the first cross couplings induced
0
0
0
•−
0
9
10
by the super-reductant e
.
aq
t / ms
t / h
Keywords: Photocatalysis; Photochemistry; Sustainable Chemistry; Radical reactions;
Hydrated electrons
15
This article is protected by copyright. All rights reserved.