1H and 15N Dynamic Nuclear Polarization Studies of Carbazole

Article abstract of DOI:10.1021/jp9938011

¹⁵N NMR experiments, combined with dynamic nuclear polarization (DNP), are reported on carbazole doped with the stable free radical 1,3 bisdiphenylene-2-phenylallyl (BDPA). Doping shortens the nuclear relaxation times and provides paramagnetic

Full text of DOI:10.1021/jp9938011

J. Phys. Chem. A 2000, 104, 4413-4420
4413
15
1H and N Dynamic Nuclear Polarization Studies of Carbazole
Jian Z. Hu,† Mark S. Solum,‡ Robert A. Wind,§ Brad L. Nilsson,| Matt A. Peterson,|
Ronald J. Pugmire,*,| and David M. Grant†
Departments of Chemistry and Chemical and Fuels Engineering, UniVersity of Utah,
Salt Lake City, Utah 84112, EnVironmental and Molecular Sciences Laboratory, Battelle Pacific Northwest
National Laboratory, P.O. Box 999, MS K8-98, Richland, Washington 99352, Department of Chemistry,
Brigham Young UniVersity, ProVo, Utah 84602
ReceiVed: October 26, 1999
15N NMR experiments, combined with dynamic nuclear polarization (DNP), are reported on carbazole doped
with the stable free radical 1,3 bisdiphenylene-2-phenylallyl (BDPA). Doping shortens the nuclear relaxation
times and provides paramagnetic centers that can be used to enhance the nuclear signal by means of DNP so
that 15N NMR experiments can be done in minutes. The factors were measured in a 1.4 T external field,
using both unlabeled and 98% 15N labeled carbazole with doping levels varying between 0.65 and 5.0 wt %
BDPA. A doping level of approximately 1 wt % produced optimal results. DNP enhancement factors of 35
and 930 were obtained for 1H and 15N, respectively making it possible to perform 15N DNP NMR experiments
at the natural abundance level.
Introduction
Moreover, if the DNP condition can be achieved by ap-
propriately doping the solid with paramagnetic centers, the
The 15N isotope has a spin 1/2 and, like 13C, the chemical
shift and chemical shift tensor principal values can be related
to molecular structure. The large chemical shift range of the
15N isotope (∼1100 ppm) provides an enhanced sensitivity to
molecular structure variations and intermolecular interactions
that exceeds other first row elements. However, the detection
of a 15N NMR signal in the solid state is very difficult due to
a combination of low natural abundance, low gyromagnetic ratio,
and the long spin-lattice relaxation time of this isotope. Hence,
only limited amounts of data have been acquired in the solid
state on the nitrogen shift and their principal shift values in
nitrogen heterocycles.1-6
nuclear relaxation time is generally reduced and an additional
reduction of the experimental measuring time is achieved.
DNP NMR spectroscopy in solids has focused mainly on
protons and 13C nuclei. Recently, preliminary 15N DNP experi-
ments were reported in doped benzamide,26 but the 15N DNP
enhancement could only be estimated (ca. 260) as the signal in
the absence of the DNP condition could only be approximated
with the instrumentation available at that time. Griffin and co-
workers have reported a 15N DNP enhancement of 50 at 40 K
in a glycerol solution of 15N Ala-labeled T4 lysozyme doped
with 4-amino-TEMPO.23
In this paper a systematic DNP/NMR study is reported of
the 1H and 15N nuclei in 15N labeled carbazole, a model
compound representing one of the common nitrogen hetero-
cycles found in fossil fuels. This investigation includes measur-
ing: (1) the changes in the 1H spin-lattice relaxation times in
the laboratory and rotating frames for carbazole after doping
the sample with various amounts of the free radical; (2) the 1H
and 15N DNP enhancement factors; (3) the effects of the doping
upon the 1H and 15N DNP polarization curves; and (4)
comparison of the 15N shift tensor principal values at 1.4 T under
conditions of DNP-CP, DNP-SP as well as CP at 4.7 T. The
spin-labeled material ensures that accurate values for the
unenhanced 15N signal can be measured. These experiments
demonstrate that large signal enhancements can be observed in
this case and that one can obtain useful powder patterns under
DNP experimental conditions that duplicate those obtained under
standard NMR conditions.
Increased sensitivity results from isotopic labeling and sample
doping with a relaxation agent to reduce the recycle delay time.
Further, a third signal enhancement alternative, dynamic nuclear
polarization (DNP), has been known7-11 for a number of years.
This technique can be applied in solids containing both magnetic
nuclei and unpaired electrons (present either by nature or by
doping the solid with a suitable stable free radical). Polarization
transfer can be induced between the electron and nuclear spin
systems by irradiating at or near the electron Larmor frequency
resulting in an enhanced nuclear polarization and, hence, a
greater NMR signal. In principle, larger NMR signal enhance-
ments may be obtained via DNP where the theoretical maximum
enhancement is the ratio of the gyromagnetic ratios of the
electrons and the nuclei. For example, for 1H, 13C, and 15N nuclei
the maximum enhancement factors are 660, 2600, and 6500,
respectively. In practice these large theoretical enhancements
are unrealized for a variety of reasons but enhancements of 1-3
orders of magnitude have been observed. Thus, the measurement
time of an experiment can be significantly reduced.10-25
Experimental Section
1. Sample Doping. The 15N labeled carbazole was doped
with the stable radical 1,3-bisdiphenylene-2 phenylallyl (BDPA)
using the solvent-evaporation method,27 i.e., both the carbazole
and BDPA were dissolved in acetone, after which the acetone
was allowed to evaporate slowly. This method of sample doping
* Author to whom correspondence should be sent.
† Department of Chemistry, University of Utah.
‡ Department of Chemical and Fuels Engineering, University of Utah.
§ Battelle Pacific Northwest National Laboratory.
| Brigham Young University.
10.1021/jp9938011 CCC: $19.00 © 2000 American Chemical Society
Published on Web 04/19/2000

4414 J. Phys. Chem. A, Vol. 104, No. 19, 2000
Hu et al.
SCHEME 1a
a Reagents: (a) (BOC)2O, K2CO3, THF-H2O (1:1); (b) i. tBuLi, THF, -78 °C to -20 °C, ii. Bu3SnCl, -78 °C to 25 °C; (c) bromobenzene,
Pd(Ph3)2Cl2; (d) 4 M HCl; (e) i. NaNO2 (aq), AcOH; ii. NaN3; (f) 180 °C.
has proven to be a very useful way to dope crystalline materials.
Four samples were prepared with doping levels of 0.65, 1.1,
2.5, and 5.0 wt % BDPA.
2. DNP Spectrometer. The DNP NMR spectrometer operates
at a field of 1.4 T, corresponding to electron, 1H, and 15N Larmor
frequencies of 40 GHz, 60 MHz, and 6 MHz, respectively. The
NMR components of the spectrometer consist of a Chemagnetics
CMX-100 dual channel console. The magnet is a small-animal
horizontal imaging 2.3 T superconducting magnet from Magnex,
in which the field was reduced to 1.4 T. This magnet has a
clear bore diameter of 22 cm, which readily accommodates the
construction of the DNP NMR probes.28 Microwave irradiation
is achieved with a Wiltron microwave frequency generator,
model 68263B, capable of producing an output of 6 dBm in
the frequency range 20-40 GHz. The output traveling wave
tube amplifier (Logimetrics model A400/KA) amplifies the
Figure 1. The 40 GHz EPR spectra for carbazole samples doped with
BDPA. The doping levels are A: 0.65; B: 1.1; C: 2.5; and D: 5.0%
microwave power to 10 W c.w. in the frequency range 26-40
GHz. Compared with fixed-frequency sources such as klystrons,
(w/w).
this setup has an important advantage. Both the EPR spectrum
and the DNP enhancement curve can be determined and
equipped with two modulation coils (a SR850 DSP lock-in
optimized by sweeping the microwave frequency rather than
amplifier from Stanford Research Systems operating at a
the field, thereby avoiding the construction and calibration of
modulation frequency of 100 kHz) which, in combination with
special field-sweep coils in the magnet. It also avoids the
the microwave frequency sweep, makes it also possible to obtain
necessity of continual adjustment of the probe tuning at the
c.w. first-derivative 40 GHz EPR spectra.
different field strengths since the NMR resonance frequencies
3. NMR and DNP NMR Experiments. Unless stated
remain unaffected by sweeping the microwave field. A further
otherwise, the experiments were performed at room temperature.
advantage is that the microwave frequency and microwave
1
The H DNP enhancement factors and the relaxation times in
pulses are under computer control through a GPIB interface board
the laboratory and rotating frames were measured with standard
and are synchronized with the NMR pulse sequence. This
single pulse wide-line NMR techniques using RF fields of 64
arrangement fully automates the measurement of the DNP
kHz. This field was also used for decoupling. 15N single (90°)
enhancement curve (i.e., the DNP enhancement factor as a
pulse and 1H-15N cross-polarization (CP) experiments, with and
function of microwave frequency). Furthermore, the capability
without microwave irradiation, were performed using 15N and1H
of generating a pulsed microwave field allows lossy samples
RF fields of 48 kHz during CP, and a 1H field of 64 kHz during
to be investigated.
decoupling. To eliminate distortions resulting from acoustic ring-
The DNP probe is home-built, and is capable of performing
DNP NMR on static samples in the temperature range of 110-
down after the 15N RF irradiation, a τ-180-τ echo sequence was
employed, with a delay time τ of 150 µs.
350 K. The NMR coil is a 5-turn solenoid coil with an inner
Finally the effects of radical doping on isotropic and
diameter of 10.5 mm and length of 11 mm made of 1 mm copper
anisotropic chemical shifts and/or its spectral resolution has been
wire. The setup can be double tuned to 1H and a second nucleus
investigated. This is done by performing standard 13C and 15N
of interest (15N in this case). This setup is a variant of previous
CP experiments, with and without MAS, on the doped and
designs.10,16,29,30 The microwave irradiation is introduced into
undoped 15N labeled carbazole samples. A Chemagnetics CMX-
the sample using a combination of a cylindrical horn antenna
200 spectrometer was used for these measurements.
and a movable reflector. The NMR coil axis is coincident with
4. Synthesis of 15N Labeled Carbazole. 15N labeled carba-
that of the horn antenna. The copper reflector is placed
approximately half a wavelength (about 3.5 mm) away from
zole was synthesized according to Scheme 1.32
the other end of the NMR coil so that the decrease in the quality
factor of the NMR coil is minimized. In this way both the
Results and Discussion
microwave and the NMR efficiency can be optimized. It is
estimated that with this design the 10 W incident microwave
power corresponds to an average amplitude of the microwave
field in the sample of about 0.04 T (rotating component). In
nonlossy samples, such as doped polystyrene, DNP enhance-
ments were obtained comparable to those previously reported
on considerably smaller samples.10,31 The DNP probe is
1. EPR and 1H DNP NMR. The first derivative EPR spectra
for the four doped carbazole samples are given in Figure 1. At
the two lowest doping levels (Figure 1A and 1B) a single ESR
line is observed with a near Gaussian line shape and a width of
approximately 26 MHz (full width at half-maximum), typical
results for diluted radicals in solids.31 At the 2.5% doping level
(Figure 1C), the EPR spectrum consists of two overlapping lines,

1
Carbazole H and 15N DNP Studies
J. Phys. Chem. A, Vol. 104, No. 19, 2000 4415
1
TABLE 1: EPR and H-NMR Parameters of the Carbazole Samples Studieda
wt %
Ne × 10-19 (g-1
)
Ne′ × 10-19 (g-1
)
T
1F (ms)
>500
12.0
T1S (s)
M0S (%)
T1L (s)
M0L (%)
Ne × T1S
Ne × T1L
E-1
0 (undoped)
0.65
1.1
0
0
137
1.6
1.0
0.33
0.12
18
34
55
58
65
1074
17.5
12.0
3.1
82
66
45
42
35
0
0
0
32
35
23
15
0.41
0.8
1.9
4.3
0.94
1.6
3.6
7.2
0.66
0.80
0.63
0.52
7.18
9.60
5.89
5.93
3.4
1.9
0.8
2.5
5.0
1.38
a Ne ) measured concentration of unpaired electrons; Ne′ ) the calculated unpaired electron concentration based on the amount of BDPA used
to dope the samples; T1S, T1L ) the proton spin-lattice relaxation time, measured at an external field of 1.4 T and without the presence of microwave
irradiation; T1F ) the average proton rotating frame relaxation time, measured in a lock field of 64 KHz and an external field of 1.4 T under the
DNP condition; E-1 ) the proton DNP enhancement factor at the frequencies of the solid state effect, observed when irradiating with a microwave
1
frequency equal to the difference between the electron and H Larmor frequencies.
TABLE 2: Parameters Relevant to Define the Diffusion
one with a width similar to that observed at the lower doping
levels, and one with a width of about 14 MHz. This is probably
Limits
Ne × 10-19
Ne × 10-19
an indication that the radicals are not homogeneously distributed
throughout the sample, and that a portion of the free radicals
are present in agglomerates where the electrons are so close to
each other that spin-exchange narrowing occurs.7 At the 5%
doping level the percentage of the component with broad line
width (∼26 MHz) decreases, and the line width of the narrow
component is reduced to about 6 MHz (Figure 1D). An EPR
line width of about 5 MHz for pure BDPA was obtained under
the same experimental conditions (the same EPR power and
modulation power). This result indicates that a substantial
portion of the radicals are clustered together at 5% doping levels.
The unpaired electron concentration, Ne, was measured on the
40 GHz DNP/ESR spectrometer using polystyrene doped with
BDPA with a known free radical concentration of 1.6 × 1019
g-1 as a calibration standard. The measured Ne is about half of
the calculated free radical concentration based on the amount
of BDPA added by the doping procedure.
wt %
(g-1
)
(cm-3
)
R (Å)
b (Å)
â (Å)
0.65
1.1
2.5
0.41
0.8
1.9
0.526
1.028
2.442
5.526
35.7
28.5
21.4
16.3
7
7
7
7
7.33
7.33
7.33
7.33
5.0
4.3
from lattice defects. These so-called mobilized species, on the
order of 1% or less, can be directly observed by delayed
acquisition MAS experiments.34 These mobilized species act
as relaxation sinks which relax proximate protons by a spin-
diffusion process. Consequently, a shorter 1H T1 value is
observed in these nuclei.
The spin-lattice relaxation data are complicated by the
presence of free radicals in the doped samples. A useful model
for describing this behavior consists of a diffusion limit defined
by the diffusion barrier, b, the scattering radius, â, and a radius,
R, that may be related to the average distance between electrons
when the relationship (4/3)πR3Ne ) 1 holds.35-38 Within the
distance, b, between a nucleus and the unpaired electron, (i.e.,
r < b) the electron-nuclear dipolar interaction is sufficiently
large that spin diffusion among the nuclei is quenched, and the
protons are unobservable. Within the scattering radius, i.e., r
< â, proton relaxation is governed primarily by direct electron-
proton interactions, whereas for r > â proton relaxation is
governed by proton-proton spin diffusion until a proton closer
to the free electron is encountered. Wind et al.38 have provided
a detailed analysis of this model. Complete spin diffusion holds
if b e â and all observable protons relax exponentially with a
uniform average relaxation rate. Limited spin diffusion (LSD)
holds if b < â < R and within this limit, for protons at a distance
b < r < â, a distribution of relaxation rates is obtained. The
long-term relaxation exhibits exponential behavior and is
governed by the protons at a distance â < r < R.
1
Table 1 contains the H spin-lattice relaxation times in the
laboratory and rotating frame, T1 and T1F, respectively, of the
undoped and doped samples. The T1 measurements were made
with and without DNP enhancement while only the T1 values
obtained without DNP are reported in Table 1. It was found
that the 1H T1 values obtained with and without DNP enhance-
ment agree with each other within the experimental error.
However, the relative contributions from the multiple relaxation
components contributing to the magnetization may change
because the DNP enhancement factor is different for different
components. Conversely, the T1F measurements were made on
the DNP-enhanced signals in order to take advantage of the
increased sensitivity available.
1
In the T1 measurements the H magnetization as a function
of the recovery time is multiexponential but can be expressed
with a high degree of confidence by the following equation
consisting of two exponential components,
Duijvestijn et al.31 have determined that b is equal to 7 Å in
polystyrene doped with BDPA. It is reasonable to assume the
same b value in the case of carbazole doped with BDPA. The
carbozole unit cell contains 4 molecules,39 with cell volume of
825.3 cubic Å and density of 1.285 g/cm3. The average volume
for one proton is thus 22.9 Å3. The diameter of the sphere for
each proton is 3.53 Å which is the average distance (lattice
constant) between the protons in carbazole. Given the Larmor
frequency (60 MHz) for protons, â is found to be 7.33 Å by
following the calculation given by Wind et al.38 Using the
experimentally determined radical concentration and assuming
a homogeneous distribution of radicals, the R-values for the four
doped samples are easily calculated and the results are given
in Table 2.
M(t) ) M0S(1 - exp(-(t/T1S))) + M0L(1 - exp(-(t/T1L)))
(1)
where T1S and T1L denote the components with short and long
T1 relaxation times, respectively, and M0S + M0L ) 1. The two
relaxation rates differ by an order of magnitude. In a related
1
study on dibenzofuran (DBF)27b,33 it was noted that the H T1
in DBF also displays multiexponential behavior in which the
value of the short T1 component is an order of magnitude smaller
than that of the longer T1 component. The relaxation mechanism
for the longer component arises from dipolar interactions among
the protons. A rigid lattice environment, i.e., ωHτc . 1,
1
commonly gives rise to a rather long H T1 value. However, a
It is known from Table 2 that the condition R . â ≈ b is
satisfied for the four doped samples. Consequently, the argument
for complete spin-diffusion35-38 is valid in the doped carbazole
short 1H T1 component is usually also noted in crystalline solids
and is thought to be due to mobilized species in the lattice arising

4416 J. Phys. Chem. A, Vol. 104, No. 19, 2000
Hu et al.
Figure 2. The proton DNP enhancement factor (E-1) as a function of microwave irradiation frequency ω - ωe for carbazole doped with BDPA
at A: 5 wt %; B: 2.5 wt %; C: 1.1 wt %; D: 0.65 wt %.
samples. To explain the biexponential spin-lattice relaxation
behavior, one must assume an inhomogeneous distribution of
the BDPA radicals in the sample. The doped samples can be
approximated by two or more types of domains, where one
domain is enriched in BDPA compared to other domains. Visual
inspection supports this postulate as one can clearly see a
heterogeneous distribution of BDPA in the most highly doped
sample. Whenever complete spin-diffusion is valid, a single-
exponential relaxation time is obtained for protons located inside
the same domain. This relaxation in the enriched domain is faster
because the proton T1 is inversely proportional to the radical
concentration. For doping levels less than 2.5%, both T1S and
T1L are inversely proportion to the radical concentration (see
Table 1). Significant deviations are found at the 5% doping level
as a consequence of radical clustering. These deviations are
consistent with EPR distortions noted in Figure 1d.
TABLE 3: Parameters Obtained on Carbazole Samples
Doped with Different Concentrations of BDPAa
sample
wt% BDPA (g-1
solid state
ωs E-1
thermal mixing Overhauser
)
∆
E-1
ωe
E-1
1/2
0.65
1.1
2.5
59.3 28 32
59.3 29 35
58.8 31 23
57.9 33 15
3
6
5
7
39.133
39.133
39.133
39.133
4
5
4
4
5.0
a ωs and ∆1/2 are in units of MHz. The unit ωe is in GHz. E-1 denotes
the absolute enhancement for each type of DNP mechanism. The value
of ωo was visually approximated at 20 MHz.
very advantageous as it reduces the pulse recycle time. However,
the reduction in T1F is less desirable, as it limits the contact
time that can be used in a CP experiment, which typically should
1
be of the order of 1-5 ms in H-15N CP experiments. Hence,
In the doped samples, T1F was also found to be nonexponential
(in the undoped sample the rotating frame relaxation time is so
long that only a lower limit could be determined). Two to three
exponential components are required to fit the rotating-frame
relaxation data. This nonexponential behavior may also be
attributed to heterogeneous domains with different radical
concentrations. This heterogeneity precludes a detailed discus-
sion of the data due to variability in the DNP enhancement
factors. Only the average values of T1F are given in Table 1.
These T1F values are approximated from a first-order decay law
from the time required for signal decay to its half-maximum
intensity and then divided by ln2.
it may be concluded that the doping levels should not exceed
approximately 1 wt % in order to minimize serious signal losses
in the CP experiment.
Figure 2 exhibits the 1H DNP enhancement curves of the four
doped samples. The enhancement curve presents the DNP
enhancement factor, E-1 (the DNP-enhanced signal minus the
thermal equilibrium signal)/ (the thermal equilibrium signal) as
a function of the applied microwave frequency. Extensive
reviews of the DNP mechanisms and the resulting enhancement
curves have been published elsewhere7-11,31 and the functional
forms of the enhancement mechanisms will not be repeated.
However, a brief summary of the main results on carbazole
provides valuable information on the factors contributing to the
enhanced signals.
Doping carbazole with BDPA reduces both the proton T1 and
T1F values by 2-3 orders of magnitude. The T1 reduction is

1
Carbazole H and 15N DNP Studies
J. Phys. Chem. A, Vol. 104, No. 19, 2000 4417
protons experience time-dependent electron-proton interactions
as well. It is possible that electron-electron spin-exchange
interactions are responsible for this effect. A third maximum
(e.g., approximately the same magnitude as the Overhauser
effect) is present in each curve at a frequency offset intermediate
between that of the solid state and the Overhauser contributions.
This component is due to the thermal mixing effect and results
from the EPR line being partially homogeneously broadened
so that polarization transfer can arise from a combination of
saturating the EPR line with the microwave irradiation, electron-
electron flip-flop interactions, and electron-proton dipolar
interactions.
1
To simulate the H DNP enhancement curves the combined
functional forms31 of all enhancement mechanisms should be
used. As a first approximation a set of Gaussian functions were
used to simulate the contributions of the visible and dominant
enhancement mechanisms. The origin of the fits were pairwise
locked to ωe ( ωs, ωe ( ωo, and ωe during the fitting process
to take account of the solid state (ωe ( ωs), direct thermal
mixing (ωe ( ωo), and Overhauser effect (ωe), respectively.
ωo is the frequency where the direct thermal mixing effect
becomes maximal. It is acknowledged that this simulation is
an approximation since the direct and indirect thermal mixing
mechanisms were not both individually included in the analysis.
The approximations employ only a semiquantitative analysis
of the mechanisms and relative contributions to the composite
shape of the enhancement curve to illustrate the complexities
of the several enhancement mechanisms. The enhancement
factors measured at ω ) ωe ( 60 MHz (the frequency of the
solid-state effect) of the four doped samples are given in Table
3.
Figure 3. Example of E-1 as a function of ω - ωe for (a) the
Overhauser effect where DNP is due to time dependent nuclear electron
interactions and (b), other interactions where DNP is due to fixed
paramagnetic centers; 1: solid-state effect; 2: direct thermal mixing
effect; 3: indirect thermal mixing effect; 4: overall enhancement. The
data for this figure were taken from the work of Wind et al.10
It is clear that the enhancement curves are essentially anti-
symmetrical around the electron Larmor frequency and become
maximal when the microwave frequency equals the sum or the
difference of the electron and the proton Larmor frequencies,
ω ) ωe ( ωH ) ωe ( 60 MHz, indicating that the major portion
of the electrons behave as fixed paramagnetic centers with
dominating static dipolar interactions with the protons. This
enhancement, referred to as the solid-state effect, is caused by
simultaneous electron-proton flip-flip and flip-flop transitions,
induced by the microwave irradiation. These transitions are
normally forbidden but become nonzero as a result of the
proton-electron dipolar interactions. The small nonzero en-
hancement observed at the electron Larmor frequency is due to
a small Overhauser effect, and indicates that a portion of the
Several interesting observations are apparent from the data
in Table 3. First, the largest contribution is that of the solid-
state effect which varies with the doping level, e.g., the
maximum value of 35 is observed at the 1.1% doping level and
decreases to 15 at the 5% level. Second, the frequency ωs at
Figure 4. The 15N DNP enhancement curve of 98% 15N labeled carbazole doped with 1.1 wt % BDPA as a function of microwave irradiation
frequency.

4418 J. Phys. Chem. A, Vol. 104, No. 19, 2000
Hu et al.
Figure 5. The room temperature 15N spectrum of 98% labeled carbazole, doped with 1.1 wt % BDPA, obtained with a single pulse experiment.
Top: normal single pulse experiment. The recycle delay time is 3000 sec and the number of scans is 48. The signal-to-noise ratio is 3.7. Bottom:
Single pulse DNP experiment. The microwave frequency was 11 MHz lower than the electron Larmor frequency. The recycle delay time is 3000
sec and the number of scans is 16. The signal to-noise ratio is 1980. In both spectra a Lorentzian line broadening of 25 Hz was applied.
which the curve has maximum intensity changes from 59.3 to
57.9 MHz as the doping level increases. This apparent frequency
shift is thought to be a manifestation of the change in relative
contributions of the solid-state mechanism (which decreases)
and that of the indirect thermal mixing mechanism which lies
upfield from the maximal point in the solid-state enhancement
curve and which (curve) overlaps with that of the solid-state
effect (see Figure 3).
This process is called DNP-CP;10 (2) directly, by enhancing the
15N polarization with the microwave irradiation without CP. In
the second instance the 15N signal is observed with a standard
90° pulse. This process is called single pulse DNP, or DNP-
SP. Both methods have advantages and disadvantages. For
instance, in DNP-CP the relatively short 1H T1 dictates the pulse
repetition rate, which shortens the measuring time of an
1
experiment. However, the interplay between the finite H-15N
1
contact time and H T1F can result in spectral distortions. This
As the doping level increases it is also noted that the half-
width of the solid state contribution increases while the
enhancement decreases. The thermal mixing enhancement was
simulated only in terms of the direct thermal mixing effect. The
enhancement factor increases from a value of 3 at the 0.65%
doping level and then is essentially constant at a value of
approximately 6. The Overhauser enhancement is estimated at
a value of 4 and appears to be independent of the doping level.
Hence, the estimated enhancement factors due to the Overhauser
and thermal mixing mechanisms appear to be largely indepen-
dent of the doping level. However, the contribution from the
solid-state effect is clearly related to the level of free radical
doping.
condition is avoided in a DNP-SP experiment. Moreover, in
materials that do not contain abundant spins, the DNP-SP
experiment is the only possible technique. For rare spins the
DNP enhancement factors are usually considerably larger than
those of protons10 and, as discussed below, this is also the case
for the 15N DNP enhancement factors in carbazole. The ultimate
consequence is a significant reduction in the number of spectral
scans needed to perform a successful DNP-SP experiment.
The 15N DNP enhancement curve, obtained via DNP-SP, for
the 98% 15N-labeled carbazole doped with 1.1 wt % BDPA is
presented in Figure 4. The curve is completely anti-symmetrical,
and the maximum enhancements occur at 39.133 ( 0.011 GHz.
It has been shown10 for rare spins that these maxima are mainly
due to the direct thermal mixing effect. To determine the
maximum 15N DNP enhancement factor, single pulse experi-
ments were performed on the same sample with and without
DNP. The results are presented in Figure 5.
2. 15N NMR and DNP NMR. In principle DNP techniques
can be used in two ways to enhance the 15N NMR signal: (1)
indirectly, by first enhancing the 1H signal, and then transferring
this enhanced polarization to the 15N nuclei by cross polarization.

1
Carbazole H and 15N DNP Studies
J. Phys. Chem. A, Vol. 104, No. 19, 2000 4419
TABLE 4: Carbazole 15N Shift Tensor Principal Valuesa
sample and experiment
δ11
δ22
δ33
δavg
DNP-CP natural abundance (1.4 T) -223 -263 -314 -266.7
DNP-SP, 15N labeled (1.4 T)
CP 15N labeled (4.7 T)
-227 -268 -309 -268.0
-233 -266 -306 -268.3
a Values are in ppm with respect to nitromethane.
time (about one minute in the DNP-CP case) since the recycle
delay time can be made relatively short. The DNP-CP enhance-
ment is sufficient that 15N spectra can be obtained in a relatively
short time (20 min) in unlabeled carbazole, as illustrated in
Figure 6c. This experiment was performed at 150 K. At this
temperature both the proton T1 and DNP enhancement factor
were the same as those observed at room temperature, but a
factor of 2 increase in sensitivity is obtained as a result of the
increased Boltzmann distribution.
Finally, the chemical shift tensor principal values of samples
doped at the 1.1% level are given in Table 4. The DNP-SP and
DNP-CP data (on a 15N labeled sample and natural abundance
sample, respectively) give a sense of the errors that might be
encountered in these two types of experiments. The maximum
difference in the principal values of these two experiments is 5
ppm while the isotropic shift differs by only 1.3 ppm. These
results are consistent within the experimental error for the
analysis of static powder pattern data. The standard CP
experiment was also performed on the 15N labeled sample at
4.7 T in order to acquire a sufficient S/N ratio to obtain principal
shift values of the nitrogen atom. These data are also given in
Table 4 and are within experimental error of that obtained by
DNP indicating that free radical doping does not significantly
distort the tensor shift values.
Figure 6. The 15N spectrum of carbazole, doped with 1.1 wt % BDPA,
obtained with CP and DNP-CP. (A) 98% 15N labeled carbazole, DNP-
CP. The microwave frequency was 60 MHz lower than the electron
Larmor frequency. The proton DNP enhancement is 35; recycle delay
time is 4 s; contact time is 0.5 ms; number of scans is 16. (B) 98% 15
N
Conclusions
labeled carbazole, CP without proton DNP enhancement. The recycle
delay time is 4 s; contact time is 0.5 ms; number of scans is 16. (C)
Natural abundance 15N spectrum, measured with DNP-CP at a tem-
perature of 150 K. The proton DNP enhancement is 33; recycle delay
time is 2 s; contact time is 0.5 ms; number of scans is 596. Different
line shapes of the powder patterns still give equivalent chemical shift
tensors.
It has been shown that doping polycrystalline carbazole with
BDPA via the solvent evaporation technique has several
advantages. First, doping reduces the nuclear relaxation time
and second, it presents the possibility of enhancing the nuclear
signals of both abundant and rare spins by means of DNP.
Proton DNP enhancements of about 35 have been obtained, a
SP 15N DNP enhancement of at least 930 has been measured,
and a comparable enhancement magnitude was found for the
13C nuclei in the doped carbazole samples. These data demon-
strate the feasibility of obtaining natural abundance 15N NMR
spectra in a reasonable time. For instance, with DNP-CP a
natural abundance 15N NMR spectrum with an adequate S/N
ratio could be obtained in only 20 min. Though the 15N
relaxation time in the doped carbazole sample is still long, such
a large 15N SP DNP enhancement factor (e.g., several hundred)
would be promising to allow the observation of nonprotonated
nitrogens at the natural abundance level in a MAS experiment.
Such 15N MAS experiments in coals, for instance, have been
very difficult40 since the proton T1F may be very short while
the cross polarization time for the nonprotonated nitrogens can
be much longer.1
For carbazole the optimum doping level is about 1 wt %
BDPA. At this level the proton T1 is reduced from about 800 s
to a very acceptable value of 2.2 s, whereas the proton T1F
becomes 3.4 ms, which is sufficiently large to allow cross
polarization experiments. At this doping level the proton DNP
signal enhancement becomes maximal and line broadening in
the rare spin spectra due to the presence of the radicals is not
observed. The absence of line broadening was confirmed by
comparing 13C and 15N CP and CPMAS data on doped (at the
1.1 wt % BDPA level) and undoped 98% 15N labeled carbazole
The signal-to-noise ratio (S/N) of the spectrum without DNP
is rather poor, resulting from a long 15N spin-lattice relaxation
time despite the doped sample. Therefore, only a few scans
could be taken in an acceptable measuring time. Actually, the
relaxation could be described rather accurately by exp[-(t/
T1N)1/2], which is typical for nuclei relaxed by paramagnetic
centers in instances where the spin diffusion among the nuclei
can be neglected.10 The nitrogen T1N was 340 s, and the recycle
delay had to be taken very long (3000 s) in order to reach
thermal equilibrium for the magnetization. It follows from the
observed S/N ratio that the maximum 15N DNP enhancement
1
factor is at least 930, about 25 times larger than the H DNP
enhancement. With this large enhancement it is possible to
obtain a 15N spectrum with excellent S/N with only one scan.
Similar DNP-SP experimental data were acquired from the 13
C
spectra of carbazole and a 13C-enhancement factor of ap-
proximately 800 was obtained.
Figure 6 shows 15N spectra of carbazole obtained via the
DNP-CP technique. Figure 6a and b displays the spectra of the
15N labeled sample, doped with 1.1 wt % BDPA, obtained with
and without proton DNP. These spectra illustrate the 15N signal
enhancement by the indirect 1H DNP enhancement factor. Even
though the 15N enhancement is considerably reduced in the
DNP-CP experiment (as compared to the DNP-SP experiment),
a spectrum with a high S/N ratio can be obtained in a short

4420 J. Phys. Chem. A, Vol. 104, No. 19, 2000
Hu et al.
(11) Wind, R. A. DNP and High-Resolution NMR in Solids. In
Encyclopedia of Magnetic Resonance; Grant, D. M., Harris, R. K., Eds.
John Wiley & Sons: London; 1996, p 1798.
(12) Duijvestijn, M. J.; Van der Lugt, C.; Smidt, J.; Wind, R. A.; Zilm,
K. W.; Staplin, D. C. Chem. Phys. Lett. 1983, 11, 25.
(13) Lock, H.; Wind, R. A.; Maciel, G. E.; Zumbulyadis, N. Solid State
Comm. 1987, 64, 41.
(14) Maresch, G. G.; Kendrick, R. D.; Yannoni, C. S.; Calvin, M. E.
Macromolecules 1988, 21, 3525.
(15) Singel, D. J.; Seidel, H.; Kendrick, R. D.; Yannoni, C. S. J. Magn.
Res. 1989, 81, 145.
on experiments carried out at 4.7 T. Within experimental error,
the chemical shift anisotropy and the isotropic chemical shifts,
as well as the line intensities and line widths, were the same in
the doped and undoped samples.
In conclusion, doping crystalline solids such as carbazole with
BDPA allows one to acquire rare spin NMR data on samples
that would otherwise be very difficult or impossible at a
magnetic field as low as 1.4 T without the enhancements
provided by DNP techniques.
(16) Afeworki, M.; McKay, R. A.; Schaefer, J. Macromolecules 1992,
25, 4084.
Acknowledgment. Support for this work was provided by
NEDO (New Energy Development Organization), Pacific
Northwest National Laboratory through contract 353465-A-5E,
and the University of Utah. Additional support was provided
by the Department of Energy under contract number DE-FG02-
94ER14452 from the Division of Chemical Sciences of the
Office of Basic Energy Sciences. RW was supported by an
internal grant from the Pacific Northwest National Laboratory,
a multi-program laboratory operated by Battelle Memorial
Institute for the U.S. Department of Energy under contract DE-
AC06-76RLO 1830.
(17) Afeworki, M.; Schaefer, J. Macromolecules 1992, 25, 4092.
(18) Afeworki, M.; Schaefer, J. Macromolecules 1992, 25, 4097.
(19) Afeworki, M.; Vega, S.; Schaefer, J. Macromolecules 1992, 25,
4100.
(20) Lock, H.; Maciel, G. E. J. Mater. Res. 1992, 7, 1.
(21) Li, L.; Qiu, J.; Hu, X.; Liu, Y.; Ye, C. Sci. China 1992, A4, 419.
(22) Zhou, J.; Li, L.; Hu, H.; Yang, B.; Dan, Z.; Qiu, J.; Guo, J.; Chen,
F.; Ye, C. Solid State NMR 1994, 3, 339.
(23) Hall, D. A.; Maus, D. C.; Gerfen, G. J.; Inati, S. J.; Becerra, L. R.;
Dahlquist, F. W.; Griffin, R. G. Science 1997, 276, 930.
(24) Gerfen, G. J.; Becerra, L. R.; Hall, D. A.; Griffin, R. G.; Temkin,
R. J.; Singel, D. J. J. Chem. Phys. 1995, 102, 9494.
(25) Becerra, L. R.; Gerfen, G. J.; Bellew, B. F.; Bryant, J. A.; Hall, D.
A.; Inati, S. J.; Weber, R. T.; Un, S.; Prisner, T. F.; McDermott, A. E.;
Fishbein, K. W.; Kreischer, K. E.; Temkin, R. J.; Singel, D. J.; Griffin, R.
G. J. Magn. Reson. 1995, A117, 28.
References and Notes
(26) Hu, J. Z.; Zhou, J.; Yang, B.; Li, L.; Qiu, J.; Ye, C.; Solum, M. S.;
Wind, R. A.; Pugmire, R. J.; Grant, D. M. Solid State NMR 1997, 8, 129.
(27) (a) Ganapathy, S.; Naito, A.; McDowell, C. A. J. Am. Chem. Soc.
1981, 103, 6011; (b) Bao-Lian, Y.; Ji-Wen, F.; Li, L.; Ye, C.; Hu, J. Z.;
Pugmire, R. J.; Grant, D. M. Acta Physica Sinica 1998, 7, 106.
(28) Details of the DNP setup and the DNP probe will be submitted
separately for publication.
(29) Wind, R. A.; Anthonio, F. E.; Duijvestijn, M. J.; Smidt, J.; Trommel,
J.; De Vette, G. M. C. J. Magn. Reson. 1983, 52, 424.
(30) Wind, R. A.; Hall, R. A.; Jurkiewicz, A.; Lock, H.; Maciel, G. E.
J. Magn. Reson. 1994, A110, 33.
(31) Duijvestijn, M. J.; Wind, R. A.; Smidt, J. Physica 1986, 138B, 147.
(32) Peterson M. A.; Nilsson, B. L. Synth. Commun. 1999, 29, 3821.
(33) Manuscript in preparation.
(34) Gerstein, B. C.; Hu, J. Z.; Zhou, J.; Ye, C.; Solum, M. S.; Pugmire,
R. J.; Grant, D. M. Solid State NMR 1996, 6, 63.
(1) Solum, M. S.; Altmann, K. L.; Strohmeier, M.; Berges, D. A.;
Zhang, Y.; Facelli, J. C.; Pugmire, R. J.; Grant, D. M. J. Am. Chem. Soc.
1997, 119, 9804.
(2) Duncan, T. M. A Compilation of Chemical Shift Anisotropies; The
Farragut Press: Chicago, 1990.
(3) Mason, J. Nitrogen NMR, In Encyclopedia of Nuclear Magnetic
Resonance; Grant, D. M., Harris, R. K., Eds.; John Wiley: London, 1996;
p 3222.
(4) Anderson-Altmann, K. L.; Phung, C. G.; Mavromoustakos, S.;
Zheng, Z.; Facelli, J C.; Poulter, C. D.; Grant, D. M. J. Phys. Chem. 1995,
99, 10 454.
(5) Strohmeier, M.; Orendt, A. M.; Facelli, J. C.; Solum, M. S.;
Pugmire, R. J.; Parry, R. W.; Grant, D. M. J. Am. Chem. Soc. 1997, 119,
7114.
(6) Hu, J. Z.; Facelli, J. C.; Alderman, D. W.; Pugmire, R. J.; Grant,
D. M. J. Am. Chem. Soc. 1998, 120, 9863.
(7) Abragam, A. The Principles of Nuclear Magnetism; Oxford
University Press: London, 1961.
(8) Goldman, M. Spin Temperature and NMR in Solids; Oxford
University Press: London, 1970.
(35) Blumberg, W. E. Phys. ReV. 1960, 119, 79.
(36) Lowe, I. J.; Tse, D. Phys. ReV. 1968, 166, 279.
(37) Tse, D.; Lowe, I. J. Phys. ReV. 1968, 166, 292.
(38) Wind, R. A.; Jurkiewicz, A.; Maciel, G. E. Fuel 1989, 68, 1189.
(39) Clarke, P. T.; Spink, J. M. Acta Crystallogr. 1969, B25, 162.
(40) Solum, M. S.; Pugmire, R. J.; Grant, D. M.; Kelemen, S. R.;
Gorbaty, M. L.; Wind, R. A. Energy Fuel 1997, 11, 491-494.
(9) Abragam A.; Goldman, M. Nuclear Magnetism: Order and
Disorder; Oxford University Press (Clarendon), London/New York, 1982.
(10) Wind, R. A.; Duijvestijn, M. J.; van der Lugt, C.; Manenschijn,
A.; Vriend, J. Prog. NMR Spectrosc. 1985, 17, 33.