31P NMR spectroscopic analysis on photooxidation of 1,n-bis(diphenylphosphino)alkanes with the aid of DFT calculations

Article abstract of DOI:10.1002/poc.4021

The chloroform-d solution of diphosphine, 1,n-bis(diphenylphosphino)alkane (Ph₂P(CH₂)_nPPh₂; n = 1-6), was photolyzed with light from a xenon lamp in air. The progress of the reaction was followed by ³¹P NMR spectroscopy. The observed spectral change showed that the diphosphine is initially oxidized to diphosphine monoxide, Ph₂P(═O)(CH₂)_nPPh₂, which is further oxidized to diphosphine dioxide, Ph₂P(═O)(CH₂)_nP(═O)Ph₂. The oxidation of the diphosphine to the diphosphine monoxide took place according to first-order kinetics with respect to the concentration of the diphosphine, the first-order rate constant, k_obs, being larger with increasing number of the methylene units in the spacer. The observation in kinetics is interpreted based on the conformation of the diphosphine radical cation intermediate initially generated by electron transfer from the photoexcited diphosphine to oxygen. Density functional theory (DFT) calculations predict that the diphosphine radical cation takes “folded” conformation where two phosphorus atoms are arranged closely to each other. The “folded” conformer of the diphosphine radical cation results from electrostatic interaction of these two phosphorus atoms. This conformer explains the observed dependency of k_obs on the length of the spacer in the diphosphine.

Full text of DOI:10.1002/poc.4021

Received: 24 April 2019
DOI: 10.1002/poc.4021
Revised: 6 August 2019
Accepted: 6 August 2019
R E S E A R C H A R T I C L E
31P NMR spectroscopic analysis on photooxidation of 1,n-
bis(diphenylphosphino)alkanes with the aid of DFT
calculations
Shinro Yasui1
| Shoko Yamazaki2
1Institute of Human and Environmental
Sciences, Tezukayama University, Nara,
Japan
Abstract
The chloroform-d solution of diphosphine, 1,n-bis(diphenylphosphino)alkane
(Ph2P(CH2)nPPh2; n = 1-6), was photolyzed with light from a xenon lamp in
air. The progress of the reaction was followed by 31P NMR spectroscopy. The
observed spectral change showed that the diphosphine is initially oxidized to
diphosphine monoxide, Ph2P(═O)(CH2)nPPh2, which is further oxidized to
diphosphine dioxide, Ph2P(═O)(CH2)nP(═O)Ph2. The oxidation of the
diphosphine to the diphosphine monoxide took place according to first-order
kinetics with respect to the concentration of the diphosphine, the first-order
rate constant, kobs, being larger with increasing number of the methylene units
in the spacer. The observation in kinetics is interpreted based on the conforma-
tion of the diphosphine radical cation intermediate initially generated by elec-
tron transfer from the photoexcited diphosphine to oxygen. Density functional
theory (DFT) calculations predict that the diphosphine radical cation takes
“folded” conformation where two phosphorus atoms are arranged closely to
each other. The “folded” conformer of the diphosphine radical cation results
from electrostatic interaction of these two phosphorus atoms. This conformer
explains the observed dependency of kobs on the length of the spacer in the
diphosphine.
2Department of Chemistry, Nara
University of Education, Nara, Japan
Correspondence
Shinro Yasui, Institute of Human and
Environmental Sciences, Tezukayama
University, Gakuen-Minami, Nara
631-8585, Japan.
Email: yasui@tezukayama-u.ac.jp
K E Y W O R D S
31P NMR spectroscopy, DFT calculation, diphosphine, phosphine radical cation, photooxidation,
steady-state photolysis
1 | INTRODUCTION
cation, Ar3Pꢀ+ (Ar = aryl), generated during laser flash
photolysis, undergoes competitively radical coupling with
Trivalent phosphorus compounds, Z3P, are easily
converted to the corresponding radical cations, Z3Pꢀ+
oxygen and electrophilic attack on the parent phosphine,
giving peroxidic radical cation, Ar3P–OOꢀ+, and dimeric
phosphine radical cation, [Ar3P–PAr3]ꢀ+, respectively.[6]
Besides, when another trivalent phosphorus compound
Z03P exists in the vicinity of Z3Pꢀ+, Z3Pꢀ+ may interact
with Z03P somehow. It is interesting to explore how the
interaction is regulated by the distance and the geometry
of these two phosphorus compounds.
,
so-called phosphoniumyl radical,[1] upon the treatment
with electron acceptors such as diazonium salts,[2]
bipyridinium dications (viologens),[3] metal complexes,[4]
and dyes in the photoexcited state.[5] The radical cation,
Z3Pꢀ+, thus formed is either a radical or a cation in prin-
ciple. We found recently that a triarylphosphine radical
J Phys Org Chem. 2020;e4021.
wileyonlinelibrary.com/journal/poc
© 2020 John Wiley & Sons, Ltd.
1 of 11
https://doi.org/10.1002/poc.4021

2 of 11
YASUI AND YAMAZAKI
To conveniently tune the arrangement of two phos-
the optimized conformation of the radical cation 1ꢀ+. The
photolysis of triphosphine, bis(2-(diphenylphosphino)
ethyl)phenylphosphine (4), is also discussed briefly to
evaluate how the third phosphine unit influences the oxi-
dation of the first phosphine unit.
phorus atoms, we take a diphosphine in this study, where
two phosphine units are covalently connected by a
spacer. By changing the length of the spacer, two phos-
phorus atoms in the diphosphine would be arranged dif-
ferently. Meanwhile, we have found that steady-state
photolysis of a phosphine with aromatic ligands in air is
a convenient tool to generate the corresponding phos-
phine radical cation. Thus, during photooxidation of tri-
arylphosphine, Ar3P, using a xenon lamp in air,
triarylphosphine radical cation, Ar3Pꢀ+, is generated as
an intermediate without giving prominent by-products.[7]
It is anticipated that when a diphosphine with aromatic
ligands is photolyzed in air, the corresponding
diphosphine radical cation is generated as an intermedi-
ate as well on the way to give the diphosphine monoxide.
Two phosphorus atoms in the resulting diphosphine radi-
cal cation would interact with each other, differently
depending on the length of the spacer. Difference in the
interaction in terms of the distance and the geometry
should result in difference in the oxidation rate.
2 | EXPERIMENTAL
2.1 | Instruments
The 31P NMR spectra were recorded using a Varian
INOVA 400 operating at 161.9 MHz. External 85% phos-
phoric acid was used as the reference standard (0.0 ppm).
UV-visible spectra were recorded using a Shimadzu UV-
2200A.
2.2 | Materials
Diphosphines, 1a-e, as well as triphosphine, 4, were com-
mercially available (Aldrich) and purified by recrystalli-
zation from ethanol except for 1c. Recrystallization of 1c
was not feasible because of rather low melting point
(mp 63-65ꢁC) compared with that of the others (ranging
from 118ꢁC to 142ꢁC). The 1c sample was subjected to
the photolysis without purification. No appreciable sig-
nals other than that from 1c were found on the 31P NMR
spectra before the photolysis. Deuterated chloroform
(CDCl3) used was a product of Euriso-Top (>99.80%).
In a preliminary report,[8] we presented the results
from steady-state photolysis of chloroform-d (CDCl3)
solution of 1,-n-bis(diphenylphosphino)alkane (1) (n =
1-6) performed by using a xenon lamp in air. Followed by
31P NMR spectroscopy, it has been found that the two
phosphine units in 1 are consecutively oxidized via
diphosphine monoxide (2), eventually to diphosphine
dioxide (3) (Scheme 1). The oxidation of 1 to 2 most likely
takes place via the diphosphine radical cation, 1•+, gener-
ated by electron transfer (ET) from 1 in the excited state
to oxygen. This mechanism allows us to evaluate how
two phosphorus atoms interact in 1ꢀ+; thus, the rate of
the oxidation of 1 to 2 would reflect the way of the P–P
2.3 | General procedure
interaction in 1ꢀ+
.
A CDCl3 solution of diphosphine (1) (2.00 × 10−3M) con-
taining triethyl phosphate as the internal standard (8.0
μL/mL) was prepared in a quartz NMR tube. The solu-
tion was irradiated by light from a Ushio xenon short arc
UXL-500D-0 lamp powered by a Ushio XB-50101AA
power supply. A Toshiba sharp-cut glass filter, UV-31,
was placed between the lamp and the reaction vessel to
control the wavelengths of the irradiating light.
In this article, we present the rate constants of the
oxidation of 1 to 2 determined by 31P NMR spectroscopy.
The results from density functional theory (DFT) theoret-
ical calculations are also reported to discuss the opti-
mized conformations of 1ꢀ+ as well as 1. The dependency
of the rate constant of the oxidation rate of 1 to 2 on the
length of the spacer in 1 is examined in connection with
SCHEME 1 The photooxidation of diphosphine 1 under the steady-state irradiation by a xenon lamp

YASUI AND YAMAZAKI
3 of 11
2.4 | Theoretical calculations
The 31P NMR spectrum of the prepared solution of 1b is
given in Figure S1-1. Figure 1 shows that the integral
ratio of the signal from 1b (Int (smp)) and that from
(EtO)3P═O (Int (std)) is linearly dependent on the con-
centration of 1b, confirming that the ratio, Int (smp)/Int
(std), for each signal represents the relative concentration
of each phosphorus species. For the sake of comparison,
we measured 31P NMR spectrum of triphosphine, 4, in
CDCl3 and acetonitrile-d3 (CD3CN). The chemical shifts
are presented in Table 2. The spectrum obtained in
CD3CN is given in Figure S1-2, where the terminal and
the central phosphorus nuclei in 4 exhibit doublet and
triplet coupling, respectively.
The geometries of the neutral diphosphines and the
diphosphine radical cations were fully optimized at the
B3LYP/6-31G(d) level of theory with normal convergence
using the Gaussian 09, Revision A.02 program.[9] No
imaginary frequencies were obtained.
3 | RESULTS
3.1 | 31P NMR spectroscopy
The deuterated chloroform (CDCl3) solution of
diphosphine (1) (2.00 × 10−2M) containing triethyl phos-
phate, (EtO)3P═O, as an internal standard (δ = 1.0 ppm
from 85% H3PO4)[10] was subjected to 31P NMR spectros-
copy. Parameters used for the 31P NMR measurement are
given in Table S1. The phosphorus nuclei in 1a-e reso-
nate in the range of −20 to −10 ppm as shown in Table 1.
3.2 | Steady-state photolysis
The steady-state photolysis using
a xenon lump
(>310 nm with a glass filter) was carried out on the
CDCl3 solution of
1
(2.00
×
10−2M) containing
TABLE 1 31P NMR chemical shifts of 1-3
Compound
1a
2a (P(III))c
2a (P(V))c
3ac
na
δ in ppm from 85% H3PO4
1
−19.9
−26.3 [50.4]
32.3
(−22.6)b
(−28.5)b
(28.9)b
(−26.5 [50.5])d
(29.8)d
27.9
(24.7)b
1b
2
3
4
6
−10.6
−9.9 [48.9]
34.9
(−12.8)b
(−12.2)b
(32.6)b
2b (P(III))c
2b (P(V))c
3bc
(−11.5 [32.3])d
(−11.6 [48.9])e
(32.3)d
(33.2)e
34.8
(32.8)b
1c
−15.2
−15.9
34.2
(−17.8)b
(−18.5)b
(31.9)b
2c (P(III))c
2c (P(V))c
3cc
(−17.2)d
(32.0)d
34.5
(32.3)b
1d
−13.7
−13.8
34.1
2d (P(III))c
2d (P(V))c
3dc
(−16.5)d
(32.2)d
34.2
1e
−14.1
−13.5
34.6
2e (P(III))c
2e (P(V))c
3ec
34.5
Note. Measured in CDCl3. The values in parentheses are literature values.
an in Ph2P(CH2)nPPh2.
bBerners-Price et al.11 In CDCl3.
cObserved during the photolysis of 1 (2.00 × 10−2M). The values in brackets denote the JP–P coupling constants in Hz.
dGrushin.12 In CDCl3.
ePetersson et al.13 In CDCl3.

4 of 11
YASUI AND YAMAZAKI
TABLE 2 31P NMR chemical shifts of 4
δ, ppm from 85% H3PO4
P(a)
−12.5 (d, 3JP–P = 30
P(b)
−15.8 (t, 3JP–P = 30
In
CD3CN
In CDCl3
Hz)
Hz)
−12.2 (d, 3JP–P = 30
−15.0 (broad)
Hz)
Note. [4] = 2.00 × 10−2M.
observed after 90-minute irradiation of 1a, exhibited
splitting of the signals (Figure S1-4). Similarly, splitting
of the signals of 2b (n = 2) was observed during the pho-
tolysis of 1b after 15-minute irradiation. The observations
confirm unsymmetrical structures of 2a and 2b. No cou-
pling was observed in 2c-e (n ≥ 3).
The 31P NMR signals from 1 were apt to shift down-
ward as the reaction proceeded. Control experiments con-
firmed the concentration shift of the signals from 1. For
example, δ values (in ppm) of the signal from 1b are
−10.6, −10.4, and −10.2 at 2.00 × 10−2M, 1.60 × 10−2M,
and 7.80 × 10−3M, respectively.
FIGURE 1 Plot of the integral ratio, Int (smp)/Int (std),
against the concentration of 1b
(EtO)3P═O in a quartz NMR tube in air. The solution
was analyzed periodically by 31P NMR spectroscopy. The
31P NMR spectrum of 1d observed after 15-minute irradi-
ation is given in Figure 2. The signal (−13.7 ppm) from
the phosphorus of the starting diphosphine, 1d, gradually
disappeared under the irradiation with the concomitant
appearance of two new signals resonated at −13.8 and
34.1 ppm. Comparing these δ values with literature
values, these new signals are judged to result from two
phosphorus atoms in diphosphine monoxide, 2d.[11–13]
Namely, the former (at higher field) and the latter
(at lower field) result from the trivalent and pentavalent
phosphorus atoms in 2d, respectively. These signals dis-
appeared slowly, being accompanied by the appearance
of the signal at 34.2 ppm assignable to the phosphorus
from diphosphine dioxide, 3d. These observations indi-
cate that the oxidation of a phosphine unit in 1 is
followed by the sluggish oxidation of the second phos-
phine unit. The time course of the photooxidation of 1d
is shown in Figure 3, where relative values of Int (smp)/
Int (std) are plotted against time.[14]
A similar spectral change was observed during the
photolysis of the other diphosphines, 1a-c,e, showing
that two phosphine units in 1 are oxidized consecutively
as shown in Scheme 1. 31P NMR spectroscopic analysis
on the photolysis of the triphosphine, 4, indicates that
one of the terminal phosphine units is initially oxidized
(Scheme 2). Time course of the photolysis of 1a-c,e, and
4 is depicted in Figures S2 and S3. The 31P NMR chemical
shifts from 1 and those from 2 and 3 observed during the
photolysis of 1 are summarized in Table 1, along with lit-
erature values. The spectrum of 2a (n = 1), which was
Several unidentified signals appeared in the 31P NMR
spectra during the photolysis of 1 in CDCl3. For example,
the signals at 23, 27, and 47 ppm appeared upon the pho-
tolysis of 1d (Figure 2). These signals also appeared in a
considerable amount when the CDCl3 solution (2.00 ×
10−2M) of 1d was photolyzed under argon atmosphere for
1 hour. Meanwhile, the photolysis of 1d in CD3CN in air
did not afford these signals but afforded exclusively the
signal at 31 ppm assignable to 3d.[16] These observations
suggest that the unidentified signals under consideration
are assignable to products resulting from abstraction of
hydrogen or chlorine atom(s) from the solvent CDCl3 by
the radical cation intermediate. Unfortunately, we have
no way to identify and quantify these products.
3.3 | Kinetics
Figure 4 shows logarithmic plots of the decrease in the
concentration of 1 (as well as 4), [1], during the steady-
state photolysis.[17] The straight line obtained in the plot
demonstrates that the decrease in [1], namely, the oxida-
tion of 1 to 2, obeys first-order kinetics with respect to
[1]. That is, Equation (1) holds for the oxidation of 1 to
2 (as well as that of 4 to 5).
−d½1ꢂ=dt ¼ kobs½1ꢂ,
ð1Þ
where kobs is the observed first-order rate constant. The dis-
appearance of 2, which corresponds to the second oxida-
tion, namely, oxidation of 2 to 3, was too slow to interfere

YASUI AND YAMAZAKI
5 of 11
FIGURE 2 31P NMR spectrum of 1d measured in CDCl3 after 15-min irradiation of light from a xenon lamp. δ in ppm from 85%
H3PO4. The signal from (EtO)3P═O at δ = 1.0 ppm included. Numbers below the scale denote the integral values normalized by taking
100 for the integral value of (EtO)3P═O. The inset is a magnification of the region enclosed by broken lines. The signals with asterisk are
unidentified. The magnified spectrum in the region of −13.4 to −14.0 ppm is presented in Figure S1-3, which shows that two signals from 1d
and 2d(III) are clearly distinguishable
with the kinetics of the first oxidation.[18] From the slope of
the plot in Figure 4, first-order rate constant, kobs, was deter-
mined for each reaction of 1a-e as well as 4, as summarized
in Table 3. The order of kobs is 1a < 1b < 1c < 1d < 1e. The
triphosphine, 4, which has the same number of the methy-
lene spacers as 1b, is oxidized more slowly than 1b.
FIGURE 3 Time course of the photolysis of 1d in CDCl3 in
air. Relative integral intensities of the 31P NMR signals are plotted
SCHEME 2 The photooxidation of triphosphine 4 under the
steady-state irradiation by a xenon lamp
against time. Irradiated by light of wavelengths over 310 nm. See
reference15

6 of 11
YASUI AND YAMAZAKI
spectra were similar to each other, showing small absorp-
tion in this range of the wavelengths. The extinction coef-
ficients ε at 310 nm for 1a-e and 4 are around 200M−1
cm−1 as listed in Table 5.[19]
3.6 | DFT calculations
The changes in the total energy of the diphosphine, 1,
and the diphosphine radical cation, 1ꢀ+, were computed
as a function of the through-space distance between two
phosphorus atoms, d(P–P), on the basis of
B3LYP/6-31G(d). For both 1 and 1ꢀ+, at least two local
minima with no imaginary frequencies were obtained
along the coordinate of d(P–P). The energy minima that
occur at conformations with shorter and longer d(P–P)
results from “folded” and “linear” conformers, respec-
tively. The structures of these conformers of 1ꢀ+ are
FIGURE 4 Logarithmic plots of the decrease in the amounts
of 1 and 4 during the steady-state photolysis in CDCl3 in air.
Irradiated by light of wavelengths over 310 nm
depicted in Figure 5 (1eꢀ+) and in Figure S4 (1aꢀ+, 1bꢀ+
,
1cꢀ+, and 1dꢀ+). The “folded” conformers of 1ꢀ+
exhibited distinct local minima in the energy diagrams,
while the minima corresponding to the “linear” confor-
mations of 1ꢀ+ were relatively shallow, as shown in
Figure 6A (1eꢀ+) and in Figure S5 (1bꢀ+, 1cꢀ+, and 1dꢀ+).
In the neutral diphosphine 1, energy minima were
shallow and ambiguous even at the “folded” conforma-
tion. The energy minima for the “folded” and “linear”
TABLE 3 First-order rate constant, kobs, of photooxidation of
1 and 4
1
kobs, 104 s−1
1.70
1a
1b
1c
1d
1e
4
2.04
2.42
4.50
TABLE 4 First-order rate constant, kobs, of photooxidation of 1
in the absence or the presence of sensitizer, Rho+, under the
irradiation with light of different wavelengths
5.02
1.29
Note. [1] = 2.00 × 10−2M in CDCl3 in air. Irradiated by light of wavelengths
over 310 nm.
kobs, 104 s−1
Conditions
k(1e)/k(1b)a
1b
1e
>310 nm,b without Rho+
>430 nm,b with Rho+
2.04c
5.02c
2.46
1.89
3.4 | Photolysis in the presence of a
sensitizer
0.710
1.34
Note. [1] = 2.00 × 10−2M. [Rho+] = 1.00 × 10−4M in CDCl3. In air.
aThe ratio of kobs of 1e over kobs of 1b.
bThe wavelengths of the irradiating light.
cData from Table 3.
The CDCl3 solution of 1b or 1e (2.00 × 10−2M) was pho-
tolyzed with light of wavelengths over 430 nm using a
Toshiba glass filter Y-43 in the presence of a sensitizer,
rhodamine 6G (Rho+) (1.00 × 10−4M), in air. 31P NMR
spectroscopic analysis showed the formation of the same
oxidation products as in the nonsensitized photolysis
with light of wavelengths over 310 nm. Following 31P
NMR spectral change, the rate constant, kobs, for disap-
pearance of 1 was obtained as listed in Table 4.
TABLE 5 Extinction coefficient e of 1 and 4 at 310 nm
1
Concentration, mM
Absorbance
0.961
ε M−1 cm−1
1a
1b
1c
1d
1e
4
4.96
5.03
5.04
4.98
5.00
4.99
194
249
223
229
237
256
1.250
1.123
1.138
3.5 | UV-visible spectra
1.186
1.275
UV-visible spectra of the CHCl3 solutions of 1a-e and
4 were recorded in the range of 300 to 400 nm. The
Note. In CHCl3.

YASUI AND YAMAZAKI
7 of 11
FIGURE 5 Optimized “linear” (A) and
“folded” (B) conformers of 1eꢀ+ computed based
on B3LYP/6-31G(d)
1
conformers of 1e are shown in Figure 6B as represen-
tative examples. Figure S5 depicts the energy diagrams
for 1b-d. Table 6 summarizes the d(P–P) values of the
optimized “linear” and “folded” conformers of 1ꢀ+ and
1. The total energy at each optimized conformation is
given in Table 7, along with difference in the total
energy between the “linear” and “folded” conformers.
The “folded” conformer is more stable than the “lin-
ear” conformer for 1ꢀ+, whereas the opposite is the
case for 1.
(−1.05 V vs Ag/Ag+).[22] As a result, 1* undergoes ET to
O2 to generate the radical cation, 1ꢀ+, as well as superox-
ide radical anion, O2 .
ꢀ− [23] The resulting 1ꢀ+ is trapped by
O2 to produce the peroxidic radical cation, 7, which even-
tually collapses to 2. Meanwhile, the starting material 1 is
1 *
regeneratedꢀt−hrough nonradiative decay of 1 and back
ET from O2 to 1ꢀ+. The mechanism is summarized in
Scheme 4, where k0 represents the sum of the rate con-
stants for any nonradiative decay processes of 11*, and kb
is the rate constant oꢀf−the back ET. Radical coupling
The total energy of dimeric radical cation, 6aꢀ+, from
ethyldiphenylphosphine, EtPh2P, was computed as a
function of an intermolecular distance between two phos-
phorus atoms, d0(P–P), to give a distinct minimum in the
energy diagram with d0(P–P) = 3.02 pm as shown in
Figure 7. The d0(P–P) value of the optimized structure of
6aꢀ+ is included in Table 6. In addition, DFT calculation
gave d0(P–P) = 2.99 pm in the dimeric radical cation,
6bꢀ+, from phosphine, H3P (Scheme 3).
between 1ꢀ+ and O2
followed by collapse of the
resulting zwitterion intermediate is a possibility to give 2.
However, if this is the case, the observed rate would fol-
low second-order kinetics because the concentrations of
1ꢀ+ and O2ꢀ− are the same, which is against our observa-
tion shown by Equation (1).[24]
The presence of 1ꢀ+ as an intermediate during the
photooxidation of 1 to 2 was confirmed by experiments
using a sensitizer. Diphosphine 1b as well as 1e was pho-
tolyzed with light of wavelengths over 430 nm, where 1b
and 1e are completely transparent. Under these photo-
chemical conditions, a catalytic amount of a sensitizer,
rhodamine 6G (Rho+), promoted the oxidation of 1b and
1e to 2b and 2e, respectively. The sensitizer, Rho+,
absorbs light of wavelengths over 430 nm to be excited to
4 | DISCUSSION
Under the present photochemical conditions, 1 is excited
1 *
to its singlet excited state, 1 . Based on the oxidation
1
potential of 1 in the ground state (0.92-1.90 V vs
Ag/Ag+)[20] and the excitation energy of 1 at 310 nm,[21]
its singlet excited state, Rho+*. The half-wave potential,
E1/2, of 1Rho+* (1.22 V vs Ag/Ag+)[5b] is very close to that
of 1b (1.27 V) or higher than that of 1e (0.92 V),[20] all-
owing 1Rho+* to act as an electron acceptor from 1b and
1 *
the oxidation potential of 1 is predicted more negative
than the reduction potential of O2 in the ground state

8 of 11
YASUI AND YAMAZAKI
FIGURE 6 The changes in the total energy
of 1eꢀ+ (A) and 1e (B) computed as a function of
the through-space distance between two
phosphorus atoms, d(P–P), based on
B3LYP/6-31G(d)
1e to generate 1bꢀ+ and 1eꢀ+. That is, once 1ꢀ+ is gener-
ated in the presence of Rho+, the photooxidation of
1 takes place to give 2. The results strongly suggest that
1ꢀ+ would be an intermediate to give 2 in the non-
sensitized photolysis as well. Table 4 shows that the rate
ratio, k(1e)/k(1b), remains almost unchanged upon
change of the reaction conditions, suggesting that both of
the nonsensitized and Rho+-sensitized oxidations of
1 take place via a common intermediate. We have found
a similar catalytic role of Rho+ in the photooxidation of
triphenylphosphine to triphenylphosphine oxide.[25]
Based on Scheme 4, the disappearance of 1 is
expressed by Equation (2).
TABLE 6 The computed through-space distance d(P–P)
between two phosphorus atoms in 1 and 1ꢀ+ at the optimized
“folded” and “linear” conformations
1
1ꢀ+
1
na
d(P–P), pmb
d(P–P), pmb
“Linear”
“Folded”
“Linear”
“Folded”
308
– d½1ꢂ=dt ¼ ϕ’Ia – k0½1*ꢂ – kb½1•þꢂ,
ð2Þ
1a
1b
1c
1d
1e
6a
6b
1
2
3
4
6
–
–
–
310
350
372
387
605
–
–
where Φ0 and Ia are the efficiency of populating 1*
(the singlet or the triplet state) and the intensity of the
light absorbed by 1, respectively. With steady-state
approximations with respect to [1*] and [1ꢀ+],
Equation (2) is transformed into Equation (3).[26]
456
571
707
960
–
447
568
705
960
–
285
285
276
310
302c
299d
–
–
–
– d½1ꢂ=dt ¼ ϕ’Iak1k2=fðk0 þk1Þðkb þk2Þg,
ð3Þ
Note. Computed based on B3LYP/6-31G(d).
an in Ph2P(CH2)nPPh2.
bThe values in the optimized conformations.
where k1 and k2 denote the rate constants given in
Scheme 4. The quantum yield for the photooxidation
of 1 is extremely low[27]; namely, the nonradiative
cd0(P–P) in 6aꢀ+
dd0(P–P) in 6bꢀ+
.
.

YASUI AND YAMAZAKI
9 of 11
TABLE 7 Total energies of 1 and 1ꢀ+ at the optimized “folded” and “linear” conformations computed based on B3LYP/6-31G(d) and
difference in the total energies between “folded” and “linear” conformers
1
1ꢀ+
1
na
1
E (linear)b
–
E (folded)c
ΔEd
E (linear)b
–
E (folded)c
ΔEd
1a
1b
1c
1d
1e
−1648.607228
−1687.920497
−1727.228447
−1766.541754
−1845.172423
–
−1648.361870
−1687.692012
−1727.009209
−1766.321251
−1844.940440
–
2
−1687.922856
−1727.235290
−1766.549711
−1845.177093
0.002359 (6.19)e
0.006843 (18.0)e
0.007956 (20.9)e
0.011649 (12.3)e
−1687.674847
−1726.988148
−1766.302769
−1844.933447
−0.017165 (−45.1)e
−0.021062 (−55.3)e
−0.018482 (−48.5)e
−0.006993 (−18.4)e
3
4
6
Note. The values are reported in a.u. unless otherwise indicated.
an in Ph2P(CH2)nPPh2.
bThe total energy at the “linear” conformation.
cThe total energy at the “folded” conformation.
dDifference in the total energy, ΔE = E (folded) − E (linear).
eIn parentheses are reported the values in kJ mol−1
.
decay (k0) is much more rapid than the chemical pro-
cess (k1) (k0 >> k1). Meanwhile, since kb is certainly
diffusion limited (ca 1010 s−1), then kb >> k2 holds.
With these approximations as well as relationships, Ia
= 2.3εI0[1]l and k0 = 1/τ0, Equation (3) is reduced to
Equation (4).
– d½1ꢂ=dt ¼ 2:3εϕ’I0τ0lðk1k2=kbÞ½1ꢂ,
ð4Þ
where ε and I0 are the extinction coefficient of 1 and the
intensity of the incident light at a given wavelength,
respectively. l is the thickness of the reaction vessel, and
τ0 is the lifetime of 1*. Comparing Equation (4) with the
experimentally determined equation, Equation (1), we
finally obtain
FIGURE 7 The change in the total energy of the dimeric
radical cation, 6bꢀ+, from EtPh2P computed as a function of d0(P–P)
based on B3LYP/6-31G(d)
kobs ¼ 2:3εϕ’I0τ0lðk1k2=kbÞ:
ð5Þ
In Equation (5), I0 and l are constant throughout the
present study, and Table 5 shows almost invariable ε
values for 1a-e and 4.[29] The quantities τ0 and Φ0 are likely
to be approximately constant for 1a-e and 4. The k1 step is
highly exothermic as discussed above, rendering k1 for 1a-
e and 4 diffusion limited. Furthermore, the back ET from
O2ꢀ− to 1ꢀ+ is such exothermic that kb is diffusion limited.
SCHEME 3 Dimeric radical cation from H3P (6aꢀ+) and
EtPh2P (6bꢀ+
)
SCHEME 4 The mechanism of the
photooxidation of 1

10 of 11
YASUI AND YAMAZAKI
That is, both k1 and kb values are constant irrespective of
1. As a result, Equation (5) is reduced to Equation (6).
importance of the steric factor in the attack of O2 on the
phosphine radical cation.
kobs ¼ const:× k2,
ð6Þ
5 | CONCLUSION
which means that the difference in kobs results from the
difference in k2. That is, the rate of the oxidation of 1 to
2 is determined by the reactivity of the k2 step, where the
radical cation 1ꢀ+ is trapped by O2 to afford the peroxidic
radical cation 7.
The steady-state photolysis of diphosphine 1 in air was
examined by 31P NMR spectroscopy. The DFT calcula-
tions revealed that the diphosphine radical cation inter-
mediate, 1ꢀ+, is stabilized by electrostatic interaction
between two phosphorus atoms in 1ꢀ+, making the con-
formation of 1ꢀ+ “folded”. This conformer of 1ꢀ+ explains
the origin of the difference in the oxidation rate of 1. The
present study has provided an opportunity to evaluate
how a radical center in a phosphine radical cation inter-
acts with a neutral phosphorus atom in a phosphine.
The DFT calculations show that the “linear” con-
former of the neutral diphosphine 1 is more stable than
the “folded” one, whereas the opposite is the case for the
cation radical counterpart, 1ꢀ+ (Table 7).[30] Stabilization
brought about by electrostatic interaction of two phospho-
rus atoms in 1ꢀ+ overcomes steric disadvantage that other-
wise could be serious in the “folded” conformer. The
calculated distances, d(P–P), in 1ꢀ+ at the “folded” confor-
mation (Table 6) are very close to or even shorter than the
calculated P–P distances in model radical cations, namely,
dimeric radical cations from EtPh2P (6aꢀ+) (302 pm) and
H3P (6bꢀ+) (299 nm) (Scheme 3). That is, the P–P interac-
tion in the “folded” 1ꢀ+ is obvious, even though these dis-
tances are still longer than the reported distance of a
P(III)–P(III) single covalent bond (222 pm).[31]
ORCID
Shinro Yasui https://orcid.org/0000-0002-8752-9227
REFERENCES
[1] Nearly three decades ago, Tordo and his coworkers proposed
the term, “phosphoniumyl radical”, to designate a trivalent
phosphorus radical cation Z3Pꢀ+. The author (S.Yas.) wanted
to be a follower to them in using this term, appreciating its
simplicity and versatility. However, this term has never gotten
popular somehow. M. Culcasi, Y. Berchadsky, G. Gronchi,
P. Tordo, J. Org. Chem. 1991, 56, 3537.
The ground state of 1 exists as a more stable conformer,
namely, the “linear” conformer. Since the excitation takes
1 *
place without conformational change, 1 also takes the
“linear” conformation. As a result, the ET takes place from
“linear” 1 to O2, which is accompanied by conforma-
[2] a) S. Yasui, M. Fujii, C. Kawano, Y. Nishimura, K. Shioji,
A. Ohno, J. Chem. Soc., Perkin Trans. 2. 1994, 177. b) S. Yasui,
K. Shioji, A. Ohno, Heteroatom Chem. 1995, 6, 223.
1 *
tional change of the initially generated “linear” conformer
of 1ꢀ+ to more stable “folded” one. The stabilization
energy (ΔE for 1ꢀ+ in Table 7) gained by the conforma-
tional change shows no correlation with kobs. In other
words, the observed order of kobs is not explained by the
stability of the “folded” 1ꢀ+. An explanation may be given
by a steric factor on the k2 step, namely, the step of cou-
pling of 1ꢀ+ with O2. A “folded” 1ꢀ+ with a longer
methylene-chain spacer would provide a larger space
around the radical cationic phosphorus, making a chance
of the coupling with O2 higher. This may be the reason
why kobs becomes larger with increasing the length of the
spacer in 1. On the other hand, if O2 couples with 1ꢀ+ in its
“linear” conformer, difference in the geometry around the
radical cationic phosphorus would be too small, if any, to
explain the difference in kob.
The triphosphine, 4, which has the same number of
the methylene spacer to 1b, is oxidized in a slower rate
than 1b. The third phosphine unit in 4 exhibits no accel-
erative but inhibitory effect on the oxidation of the first
phosphine unit. That is, the third phosphine unit works
as a blockage to prevent O2 from attacking the radical
cationic center in 4ꢀ+. This finding suggests again the
[3] a) S. Yasui, K. Shioji, M. Tsujimoto, A. Ohno, J. Chem. Soc.,
Perkin Trans. 2 1999, 855. b) S. Yasui, K. Itoh, A. Ohno,
N. Tokitoh, Org. Biomol. Chem. 2006, 4, 2928.
[4] S. Yasui, K. Itoh, M. Tsujimoto, A. Ohno. Bull. Chem. Soc. Jpn.
2002, 75, 1311.
[5] a) S. Yasui, M. Tsujimoto, K. Itoh, A. Ohno, J. Org. Chem.
2000, 65, 4715. b) S. Yasui, M. Tsujimoto, J. Phys. Org. Chem.
2013, 26, 1090.
[6] S. Yasui, S. Kobayashi, M. Mishima, J. Phys. Org. Chem. 2016,
29, 443.
[7] S. Yasui, Y. Ogawa, K. Shioji, M. Mishima, S. Yamazaki, Bull.
Chem. Soc. Jpn. 2014, 87, 988.
[8] S. Yasui, S. Yamazaki, Chem. Lett. 2015, 44, 422.
[9] Gaussian 09, Revision A.02, M. J. Frisch, G. W. Trucks, H. B.
Schlegel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman,
G. Scalmani, V. Barone, B. Mennucci, G. A. Petersson,
H. Nakatsuji, M. Caricato, X. Li, H. P. Hratchian, A. F.
Izmaylov, J. Bloino, G. Zheng, J. L. Sonnenberg, M. Hada,
M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida,
T. Nakajima, Y. Honda, O. Kitao, H. Nakai, T. Vreven, J. A.
Montgomery, Jr., J. E. Peralta, F. Ogliaro, M. Bearpark, J. J.
Heyd, E. Brothers, K. N. Kudin, V. N. Staroverov,
R. Kobayashi, J. Normand, K. Raghavachari, A. Rendell, J. C.
Burant, S. S. Iyengar, J. Tomasi, M. Cossi, N. Rega, J. M. Mil-
lam, M. Klene, J. E. Knox, J. B. Cross, V. Bakken, C. Adamo,

YASUI AND YAMAZAKI
11 of 11
J. Jaramillo, R. Gomperts, R. E. Stratmann, O. Yazyev, A. J.
[23] There is no evidence to eliminate a possibility that 11*
Austin, R. Cammi, C. Pomelli, J. W. Ochterski, R. L. Martin,
K. Morokuma, V. G. Zakrzewski, G. A. Voth, P. Salvador, J. J.
Dannenberg, S. Dapprich, A. D. Daniels, O. Farkas, J. B. Fore-
sman, J. V. Ortiz, J. Cioslowski, and D. J. Fox, Gaussian, Inc.,
Wallingford CT, 2009.
undergoes the intersystem crossing to its triplet state before the
3
ET. Nevertheless, the oxidation potential of 1* would be only
slightly less negative (1 V at most) than that of 11*, and it should
be still more negative than the reduction potential of O2.
[24] S. Yasui, S. Tojo, T. Majima, J. Org. Chem. 2005, 70, 1276.
[25] S. Yasui, J. Contemporary Human Life Science, Tezukayama
University 2015, 11, 23.
[10] R. L. Brainard, G. M. Whitesides, Organometallics 1985, 4, 1550.
[11] S. J. Berners-Price, R. E. Norman, P. J. Sadler, J. Inorg. Bio-
chem. 1987, 31, 197.
[26] According to steady-state approximations with respect to [1*]
and [1ꢀ+],
[12] V. V. Grushin, Organometallics 2001, 20, 3950.
[13] M. J. Petersson, W. A. Loughlin, I. D. Jenkins, Chem.
Commun. 2008, 4493.
[14] A value of Int (smp)/Int (std) for 1 at t = 0 was normalized to
unity. The factor used in this normalization was applied to the
Int (smp)/Int (std) values obtained for all other signals to
determine the relative values.
[15] As seen in Table 3 and Tables S2-1-4, Int (smp)/Int (std) values
for 2(P(III)) and 2(P(V)) are sometimes inconsistent with each
other, which may result from experimental errors in determin-
ing the integral areas of very small signals.
d[1*]/dt = ϕ0Ia − (k0 + k1)[1*] = 0,
and
(7)
(8)
d[1ꢀ+]/dt = k1[1*] − (kb + k2)[1ꢀ+] = 0,
respectively. Then,
[1*]= ϕ0Ia/(k0 + k1).
(9)
[16] Because of poor solubility of 1 in CD3CN, the measurement
was performed at [1d] = 2.70 × 10−3M. The signal from the
product, 3d, appeared at slightly upper field than that observed
in the photolysis in CDCl3 at [1d] = 2.00 × 10−2M (34 ppm).
This is because of difference in solvent and/or concentration.
[17] The ordinate represents logarithm of the normalized Int
(smp)/Int (std) value.
[18] It was impossible to determine a precise value of the rate con-
stant for disappearance of 2, because of inevitable side-
reactions during the long period irradiation.
Accordingly,
[1ꢀ+] = ϕ0Iak1/{(k0 + k1)(kb +k2)}.
(10)
Substituting Equations (9) and (10) into Equation (2),
Equation (3) is obtained.
[27] Based on our previous chemical actinometric study,[28] the
incident light received by the surface of our reaction vessel
from our xenon lamp is roughly 10−6 einstein s−1. Using this
value, we estimated the quantum yield of the photooxidation
of 1 to be in the order of 10−4
[28] S. Yasui, M. Tsujimoto, K. Shioji, A. Ohno, Chem. Ber./Recueil
1997, 130, 1699.
[29] A smaller value of λ for 1a at 310 nm than the others may
explain the smaller value of kobs for 1a.
[19] The absorption of 1 at 310 nm is a tail of the absorption with
λmax around 250 nm. The extinction coefficients ε of 1b, 1d,
1e, and 4 at λmax are 17 200M−1 cm−1 (λmax; 252 nm), 14 000
(252), 7600 (251), and 22 400 (252) in CH3CN, respectively.
.
1 exhibits also an intense absorption (ε >40 000M−1 cm−1
around 210 nm certainly because of the π-π* transition.
)
[20] The potential of 1 was measured by cyclic voltammetry
(CV) in MeCN under an argon atmosphere at [1] = 2.00 ×
10−3M using tetraethylammonium tetrafluoroborate (0.100M)
as a supporting electrolyte. Unfortunately, the irreversible
character of 1 in CV allowed us to only determine the peak
oxidation potentials Epox. Based on the theory that E1/2 is
[30] Because of lack of the flexibility, the “folded” conformers of 1a
and 1aꢀ+ are very similar to each other in the geometry, giving
the almost same value of d(P–P).
[31] D. G. Gilheany, in The Chemistry of Organophosphorus
Compounds ed. by F. R. Hartley, Wiley, New York, 1990,
Vol. 1, Chap. 2, p.15.
ox
0.03 V lower than Ep when one electron transfers in
a
reversible way at room temperature, we converted
ox
SUPPORTING INFORMATION
Additional supporting information may be found online
in the Supporting Information section at the end of this
article.
the measured Ep values to the half-wave potentials E1/2. E1/2
of 1a, 1b, 1c, 1d, 1e, and 4 are 1.40, 1.27, 1.35, 1.05, 0.92, and
1.32 V vs Ag/Ag+, respectively. Uncertainties in these values,
if any, are too small to cause serious problems in our
discussion.
[21] Excitation at 310 nm results in 4.00 V negative shift of an oxi-
dation potential. Since the oxidation potentials of 1 are
How to cite this article: Yasui S, Yamazaki S. 31P
NMR spectroscopic analysis on photooxidation of
1,n-bis(diphenylphosphino)alkanes with the aid of
DFT calculations. J Phys Org Chem. 2020;e4021.
https://doi.org/10.1002/poc.4021
1
0.92-1.40 V vs Ag/Ag+, the oxidation potentials of 1* are esti-
mated to be −3.08 to −2.60 V. These values are 2.03-1.55 V
more negative than the reduction potential of O2 (−1.05 V).
[22] a) D. S. Steichen, C. S. Foote, J. Am. Chem. Soc. 1981, 103,
1855. b) K. Mizuno, T. Tamai, I. Hashida, Y. Otsuji, J. Org.
Chem. 1995, 60, 2935. c) N. Somasundaram, C. Srinivasan,
J. Org. Chem. 1996, 61, 2895.