Article

pubs.acs.org/jchemeduc

Marcus Theory: Thermodynamics CAN Control the Kinetics of

Electron Transfer Reactions
Todd P. Silverstein*
Chemistry Department, Willamette University, Salem, Oregon 97302, United States
*
S Supporting Information

ABSTRACT: Although it is generally true that thermodynamics do not influence kinetics, this is NOT the case for electron
transfer reactions in solution. Marcus Theory explains why this is so, using straightforward physical chemical principles such as
transition state theory, Arrhenius’ Law, and the Franck-Condon Principle. Here the background and applications of Marcus
Theory are presented, with special emphasis given to biological redox reactions. Interesting conclusions that arise from the theory
are also discussed. These include the nonlinear contribution of reorganization energy (λ) to the activation free energy (ΔG⧧), as
well as the existence of a defined “normal range” of reaction cell potential (ΔE°): Only if ΔE° is within this range does a reaction
of lower λ have a lower ΔG⧧; outside of this range, the lower-λ reaction will have the higher ΔG⧧.
KEYWORDS: Upper-Division Undergraduate, Biochemistry, Physical Chemistry, Inorganic Chemistry, Aqueous Solution Chemistry,
Bioinorganic Chemistry, Biophysical Chemistry, Catalysis, Electrochemistry, Enzymes, Kinetics, Oxidation/Reduction, Thermodynamics

■ BACKGROUND
Almost all General/Introductory Chemistry textbooks and
instructors at one point make the assertion that reaction
thermodynamics do NOT influence kinetics: A spontaneous
reaction can be slow, whereas a nonspontaneous reaction can
be fast. The point of this paper is to describe a common
exception to this general assertion that kinetics are independent
of thermodynamics. Anderson’s “90% rule” states that good
rules in chemistry are correct over 90% of the time, and thus
deserve their place in the teaching of General/Introductory
Chemistry.1 However, even good rules have exceptions, and as
Emeric Schultz has pointed out,1 “some of the most interesting
stuff is in the 10% exceptions.” These exceptions are worth
mentioning, but not teaching, at the introductory level;1 on the
other hand, exceptions often include fascinating chemistry for
students at the upper division undergraduate, and graduate
level. Figure 1. Potential energy curves for a nonspontaneous reaction (R →
First the 90%: For most reactions there is no relationship P, upper blue curve), and one that is more spontaneous (R → P′,
between the standard free energy differences between reactants lower red curve).
and products (ΔG°) on one hand, and between reactants and
transition state (ΔG⧧, the activation free energy) on the other.
The reasons underlying this rule are discussed in most physical reactions, a three (or more!) dimensional plot showing
chemistry texts (e.g., ref 2) under the heading “potential energy potential energy contour surfaces must be employed. Because
surfaces.” Briefly, using simple two-dimensional parabolic of the complexity of these multidimensional potential energy
potential energy curves for the reaction R → P (upper curve contours, and the different nuclear separation distances of
in Figure 1), the intersection between the reactant and the different transition states, for most reactions there is in fact no
product curves gives the activation free energy (E in Figure 1). discernible relationship between ΔG⧧ and ΔG°.
One might be tempted to predict that for a more spontaneous This rule works for most reactions (the 90%), but there are
reaction, for example, R → P′ (lower curve, Figure 1), the some notable exceptions (the 10%). For organic reactions,
activation free energy must be lower, that is, E′ < E. However, Hammond’s postulate states that the transition state structure
there are two problems with this prediction: First, Figure 1 resembles that of the species nearest to it in free energy: In
makes the unwarranted assumption that for the two reactions, other words, all other things being equal, the free energy of
the transition states occur at identical nuclear separation distances. activation should be higher for an endergonic reaction, and
Furthermore, the two-dimensionality of Figure 1 is an lower for an exergonic reaction.
oversimplification. Because of the number of different
interatomic distances and collision angles sampled in most Published: July 11, 2012
© 2012 American Chemical Society and
Division of Chemical Education, Inc. 1159 dx.doi.org/10.1021/ed1007712 | J. Chem. Educ. 2012, 89, 1159−1167
Journal of Chemical Education Article

The most well-characterized exception to the thermody- For example, in a paper analyzed below, Candeias et al.17
namics/kinetics independence rule is the solution-phase “outer measured the kinetics of peroxidase reduction (E° = +0.88 V;
sphere” redox reaction,3 that is, that between solvated donors see below “What is Marcus Theory Good For?”) by a series of
and acceptors that exchange electrons. For his detailed phenols with E°s ranging from +0.42 to +1.17 V, giving cell
characterization of this particular class of “exceptions to the potentials of −0.29 to +0.46 V. Changes in reaction kinetics
rule”, Rudolph A. Marcus won the Nobel Prize4 in 1992. could result from differences in three distinct kinetic
Marcus initially developed his theory to describe electron parameters: (1) ka and k−a, due to association (i.e., Kd); (2)
transfer between small solvated donors and acceptors, and he AET, due to differences in donor−acceptor orientation or
was later able to expand it to include electrode reactions and distance in the association complex; and (3) ΔG⧧. It is the
biological electron transfers.5 Previous papers in this Journal latter property, the dependence of ΔG⧧ on ΔG° (or
(most published over 20 years ago) have discussed Marcus alternatively, ΔE°), that is explained by Marcus Theory:18
Theory in broad, general terms,6,7 or have dealt only with the
λ ⎛⎜ ΔG°′ ⎞⎟
2
so-called “cross-relation” aspect of the theory,8−10 or with (λ + ΔG°′)2
ΔG⧧ = wrct + 1+ = wrct +
specific experimental systems.6,8,11 As we shall see, Marcus 4⎝ λ ⎠ 4λ
Theory can be explained by relatively straightforward concepts (2)
in physical chemistry (especially transition state theory and the
Franck−Condon Principle). Furthermore, a number of where w = electrostatic work done to bring reactants (wrct), or
interesting conclusions arise from the theory, including the products (wprod), together to form the donor−acceptor
“inverted region”: redox reactions get faster as they get more association complex, and ΔG°′ = ΔG° + wprod − wrct. The
spontaneous only up to a certain point; beyond this point, in parameters λ, w, ΔG°, ΔG⧧ are all energy terms with units of
the inverted region, they get slower for more negative ΔG°. kJ/mol (or kcal/mol). Often in the literature, the assumption is

■
made that wprod ≈ wrct, in which case ΔG°′ ≈ ΔG° = −nFΔE°,
MARCUS THEORY and eq 2 can be written as
An outer-sphere redox reaction can be seen as a series of three nF(λ − ΔE°)2
steps: association, electron transfer, and dissociation (Scheme ΔG⧧ = wrct +
4λ (3)
1).
Here λ is in eV and ΔE° is in V; multiplying by nF converts to
Scheme 1 kJ/mol.
Equation 2 shows that for outer sphere redox reactions there
is a mathematical relationship between ΔG⧧ and ΔG°. The
reason why this is so will be discussed below in the “Quantum
Mechanical Explanation” section. Suffice it to say here that
because the electron is so light and transfers so quickly, nuclear
• Step 1 (association): formation of the donor−acceptor separation distances remain constant as the electron is
(D•A) association complex transferred, and Figure 1 becomes a reasonable approximation
K assocn = ka /k −a = 1/Kdissociation = 1/Kd of potential energy changes during the reaction.
Equation 2 introduces an important new component of ΔG⧧,
• Step 2 (electron transfer): electron is passed from donor the “reorganization” energy, λ. Marcus describes λ as the energy
(D) to acceptor (A)12 input necessary to alter the structure of solvated reactants to
• Step 3 (release): separation of the product (D+•A−) match that of solvated products, before electron transfer takes
complex place.4,5 It can be broken up into an inner shell component
All rate constants in Scheme 1 obey the Arrhenius/Eyring (λin) involving changes in bond lengths and angles, and an
Laws, but in Marcus Theory, kET is especially significant: outer shell component (λout) involving reorientation of the
kET = AETe−ΔG
⧧
/ RT surrounding solvent shell. The two components are always
(1)
positive, and are additive: λ = λin + λout.
The frequency (or pre-exponential) factor, AET, is the Equations 1−3 show that for outer sphere redox reactions,
activation-less rate constant, approached as ΔG⧧ → 0 (or T reaction thermodynamics (ΔG° or ΔE°) really do influence
→ infinity). As in standard Arrhenius Theory, AET gives the reaction kinetics. Figure 2 depicts how ΔG⧧ (solid, black curve)
frequency of ef fective associations at infinite temperature.13 varies with cell potential (ΔE°), for λ = 0.5 eV and wrct = 5 kJ/
Additionally for redox reactions, AET declines exponentially mol.
with the edge-to-edge separation distance (r) between donor Because eq 3 is a second order polynomial in ΔE°, the ΔG⧧
and acceptor in the effective association complex: AET(r) = (solid) curve is a parabola; its minimum occurs at ΔE° = 0.5 eV
Zκ(r), where Z has units of collision frequency, and κ(r) = = λ, and ΔG⧧min = wrct. At this point, the free energy supplied
κ0e−βr is the transmission coefficient, or the probability that by the spontaneous reaction is equal and opposite to the
electron transfer will occur once the reactant association reorganization energy; the second term in eq 3 disappears here
complex reaches the transition state. The distance sensitivity, β, (i.e., λ − ΔE° = 0), so only wrct contributes to ΔG⧧.
tends to be restricted to a narrow range (10−15 nm−1),14 and Two inferences can be drawn from the curves in Figure 2:
for an adiabatic reaction, that is, one with strong electronic The left side shows that as the redox reaction gets less
coupling between reactants,5,15 κ ≈ 1. spontaneous (i.e., ΔE° < λ) the reaction gets slower (higher
Consider redox reactions involving a single donor and a ΔG⧧), which makes some intuitive sense (even though we
series of structurally related acceptors16 with different standard implore our students NOT to draw conclusions like this!). By
reduction potentials (E°s), and thus different cell potentials the same token though, the right side of the parabola tells us
(ΔE° = E°A − E°D) and free energy changes (ΔG° = −nFΔE°). that as the redox reaction gets more spontaneous (i.e., ΔE° > λ)
1160 dx.doi.org/10.1021/ed1007712 | J. Chem. Educ. 2012, 89, 1159−1167
Journal of Chemical Education Article

Figure 3. Reaction coordinate (lef t) and potential energy (right)

diagrams for the endergonic reaction R → P1. The red dotted line
follows the trajectory for reactants converting to products; ΔG⧧1 = g1
− gR > ΔG°.

In Figure 4, the endergonic R → P1 reaction from Figure 3 is

compared to more spontaneous reactions. For the reaction R
Figure 2. Marcus Theory predictions for the variation with cell
potential (ΔE°) of activation free energy (ΔG⧧, solid curve, left scale,
from eq 3); kET, long dashed curve, left scale, from eq 1 and 3; and
ln(kET), short dashed curve, right scale. Parameter values used are: λ =
0.5 eV, wrct = 5 kJ/mol, T = 25 °C, and AET = 100 s−1. At ΔE° = 0.5 eV
= λ, ΔG⧧min = wrct, and kET(max) = 13.3 = AET e−w(rct)/RT.

the reaction also gets slower (higher ΔG⧧). Marcus refers to

this right side of the parabola as the “inverted region”.4
Using the same parameter values (λ = 0.5 eV and wrct = 5 kJ/
mol) plus T = 25 °C and AET = 100 s−1, we see that the kET vs
ΔE° plot gives the bell-shaped blue (long dashes) curve. As
with the ΔG⧧ minimum in Figure 2, the maximum value of kET
occurs at ΔE° = 0.5 eV = λ. Interestingly, the kET vs ΔE° blue
curve resembles a condensed phase UV−vis absorbance peak:
There is a particular energy that yields the highest y-value, with Figure 4. Potential energy diagrams for conversion of reactants R to
higher or lower energy values giving successively lower y-values. four different products: P1 (black), less stable than R (as in Figure 3);
As with absorbance, the shape of the curve is explained by P0 (blue), ΔG° = 0; P−1 (red), exergonic with ΔG⧧−1 ≈ 0 (ΔG°−1 =
−λ); and P−2 (green), more stable than P−1. Based on Figures 1, 3, and
quantum mechanical considerations (discussed below).
5 of ref 5.
It is worth mentioning here that Marcus Theory analyses are
often depicted with ln(k) vs ΔE° (or ΔG°) curves (Figure 2,
short dashed curve). Such semilog plots have advantages (they → P0 (Figure 4, blue curve), ΔG° = 0, and ΔG⧧0 = g0 − gR.
stretch apart small y-values and compress large ones together), Because g0 is lower than g1, ΔG⧧0 < ΔG⧧1, and the more
but they tend to give undue weight in any fitting procedure to spontaneous reaction (R → Po) is faster.
the lowest, most error-prone values. The exergonic reaction R → P−1 (Figure 4, red curve) is a

■ QUANTUM MECHANICAL EXPLANATION

Marcus explains the inverted region on the basis of the Franck−
special case: The P−1 curve crosses the R curve near its lowest
point (g−1 is only slightly above gR). This reaction is nearly
barrierless (ΔG⧧−1 ≈ 0), and it reacts at close to the diffusion-
Condon Principle and corresponding potential energy dia- controlled limiting rate: k ≈ AET. The rate is maximal because
grams. Because the electron is such a light particle, the Franck− the P−1 curve intersects the R curve near its most highly
Condon Principle applies even to “weak overlap” electron populated (i.e., lowest energy) point, so the crossing probability
transfers.5 During the fast electron transfer process, nuclei have is highest. Nevertheless, even for this optimal reaction, an
no time to shift position, thus the electron transfer occurs association complex must still form, and reactant and solvent
mainly at or near the intersection of the reactant and product molecules must still reorganize themselves to reach and cross
potential energy curves19 as depicted in Figure 3, right. It is through the transition state on the way to products. Hence the
here that the overlap of wave functions for reactant and product reorganization energy is not zero. From eq 2 it follows that
potential energy curves is maximal. In Figure 3 this intersection ΔG⧧−1 is minimal because ΔG° + λ ≈ 0. One way to envision
point is marked as g1, which also denotes the position and this situation is that the free energy released as the reaction
energy of the transition state. In the endergonic reaction proceeds (ΔG° ≈ −λ) is “invested” in the reorganization
depicted in Figure 3, for a reactant molecule to cross from the process, such that the activation free energy is minimized.
R potential energy curve to P1 (red dotted line in Figure 3, Considering the three reactions discussed above, R → P1 vs
right), the highest energy point it passes through is g1. In this P0 vs P−1, as ΔG° gets more negative, ΔG⧧ decreases and the
case, the free energy of activation is g1 − gR = ΔG⧧1, which reaction goes faster. This Marcus Theory prediction is observed
exceeds ΔG°1. on the left side of the curves in Figure 2. However, ΔG° = −λ is
1161 dx.doi.org/10.1021/ed1007712 | J. Chem. Educ. 2012, 89, 1159−1167
Journal of Chemical Education Article

the optimal point: Any reactions more spontaneous than this declines); this is the situation envisioned by many authors.
actually go slower. As we saw in our discussion of Figure 2 However, for λ ≤ |ΔE°|, as λ decreases, ΔG⧧ rises (and kET
above, this situation can be likened to a condensed phase UV− declines). Hence to predict correctly the effect on ΔG⧧ of a
vis absorbance peak: higher energy incident photons lead to change in λ, one must know where both reorganization energies
higher absorbances only up to a certain point, that point being lie relative to the cell potential of the reaction.20 Although the
the energy difference between the ground and excited states. significance of the Marcus Theory inverted region is widely
Incident photons above this energy are absorbed less. recognized, this mathematical complexity in how λ impacts
Analogously, as reaction free energies decrease below −λ, ΔG⧧ is often overlooked, and is worth further clarification.
reactions get slower. Figure 6 uses a specific comparison to view the complex
This effect can be seen by comparing the green P−2 curve to mathematical relationship between λ and ΔG⧧: Here we plot
the red P−1 curve in Figure 4. Clearly, ΔG° (R → P−2) is more
negative than ΔG° (R → P−1); on the other hand, because g−2
is higher than g−1, ΔG⧧−2 exceeds ΔG⧧−1, and the R → P−2
reaction is slower. Thus decreases in ΔG° below −λ feature
higher ΔG⧧, as predicted by the Marcus Theory “inverted
region” and observed on the right side of the curves in Figure 2.

■ REORGANIZATION ENERGY
As discussed above, the reorganization energy (λ) is generally
described as the energy necessary to reorganize reactant bonds
and surrounding solvent to match those found in the product.
Using simple parabolas for the reactant and product potential
energy curves (PE = kx2), it can be shown that the
reorganization energy is identical to kΔr2, where Δr is the
nuclear coordinate separation distance between reactants and
products (see Supporting Information B). This lends a
quantum mechanical meaning to the above description, namely,
that λ is the energy input necessary to move reactants along Figure 6. Marcus Theory predictions for the variation of activation
their parabolic potential energy curve to a nuclear coordinate free energy (ΔG⧧) with cell potential (ΔE°), based on eq 3. λ = 0.5 eV
separation equal to Δr, corresponding to the most stable (blue curve) or 0.2 eV (red curve); wrct = 5 kJ/mol; T = 25 °C.
configuration for the products.
Another important point to make is that the reorganization
energy contributes to the activation free energy, but not in a
simple, additive way. As we shall see below, authors often infer ΔG⧧ as a function of ΔE° for two different reactions, one with
that as λ increases, ΔG⧧ increases; however, this is not always λ = 0.5 eV (blue curve, as in Figure 2) and another with λ = 0.2
the case. In fact, from eq 3 it is clear that for large values of λ (λ eV (red curve). We can see that the small green-shaded region
≫ ΔE°), ΔG⧧ increases linearly with λ, but for small values of λ is the only region in which the red curve (lower λ = 0.2 eV)
(λ ≪ |ΔE°|), ΔG⧧ is inversely proportional to λ (Figure 5). features an activation free energy that is lower than that for the
Thus for λ ≥ |ΔE°|, as λ increases, ΔG⧧ rises (and kET blue curve (higher λ = 0.5 eV). Outside of this “normal” region,
the opposite is the case: the red curve has a higher ΔG⧧ than
the blue curve. The green-shaded “normal” region stretches
from ΔE° = −0.32 to +0.32 V. In fact, it can be shown20 from
eq 3 that this region is bounded by ΔE° = ±(λlowλhigh)1/2.
Similarly, Figure 7 plots kET as a function of ΔE° for both λ =
0.5 eV (blue curve, as in Figure 2) and for λ = 0.2 eV (red
curve). Again, within the green-shaded “normal” region
bounded by ΔE° = ± 0.32 V (= ± (λlowλhigh)1/2),21 kET is
higher for the red curve (lower λ, 0.2 eV). Outside of this
region, the red curve (lower λ) has a lower kET than the blue
curve. This notion of a “normal” range of cell potentials,
bounded by ± (λlowλhigh)1/2, is an important aspect of Marcus
Theory: Within this range, the high-λ reaction is slower, but
outside of this range, the high-λ reaction is faster.
We also see from Figure 7 that the reorganization energy, λ,
not only sets the location of the maximum on the cell potential
axis (ΔE° = λ), but it also controls the width of the kET vs ΔE°
Figure 5. Dependence of activation free energy (ΔG⧧) on peak. Specifically, the lower-λ red peak is narrower than the
reorganization energy (λ), from Marcus Theory, eq 3. Parameter higher-λ blue peak. In fact, from eq 3 it can be derived that the
values used are: ΔE° = +1.0 V, wrct = 5 kJ/mol. ΔG⧧min = wrct, at λ =
1.0 V = ΔE° (point a). Points b and b′ (λ = 2 eV, 0.5 eV, respectively) width of the kET vs ΔE° peak at half-height, Δw1/2, is
have identical ΔG⧧ values, as do points c and c′ (λ = 4 eV, 0.25 eV, proportional to √λ.22 The fact that λ figures into both the
respectively). From eq 3, ΔG⧧(λlow) = ΔG⧧(λhigh) when λlow = ΔE°2/ location and the width of the kET vs ΔE° peak puts an
λhigh. important constraint on fitting experimental results.
1162 dx.doi.org/10.1021/ed1007712 | J. Chem. Educ. 2012, 89, 1159−1167
Journal of Chemical Education Article

Figure 8. Association = rate-determining step; Marcus Theory

prediction for the varia-tion of kobs as a function of cell potential. λ
Figure 7. Marcus Theory predictions for the variation of kET with cell = 0.5 eV; ka = 20 M−1 s−1; k−a = 0.02 s−1; AET = 100 s−1; wrct = 5 kJ/
potential (ΔE°), based on eqs 1 and 3; λ = 0.5 eV (blue curve) or 0.2 mol; and T = 25 °C. Note the wide, flat-topped peak: kobs,max saturates
eV (red curve); AET = 100 s−1; wrct = 5 kJ/mol; T = 25 °C; for the blue at ka = 20 M−1 s−1. From note in ref 22, Δw1/2 = 1.15 V = Δw1/2 =
curve, Δw1/2 = 0.38 V = (0.534 V1/2)·(0.5 V)1/2 (see note in ref 22). 4(λ·((−wrct + RT ln(AET/k−a))/(nF))1/2).

■ RATE-DETERMINING STEP: ASSOCIATION VS

ELECTRON TRANSFER
others have also expanded the theory to include electro-
chemical reactions at solid electrodes.4,5,23,24
In Scheme 1 above, we saw that the complete redox reaction Another area where Marcus Theory is having an ongoing
comprises three steps, each with its associated rate constants. impact is in the field of biological electron transfer.4,5 The
The observed reaction rate with its overall rate constant, kobs, in theory has been applied to a number of redox enzymes,
turn depends on rate constants for all three steps: association including peroxidase,25−31 laccase,30−32 glucose oxidase,33−37
(ka/k−a), electron transfer (kET), and separation (krel). Using the cytochromes,5 and photosynthetic reaction centers.4,5,38 The
steady state approximation (i.e., assume that kET is the rate- general idea in these studies is to fit kobs (or kcat/Km) vs ΔE°
determing step; see Supporting Information C for derivation), data to eqs 1−3 and obtain best-fit values for λ and AET/Kd. A
range of cell potentials is usually obtained by using a series of
kobs = kETka /(kET + k −a) = ka /(1 + k −a /kET) (4) structurally related substrates; depending on the enzyme these
Given the Arrhenius/Eyring equation (eq 1: k ET = could be electron donors or acceptors. Less commonly, the
AETe−ΔG‡/RT) and the Marcus equation (eq 3), eq 4 has a same substrate can be used with a series of mutant enzymes of
total of three fittable parameters: ka, k−a/AET, and λ. Typically, varying redox potential.
either association or electron transfer will be the rate- Marcus Theory analysis can answer a number of interesting
determining step: For reactions in which electron transfer is questions regarding redox enzyme kinetics. Here I summarize
recent results from just two redox enzymes, peroxidase and
rate-limiting, that is, kET ≪ k−a, eqs 1 and 4 reduce to eq 5
glucose oxidase. The Fe3+-heme in peroxidase is initially
⧧
kobs ≈ kET/Kd = (AET /Kd)e−ΔG / RT
(5) oxidized by peroxide in a two-electron process (k1), and then
rereduced by two substrate molecules in a sequence of two
which has only two fittable parameters: (AET/Kd) and λ, the consecutive one-electron steps. The first of these one-electron
latter arising from using eq 3 to express ΔG⧧. In this case, plots steps (k2, reduction of Fe4+-heme·+ to Fe4+-heme) is always
of kobs vs ΔE° feature sharp peaks resembling Figure 2 and 5b, significantly faster than the second step (k3, reduction of Fe4+-
with the only difference being that kobs is a second order rate heme to Fe3+-heme). Folkes and Candeias25 set out to answer
constant, with units of M−1 s−1 (as opposed to kET which is first this question: Is k2 faster than k3 because of more optimal
order). substrate binding (ka/k−a = 1/Kd), orientation/distance (AET),
The opposite case is more complex, but where association is or reorganization energy (λ)? Using horseradish peroxidase at
rate-limiting (i.e., kET ≫ k−a), eq 4 reduces to: kobs ≈ ka. On pH 7 and a series of phenol reductants, they found (Figure 9
the other hand, kET declines rapidly for ΔE° values far from λ, and Table 1) that the reorganization energy was ≈0.5 eV for
to the point where ET inevitably becomes rate-limiting. This both k2 and k3 (range: 0.35−0.5 eV). This is perhaps not
yields a kobs vs ΔE° curve which is wide and flat-topped (Figure surprising, as the substrate is identical in the two steps, and only
8): For ΔE° ≈ λ, kobs saturates at ka, but for ΔE° further from λ, the redox state of the heme is different. In fact, studies have
kobs declines to zero. An important implication here is that one shown no detectable structural changes occur near the heme
can tell from the shape of the kobs vs ΔE° curve whether the upon reduction of Fe4+-heme·+ to Fe4+-heme.39,40 On the other
rate-limiting step is association (wide flat-topped curve) or hand, AET/Kd was over 10-fold lower for k3 relative to k2 (Table
electron transfer (narrow sharp peak).

■
1); this can be seen in Figure 9 in the much smaller peak height
of k3. Thus the k2/first heme reduction step (from Fe4+-heme·+
WHAT IS MARCUS THEORY GOOD FOR? to Fe4+-heme) causes a conformational change at the substrate
Marcus Theory has been used successfully to predict rate binding site or the active site: (a) At the substrate binding site,
constants and solvent effects for simple outer-sphere redox a decrease in binding affinity would raise Kd (and lower k3); (b)
reactions4,5 such as the so-called “cross-reactions” between at the active site, rotating the substrate or heme into a
inorganic redox couples (Aox + Bred → Ared + Box). Marcus and suboptimal orientation would lower AET (thus lowering k3);
1163 dx.doi.org/10.1021/ed1007712 | J. Chem. Educ. 2012, 89, 1159−1167
Journal of Chemical Education Article

“normal” range20 is ΔE° between +0.04 and −0.04 V. Since all

but one of the indole points lie outside of this range, at pH 7 the
45-fold lower reorganization energy of indoles (relative to
phenols) causes the reaction rate to be slower, not faster. Of
course, the 6-fold lower value of AET/Kd for indoles (relative to
phenols, Table 1) also contributes to reaction sluggishness.
For indoles, over much of the cell potential range enzyme
reduction is faster at pH 5 compared to pH 7 (Supporting
Information, Figure S1a): AET/Kd (which is 8-fold higher at pH
5) accounts for some but not all of this effect. At pH 5, the
indole reorganization energy is 0.049 ± 0.006 eV (Table 1),
almost five times higher than at pH 7. Because the “normal”
range comparing these two reactions stretches from ΔE° =
+0.02 to −0.02 V, most of the data lie outside of this range. As
with the previous comparison above, the higher λ at pH 5 helps
to account for the higher rate constants and faster reaction.
Figure 9. Horseradish peroxidase reduction by phenols and indoles.
Phenol substrates, pH 7 (data from ref 25): first enzyme reduction
A few comments are in order here concerning the peroxidase
step (compound I reduction, k2), blue diamonds; second enzyme results discussed above. First of all, there is an interesting
(compound II) reduction, k3, black circles. Data are fit to eqs 3 and 5, difference between the phenols k2 vs k3 comparison (Figure 9)
assuming electron transfer is rate-limiting. Solid lines use best-fit values and the phenols vs indoles k2 comparison (Supporting
for λ and AET/Kd from nonlinear regression (Kaleidagraph) listed in Information, Figure S1a). For phenols at pH 7, k2 is
Table 1. Although three-parameter fits (λ, ka, and AET/k−a) to eqs 1, 3, consistently higher than k3, over the entire range of cell
and 4 are marginally better, P-values for the fit parameters (obtained potentials tested, even at very low ΔE° (Figure 10). On the
from PeakFit) show that use of the model is not statistically justified
by the data points, which are too few and too scattered.

alternatively, (c) increasing the distance between substrate and

heme would also lower AET, however, NMR data suggest41 that
the heme-substrate distance in peroxidase is consistently about
10 Å.
Candeias et al. also studied27,28 the kinetics of the first
enzyme reduction step (k2) using phenols at pH 7, and indoles
at pH 7 and pH 5. Enzyme reduction at pH 7 is generally
slower using indole substrates relative to phenols, except at ΔE°
≈ +30 mV. The indoles at pH 7 (Supporting Information,
Figure S1a and Table 1) had a surprisingly low reorganization
energy (0.0108 ± 0.0009 eV), showing that the indole and the
enzyme heme are bound “in a geometry ideal for electron
transfer [similar to] optimized systems such as the bacterial
photosynthetic reaction center.28” Note that as pointed out
above, a low reorganization energy does not necessarily mean
fast electron transfer. In fact, when comparing phenols vs Figure 10. A subset of data from Figure 9.
indoles at pH 7 (λ = 0.35 vs 0.01 eV, respectively), the

Table 1. Fitted Kinetic Parameters for the Reduction of Horseradish Peroxidase (HRP) by Phenols or Indoles, at pH 7 or 5a
λ (eV) AET/Kd (μM‑1s‑1) Relative rate ΔE° range (V) “Normal” range (V) R2
HRP k3, phenols, pH 7, (ref 7) 0.50 ± 0.05 28.2 ± 2.2 Low −0.3 to +0.5 0.95
From semilog slope: 0.44 ± 0.23 29 ± 22 0.90
HRP k2, phenols, pH 7, (ref 7) 0.346 ± 0.016 330 ± 40 High −0.3 to +0.5 ±0.47 0.85
From semilog slope: 0.6 ± 0.3 0.95
HRP k2, indoles, pH 7, (ref 9) 0.0108 ± 0.0009 52 ± 4 Low43 −0.1 to +0.07 ±0.04 0.993
From semilog slope: 0.012 ± 0.005 36 ± 12 0.90
HRP k2, indoles, pH 5, (ref 10) 0.049 ± 0.00644 430 ± 4044 Moderate45 −0.2 to +0.03 ±0.02 0.999
From semilog slope: 0.052 ± 0.024 500 ± 400 0.97

a
For peroxidase reduction, k2 refers to the one-electron reduction of compound I (Fe4+-heme·+) to compound II, and k3 to the one-electron
reduction of compound II (Fe4+-heme) to resting enzyme (Fe3+-heme). Bolded parameters (AET/Kd, λ, or both) are responsible for the differences
in relative rates. Data from Figure 9 and Supporting Information, Figure S1a are fit to eq 3 and 5, assuming wrct ≈ 0 and electron transfer is rate-
limiting (i.e., AET ≪ k−a).42 Two-parameter fits (λ and AET/Kd) are obtained by nonlinear regression (Kaleidagraph), and judged to be statistically
significant by P-values from PeakFit. Values for λ listed in the “from semi-log slope” row are obtained from the slopes of the lines in Figure 11 and
Supporting Information, Figure S1b, using eq 8. AET/Kd values in this row are averages of values obtained by using eq 3 and 5, applied to each (cell
potential, kobs) data point in the linear range, along with λ obtained from the linear range slopes in Figure 11 and Supporting Information, Figure S1b
(eq 8).

1164 dx.doi.org/10.1021/ed1007712 | J. Chem. Educ. 2012, 89, 1159−1167

Journal of Chemical Education Article

other hand, when comparing phenols to indoles (Supporting curve, in the semilog plot, λ controls both the shape (width)
Information, Figure S1a), one cannot unequivocally rank and the placement along the cell potential axis; AET/Kd controls
reaction rates, which vary as a function of cell potential as placement along the ln(k) axis. From the fact that the k2 curve
expected from Marcus Theory. In fact, at a cell potential of closely resembles the k3 curve displaced vertically upward
about +30 mV, k2 is roughly the same for all three data sets. (Figure 11), Folkes and Candeias25 concluded (correctly) that
Again, this is predicted by Marcus Theory, because ±30 mV the main difference between the two steps lay not in λ, but in
marks the edge of the “normal” region. AET/Kd.
Second, the notion of the “normal” region is one that escapes Quantitative results can also be extracted from semilog plots.
some authors. For example, Candeias et al.28 assumed that a By differentiating eqs 1 and 2, Marcus showed5 that small
lower reorganization energy always yields a lower activation free regions of the semilog plot are linear.46 For the ln(kET) vs ΔG°
energy and a higher rate constant. They concluded erroneously plot, the slope has units of mol/kJ, and is a function of ΔG°/λ:
that indoles at pH 7 react more slowly than phenols solely 1
because of the lower AET/Kd (referred to as k0/Km in their −RT × slope = (1 + ΔG°/λ)
2 (6)
paper, ref 28). As discussed above, part of this sluggishness is in −1
fact caused by the 5-fold lower reorganization energy at pH 7. For the ln(kET) vs ΔE° plot, the slope has units of V , and is a
Finally, a word is in order concerning experimental error and function of ΔE°/λ:
uncertainty in nonlinear fitting. Of the four data sets in Figure 9 1
and Supporting Information, Figure S1 (pH 7 phenols appear (RT /F ) × slope = (1 − ΔE°/λ)
2 (7)
in both figures), two do not extend beyond the inverted region,
and none contain enough points to clearly define the peak of Using eq 7, one can obtain a value for λ independent of that
the kET vs cell potential curve. This is often a problem in derived from nonlinear fitting of the k vs cell potential curve:
o
Marcus analyses: the range of cell potentials tested is limited by ΔEmid
the reactants that are available. Furthermore, experimental error λ=
[1 − (2RT /F ) × slope] (8)
in kinetic results can be an order of magnitude or more, hence
the scatter in data points is often high (for example, see the where ΔE°mid is the midpoint of the linear range of cell
blue k2 points between −0.1 and 0.0 V in Figure 9). For these potentials selected. For the Candeias et al. peroxidase results in
reasons, nonlinear fits using the Marcus Equations can Figure 9 and Supporting Information, Figure S1, λ obtained
sometimes seem unconvincing, or, as one colleague put it, from eq 8 and semilog slopes agrees with that from nonlinear
the fitted lines seem like “wishful thinking.” Even so, all of this fitting of eq 3 and 5 within ≈10% for three of the four data sets
is recognized and accepted in the peer-reviewed electro- (Table 1). This adds to our confidence in the accuracy of the
chemistry literature. At the same time, I have magnified these best-fit parameters derived from these measurements.
fitting “warts” by plotting k vs cell potential. In the literature Other recent Marcus analyses30−32 have compared sets of
authors almost always plot ln(k) vs cell potential, thus organic reductant substrates (e.g., phenols, phenothiazines,
minimizing the appearance of scatter. However, the semilog phenylenediamines) interacting with peroxidase and laccase
plot underweights the importance of high rates in the fitting enzymes from various species. In general, differences in rate
process, and overweights the importance of low rates, which constants are due either to AET/Kd alone (as in Figure 9), or to
can lead to inaccurate fitting results. a combination of both λ and AET/Kd (as in Supporting
Even with this drawback, the ln(k) vs cell potential semilog Information, Figure S1). Rarely, it seems, do changes in λ alone
plot can serve a useful purpose, especially when the range of cause a dramatic shift in reaction rate, but Roth and Klinman33
available cell potentials is limited and data scatter is substantial. recently made just such an argument for glucose oxidase.
The semilog version of k2 and k3 for peroxidase reduction by Glucose oxidase works via a “ping pong” mechanism in which
phenols is shown in Figure 11. As with the k vs cell potential the flavoenzyme is first reduced by glucose in a two-electron
process (glucose-CHOH + FAD → gluconolactone=O +
FADH:− + H+), and then oxidized by dioxygen in a sequence of
two one-electron steps:47,48
The first of the steps in Scheme 2, creating the
flavosemiquinone and superoxide radicals, is rate-limit-

Scheme 2

ing.33,49,50 Also, the active site histidine516 is a critical residue:

In the initial enzyme reduction step the deprotonated imidazole
side chain on his516 base-catalyzes proton removal from
glucose’s C1H−OH hydroxyl group;50 in the subsequent
enzyme oxidation step depicted in Scheme 2, the protonated
his516 group serves a key catalytic function.33,50 One would
Figure 11. Semilog plot of data in Figure 9, with straight line drawn ordinarily expect this latter function to involve acid-catalysis,
through first four points: ln[k2, M−1 s−1]), blue diamonds, slope = 26 however, the rate-determining step of enzyme oxidation
± 6 V−1, intercept = 16.8 ± 1.2, R2 = 0.90; ln(k3, M−1 s−1), black (creation of the flavin and superoxide radicals) involves
circles, slope = 28 ± 4 V−1, intercept = 13.1 ± 0.7, R2 = 0.95. electron transfer only.
1165 dx.doi.org/10.1021/ed1007712 | J. Chem. Educ. 2012, 89, 1159−1167
Journal of Chemical Education

■
Article

This and other lines of evidence51 suggest that the AUTHOR INFORMATION
protonated his516H+ optimizes enzyme oxidation/dioxygen Corresponding Author
reduction via its positive charge rather than its acidity.33 The
importance of establishing a “favorable electrostatic environ- *E-mail: [email protected].
ment” to lower the activation energy of enzyme-catalyzed Present Address
hydrogen-transfer reactions has recently been discussed.52 In School of Chemistry, National University of Ireland, Galway,
fact, in their temperature studies of glucose oxidase activity at Ireland.
high pH and low pH, Roth and Klinman33 showed that the
enhanced activity of the low pH form of the enzyme is due to Notes
its lower activation energy (3.0 vs 9.0 kcal/mol), and not to any The authors declare no competing financial interest.
large difference in frequency factor (0.5−4 nM−1 s−1 at the pH
extremes). The authors concluded that the positive charge of
the protonated his516 imidazole lowered the activation energy
■ ACKNOWLEDGMENTS
I wish to thank my colleagues Donal Leech for his helpful
for electron transfer from reduced FADH:− to dioxygen33 by suggestions and encouragement, and J. Charles Williamson,
lowering the reorganization energy of the one-electron transfer who went above and beyond the call of duty in his close reading
reaction. They calculated values of λ = 23 kcal/mol (1.0 eV) at and constructive criticisms of this paper.

■
low pH vs 42 kcal/mol (1.8 eV) at high pH. Although the
magnitude of this difference may be affected by corrections to REFERENCES
calculation errors,53 the conclusion that the positive charge of
his516H+ at the active site serves to aid catalysis by orienting (1) Schultz, E. J. Chem. Educ. 2010, 87, 472−473.
(2) Silbey, R. J.; Alberty, R. A.; Bawendi, M. G. Physical Chemistry,
solvent in the active site environment and lowering 4th ed.; J.Wiley & Sons: New York, 2005; pp 690−692.
reorganization energy seems secure. This echoes the effects of (3) In “inner sphere” redox reactions, the donor and acceptor are
low reorganization energies seen for electron transfer within the connected by chemical bonds or bridges, and through-bond electron
photosynthetic reaction center.38 transfer may occur.

■ CONCLUSION
Marcus Theory has been applied usefully to inorganic and
(4) Marcus, R. A. (1992) “Rudolph A. Marcus - Nobel Lecture”.
Nobelprize.org. June 24, 2010; http://nobelprize.org/nobel_prizes/
chemistry/laureates/1992/marcus-lecture.html.
(5) Marcus, R. A.; Sutin, N. Biochim. Biophys. Acta 1985, 811, 265−
biological electron transfer reactions. Generally these analyses 322.
are carried out on a set of rate constants for a series of related (6) Scott, R. A.; Mauk, A. G.; Gray, H. B. J. Chem. Educ. 1985, 62,
redox enzyme reactions with differing cell potentials. The series 932−938.
can comprise either a single enzyme reacting with structurally (7) Van Houten, J. J. Chem. Educ. 2002, 79, 1055−1059.
related substrates (e.g., phenols, indoles, phenylenediamines, (8) Martins, L. J. A.; da Costa, J. B. J. Chem. Educ. 1988, 65, 176−
phenothiazines) or a single substrate reacting with a series of 178.
related enzymes. Not all such cell potential, rate constant data (9) Newton, T. W. J. Chem. Educ. 1968, 45, 571−575.
sets show the bell-shaped peak expected from Marcus Theory. (10) Lewis, N. A. J. Chem. Educ. 1980, 58, 478−483.
(11) Neta, P. J. Chem. Educ. 1981, 79, 110−113.
There are at least five reasons why this might be the case: (1) (12) For electron transfer at electrodes, either D or A could represent
The redox enzyme-catalyzed reaction is inner-sphere (unusual the electrode; for biological electron transfer reactions, either D or A
for enzymes). (2) The rate-determining step is product release could represent a redox enzyme.
rather than electron transfer (i.e., krel ≪ kET);54,55 in this case, (13) Because of the difference between the Arrhenius activation
because product release involves nuclear motion, Figure 1 energy (Ea) and the Eyring activation free energy (ΔG⧧), AET in eq 1
would be a gross oversimplification and there should be no actually equals the Arrhenius frequency factor (A0) divided by
discernible relationship between ΔG° and ΔG⧧. Or, the series e(1+ΔS‡/R) (derivation in Supporting Information A).
of steps in Scheme 1 may feature inconstant (3) binding (14) Interestingly, β values measured for nonphotochemical adiabatic
parameters (ka, k−a, Kd); (4) binding orientation/electron systems all seem to lie in the 10−15 nm−1 (1.0−1.5 Å−1) range, with
transfer distance (AET); or (5) reorganization energy (λ). nonbiological reactions5 centered around 12 ± 2 nm−1, and biological
electron transfers closer to 14 ± 2 nm−1. The exponential distance
However, for reactions that are successfully described by dependence factor, β, can be interpreted roughly in terms of electron
Marcus Theory, informative comparisons can be made tunnelling through a square potential barrier5,14; values of β = 11 and
regarding differences in reorganization energy (λ) vs binding 15 nm−1 represent barrier heights of 1.1 and 2.1 eV, or 110 and 200
(ka, AET/k−a, or AET/Kd). Furthermore, Marcus Theory kJ/mol, respectively5. For a particular redox reaction, κ(r) is assigned
illustrates two points of interest to upper division chemistry an average value of κ over the relevant range of electron transfer
undergraduates: thermodynamics do control kinetics for outer- distances; for an adiabatic reaction, that is, one with strong electronic
sphere electron transfer reactions (unlike most other chemical coupling between reactants, κ ≈ 1.
reactions); and the reason for this can be explained with (15) Marcus, R. A. Annu. Rev. Phys. Chem. 1964, 15, 155−196.
straightforward applications of common physical chemical (16) The converse works just as well: a single acceptor with a series
concepts like transition state theory, Arrhenius’ Law, and the of related donors.
(17) Candeias, L. P.; Folkes, L. K.; Porssa, M.; Parrick, J.; Wardman,
Franck−Condon Principle.

■
P. Biochemistry 1996, 35, 102−108.
(18) A simplified derivation of eq 2 is included in the Supporting
ASSOCIATED CONTENT Information B.
*
S Supporting Information (19) Marcus explains the situation thusly5: because of the “small
overlap of the electronic orbitals, it then follows that [electron]
Derivation of AET vs A0, derivation of the Marcus Equation, and transfer will occur only at or near nuclear configurations for which the
steady state derivation of eq 4, and Figures S1a, S1b, and S2. total potential energy of the [solvated] reactants is equal to that of the
This material is available via the Internet at http://pubs.acs.org. [solvated] products.”

1166 dx.doi.org/10.1021/ed1007712 | J. Chem. Educ. 2012, 89, 1159−1167

Journal of Chemical Education Article

(20) From eq 3, ΔG⧧(λlow) = ΔG⧧(λhigh) when λlow = ΔE°2/λhigh. (46) This is essentially a kinetic Linear Free Energy Relationship
Hence, as long as λlow > ΔE°2/λhigh, ΔG⧧(λlow) < ΔG⧧(λhigh) and applied to electron transfer. It is an application of the Bell−Evans−
kET(λlow) > kET(λhigh), i.e., the behavior assumed by many authors. Polanyi Principle, also known as the Bronsted Catalysis Law (see Roy,
However, if λlow < ΔE°2/λhigh, the opposite holds true. In other words, S.; Goedecker, S.; Hellmann, V. Phys. Rev. E 2008, 77, 056707 , and
if ΔE° = 1 eV and λhigh = 2 eV (point b in Figure 3), then as long as 0.5 references cited therein).
< λlow < 2 eV (point b′), ΔG⧧(λlow) < ΔG⧧(λhigh). However, if λlow < (47) Gibson, Q. H.; Swoboda, B. E. P.; Massey, V. J. Biol. Chem.
0.5 eV, then ΔG⧧(λlow) > ΔG⧧(λhigh). Similar conclusions can be 1966, 239, 3927−3934.
drawn concerning λhigh = 4 eV (point c in Figure 3): The cutoff in this (48) Weibel, M.; Bright, H. J. J. Biol. Chem. 1971, 246, 2734−2744.
case is at λlow = 0.25 eV (point c′). (49) Palfey, B. A.; Ballou, D. P.; and Massey, V. In Active Oxygen in
(21) This is true for the special case of AET (low λ) = AET (high λ). If Biochem-Biochemistry; Valentine, J. S., Foote, C. S., Greenberg, A.,
AET (low λ) ≠ AET (high λ), then the “normal” range is bounded by ± Liebman, J. F., Eds.; Chapman & Hall: New York, 1995; pp 37−83.
(50) Leskovac, V.; Trivic, S.; Wohlfart, G.; Kandrac, J.; Pericin, D. Int.
f√λlowλhigh, where f = {1−[4RT/(nF(λhigh − λlow)] ln(AET,λhigh/ J. Biochem. Cell Biol. 2005, 37, 731−750.
AET,λlow)}1/2. (51) For example, the kinetic isotope effect is weak, as is the rate
(22) From eqs 1−3, Δw1/2 = 4[λ·(−wrct + RT ln(AET/k−a) + 2 dependence on solvent viscosity; also, the kinetic isotope effect is
ewrct/RT)/(nF)]1/2). If binding is rate-limiting (i.e., k−a ≪ AET), then this unaltered by pH and unaltered by removal of the imidazole group in
reduces to Δw1/2 = 4[λ·(−wrct + RT ln(AET/k−a))/(nF)]1/2). If the his516→ala516 mutant.
electron transfer is rate-limiting (i.e., k−a ≫ AET), then this reduces to (52) Hammes-Schiffer, S.; Benkovic, S. J. Annu. Rev. Biochem. 2006,
Δw1/2 = 4[λ·(RT ln(2))/(nF)]1/2), which for n = 1 and T = 25 °C 75, 519−541.
gives Δw1/2 = (0.534 V1/2)√λ. (53) After corrections to values of ΔG⧧ and wrct, I get λ(low-pH) =
(23) Weaver, M. J. J. Phys. Chem. 1980, 84, 568−576. 56 kcal/mol vs λ(high-pH) = 66 kcal/mol.
(24) Saveant, J.-M.; Tessier, D. Faraday Discuss. Chem. Soc. 1982, 74, (54) Cleland, W. W. Acc. Chem. Res. 1975, 8, 145−151.
(55) Cleland54 argues that for many (perhaps even most) enzymes,
57−72.
product release and not catalysis is rate-limiting. Regarding redox
(25) Folkes, L. K.; Candeias, L. P. FEBS Lett. 1997, 412, 305−308.
(26) Khopde, S. M.; Priyadarsini, K. I. Biophys. Chem. 2000, 88, 103− enzymes, he gives evidence that this is the case for a number of
dehydrogenases (glyceraldehyde-3-phosphate, alcohol, and lactate, but
109.
NOT glutamate) and oxidative decarboxylases (malate and isocitrate).
(27) Candeias, L. P.; Folkes, L. K.; Porssa, M.; Parrick, J.; Wardman,
P. Biochemistry 1996, 35, 102−108.
(28) Candeias, L. P.; Folkes, L. K.; Wardman, P. Biochemistry 1997,
36, 7081−7085.
(29) Kulys, J.; Deussen, H.-J.; Krikstopaitis, K.; Lolke, R.; Schneider,
P.; Ziemys, A. Eur. J. Org. Chem. 2001, 2001, 3475−3484.
(30) Krikstopaitis, K.; Kulys, J.; Pedersen, A. H.; Schneider, P. Acta
Chim. Scand. 1998, 52, 469−474.
(31) Kulys, J.; Krikstopaitis, K.; Ziemys, A. JBIC 2000, 5, 333−340.
(32) Xu, F. Biochemistry 1996, 35, 7608−7614.
(33) Roth, J. P.; Klinman, J. P. Proc. Natl. Acad. Sci. U.S.A. 2003, 100,
62−67.
(34) Forrow, N. J.; Sanghera, G. S.; Walters, S. J. J. Chem. Soc., Dalton
Trans. 2002, 2002, 3187−3194.
(35) Zakeerudin, S. M.; Fraser, D. M.; Nazeerudin, M.-K.; Gratzel, M.
J. Electroanal. Chem. 1992, 337, 253−283.
(36) Zakeerudin, S. M.; Gratzel, M.; Fraser, D. M. Biosens. Bioelectron.
1996, 11, 305−315.
(37) Zhang, C.; Haruyama, T.; Kobatake, E.; Aizawa, M. Anal. Chim.
Acta 2000, 408, 225−232.
(38) Jia, Y.; DiMagno, T. J.; Chan, C.-K.; Wang, Z.; Du, M.; Hanson,
D. K.; Schiffer, M.; Norris, J. R.; Fleming, G. R.; Popov, M. S. J. Phys.
Chem. 1993, 97, 13180−13191.
(39) Penner-Hahn, J. E.; Eble, K. S.; McMurry, T. J.; Renner, M.;
Balch, A. L.; Groves, J. T.; Dawson, J. H.; Hodgson, K. O. J. Am. Chem.
Soc. 1986, 108, 7819−7825.
(40) Schultz, C. E.; Devaney, P. W.; Winkler, H.; Debrunner, P. G.;
Doan, N.; Chiang, R.; Rutter, R.; Hager, L. P. FEBS Lett. 1979, 103,
102−105.
(41) Veitch, N. C. Biochem. Soc. Trans. 1995, 23, 232−239.
(42) For indoles at pH 5 (Supporting Information, Figure S1a), the
data could only be fit using the three-parameter fit from eqs 1, 3, and
4, suggesting that binding (not electron transfer) is rate-limiting.
(43) Except at ΔE° = 0.03 V ≈ λ. Here all three rate constants (k2 for
phenols at pH 7, indoles at pH 7, and indoles at pH 5) are about the
same.
(44) This data set could only be fit by assuming that binding (not
electron transfer) is rate-limiting. Best-fit results are: ka = 38.3 ± 0.9
μM−1 s−1, and AET/k−a = 11.1 ± 1.0 (unitless); thus AET/Kd = ka(AET/
k−a) = 430 ± 40 μM−1 s−1.
(45) Except at ΔE° = 0.03 V ≈ λ. Here, saturation due to the rate-
limiting binding step keeps the rate constant low.

1167 dx.doi.org/10.1021/ed1007712 | J. Chem. Educ. 2012, 89, 1159−1167