EPR and ENDOR Studies of 1H and 14N Hyperfine and Quadrupolar Couplings in Crystals of L-O-Serine Phosphate after X-Irradiation at 295 K

Article abstract of DOI:10.1021/jp960063x

Single crystals of L-O-serine phosphate, (HOOC)CH(NH3(1+))CH2OPO3(1-)H, were X-irradiated at 295 K and studied using EPR, ENDOR, and FSE techniques.At this temperature, three carbon-centered radicals were identified.Radical I, the deamination product (HOOC)CH(1.)CH2OPO3(1-)H, is shown to have undergone a major molecular reorientation upon formation.Radical II is identified to be the product (HOOC)CH(NH3(1+))CH(1.)OPO3(1-)H.This species exhibits a nonplanar site for the lone electron density.This deviation from planarity is ascribed to electrostatic interaction between the lone electron orbital and lone pairs centered on the neighboring oxygen atom.Hyperfine interaction with the β-nitrogen of the amino group is observed.Both the 14N hyperfine and quadrupole coupling tensors are determined and the signs of the hyperfine coupling tensor principal values are deduced from interpretation of the quadrupole tensor employing the Townes-Dailey approach.The β14N isotropic interaction is well described by a Heller and McConnell type cos²θ rule.The results found in the present work, together with other recent observations, yield estimated values of the constants B₀ = 1.0 MHz and B₂ = 34.5 MHz.Radical III exhibits the structure (HOOC)CH(NH3(1+))CH2(1.), formed by scission of the phosphate-ester bond at the carbon side.No β14N coupling is observed for this radical due to a dihedral angle close to 90 deg.Possible mechanistic routes for the formation of these radicals are discussed in comparison with previously published data on serine and other alkyl phosphate derivatives.

Full text of DOI:10.1021/jp960063x

J. Phys. Chem. 1996, 100, 9545-9555
9545
1
14
EPR and ENDOR Studies of H and N Hyperfine and Quadrupolar Couplings in Crystals
of L-O-Serine Phosphate after X-Irradiation at 295 K
Audun Sanderud* and Einar Sagstuen
Department of Physics, UniVersity of Oslo, P.O. Box 1048 Blindern, N-0316 Oslo, Norway
ReceiVed: January 2, 1996; In Final Form: March 7, 1996X
+
-
Single crystals of L-O-serine phosphate, (HOOC)CH(NH3 )CH2OPO3 H, were X-irradiated at 295 K and
studied using EPR, ENDOR, and FSE techniques. At this temperature, three carbon-centered radicals were
-
identified. Radical I, the deamination product (HOOC)C˙ HCH2OPO3 H, is shown to have undergone a major
molecular reorientation upon formation. Radical II is identified to be the product (HOOC)CH(NH3+)C˙ HOPO3-H.
This species exhibits a nonplanar site for the lone electron density. This deviation from planarity is ascribed
to electrostatic interaction between the lone electron orbital and lone pairs centered on the neighboring oxygen
atom. Hyperfine interaction with the â-nitrogen of the amino group is observed. Both the 14N hyperfine and
quadrupole coupling tensors are determined and the signs of the hyperfine coupling tensor principal values
are deduced from interpretation of the quadrupole tensor employing the Townes-Dailey approach. The â14N
isotropic interaction is well described by a Heller and McConnell type cos2 θ rule. The results found in the
present work, together with other recent observations, yield estimated values of the constants B0 ) 1.0 MHz
+
and B2 ) 34.5 MHz. Radical III exhibits the structure (HOOC)CH(NH3 )C˙ H2, formed by scission of the
phosphate-ester bond at the carbon side. No â14N coupling is observed for this radical due to a dihedral
angle close to 90°. Possible mechanistic routes for the formation of these radicals are discussed in comparison
with previously published data on serine and other alkyl phosphate derivatives.
1. Introduction
Sørnes et al.8 have investigated room-temperature radical
formation by X-irradiation of aminoethyl hydrogen sulfate
The mechanisms leading to radiation-induced breaks of the
phosphodiester bond in DNA has been subject to numerous
investigations, in part by EPR and ENDOR spectroscopy.1
Radiation damage by direct effects to DNA has been shown
mainly to produce primary ionization of the aromatic bases, but
phosphate radicals have not been observed in irradiated DNA.2,3
There is, however, experimental evidence that the primary ionic
sites on DNA are precursors to strand breaks.2,4 The mecha-
nisms by which base damage is transferred to the sugar-
phosphate region of DNA resulting in breakage of the sugar-
phosphate bonds are still largely unknown. It is believed that
radical transfer to the C4’ position of the deoxyribose moiety is
involved, but few details about this process are known.1
Several model systems have been investigated to elucidate
the influence of various substituents to alkyl phosphate deriva-
tives on the release of inorganic phosphate or phosphate-centered
radicals.5-8 Bungum et al.5 studied X-irradiated disodium and
dipotassium salts of glucose-1-phosphate, finding evidence for
secondary loss of phosphoryl radicals, probably Via a carbon-
centered radical precursor. The same products were found6 in
the barium salt of glucose-6-phosphate and the disodium salt
of â-glycerol phosphate. However, in X-irradiated disodium
and barium salts of ribose-5-phosphate, the disodium salt of
R-glycerol phosphate, and the monosodium salt of glucose-6-
phosphate, phosphoryl radicals were not detected.
(AES), where the phosphate group of PEA has been replaced
by the slightly more electronegative sulfate group. The studies
reveal the presence of a sulfite radical, while cleavage of the
(NH3+)CH2CH2-OSO3 oxygen-ethyl bond is not observed.
-
This difference between AES and PEA is mainly ascribed to
the increased electron affinity of sulfur as compared to
phosphorus.
L-O-serine phosphate (SP), (HOOC)CH(NH3+)CH2OPO3-H,
consists of the PEA system with a carboxyl group added.
From previous studies of amino acids, the carboxyl group is
known to be an electron sink just as efficient as the aromatic
bases in DNA. Our recent experiments have shown that, as in
DNA, no phosphate-centered radicals are observed in SP.6 It
was of interest to see if net dephosphorylation still occurs,
and in fact, the present work shows the cleavage of the
(HOOC)CH(NH3+)CH2-OPO3-H phosphate ester bond takes
place. Thus, it seem that the presence of an electron scavenger
in the system prevents the formation of phosphate-centered
radicals, but not cleavage of the phosphate ester bond at the
carbon side.
In the present paper three carbon-centered radicals formed
by X-irradiation in SP at room temperature are identified. The
corresponding radicals are observed in serine,9,10 and it is
suggested that similar radical reactions lead to the observed
room-temperature radicals in the two compounds.
In X-irradiated O-phosphorylethanolamine (PEA), (NH3+)-
CH2CH2OPO3-H, Fouse et al.7 have reported the presence of
secondary phosphoryl radicals, resulting from cleavage of the
(NH3+)CH2CH2O-PO3-H phosphate ester bond. Dissociative
electron capture at the bond is the assumed mechanism. In the
same system Fouse et al.7 reports possible cleavage of the
(NH3+)CH2CH2-OPO3-H bond, resulting in a carbon-centered
radical. The concentration of the two radicals is equal.
2. Experimental Section
Single crystals of L-O-serine phosphate (SP) were grown from
warm (40 °C) saturated aqueous solutions by slow cooling.
Partially deuterated samples were similarly prepared by repeated
recrystallization from 99.8 g/100 g purity deuterium oxide
solutions. The crystal structure of SP has been analyzed by
Sundaralingam and Putkey.11 The unit cell is orthorhombic with
space group P212121 and Z ) 4.
* Author to whom all correspondence should be addressed.
X Abstract published in AdVance ACS Abstracts, May 1, 1996.
S0022-3654(96)00063-9 CCC: $12.00 © 1996 American Chemical Society

9546 J. Phys. Chem., Vol. 100, No. 22, 1996
Sanderud and Sagstuen
The crystals were irradiated at 280 K using X-rays from a
tungsten target tube operated at 60 kV and 50 mA, such that
each sample received a total dose of 50 kGy at about 30 kGy/
h. The crystals were aligned along each crystallographic axis
using a Weissenberg X-ray diffraction camera and subsequently
transferred to a quartz sample holder without loss of axis
alignment. First-derivative room-temperature data were col-
lected at X-band microwave frequencies with a Bruker ER-200
D-SRC spectrometer with and EN-200 ENDOR unit equipped
with a 100 W ENI rf-amplifier, in 5° intervals through more
than 90° in each plane. At each orientation the ENDOR
frequency was swept from 1 to 75 MHz in 10 or 50 MHz
intervals.
Figure 1. First derivative EPR spectra (width 20.0 mT, centered at ge
) 2.0023) of L-O-serine phosphate single crystals, X-irradiated at 280
K and then measured at room temperature. The spectra in part a are
recorded using crystals grown in water, in part b the spectra are recorded
using partially deuterated crystals, and in part c they are recorded using
crystals grown in water stored for 4 months after X-irradiation. The
spectra are measured with the external magnetic field along the indicated
crystal axes.
The proton hyperfine coupling (hfc) tensors were determined
using the program MAGRES,12 while the 14N hyperfine and
quadrupolar coupling (nqc) tensors were found using the
program NQENDFIT.8 In both cases isotropic g-tensors were
assumed (see below).
3. Results
3.1. EPR Spectroscopy. Figure 1 shows the EPR spectra
obtained with the magnetic field directed along each of the
crystallographic axes. The spectra in Figure 1a are from crystals
grown in aqueous solution, while spectra in Figure 1b are from
partially deuterated samples. There are only small differences
between the spectra from the two kinds of crystals, indicating
that the major hyperfine couplings are due to interaction with
carbon-bonded protons or the nitrogen atom. When the crystals
were stored at room temperature for 3-4 months, the resonance
shown in Figure 1a disappeared and the spectra shown in Figure
1c were observed. The ENDOR experiments show that the
resonance is present initially, but hidden under the far more
intense resonance depicted in Figure 1a.
Figure 2. First derivative ENDOR spectra of L-O-serine phosphate
single crystals freshly X-irradiated at 280 K and measured at room
temperature. The spectra are recorded with the external magnetic field
oriented along the crystallographic b-axis. Spectrum a is from a crystal
grown in water, while b is obtained from a partially deuterated crystal.
The high-branch (+) and low-branch (-) ENDOR transitions from
which hfc or nqc tensors were calculated are indicated by numericals.
Line #1+ is not visible in part b because the spectrum has a slightly
shorter range than the spectrum in part a.
Since the EPR spectra contain contributions from several
radicals, they were difficult to analyze. The ENDOR analyses,
however, gave good results, and radical structures were deter-
mined without further use of the EPR results other than
establishing that the resonances exhibit virtually isotropic
g-tensors.
3.2. ENDOR Spectroscopy. The ENDOR spectra revealed
eight hyperfine couplings which could be followed through all
three rotation planes. The angular variations of these couplings
are shown in Figures 4, 5, and 8. From these data seven proton
hfc tensors and one 14N hfc tensor with an nqc tensor were
determined. FSE (field-swept ENDOR) experiments showed
that these couplings could be ascribed to three different
radicals: one major radical (radical I) and two less abundant
species (radicals II and III).
Figure 2 shows the ENDOR spectra obtained with the
magnetic field along the crystallographic b-axis both for
protiated (Figure 2a) and for partially deuterated crystals (Figure
2b). The spectrum in Figure 2a contains resonance lines from
couplings for which the hfc tensors could not be determined
because of difficulties following the resonance lines through
all three rotation planes. In the spectrum from the deuterated
crystal (Figure 2b) some of the lines present in Figure 2a are
missing. These resonance lines are thus due to couplings with
easily exchangeable protons. For one of these exchangeable
couplings (designated #6) the hfc tensor was determined
(Table 2).
Hyperfine coupling tensors based on data from rotations
around three orthogonal axis leave an ambiguity in the signs of
the off-diagonal tensor elements. Schonland13 pointed out that
this could be resolved by investigating a plane of data using a
skewed axis of rotation. This has been done for most of the
tensors obtained in this work. Figure 3 shows typical results
from a skewed axis rotation plane (θ ) 45°, φ ) 90°) displaying
results for hfc tensor #4. Here, the fully drawn and stipled
curves are calculated from the two different hfc tensors given
by the Schonland ambiguity. It is clear that only one of the
two tensor alternatives reproduces the experimental result. This
method was used for tensors #1, #3, #4, #8, and #9 in this work.
For the other tensors, the differences between the two tensor
alternatives were too small and indirect arguments were used
to settle on those chosen in Table 1 (see below).
3.3. Radical I. FSE was used to assign the ENDOR lines
in Figure 2 to three different radicals. Radical I is the major
species. As described above, the EPR due to this radical slowly
disappeared upon storage at room temperature. Three proton
ENDOR resonance lines (#1, #2, and #3) were assigned to this
radical, as demonstrated in Figure 4. These ENDOR lines were

Crystals of L-O-Serine Phosphate
J. Phys. Chem., Vol. 100, No. 22, 1996 9547
Figure 3. Calculated and observed ENDOR frequency variation
through the skew axis rotation plane for hfc tensor #4 from L-O-serine
phosphate single crystals freshly X-irradiated at 280 K. This plot allows
the determination of the signs of the off-diagonal elements of the hfc
tensor. The axis of rotation is given by the polar angles θ ) 45° and
æ ) 90°. The fully drawn and the dotted lines represent the calculated
variations for the two different sets of sign alternatives obtained from
the tensor given in Table 2. The filled and open circles represent
experimental data for the two different molecular sites.
TABLE 1: Hyperfine Coupling Tensors for Radical I in
Single Crystals of L-O-Serine Phosphate X-Irradiated at 280
Ka
principal
values
(MHz)
isotropic
value
(MHz)
Figure 4. Angular variations of the high-branch (+) ENDOR
transitions from the proton couplings #1, #2, and #3 of radical I in
L-O-serine phosphate single crystals freshly X-irradiated at 280 K. The
FSE spectra (width 11.8 mT, centered about ge ) 2.0023) connecting
these couplings to one radical are shown. The solid curves represent
the angular variations calculated from the coupling tensors listed in
Table 1.
direction cosines
tensor
a
b
c
RH
(#1)
-86.5
-50.9
-25.2
-54.2
104.3
57.4
0.559
0.829
0.012
-0.819
0.549
0.164
-0.130
0.102
-0.986
âH
(#2)
111.5
102.3
99.0
0.063
0.786
0.616
-0.936
0.261
-0.237
0.349
0.561
-0.752
âH
(#3)
65.5
53.5
53.2
0.773
0.633
0.049
-0.617
0.767
-0.150
0.107
-0.177
0.983
a The errors involved in the principal values and eigenvectors are
about (0.2 MHz and (3°, respectively.
all characterized by fading upon storage at room temperature.
The hfc tensors for these three couplings are given in Table 1.
The three proton hfc tensors were used to simulate the EPR
absorption spectra measured along the crystallographic a- and
b-axes. These simulated spectra were compared with the
experimental FSE spectra at these orientations, and the three
proton couplings fully reproduce the splitting pattern in the FSE
spectra.
3.4. Radical II. After the disappearance of radical I, the
dominant radical responsible for the spectra in Figure 1c is
radical II. The FSE spectra were used to assign two proton
hfc’s (#4, #6) and the 14N coupling (#5) to this radical, as shown
in Figure 5. Coupling #6 is to a proton which is exchanged
with deuterium in the deuterated crystals. The proton and
nitrogen hfc tensors are given in Table 2.
The FSE spectrum from the high-frequency ENDOR line of
coupling #4 at the c-axis orientation clearly shows seven lines.
A stick diagram was constructed from the hfc tensors of radical
II, Table 2. To account for this FSE spectrum, it was necessary
to include one extra proton coupling. This additional coupling
was calculated from the FSE spectra to be 0.62 mT at this
orientation of the crystal. In the same way it was necessary to
include an extra coupling to explain the FSE along the other
two axes, a and b. At these orientations, the extra coupling
was calculated to be 1.39 and 0.63 mT, respectively. For the
b-axis orientation this corresponds to ENDOR lines at 5.8 and
23 MHz. In Figure 2 there clearly are ENDOR lines present
which fit these expectations. It was not possible to obtain a
full hfc tensor for this coupling, but from a few observations
Figure 5. Angular variations of the ENDOR transitions from proton
couplings #4 and #6 and the 14N coupling #5 of radical II in L-O-
serine phosphate single crystals X-irradiated at 280 K. The FSE spec-
tra (width 11.8 mT, centered about ge ) 2.0023) connecting these
couplings to one radical are shown. The solid curves represent the
angular variations calculated from the coupling tensors listed in
Table 2.
an estimate for the coupling was obtained. This estimated tensor
is included in Table 2 (#7). The coupling is nonexchangeable
and assumed to be of â-type.

9548 J. Phys. Chem., Vol. 100, No. 22, 1996
Sanderud and Sagstuen
TABLE 2: Hyperfine Coupling Tensors and the
Quadrupolar Coupling Tensor of the 14N Coupling for
Radical II in Single Crystals of L-O-Serine Phosphate
X-Irradiated at 280 Ka
principal
values
isotropic
value
direction cosines
tensor
(MHz)
(MHz)
a
b
c
RH
(#4)
-85.5
-47.2
-20.3
-51.0
0.799
0.107
0.598
-0.177
0.581
0.153
-0.799
0.982
0.059
âN(#5)
hfc
17.95
15.62
14.07
15.89
0.613
0.738
0.282
0.411
0.007
-0.911
0.675
-0.675
0.299
âN(#5)
nqc
0.904
-0.281
-0.623
0.529
0.078
0.845
0.378
-0.913
-0.152
0.760
0.400
-0.512
γH
16.0
1.8
0.1
5.9
25
0.773
0.633
0.049
-0.617
0.767
-0.177
-0.150
0.107
0.983
(#6)
âH
(#7)
39
18
17
∼a-axis
∼b-axis
∼c-axis
Figure 6. EPR/ENDOR test. To the left the first derivative EPR
spectrum from L-O-serine phosphate single crystals X-irradiated at 280
K is shown. The magnetic field is directed along the crystallographic
c-axis. The field positions 1, 2, and 3 where the magnetic field has
been locked during the recording of the ENDOR spectra 1, 2, and 3 to
the right are indicated. In spectrum 2, recorded off the central position
of the EPR spectrum, all four ENDOR lines are present, while in
spectrum 1 only the two inner ENDOR transitions and in spectrum 3
only the two outer ENDOR transitions are present. This experiment
shows that the signs of the hyperfine and quadrupolar coupling are the
same at this orientation of the crystal. The ENDOR line at 7.18 MHz
is due to an unidentified coupling.
a The errors involved in the principal values and eigenvectors are
about (0.2 MHz and (3°, respectively, for the proton tensors. For
the 14N coupling tensors the corresponding errors are (0.01 MHz and
(1°, respectively.
3.5. The ENDOR/FSE Test. The FSE spectra obtained by
monitoring the ENDOR lines due to the 14N coupling show only
six lines at the c-axis orientation, while the FSE spectra obtained
by monitoring the ENDOR lines due to the proton couplings
of the same radical show seven lines. This is due to the fact
that 14N has nuclear spin I ) 1. Some of the EPR transitions
are therefore not present in the FSE spectra, dependent on which
ENDOR transitions are monitored and upon the relative signs
of the quadrupole and the hyperfine coupling of 14N . For the
same reason, some ENDOR transitions are not present depend-
ing on which EPR transitions of the 14N manifold are monitored.
This was discussed in detail in a previous paper by Sørnes et
al.8
Figure 6 shows ENDOR spectra obtained by monitoring
different EPR lines with different 14N spin quantum numbers
(positions 1 and 3 correspond to mI ) (1, and position 2
corresponds to mI ) 0, (1), and Figure 7 shows FSE spectra
obtained by monitoring different ENDOR transitions. Both
these experiments lead to the conclusion that the quadrupolar
and hyperfine couplings must have different signs at this
orientation. The analysis of the nqc tensor given below suggests
that the nitrogen hfc is positive. Together, this yields the hfc
and nqc tensors presented in Table 2.
3.6. Radical III. The two last proton hfc’s for which
coupling tensors were determined (#8 and #9) exhibit ENDOR
lines which easily were followed through the three planes of
rotation. The FSE spectra were characteristic, as shown in
Figure 8, and unambiguously connect the two ENDOR lines to
the resonance of one radical. The hfc tensors obtained, assigned
to radical III, are given in Table 3.
The FSE spectra recorded along, for example, the b-axis
exhibit five lines. To accomplish this, at least three proton hfc’s
are needed. Consequently, one further coupling must be
associated with radical III, which has not been possible to obtain
from the ENDOR data. The FSE spectrum along the b-axis
was reconstructed by adding either one additional proton
coupling of 2.55 mT or a 14N coupling of 1.28 mT. Both these
alternatives yield a satisfactory reconstruction of the FSE spectra.
The proton coupling should give rise to ENDOR lines at 50.3
and 21.1 MHz, and the 14N coupling should (if quadrupolar
Figure 7. ENDOR/FSE test. In the ENDOR spectrum to the left the
positions 1, 2, 3, and 4 indicate the monitoring frequencies used while
recording the FSE spectra 1, 2, 3, and 4 to the right. For reference,
spectrum 5 is the entire FSE spectrum recorded off the high-frequency
ENDOR line of coupling #4. In the FSE spectra 2 and 3, recorded off
the inner 14N ENDOR transitions, the high-field line of the FSE
spectrum is missing, while in the FSE spectra 1 and 4 the low-field
line is missing. This shows, similarly to the results in Figure 6, that
the signs of the hyperfine and quadrupolar coupling are the same at
this orientation. The different lines in the stick diagram represent
different mI quantum numbers, and for a positive nitrogen hyperfine
coupling the solid lines represent mI ) +1; the dotted lines, mI ) 0;
and the stippled lines, mI ) -1.
splitting is neglected) give ENDOR lines at 19.0 and 21.2 MHz.
In Figure 2 there is an ENDOR line at ∼21 MHz, but it has not
been possible to follow this line through the three rotation
planes so that a tensor could be calculated. However, it is

Crystals of L-O-Serine Phosphate
J. Phys. Chem., Vol. 100, No. 22, 1996 9549
TABLE 4: Crystallographic Directions As Determined by
Sundaralingam and Putkey10 between C2 and H2, H3, H4,
and N in Single Crystals of L-O-Serine Phosphatea
experimental
difference
direction crystallographic
#3
#2
#3
#2
C2-H2
C2-H3
C2-H4
C2-N
0.800
-0.590
0.111
0.773
-0.617
0.150
0.043
-0.938
0.345
2.68° 51.2°
0.897
0.241
0.370
0.7773
0.617
0.150
0.043 26.1°
0.938
0.345
66.9°
-0.442
-0.276
0.853
-0.012
26.9°
44.5°
-0.164
0.986
0.418
0.530
0.738
0.829
0.549
0.102
a These directions are compared with corresponding experimental
eigenvectors for the hfc tensors of radical I. In each case, the signs of
the direction cosine elements of the eigenvectors are chosen so as to
minimize the angular deviation.
Upon deamination of SP, the C2 atom is assumed to
rehybridize to a planar sp2 configuration, mostly by reorientation
of H4. An indication of the degree of rehybridization may be
obtained by comparison of two methods for finding the 2pπ
spin density at C2 from the R-coupling tensor #1. One method
consists of using the isotropic value, aiso ) -54.2 MHz, in the
McConnell relation14 with QCHH ) -73.4 MHz15 for a planar
sp2 configuration.16 This gave Fiπso ) 0.738. The other
method consists of using the dipolar part of the R-coupling
Figure 8. Angular variations of the high-branch (+) ENDOR
transitions from the proton couplings #8 and #9 of radical III in L-O-
serine phosphate single crystals X-irradiated at 280 K. The FSE spectra
(width 11.8 mT, centered about ge ) 2.0023) connecting these couplings
to one radical are shown. The solid curves represent the angular
variations calculated from the coupling tensors listed in Table 3.
TABLE 3: Hyperfine Coupling Tensors for Radical III in
Single Crystals of L-O-Serine Phosphate X-Irradiated at
280 Ka
tensor #1 in the Gordy-Bernhard method,17 giving Fdπip
)
principal
values
isotropic
value
0.749 with Qdzip ) 38.7 MHz. The similarity of these two
values confirms sp2 configuration at C2.
direction cosines
tensor
(MHz)
(MHz)
a
b
c
The dihedral angles θ2 and θ3 of the two â-protons (tensors
#2 and #3) with respect to the LEO are found by the Heller-
McConnell relation18
RH
(#8)
-69.1
-48.6
-23.5
-47.1
0.829
0.254
0.499
0.114
0.795
-0.595
0.548
-0.550
-0.630
RH
(#9)
-75.4
-50.3
-26.6
-50.8
0.812
0.214
0.543
-0.475
0.782
0.404
-0.338
-0.586
0.736
aiso ) Fπ(B0 + B2 cos2 θ)
(1)
a The errors involved in the principal values and eigenvectors are
about (0.2 MHz and (3°, respectively.
For aliphatic systems the constants B0 and B2 in this relation
are given by a B0 between (14 MHz and B2 ≈ 126 MHz.19
The isotropic value of coupling #2 is very large, and to obtain
θ2 ) 0° with a spin density as found above, B0 must be set to
13.3 MHz. This in turn gives the dihedral angle for coupling
#3, θ3 ) 135.2°. The C3 atom is expected to maintain its near
sp3 hybridization configuration, and the difference between the
two dihedral angles θ2 and θ3 should therefore be 120°. Using
this as a constraint, the 2pπ spin density at C2 according to the
Heller-McConnell relation should be as large as 0.782, and
the corresponding dihedral angles become θ2 ) 13.3° and θ3
) 133.3°. This probably implies that B2 is somewhat under-
estimated, but nevertheless gives a useful estimate of the dihedral
angles.
The crystallographic unit cell space group P212121 with Z )
4 implies that each of the four possible unique combinations of
signs of each direction cosine element of an eigenvector {(l,m,n),
(l,m,-n), (l,-m,n), and (l,-m,-n)} corresponds to one of the four
molecular sites in the crystal. For a structural discussion, it is
important to be certain that the set of eigenvectors used for all
hfc tensors of a given radical are referred to the same molecular
site. In the absence of more detailed experimental observations,
the sets of eigenvectors to be chosen are obtained by comparison
with specific crystallographic directions. The guidelines used
are that the eigenvector for the R-proton minimum principal
value corresponds to the C-HR bond direction, the eigenvector
clear that one extra coupling in addition to the two found by
ENDOR is required. This will be discussed further below
(section 4.4).
4. Discussion of Radical Structure
4.1. Radical I. Three hfc tensors, designated #1, #2, and
#3 in Table 1, have been shown to be associated with radical I.
Tensor #1 exhibits principal values with typical R-character,
while tensors #2 and #3 are typical â-type tensors. All couplings
are clearly due to protons which are not exchanged with
deuterons in the partially deuterated crystal. In SP the only
nonexchangeable protons are H2, H3, and H4. Therefore, the
only possible location for the main density of the lone electron
orbital, LEO, is C2. Furthermore since no R-nitrogen hfc is
observed, it is assumed that radical I is formed by a net
deamination process and exhibits the structure

9550 J. Phys. Chem., Vol. 100, No. 22, 1996
Sanderud and Sagstuen
TABLE 5: Mutual Angles between the Directions from C2
to H2, H3, and C3 for Single Crystals of L-O-Serine
Phosphate, As Determined from the Crystallographic Work
of Sundaralingam and Putkey10 and from the Experimental
Eigenvectors for the hfc Tensors of Radical I
H4-C2-C3 ) 120°
angle
crystallographic
experimental
difference
H2-C2-H3
H2-C2-C3
H3-C2-C3
51.9°
27.7°
30.9°
48.5°
23.9° #2
25.4° #3
3.4°
3.8°
5.5°
Figure 9. Structure of the O6-C1-C2-N fragment in L-O-serine
phosphate single crystals. The left side illustrates that the O6, C1,
C2, and N atoms are situated in the same plane and highlights the
2pz orbitals of O6 and C1. Upon deamination forming radical I, the
right side shows how the LEO will tend to get parallel with the
two 2pz orbitals so as to increase the conjugation between C2 and
C1-O6.
The new C2-C3 bond direction found from the hfc tensors
of radical I makes it possible to calculate the expected dihedral
angles of the â-protons H2 and H3. This calculation yields θ2
) 6.70° and θ3 ) 141.6°. The agreement with the values
estimated on the basis of the hfc isotropic values above is
satisfactory.
4.2. Radical II. From the FSE experiments, three of the
ENDOR determined hfc tensors have been ascribed to radical
II. Tensor #4 represents a coupling to a nonexchangeable
proton and is clearly of R-type. Combined with the observation
that the g-tensor is virtually isotropic, this requires that the LEO
is located at one of the carbon atoms C2 or C3.
The exchangeable proton responsible for the hfc coupling
#6 must be one of the amino protons, but the coupling tensor
does not, at first look, give an answer if this is a â- or γ-type
hfc. Also, the nitrogen hfc tensor #5 does not directly indi-
cate the localization of the LEO. However, knowledge of the
relative signs of the 14N hyperfine and quadrupolar tensors,
combined with an estimate of the absolute sign of the quadru-
polar tensor, yields the sign of the 14N hfc tensor, which in turn
tells if it is an R- or â-type tensor. This will be done below
(section 4.3).
for the intermediate principal value for the R-proton corresponds
to the direction of the LEO, and the eigenvector for the
maximum principal value for a â-proton nearly corresponds to
the C‚‚‚Hâ direction. In Table 4 the crystallographic direction
for the C2-H2 bond and the crystallographic C2-H3, C2-
H4, and the C2-N directions are compared with experimental
eigenvectors. In each case the sign of each direction cosine is
chosen so as to minimize the calculated angle of deviation. The
crystallographic directions (except C2-H4) are expected to be
conserved in the radical if reorientation of radical I is restricted
to rehybridization at C2.
Except for the case of tensor #3 and the C2-H2 direction,
all other crystallographic directions fit poorly with the experi-
mental eigenvectors. A more extensive molecular reorientation
must therefore be assumed to occur upon radical formation to
explain the hfc tensors. A larger reorientation has also been
observed for the corresponding radical in single crystals of
serine.10
Regardless of the localization of the LEO, the spin density
at the central atom can be calculated from tensor #4. Using
the McConnell relation14 with QCHH ) -73.4 MHz,15 the
isotropic value aiso ) -51.0 MHz gives Fπiso ) 0.695. The
McConnell relation is valid only if the C atom has rehybridized
to a planar sp2 configuration.16 The Gordy-Bernhard method17
gives Fπdip ) 0.793 from the dipolar tensor, which is 14% larger
than Fπiso. This indicates that the configuration of the central
atom deviates from pure sp2 hybridization and that Fπdip is the
better estimate for the LEO spin density.
Firstly, assume that the LEO is located at C2. In addition to
the hfc tensors #4, #5, and #6, the FSE spectra show that at
least one more coupling, almost undetectable in ENDOR, must
be present. If it is just one significant coupling, this interaction
must have tensor elements close to those of tensor #7 in Table
2, which is a â-type coupling to a nonexchangeable proton.
Tensor #7 has aiso ) 25 MHz, a value, however, with
considerable uncertainty. With a spin density of Fπ ) 0.793,
the Heller-McConnell relation18 gives a dihedral angle of θ )
68.1° with B0 ) +14.0 MHz and B2 ) 126 MHz.19 (For the
present discussion, nonplanarity of the radical center is not
taken into account for the discussion of proton â-coupling
interactions.) If the LEO is localized on C2, the #7 coupling
must be a coupling with either H3 or H2. The difference
between the dihedral angles of these two protons is 120°. From
the direction of the LEO, assumed to be parallel to the
eigenvector for the intermediate principal value of tensor #4, it
is found that the dihedral angles of H2 and H3 (using the
crystallographic bond directions) are 1.9° and 129.0° (51.0°).
A â-coupling with dihedral angle close to 0° would give a
large coupling which easily should be observed in the FSE
spectra. The failure to do so argues against the LEO being
localized on C2.
This reorientation can be explained from the hydrogen-
bonding pattern in the SP crystal. The crystal structure shows
five unique hydrogen bonds11 (hb’s), so that every SP molecule
is involved with 10 hb’s. Three of these are associated with
the amino group, five with the phosphate group, and two with
the carboxyl group. The amino group is therefore important
as an anchor, holding the molecule fixed in the crystalline lattice.
When this group is lost, as in radical I, a displacement of the
C2 atom becomes possible.
A close examination of the structure of SP reveals that a
reorientation is not only possible but also highly likely. The
atoms of the fragment N-C2-C1-O6 are situated almost in
the same plane (the torsion angle between planes N-C2-C1
and C2-C1-O is only 2.6°), and one of the bonds between
O6 and C1 is a 2pπ bond, as shown in Figure 9a). When the
amino group is lost, the resulting LEO at C2 will tend to become
parallel with the two 2pπ orbitals and increase the conjugation
with C1-O6, as illustrated in Figure 9b.
If this reorientation occurs with the remaining hb’s intact, it
must take place by torsion about the P-O1 and the O1-C3
bonds and rotation about the O5-C1 bond. The sp3 configu-
ration of C3 will by this be preserved and thereby also the angles
between C2 and the protons H2 and H3. This constraint is thus
used to determine the choice of signs of eigenvector components
for the three hfc tensors involved (Table 1). In Table 5 the
resulting angles are shown, together with those calculated from
crystallographic data. The experimental angles are calculated
on the basis that C2 exhibits a perfect sp2 configuration with
C3-C2-H4 ) 120°. It should be noted that these results
predict a fairly large reorientation of a part of the SP molecule
upon radical formation. The new and the old C2-C3 bond
directions differ by 52.8°, whereas the new and the old C2-
C1 bond directions differ by 36.1°.

Crystals of L-O-Serine Phosphate
J. Phys. Chem., Vol. 100, No. 22, 1996 9551
Figure 10. C2-C3-O1 fragment in L-O-serine phosphate single
crystals as seen along the C3-O1 bond. (a) The projection of the lone
pairs (1p, straight arrows) at O1 into the plane of the paper, together
with the vector n, the normal to the plane spanned by C2-C3-O1,
and the C3-H2 and C3-H3 bonds before radical formation. (b) The
same situation in radical II with H2 removed and replaced by LEO
rotated into its stable orientation.
Figure 11. Geometry around the nitrogen atom in L-O-serine phosphate
with the reference coordinate system (x,y,z) and the definition of the
angles ꢀ, δ, and η used for the discussion of quadrupolar couplings in
radical II.
pair (lp) orbitals at the neighboring oxygen atom which interact
with the LEO by electrostatic repulsion. In Figure 10a the
situation in SP as seen along the C3-O1 bond is illustrated.
One of the lp’s of O1 is oriented between the perpendicular to
the C2-C3-O1 plane and the broken C-H bond and thereby
prohibits the LEO from attaining the sp2 configuration. The
result is a pyramidal structure as visualized in Figure 10b, which
is the situation in radical II. This argument is valid only if it
is the H2-C3 bond that breaks upon radical formation. This
appears to be so from the following argument.
If the LEO is localized at C3, the LEO direction from tensor
#4 and the crystallographic data show that the dihedral angle
of H4 is 59.7°, which is in reasonable agreement with the
calculated value, 68.1°, for tensor #7. The FSE spectra also
show that the maximum value of this coupling occurs close to
the a-axis. The H4-C3 direction is only 21° from this axis.
The eigenvector of the intermediate principal value of tensor
#4 defines the direction of the LEO.8 The angle between this
direction and the direction perpendicular to the plane defined
by C2-C3-O1 is 19.5°. If the configuration of C3 was pure
sp2, these two directions should have been parallel. If the LEO
is localized at C3, the radical must have been formed upon
abstraction of either H2 or H3. It is assumed that when the
C-H bond is broken, the initially sp3-hybridized C3 atom will
try to attain an sp2 configuration by rotation of the LEO in the
plane defined by H2-C3-H3. This is consistent with the fact
that the LEO and the eigenvector of the smallest principal
value of tensor #4 are both almost in this plane (the angles
with the perpendicular to the plane are 83.3° and 87.8°,
respectively). Upon further discussion it is assumed and con-
firmed that the LEO is localized at C3 and that radical II exhibits
the structure
Radical II is the result of removing one of the H atoms
bonded to C3. It is assumed that the LEO is closest to the
direction of the C-H bond from which it originates. The H2-
C3 bond direction and the LEO form an angle of 20.0°, while
the corresponding angle between the LEO and the C3-H3 bond
direction is 44.4°. It therefore seems reasonable that radical II
is formed by the net loss of H2 from SP.
4.3. The 14N Hyperfine and Quadrupolar Coupling. The
Townes-Dailey approximation22 yields a method to estimate
the quadrupolar tensor. For the nitrogen atom, only the 2px,
2py, and 2pz orbitals are to be considered in the Townes-Dailey
approximation. With a basis as shown in Figure 11, Bo¨ttcher
et al.23 (Sørnes et al.8 have called attention to some errors
in ref 23, which are corrected in eq 2) defined the bonding
orbitals of the nitrogen atom and thereby the tensor elements
to be
e2Q
Qxx ) -
a0[-1/2 sin2 RσC - 1/2 sin2
4πꢀ02I(2I - 1)h
â(3 sin2 δ - 1)σH6 + 1/2 sin2 γ(3 cos2 (ꢀ/2)
sin2 η - 1)(σH7 + σH8)]
It seems clear that C3 does not exhibit a pure sp2 configu-
ration. Thus, the radical center is slightly pyramidal and the
LEO exhibits a small 2s orbital character. Incomplete rehy-
bridization is not unusual in carbon-centered radicals with
oxygen atoms as nearest neighbors. Dobbs et al.16,20 have
observed this in oxy-substituted alkyl radicals and have re-
ported smaller than expected numerical values for isotropic
values for R-couplings. This was explained by a 2s orbital
contribution to the LEO because of the deviation from planar
hybridization. The same radical structure as that suggested for
radical II observed in this work was observed by Sørnes et al.8
in AES.
e2Q
Qyy ) -
a0[-1/2 sin2 RσC -
4πꢀ02I(2I - 1)h
1/2 sin2 âσH6 + 1/2 sin2 γ(3 sin2 (ꢀ/2) - 1)(σH7 + σH8)]
e2Q
Qzz ) -
a0[sin2 RσC - 1/2 sin2 â(cos2 δ +
4πꢀ02I(2I - 1)h
1)σH6 - 1/2 sin2 γ(cos2(ꢀ/2) cos2 η + 1)(σH7 + σH8)]
(2)
e2Q
Qxy ) -
a0[-3/4 sin2 γ sin η
4πꢀ02I(2I - 1)h
Worth and Richards21 have explained this effect by ab initio
molecular-orbital calculations with the presence of the lone-
sin ꢀ(σH7 - σH8)]

9552 J. Phys. Chem., Vol. 100, No. 22, 1996
Sanderud and Sagstuen
e2Q
yields σH7 ) b -0.067 and σH8 ) b + 0.062. (The Qxy and Qyz
elements of eq 4 give different answers for the difference
between σH7 and σH8: -0.089 and -0.1743. The mean value
is used in this work.) These differences are both small and
indeed possible, considering the variations in bond lengths and
hydrogen bonding. The diagonal elements of eq 4 give an
estimate for the differences in electron population in the C-N
bonding orbital. With σC ) a, the results are a - b ) -0.276
for Qxx, a - b ) -0.236 for Qyy, a - b ) -0.256 for Qzz, and
the mean is a - b ) -0.256.
The result that b is greater than a is explained by taking into
account that the electron distribution in a bond depends on the
electronegativity of the different atoms constituting this bond.8
The electronegativity is for N 3.0, for C 2.5, and for H 2.1.27
This results in the carbon atom taking more of the electrons in
the C-N bond than the protons in the H-N bonds, and thus
the electron populations in the bonding orbitals of N to H
become greater than that in the orbital of N to C. This confirms
the sign of the quadrupolar tensor used in Table 2. Furthermore,
this confirms the positive isotropic value of the nitrogen hfc
and consequently the localization of the LEO to C3.
Qyz ) -
a0[-3/4 sin2 γ cos η
4πꢀ02I(2I - 1)h
sin ꢀ(σH7 - σH8)]
a0[3/4 sin2 â sin 2δσH6
e2Q
Qzx ) -
-
4πꢀ02I(2I - 1)h
3/4 sin2 γ sin 2η cos2(ꢀ/2)(σH7 + σH8)]
where σk denotes the valence orbital coefficient (electron
population) in the bonding molecular orbital modeling the kth
bond (assuming a RHF model), the angles ꢀ, δ, and η are defined
in Figure 11, and the angles R, â, and γ are obtained by
orthogonality requirements to be
cos(δ + η)
tan2 R ) -
cos η cos δ
cos η
cos(δ + η) cos δ
tan2 â ) -
(3)
The dihedral angle of the N atom with respect to the direction
of the LEO is found to be θN ) 42.1°, using the crystallographic
data combined with the experimental direction of the LEO. In
the same way as for the Heller-McConnell relation for
â-protons (eq 1 above) this value should be associated with the
nitrogen isotropic value of aiso ) 15.89 MHz. While the
constants B0 and B2 are well-known for â-protons, they are far
less known for â-nitrogen couplings, although some theoretical
estimates using INDO calculations have been done by Close et
al.28 The corresponding values observed in an equivalent radical
in AES8 is θN′ ) 10.5°, aiso′ ) 26.3 MHz, and F′dip ) 0.765. If
a model as simple as aiso ) B2 cos2 θFπ is used, the two systems
give the constants B2 ) 36.4 MHz and B2′ ) 35.6 MHz. If a
constant B0Fπ is added to the model, as in the Heller-McConnell
relation (eq 1), the two systems together give B0 ) 1.0 MHz
and B2 ) 34.5 MHz. Due to the nonplanar nature of the radical
center, a first-order term is expected29 to contribute in the
expression, which then will take the form
cos δ
tan2 γ ) -
cos(δ + η) cos η cos2(ꢀ/2)
The factor -[e2Q/4πꢀ02I(2I - 1)h)]a0 has a value between -5.0
and -4.0 MHz.24 In this work -4.5 MHz has been used.
In a perfectly tetrahedral amino group ꢀ and δ are 109.47°
and η is 125.26°. Equation 2 then gives an axially symmetric
quadrupolar tensor which only depends on differences in the
electron populations of the bonding orbitals at the nitrogen atom.
However, deviation from tetrahedral structure will also make
actual populations significant variables and axial symmetry may
disappear.
Sørnes et al.8 have observed an axially symmetric quadrupolar
coupling for the nitrogen atom of the amino group in AES. The
asymmetry parameter η was only 0.095. (The asymmetry
parameter is defined by McDowell and Naito25 as (Qaa - Qbb)/
Qcc with |Qaa| e |Qbb| e |Qcc|.) In the present work, Table 2
shows that η ) 0.379, far too large to categorize the quadrupolar
tensor as axially symmetric. In glycine Deigen et al.26 have
found a quadrupolar tensor for the amino group with an even
larger asymmetry. Several reasons for this behavior may be
suggested and tested. One possibility is that the amino group
has rehybridized to sp2 upon deprotonation. A deprotonation
can also occur without any rehybridization since the hydrogen
bonding in the amino group could conserve the sp3 configura-
tion. In both cases a lone-pair orbital is formed which will be
the dominating direction and remove the axial symmetry of the
quadrupolar tensor. Both models have been tested without
obtaining convincing agreement with the experimental data.
Another possible reason for the nonaxial symmetry of the
quadrupolar tensor is a deviation from tetrahedral configuration
of the amino group by differences in the electronic populations
of three H-N bonds. If the quadrupolar tensor is rotated to a
basis as shown in Figure 11, the following result is obtained:
aiso ) Fπ(B0 + B1 cos θ + B2 cos2 θ)
(5)
The present data combined with those for AES8 yield only two
parameters of this equation. Since B0 is expected to be small,
it is arbitrarily set to zero. The calculations then yield estimated
values of the constants B1 ) 2.6 MHz and B2 ) 32.9 MHz.
The dipolar tensor of the 14N coupling (2.06, -0.27, -1.82)
MHz deviates from axial symmetry with an asymmetry param-
eter of η ) 0.752. In addition, the directions of the C3-N
bond and the eigenvector for the largest principal value of the
14N hfc tensor deviate by 21.8°. The explanation must be that
the dipolar coupling to an N atom as a part of an amino group
cannot necessarily be compared with the dipolar coupling to
an H atom. This is illustrated by the following attempt to
reconstruct the dipolar coupling tensor.
The spin at C3 contributes to an axially symmetric â-type
tensor with principal elements (0.62, -0.31, -0.31) MHz. This
is calculated from the LEO spin density of 0.793, using the
point-dipole model for proton â-couplings suggested by Zeldes
et al.30 and scaled with a factor 0.0731 according to the ratio of
the nuclear g-factors of N and H.
-0.469 -0.174
P ) -0.174 -0.390 -0.240
0.004
(4)
[
]
0.004 -0.240
0.858
This tensor has an asymmetry parameter of only η ) 0.092.
With values of ꢀ, δ, and η corresponding to a perfect tetrahedral
structure, eq 2 shows that the nondiagonal elements of the
quadrupolar tensor only depend on differences in the electron
population of the H-N bonds. With a value of σH6 ) b in eq
2, the expression for the nondiagonal elements of P in eq 4
The spin localized in the 2p orbitals of the nitrogen will also
contribute to the dipolar tensor. The spin density in these 2p
orbitals is calculated as follows. From the isotropic value 15.89
MHz the nitrogen 2s population is obtained by scaling with the
constant 1811 MHz32 and becomes 0.008 77. The s/p ratios

Crystals of L-O-Serine Phosphate
J. Phys. Chem., Vol. 100, No. 22, 1996 9553
for the valence orbitals are obtained from the model of Bo¨ttcher
et al. (eq 2) and give 2p spin densities of 0.0332 for the C2-N
bond and 0.0250, 0.0265, and 0.0265 for the H6-N, H7-N,
and H8-N bonds, respectively. With unit spin density in a 2p
orbital the dipolar tensor is (95.6, -47.8, -47.8) MHz33 with
the symmetry axis along the 2p orbital direction. In the
laboratory frame of reference, the complete dipolar tensor
composed of all nitrogen 2p orbital contributions above then
becomes
tensor #8 the corresponding angle with C3-H2 is 56.1°. If
minimum reorientation of the molecule upon radical formation
is assumed, tensor #8 must represent the interaction with H2,
while tensor #9 represents the interaction with H3.
0.164
0.730 -0.004
0.730
0.267
0.055
(6)
[
]
0.267
0.055 -0.165
Taken together with the contribution from the carbon atom this
gives a diagonalized dipolar tensor with principal values of
(1.36, -0.41, -0.95) MHz. This tensor has principal values
about half of those for the experimental ones. Furthermore,
the eigenvector of the largest value deviates from the observed
direction by 30°.
From the radical structure, couplings to the amino group and
to H4 are expected. The McConnell relation and the isotropic
value of hfc tensor #8 give a spin density at C3 of Fπiso ) 0.690.
The Gordy-Bernhard method17 gives Fdπip ) 0.672 from the
dipolar tensor, which is close to Fπiso and confirms the sp2
configuration. Using the experimental direction for the LEO
and crystallographic data for the coordinates of H4, the dihedral
angle of H4 is θ ) 33.5°. By the Heller-McConnell relation
(eq 1) with Fπ ) 0.690, B0 ) 0, B2 ) 126 MHz, and θ ) 33.5°
an expected isotropic value of 73.0 MHz is calculated for the
coupling to H4. The expected principal dipolar tensor elements
are (13.3, -6.6, -6.6) MHz using the point-dipole approxima-
tion.30 If the extra coupling, missing in the ENDOR spectra
but clearly present in the FSE spectra, is a coupling to a proton,
it should be 71.4 MHz for the b-axis orientation, which fits
nicely with the above calculations for the expected â-coupling
to H4. On the other hand, the dihedral angle of N is 84.9°, and
any significant coupling to this atom is not expected. This
leads to the conclusion that the missing coupling in the FSE
spectra from radical III is a â-coupling to H4. It is interesting
that also in radical II the â-coupling to H4 was difficult to
observe by ENDOR. This probably reflects particularly favor-
able relaxation properties due to a particular environment of
this nucleus, which may be probed by studies at lower
temperature.
However, by varying the spin densities in the N-H bonds,
it is found that with 0.0175, 0.0186, and 0.0106 as 2p
contributions in the H6-N, H7-N, and H8-N bonds a dipolar
tensor of (2.61, -0.66, -1.95) MHz is obtained, with eigen-
vectors which deviated from the observed ones by only 1.6°,
1.8°, and 3.1°, respectively. This shows that small variations
in the 2p spin densities induced by the environment can give a
dipolar tensor which reproduces that observed.
4.4. Radical III. From the FSE experiments, two of the
ENDOR determined hfc tensors were associated with radical
III. Both couplings are due to nonexchangeable protons of the
R-type. In addition the FSE spectra reveal that at least one more
coupling contributes to the resonance. The two R-couplings
are similar and of a magnitude such that the responsible protons
both must be bonded to the same carbon atom. It is conse-
quently assumed that the LEO is localized at C3.
Radical III is suggested to be the result of scission of either
the C3-O1 bond or the C2-C3 bond. This will, in both cases,
lead to rehybridization at C3 from an sp3 to an sp2 configuration.
The LEO will be perpendicular to the plane defined by the sp2
hybrid orbitals. The eigenvectors for the intermediate principal
values of the two R-tensors, which represent the direction of
the LEO, deviate only 3.3° from each other. This confirms that
the two protons are bonded to the same atom. In the following
discussion, the eigenvector for the intermediate principal value
of tensor #8 is taken to denote the direction of the LEO.
As mentioned above, radical III may be formed by the
scission of either the C2-C3 bond or the C3-O1 bond. In the
case of breakage of the C2-C3 bond, the angle between LEO
and C3-O1 should be close to 90° if the geometry of the C3-
O-PO3-H fragment is preserved. However, this angle (which
depends upon the signs of the eigenvectors used) can not be
smaller than 124°. On the other hand, breakage of the C3-O1
bond gives the corresponding angle between the LEO and the
C2-C3 bond as small as 96.3°. The eigenvectors for the
smallest principal values of the R-tensors are expected to be
along the directions of the C-H bonds. From this, the angle
between the C-H direction of tensor #8 and the crystallographic
C3-C2 bond direction is 124.6°, the corresponding angle for
tensor #9 is 119.6°, and finally, the angle between the two C-H
directions given by #8 and #9 is 115.7°. The sum of these three
angles is 359.9°, which confirms the sp2 configuration at C3.
These considerations all lead to the conclusion that radical
III is a result of the cleavage of the C3-O1 bond. The angle
between the eigenvector of the smallest principal value of tensor
#9 and the crystallographic C3-H3 direction is 17.3°, and for
5. Mechanistic Aspects
All three radicals are localized at the serine part of SP, and
the equivalent species have been observed in irradiated single
crystals of serine.9,10 This indicates that the phosphate group
in SP does not significantly affect the radiation chemistry of
serine. This must be ascribed to the presence of the carboxyl
group which apparently acts as an effective electron, as well as
hole, scavenger. The formation mechanisms of the correspond-
ing radicals in serine have been thoroughly investigated,9,10 and
there are no experimental indications that the mechanisms are
not essentially the same in SP. Thus, the radicals observed
in the present work are suggested to be formed as outlined
below.
5.1. Radical I. A radical with structure equivalent to that
of radical I in SP was found in serine by Lebedev and Almanov
at 300 K.9 This was later conformed by Castleman and
Moulton.10 These authors used temperature variation experi-
ments to show that this radical is a secondary product of
the primary reduction species. On the basis of this, the
formation mechanism of radical I is suggested to be as shown
in Figure 12.
5.2. Radical II. A radical of structure equivalent to radical
II in SP was observed in serine by Castleman and Moulton.10
In serine this species is believed to be formed both as a
secondary product of the equivalent of radical I, which decays

9554 J. Phys. Chem., Vol. 100, No. 22, 1996
Sanderud and Sagstuen
Figure 12. Proposed mechanism of formation of radical I.
Figure 14. Proposed mechanism of formation of radical III.
irradiation and observation at 77 K. Thus, this eventually
formed phosphoranyl radical must be unstable and must decay
prior to observation at 77 K by phosphate elimination. Ac-
cording to this model, a formation mechanism as that shown in
Figure 14 is suggested. This model should be tested by low-
temperature experiments.
Acknowledgment. The writing of this paper was facilitated
by a travel scholarship to A.S. granted by the Nordic Academy
for Research Education, NorFA. Prof. A. Lund is thanked for
his hospitality during the 2 month visit in Linko¨ping, Sweden.
Prof. D. M. Close is acknowledged for suggesting this prob-
lem. Mr. A. R. Sørnes is acknowledged for making the analysis
of the quadrupolar couplings possible and for helpful discus-
sions.
References and Notes
(1) Becker, D.; Sevilla, M. D. AdV. Radiat. Biol. 1993, 17, 121.
(2) Bernhard, W. A.; Barnes, J.; Mercer, K. R.; Mroczka, N. Radiat.
Res. 1994, 140, 199.
(3) Nelson, D. J.; Symons, M. C. R.; Wyatt, J. L. J. Chem. Soc.,
Faraday Trans. 1993, 89, 1955.
(4) Boon, P. J.; Cullis, P. M.; Symons, M. C. R.; Wren, B. W. J. Chem.
Soc., Perkin Trans. 2 1984, 1393.
(5) Bungum, B.; Hole, E. O.; Sagstuen, E.; Lindgren, M. Radiat. Res.
1994, 139, 194.
(6) Sanderud, A.; Sagstuen, E. J. Chem. Soc., Faraday Trans. 1996,
92, 995.
(7) Fouse, G. W.; Bernhard, W. A. J. Chem. Phys. 1979, 70, 1667.
Fouse, G. W. Thesis submitted for the degree Ph.D. at the Department of
Radiation Biology and Biophysics, University of Rochester, Rochester, New
York, 1979.
(8) Sørnes, A. R.; Sagstuen, E. J. Phys. Chem. 1995, 99, 16857. Sørnes,
A. R.; Sagstuen, E.; Lund, A. J. Phys. Chem. 1995, 99, 16867.
(9) Lee, J. Y.; Box, H. C. J. Chem. Phys. 1973, 59, 2509. Lebedev,
Ya.; Almanov, G. A. Biofizika 1967, 12, 338.
(10) Castleman, B. W.; Moulton, G. C. J. Chem. Phys. 1971, 55, 2598.
Castleman, B. W.; Moulton, G. C. J. Chem. Phys. 1972, 57, 2762.
(11) Sundaralingam, M.; Putkey, E. F. Acta Crystallogr. 1970, B26, 790.
(12) Nelson, W. H. J. Magn. Reson. 1980, 38, 71.
(13) Schonland, D. S. Proc. Phys. Soc. (London) 1958, 73, 788.
(14) McConnell, H. M. J. Chem. Phys. 1965, 24, 764.
(15) Fessenden, R. W.; Schuler, R. H. J. Chem. Phys. 1963, 39, 2147.
(16) Dobbs, A. J.; Gilbert, B. C.; Norman, O. C. J. Chem. Soc. A 1971,
124.
Figure 13. Proposed mechanism of formation of radical II.
upon heating, and as a secondary radical from the primary
oxidation product. In the latter case, the secondary radical is
formed already at 153 K. Both mechanisms proceed by
intermolecular hydrogen abstraction. Thermal decay of radical
I resulting in radical II is not observed in the present work.
This makes it most probable that radical II is a secondary radical
from the primary oxidation product, and a formation mechanism
is suggested as shown in Figure 13.
5.3. Radical III. Radical III has no equivalent radical in
serine as judged from the 300 K studies.10 However, Lee and
Box9 observed a third primary product with the same structure
as radical III by X-irradiation and observation at 4.2 K. In SP
there is little experimental evidence for the precursor of this
radical. One possibility is that the formation of radical III in
SP may follow from a primary phosphoranyl radical, as observed
formed in other dialkyl or dihydroxyalkyl phosphates3,6,7,34
following reduction of the phosphate group. The phosphoranyl
radical was, however, not observed in polycrystalline SP after
(17) Gordy, W. Theory and applications of spin resonance; John Wiley
& Sons: New York, 1980. Bernhard, W. A. J. Chem. Phys. 1965, 24,
764.
(18) Heller, C.; McConnell, H. M. J. Chem. Phys. 1960, 32, 1535.
(19) Morton, J. R. J. Chem. ReV. 1964, 64, 453.
(20) Dobbs, A. J.; Norman, O. C.; Gilbert, B. C. J. Chem. Soc., Perkin
Trans. 2 1972, 786.
(21) Worth, G. A.; Richards, W. G. J. Am. Chem. Soc. 1987, 116, 239.

Crystals of L-O-Serine Phosphate
J. Phys. Chem., Vol. 100, No. 22, 1996 9555
(22) Townes, C. H.; Dailey, B. P. J. Chem. Phys. 1949, 17, 782.
(23) Bo¨ttcher, R.; Metz, H.; Windsch, W. J. Chem. Phys. 1985, 93, 137.
(24) Schweiger, A. Struct. Bonding 1982, 51.
(25) McDowell, C. A.; Naito, A. J. Magn. Reson. 1981, 45, 205.
(26) Deigen, M. F.; Krivenko, V. G.; Pulatova, M. K.; Ruban, M. A.;
Teslenko, V. V.; Kayushin, L. P. Biophysics 1973, 242.
(27) Atkins, P. W. General Chemistry (International Student Edition);
Scientific American Books: W. H. Freeman & Co.: New York, 1989.
(28) Close, D. M.; Fouse, G. W.; Bernhard, W. A.; Andersen, R. S. J.
Chem. Phys. 1979, 70, 2131.
(30) Zeldes, H.; Trammell, G.; Livingston, R.; Holmberg, R. J. Chem.
Phys. 1960 32, 618.
(31) Weil, J. A.; Preston, K. F. BRUKER EPR/ENDOR Frequency Table,
BRUKER EPR spectrometer accompaniment.
(32) Morton, J. R.; Preston, K. F. J. Magn. Reson. 1978, 30, 577.
(33) Morton, J. R.; Rowlands, J. R.; Whiffen, D. H. Atomic Properties
for Interpreting ESR Data, National Physics Laboratory, BPR 13, 1962.
(34) Nelson, D. J.; Symons, M. C. R. J. Chem. Soc., Perkin Trans. 2
1977, 286.
(29) Colson, A. O.; Sevilla, M. D. Int. J. Radiat. Biol. 1995, 67, 627.
JP960063X