Design and Use of Background-Reduced 27A1 NMR
Probes for the Study of Dilute Samples from the Environment
JANET S. MacFALL,* ANTHONY A. RIBEIRO,* GARY P. COFER,
KO-HSIU G. DAI, WILLIAM LABIOSA, BRUCE C. FAUST, and
D. D. RICHTER
Duke NMR Spectroscopy Center (A.A.R.) and Department of Radiology (J.S.M., A.A.R. G.P.C.), Duke University Medical Center,
Durham, North Carolina 27710; and School of the Environment (J.S.M., K.-H.G.D., W.L., B.C.F., D.D.R.), Duke University,
Durham, North Carolina 27708
Development of methods for the detection and measurement of aluminum
(AI) is crucial for our understanding of AI(III) chemistry and toxicity
in natural waters, soil solutions, and environmental samples. Traditional
colorimetric assays, by their very nature, alter solution AI(III) chemistry,
potentially biasing measurements. Methods based on VAl NMR spec-
troscopy have the advantage of being nondestructive and of not altering
the chemistry of the solution. Standard commercial NMR probes and
sample tubes, unfortunately, are constructed from aluminum-containing
components. These materials give substantial background signal, which
is detected as a large, broad hump, overwhelming signals from dilute
samples. We describe here the construction of two novel NMR probes
and a sample container built from a variety of materials with low AI
content. The designs feature the use of transversely mounted solenoid
coils with aluminum-free sample holders. The sample container features
a second chamber which can be filled with an external reference solution.
These novel ~7AI NMR probes are being used for the NMR spectroscopic
investigation and quantitation of natural, dilute (I0 ~ M) AI(III) samples
from the environment.
Index Headings: 27A! NMR; NMR probe background; Environmental
Some forms of AI(III) have been demonstrated to ex-
hibit toxic effects in aquatic and terrestrial ecosystems,
affecting fish, plants, mammals, and other organisms? -7
Chemical analysis of aluminum by complex formation 8-12
gives specific information on the complexed aluminum
species, but may not provide information on other AI(III)
forms present in the sample. Additionally, they alter the
chemistry of the sample as part of the method of detec-
27A1 nuclear magnetic resonance (NMR) spectroscopy
(spin 5/2, 100% natural abundance), in contrast, allows
nondestructive detection without alteration of the chem-
ical characteristics of the sample. However, while alu-
minum exists as the most abundant element in the earth's
crust, its speciation in natural (e.g., lakes, rivers) and
biological (e.g., plants, fish) samples occurs at dilute lev-
els, on the order of 10-5-10 -6 M. 6-12 This is about one
or two orders of magnitude below the sensitivity of most
commercial NMR probes. A second difficulty exists with
the use of most vendor-supplied NMR probes and sample
tubes for the 27A1 NMR spectroscopic examination of
dilute solutions. The commercial NMR probes and NMR
sample tubes are designed for multinuclear NMR and
Received 6 June 1994; accepted 17 October 1994.
* Author to whom correspondence should be sent.
contain appreciable amounts of aluminum. This circum-
stance results in a significant probe background envelope
that masks the desired signals from the natural AI(III)
sample. ' 3-, 6
Figure 1A illustrates the dominance of the probe back-
ground signal when a commercial multinuclear probe is
used to study a dilute AI(III) solution. This 27A1 NMR
spectrum was recorded over a 15,872-Hz spectral window
on a General Electric GN 300WB NMR spectrometer
with the use of a vendor-supplied 20-mm NMR probe
with 20 mL of 1 x 10 -4 M A1C13 in a standard Pyrex ®
borosilicate glass 20-mm NMR tube (Wilmad, Buena,
NJ). The probe background signal centered around 56
ppm dominates the NMR spectrum, while the resonance
of the hexaqua-Al(III) ion derived from the 10 -4 M A1C13
is the small signal at 0 ppm. On spectral integration, the
of the intensity of the commercial probe background. At
a 40,000-Hz spectral window, we found that, when the
signal from the AICI3 sample is phased for positive ab-
sorption, the probe background is reproducibly phased
to negative absorption. Since the AI(III) ion resonance
derives from the A1CI3 sample inside the coil, this finding
suggests that the background signal derives from alumi-
num components that are located outside the vendor-
supplied Helmholtz coil of the commercial probe. When
a 40-kHz spectrum was recorded with the sample and
tube removed, the background signal retained the same
negative phase, confirming that the background indeed
originates from aluminum contained in the vendor-sup-
plied probe. When a 40-kHz spectrum is acquired with
an empty 20-mm round-bottomed NMR tube (#20-9,
Wilmad, Buena, NJ) in the commercial probe, the back-
ground signal shifts up field by about 2 ppm, suggesting
the presence of additional detectable aluminum in the
glass NMR tube.
27A1 probe background signals are also seen with Bruker
WH400 probes, '4 Jeol GX-270 probes, '5 and Varian Uni-
ty 500 probes. '6 The probe background of these com-
mercial probes commonly extends from - 50 to 150 ppm.
It occurs exactly over the frequency range of tetra-, penta-,
and hexa-coordinate AI(III) species of biological and en-
vironmental interest.' 7.,8
Techniques to circumvent interference from the probe
background include (1) study of solutions with concen-
trations > 10 -3 M; (2) filtering the broad background sig-
nal by increasing the preacquisition delay from a few us
to 150-200 us; '5 (3) use of a spin-echo pulse sequence to
edit out the background; and (4) use of difference spectra
4 M AI(III) ion resonance is found to have about 0.2%
156 Volume 49, Number 2, 1995
© 1995 Society for Applied Spectroscopy
' I ' ' ~ I ' ' ' I ' ' ' I ' ' ' I ' ' ' I ' ' ' I ' ' ' I
100 80 60 40 20 0 -20 -40
FIG. 1. (A) 27A1 NMR spectrum of 1 × 10 -4 M pH 3.0 AICI 3 in water
acquired with a commercial 20-mm NMR multinuclear probe and 20-
mm glass NMR tube on a General Electric GN-300WB spectrometer
tuned to 78 MHz (8000 acquisitions, 15,872-Hz spectral width, 10-Hz
line broadening). The probe background centered near 56 ppm domi-
nates the NMR spectrum, while the 10 4 M AICI3 signal appears as the
low-intensity resonance at 0 ppm. (B) 27A1 NMR spectrum of 1 x 10 5
M pH 3.0 AICI3 (outer chamber) and 8 x 10 -4 M pH 13 AI(OD)4
(inner chamber) in water recorded with the use of capacitively tuned
solenoid probe and dual-chamber sample holder constructed with low-
aluminum materials (100,000 scans, 11,764-Hz spectral width, 10-Hz
line broadening). The 10 5 M AICI3 signal appears at 0 ppm and the 8
× 10 4 M AI(OD)4 signal appears at 79 ppm. The probe background is
now centered near 30 ppm and is of lower amplitude than the 10 -5 M
to digitally subtract out the probe background. Since sam-
ples of biological and environmental relevance generally
have solution A1 levels well below 10 -3 M, studies at high
concentrations are impractical for natural samples. Fil-
tering and spin-echo methods can sometimes be effective
to distinguish sharp sample resonances from broad back-
ground but fail when the background and resonance-
of-interest have similar linewidths. In particular, broad
resonances from AI(III) complexes with biologically im-
portant ligands (e.g., organic acids, phosphate, hydroxy
ligands) could go undetected. In difference spectroscopy,
the low levels of detectable AI in natural samples make
subtraction uncertain; incomplete background subtrac-
tion and "baseline roll" would result.
The goal of the present study was therefore to design
and construct NMR probes for 27A1 NMR spectroscopy
with reduced to undetectable probe background, and
maximal sensitivity for the detection of 27A1 in solution.
As a secondary goal, the probes would feature capability
to detect a reference solution while a sample was being
studied. The successful construction of such probes would
provide new tools for the study of AI(III) speciation in
dilute, natural samples collected from the environment.
Some aspects of this work have been previously present-
Fie. 2. Schematic diagram of the background-reduced 27A1 NMR probe
with inductively tuned solenoidal coil. The two small loops on either
side of the main coil are omitted in the capacitively tuned probe. The
dimensions of the schematic diagram are not to scale since the purpose
here is to illustrate the relative placement of the principal components
of the probe.
MATERIALS AND METHODS
A representative schematic diagram of our "back-
ground-reduced", higher-sensitivity NMR probes for the
detection of dilute AI(III) solutions is shown in Fig. 2.
The design and construction feature three primary com-
ponents, each containing as little contaminating detect-
able aluminum as possible: (1) a probe body replacing the
aluminum components of the housing in commercial
probes; (2) a radio-frequency (r0 coil designed for in-
creased sensitivity to 27A1 from the sample, but with re-
duced background; and (3) a sample holder specifically
machined for the rf coil with separate chambers for the
sample and reference, and which is nonreactive with the
sample and reference solutions. Our probes are con-
structed for use in a vertical 89-ram 7.0 Tesla supercon-
Material Selection. Material selection for probe con-
struction was based on neutron activation analysis (NAA).
The aluminum content of various plastics based on NAA
are compared to the literature values for glass and quartz
in Table I. NAA for the aluminum content of materials
was performed by the Nuclear Energy Service, Depart-
ment of Nuclear Engineering, North Carolina State Uni-
versity, Raleigh, North Carolina, with the use of an Ortec
GeLi detector coupled to an ND6700 Gamma detection
system. Materials for analysis were first soaked overnight
in a solution of 10- ~ N HCI, rinsed with deionized water,
air-dried, and then analyzed.
Probe Body. The central cylinder of the probe body
(Fig. 2) was constructed from polycarbonate (1.8 ug/g AI)
(Mobay Macralon extrusion grade, Atlantic Plastics, Ra-
leigh, NC) and Lucite ® (0.12/zg/g A1) (methyl methac-
APPLIED SPECTROSCOPY 157
TABLE I. Aluminum content of materials, a
0zg A1/g) Sample
Coming Glass 7740 b
Coming Glass 7052 b
Suprasil (synthetic quartz) d
" From Neutron Activation Analysis (see text).
b From Ref. 20.
c From Ref. 21.
d From Ref. 22.
0.7795 ± 10%
4.5091 +_ 2.5%
0.2626 _+ 13.9%
221.46 _+ 18.0%
0.9697 _+ 1.9%
0.6101 +_ 12.4%
1.7636 _+ 4.4%
0.1232 _+ 14.3%
rylate poly-acrylic, Atlantic Plastics, Raleigh, NC). The
base of the probe body was machined from Delrin ® (0.8
tzg/g AI) (Atlantic Plastics, Raleigh, NC). These materials
provide a robust and sturdy probe housing with a low
aluminum content. The housing has an interior passage
for cooling air, rf cables, and temperature-monitoring
equipment, and a flange on the base for reproducible
positioning and securing of the probe in the magnet. Two
inlets for the introduction of cooling air were placed in
the baseplate of the probe housing. In addition, a variable-
temperature thermocouple (part #483-000-100, General
Electric Corp., Fremont, CA) was placed in the probe
housing interior, with the thermocouple positioned just
below the rf coil. The electrical connections for the ther-
mocouple and the coaxial cable were mounted in the
baseplate of the probe body.
The rf coil (Fig. 2) was mounted on a tubular insert
constructed of TFE Teflon ® (0.26 ug/g A1) (Atlantic Plas-
tics, Raleigh, NC). The tubular insert holding the coil was
secured onto an interior Lucite ® tube (Fig. 2) that serves
as a coil cradle, positioning the coil transverse to the
direction of the magnetic field. An exterior polycarbonate
tube, parallel to the magnetic field and the magnet bore,
covers the coil cradle and serves as the top piece of the
Radio-Frequency Coil Construction. The vendor-sup-
plied probes for multinuclear NMR spectroscopy use a
Helmholtz (or saddle-shaped) coil where the main mag-
netic field (Bo) is parallel to the spinning z axis of the
sample. This design is convenient for removal of cylin-
drical samples from vertical superconducting magnets.
The B~ field of the Helmholtz coil is, however, not con-
fined to the interior of the coil geometry, and this coil
may be sensitive to end effects and susceptibility artifacts
from the solvent-to-air, solvent-to-glass, and coil-to-air
interfaces? 3,24 In contrast, the B~ field in a uniformly wound
solenoid coil, or, as in this case, a single-turn fiat-strip
solenoid coil, can be expected to not extend far past the
interior of the coil, as the B~ field exterior to the solenoid
decreases rapidly as the reciprocal of the radius squared.23,24
In addition, the B~ field within the confines of the solenoid
is very homogeneous. 23,24 The Helmholtz coil is also re-
ported to be inferior to the solenoid coil by a loss, by a
factor of two to three, in the potentially achievable signal-
to-noise (S/N) ratio. 23,24 A number of low-sensitivity nu-
clei such as 43Ca and 25Mg have been profitably studied
with the use of solenoid coils? 5
Given these considerations, we designed and built our
27A1 NMR probes with a solenoid coil mounted in a trans-
verse position to the primary magnetic field (Fig. 2). Since
our objective is to routinely study dilute AI(III) samples,
we constructed a solenoid coil with the maximum volume
possible in our 89-mm wide-bore vertical magnet. The
solenoid coil was assembled around a 3.6-cm-o.d. x 5.10-
cm cylindrical Teflon ® insert (Fig. 2) as described in the
probe housing section above. Two solenoid coil designs
were constructed, one with two variable, tunable air-trim-
mer capacitors and one which did not use the variable
Traditionally, solenoids are constructed with several
turns of wire. 26-29 In our design, we etched a copper-
Teflon ® laminate to construct a single-turn flat-strip so-
lenoid coil. A 3.1-cm × 1 1.0-cm sheet of 5-mil-thickness
Cu-Flon ® (copper-Teflon ® laminate) PTFE microwave
substrate (Polyflon, New Rochelle, NY) was etched to
entirely remove the copper from one side of the Cu-Flon ®.
This strip was then wrapped around the coil form, over-
lapping itself. The overlapped regions were bound to-
gether with two 10-60 pfvafiable air-trimmer capacitors
(Tfimtronics, Cazenovia, NY) soldered along the edge.
This construction created a coil for 78-MHz resonance
frequency with a tuning range of about 9 MHz.
In the second rfcoil design, the variable capacitors were
eliminated as a potential source of background aluminum
signal. A 3-cm × 1 1.0-cm sheet of 2-mil-thickness mi-
crowave substrate (Rogers Corporation, Microwave Ma-
terials Division, Chandler, AZ) was etched to remove
about two-thirds of the copper from one side of the strip,
leaving all the copper in place on the other side. This
strip was soldered to form a coil sliding freely on the coil
support. Two additional copper loops, each 1 mm in
width, were placed on the coil form on either side of the
main coil. One of these was attached to the coaxial cable
to serve as an inductively coupled pickup coil. Tuning
and matching were accomplished by adjusting the relative
distances between the two small copper loops and the
center coil on the coil form. This inductively coupled coil
had a tuning range of about 2 MHz. Inductive coupling
has previously been used in loop-gap resonators for
ENDOR 3° and in vivo 31p NMR.31.32 This design strategy
eliminates the need for additional variable capacitors
which contain additional aluminum.
Coaxial cables for connecting electronic components in
NMR probes feature aluminum braid as rf shielding. 33
To avoid pickup of aluminum signal from braid in the
coaxial cable, 33 we used an unshielded rf cable inside the
probe. To insulate against rf leakage, we wrapped the
entire probe housing in copper foil.
Sample Containers. In studying natural samples from
the environment, it is desirable to observe a standard
reference which would serve simultaneously as a chemical
shift and intensity standard. An initial sample holder (3.18-
cm o.d., 5.2-cm length) was constructed from high-den-
sity polyethylene (0.61 #g/g A1) with an outer chamber
volume of 20 mL and a simple coaxial inner reference
Volume 49, Number 2, 1995
chamber of 0.6 mL. This sample holder was used in the
solenoid coil with air-trimmer capacitors (capacitive tun-
ing) to obtain the 27A1 NMR spectra of Figs. 1B and 3A.
As a reference, a solution of pH 13 AI(OD)4- was used.
A second sample holder and cap were constructed of
milled FEP Teflon ® (0.26 ug/g A1) (Golden Rule Plastics,
Inc., Haw River, NC). In this design the cap was ma-
chined with a wedge-shaped, tapering, threaded connec-
tion which is tightened by hand to the sample holder,
creating a leak-proof seal. The cap has a center cylinder
which extends the length of the outer sample chamber.
We found that pH 13 AI(OH)4 reference solutions were
unstable on standing in the laboratory after several days
in the inner chamber of this Teflon® holder because of
reaction with atmospheric CO2 that diffused into the high-
pH AI(OH)4- reference solution. A second hollow cylin-
drical container was therefore constructed of a low-po-
rosity, ultra-high-molecular-weight polyethylene (Atlan-
tic Plastics, Raleigh, NC), with dimensions to fit into the
cap tube. This second container was filled with the ex-
ternal reference, sealed, and placed into the center cyl-
inder of the cap of the sample bottle. It was secured in
place by a high-density polyethylene screw closing the cap
center cylinder. This design keeps the sample solution
only in contact with Teflon ® surfaces and keeps the high-
pH [pH 13, AI(OH)4 or AI(OD)4-] reference solution
physically isolated from the sample solution and from
atmospheric CO> The possibility of the formation of
Al(III)/carbonate species within the external reference is
minimized with this configuration. The volume of sample
solution contained in the sample chamber was 18.7 mL.
The volume of the external reference solution in the insert
was 0.14 mL.
In the procedure to change samples, the probe is pulled
out of the magnet, the top outer polycarbonate tube of
the probe housing is removed, and the coil is refilled with
another sample bottle and retuned for resonance. We have
found sample changes to take about 15 min with this
Spectral Acquisition. Spectra were acquired on a Gen-
eral Electric GN300 spectrometer tuned to 78.199 MHz
at 22°C, without lock. Typical acquisitions included 36-
or 40-us pulses, with a 348-ms recycle delay, a 51-~s
readout delay, 8192 points, 11,764-Hz sweepwidth, and
10-Hz line broadening. The number of transients col-
lected varied between samples, depending on aluminum
Studies of Natural Samples. Surface waters were col-
lected from a mine drainage site near southeastern Ken-
tucky. Samples were stored in Teflon ® bottles at ambient
air temperature until analysis. Colorimetric, NMR, and
pH analyses were performed initially 20 days following
collection. Chemical analysis was done by 8-hydroxy-
quinoline complexation. 34,35
Fresh pine needles were harvested from a white pine
(Pinus strobus L.) growing on the Duke University cam-
pus. Needles were cut into lengths of approximately 1.5
cm and placed into the sample bottle. Spectra were ac-
quired as for solutions under study.
Coil Characteristics. Quality factors (Q), signal-to-noise
(S/N) data, and 90 ° pulse lengths for the 20-mm com-
mercial Helmholtz and two solenoid coils are presented
, , i , , , i , , , i , , , i , , , i , , , t , , , i , ,.q-q
100 80 60 40 20 0 -20 -40
FiG. 3. (A) 27A1 NMR spectrum of 1 x 10 s M pH 3.0 AIC13 (outer
chamber) and 8 x 10 -4 M pH 13 At(OD)4 (inner chamber) in water
recorded with the use of capacitively tuned solenoid probe and dual-
chamber sample holder constructed with low-aluminum materials
(100,000 scans, 11,764-Hz spectral width, 10-Hz line broadening). The
probe background has been removed by increasing the preacquisition
delay to 220 ~s. (B) Typical 27A1 NMR spectrum of 1 x 10 -6 M pH
3.0 AI(CI3) recorded with inductively tuned solenoidal coil and dual
sample holder constructed with low-aluminum materials (156,000 ac-
quisitions, 11,764-Hz spectra width, 10-Hz line broadening).
in Table II. Quality factors were measured unloaded for
the three coils. Under these conditions the capacitively
tuned solenoid measured at 650 and the inductively cou-
pled solenoid measured at 355, while the commercial
Helmholtz coil measured at 260.
For 90 ° pulse and S/N determinations with the capac-
itively tuned solenoid coil, the polyethylene sample bottle
was filled with ~20 mL ofpH 3.0 10 -3 M A1C13 solution
in the outer chamber, while the interior reference cham-
ber was left empty. The solenoid coil length is 3.1 cm
and the container length is 5.1 cm. Given that the interior
reference chamber of the bottle coaxially spans the coil,
the effective volume of the exterior sample chamber is
the annular volume bounded by the external radius of
the inner chamber (ri = 0.5 cm), the internal radius of
the outer chamber (r0 = 1.64 cm), and the coil length
(3.11 cm). This configuration calculates to an estimated
volume of ~ 13.5 mL within the coil in the outer chamber
of the sample container. Similar calculations with the
Teflon ® sample holder in the inductively coupled sole-
noid yield an estimated volume of ~ 11 mL within the
coil for the outer chamber. For work with the 20-mm GE
probe, a standard round-bottom 20-ram NMR tube was
filled to a height of ~ 55 mm, corresponding to a measured
volume of 13.5 mL. The standard 20-mm NMR tube has
an interior radius of 0.9 cm, and the height of the GE
Helmholtz coil is ~ 3.5 cm. The estimated volume of the
APPLIED SPECTROSCOPY 159
TABLE II. Dimensions and properties of aluminum coils.
active 90 ° S/N
Length volume Pulse"
(ram) (mL) Type (us) Q" S/N c volume d
10.05 ° 350 e 11.1 f 68 260 276 24.9
18 311 13.5 s 36 650 563 41.9
18 300 11.0~ 44 355 388 35.3
" The 90 ° pulse lengths were obtained from 180 ° null pulses on the same
stock of AICI3 solutions in the three coils at constant rf power in all
b The quality factor Q was measured at 78.2 MHz for unloaded coils
with probe shield in place using a tuning circuit assembled with a
reflectance bridge (Wiltron, Mountain View, CA), a calibrated sweep
generator (Wavetek 1080, Indianapolis, IN), and a display oscilloscope
(Tektronix 2201, Beaverton, OR).
c The root-mean-square signal-to-noise (S/N) was measured for a 10 3
M AIC13 in H20 in all cases. The S/N was obtained for 512 scans with
a 90 ° pulse, an acquisition time of 1.39 s, and a delay of 500 us. The
acquisition time is sufficient for full recovery of magnetization between
d S/N unit volume is calculated per milliliter.
° The radius and height, respectively, for the conventional General Elec-
tric mutinuclear 20-mm Helmholtz coil.
r The volume was estimated by measuring the volume of a commercial
20-mm cylindrical tube (round-bottomed) filled to a height of 55 mm.
This is the active volume inside the solenoid coil from the outer
chamber of the sample holders machined from polyethylene or Tef-
lon ® (see text).
A1C13 solution in the interior of the Helmholtz coil is
The 90 ° pulse lengths in all coils were measured from
spectra acquired of a stock solution of 10 3 M AICI3, pH
3.0, and were calculated from 180 ° null pulses. All pulse
lengths were measured at a constant rf power level (1 dB
below maximum power) with identical acquisition times
and pulse delays. The 90 ° pulse for the solenoid with two
air-trimmer capacitors calibrated at 36 us, and the sole-
noid constructed without variable capacitors calibrated
at 44 us, while the 90 ° pulse for the commercial Helmholtz
coil was achieved with a 68-~s pulse.
The same stock ofpH 3.0 10 -3 M A1C13 solution was
used for the S/N measurements. The S/N was measured
after 512 scans with the use of the respective 90 ° pulses
for each coil with an acquisition time of 1.39 s and a delay
of 500 us. The acquisition time is sufficient for full re-
covery of aluminum magnetization between pulses. Iden-
tical regions of noise (-30 to -40 ppm) were taken rel-
ative to the AI(H20)63+ signal at 0 ppm for the S/N
measurements. The measured S/N for the three coils is
listed in Table II. Since the volumes of the samples differ,
we calculate an S/N per unit volume for each coil. If B~
homogeneity is reasonably uniform over the same sample
volume (or unit volume), the calculations of Hoult 23,24
suggest a possible improvement in S/N by about a factor
of two for the use of a solenoid in comparison to a Helm-
~H NMR imaging provides a method to test the ex-
perimental Bl performance of an NMR coil. Since the
capacitively tuned solenoid coil could be tuned to 85
MHz, the outer chamber of the polyethylene container
was filled with distilled water, and ~H images were ac-
quired on a General Electric 2.0T horizontal CSI imaging
spectrometer. The ~H cross-sectional image showed little
variance in magnetic field across the sample, indicating
that the interior of this solenoid coil indeed had a very
uniform B~ field excitation profile. A second image across
the length of the coil and sample container showed a
decrease in signal intensity only near the bottom of the
sample bottle, indicating that the Bt homogeneity is quite
uniform over most of the sample bottle length. The drop-
off in Bl homogeneity at the end of the sample container
indicated by the ~H image is the behavior desired for
suppression of effects at the ends of the coil.
RESULTS AND DISCUSSION
The strategy pursued here (and reported earlier in pre-
liminary form ~9) to reduce or eliminate the A1 probe back-
ground is the systematic replacement of probe compo-
nents that contribute to the aluminum background. The
NMR instrument manufacturers routinely use anodized
aluminum in their probe housing, while our probe hous-
ing is constructed from materials specifically selected for
low aluminum content (see Materials and Methods). Sec-
ond, glass and/or quartz components that contain appre-
ciable amounts of aluminum oxide (Table I) are omni-
present in the commercial apparatus. The NMR sample
tubes commercially available are manufactured from Py-
rex ® borosilicate glass (Coming 7740). 36,37 Borosilicate
glass is also used in the inserts, coil supports, and tem-
perature-control Dewars of commercial NMR probes.
Coming 7740 borosilicate glass contains 22,000 gg/g A1
as aluminum oxide, while in other glasses (for example,
Coming 7052) the aluminum content may be as high as
77,000 #~g AI? °
These glass components were replaced by the low-alu-
minum materials in our first design, the solenoid with
capacitive tuning. As shown in Table II, this coil had a
very high quality factor, and essentially a factor-of-two
improvement in signal-to-noise performance in compar-
ison to the commercial 20-mm probe with Helmholtz
coil. Figure 1B illustrates the improvement provided by
the new probe. This figure shows the 27A1 NMR spectrum
obtained with a 20-mL solution of 1 × 10 -5 M A1CI3 in
the outer chamber and a 0.6 mL solution of 8 × 10 -4 M
AI(OD)4 in the reference chamber of the polyethylene
bottle. The probe background is now below the level of
the dilute 10 -5 M AI resonance. We estimate that the
probe background has been reduced by at least a factor
of 40 in comparison to results for the commercial probe.
At this very low level of probe background, filtering or
digital subtraction methods become effective for the re-
moval of the remaining probe background. Figure 3A
illustrates the removal of the residual probe background
from the capacitively coupled solenoid. The 10 -5 M A1C13
and 8 × 10 4 M AI(OD)4 resonances are now well re-
solved and can be phased with no "baseline roll". The
capacitively tuned probe thus allows the study of rela-
tively narrow A1 resonances with high sensitivity. A stan-
dard commercial probe has recently been modified by the
substitution of quartz (50 ug/g A1) zt in place of glass. 38
With the use of commercially available quartz NMR
tubes, 36 a reduction by a factor of about 30 in probe
background was achieved, and 10-2 M AI(III) signals could
160 Volume 49, Number 2, 1995
,~, 12.0 ', ' I ' ~ ' ' I ' ~ ." '
z 10.0 ..y"Y.. "i- .......... 'Y"
Z~ ..'"'"' 0."'"'" '
2.0 .,," /." pH = 2.00
AI(III) CONCENTRATION FROM MASS BALANCE (p.M)
2.0 4.0 6.0 8.0 10.0 12.0
FIG. 4. Plot of the linear relationship between integrated peak area for
the 27A1 NMR resonance of hexaquaaluminum and the aluminum con-
centration of defined solutions. This figure demonstrates the quantita-
tive ability of the solenoidal probes for direct measurement of aluminum
~q-I ' ' ' I
PP~ 100 80
be recorded with low probe background. 3s It is conceiv-
able that further reductions in the background of the mod-
ified commercial probe might be achieved by replacement
of natural quartz (50 ug/g A1) with synthetic quartz (fused
silica) such as Suprasil (0.1 ug/g A1) (Table I).
The remaining residual sources of background A1 signal
in capacitively tuned probes are in the probe electronic
components themselves. Most nonmagnetic trimmer ca-
pacitors available as popular choices in NMR probes fea-
ture annular or embedded glass or sapphire (aluminum
oxide) as their dielectric material? 9 Since these alumi-
num-containing capacitors are located near or on the NMR
coil, they easily contribute to background signal. The low-
er background in the modified standard probe was at-
tained with quartz capacitors with their lower residual
aluminum content. 38
Our second solenoid coil was therefore constructed
without commercial capacitors by using inductive
coupling 3°-32 for the probe tuning and matching. This in-
ductively coupled coil has an intermediate quality factor
and sensitivity between that of the capacitively tuned
solenoid and the commercial Helmholtz coil (Table II).
Nevertheless, the combination of low-aluminum mate-
rials and inductively coupled solenoid coil allows 27A1
NMR spectroscopy of aluminum solutions decreasing in
molarity to 1 uM with little to no interference from back-
ground signal. Figure 3B shows the 27A1 NMR spectrum
recorded with l0 -6 M A1Cla in the outer chamber and
the internal, sealed AI(OD)4- reference in the Teflon®
holder. The resonances are well resolved, with flat base-
line between sample and reference signal. This quality in
probe performance now allows direct quantitative sample
measurements to be made from the 27A1 NMR spectra of
I' ' ' I '
' ' I '
I ' '' }
FI6. 5. 27A1 NMR spectra of surface water sample collected from mine
drainage site with the use of solenoidal coil (11,764-Hz spectral width,
10-Hz line broadening). The first spectrum (140,322 acquisitions) with
a sample pH of 4.33 was collected immediately following arrival at the
laboratory. The second spectrum (118,744 acquisitions) was collected
seven days later, when the sample pH had decreased slightly to 3.74.
The reference chamber had a l0 #M Al equivalence to the sample
dilute solutions, as shown by the linearity between peak
areas and sample concentration in the uM region (Fig. 4).
Figure 5 shows 2VA1 NMR spectra of surface waters
collected from a mine drainage site. These spectra illus-
trate the low levels of aluminum that are NMR-detectable
and which can now be quantitatively compared with col-
orimetric analyses. Only hexaquaaluminum was detected
in the surface water samples by NMR spectroscopy. At
the time of the first NMR spectral acquisition (Fig. 5A),
the pH of the solution was 4.33. Integration of the sample
peak at 0 ppm and the reference signal [ 10 uM equivalent
AI(OD)4] showed the sample to be about 1.9 ~tM A1 in
concentration. This was less detectable A1 than was in-
dicated by colorimetric assay using 8-hydroxy- quino-
line, 34,3s which suggested a 9.34 tiM AI concentration.
After seven days, the pH of the solution had dropped
slightly to 3.74. Analysis by NMR at that time (Fig. 5B)
showed an increase in detectable aluminum, with the con-
centration measured to be about 3.3 uM.
As a test of the ability to detect broad aluminum res-
onances in natural samples, we acquired the in vivo alu-
minum NMR spectra of AI(III) within intact pine needles
(Fig. 6). No peak was observed at 0 ppm, indicating the
absence of hexaqua-A1 species. However, a broad peak
resonating in the region described for hexacoordinated
AI ligands can clearly be distinguished. The asymmetry
of the peak suggests the presence of two components, one
centered around 12 ppm and a larger peak centered around
16 ppm. Spectra acquired of dilute A1 (10 uM) and oxalate
(250 #M) solutions at low pH (pH 3.48) have shown a
APPLIED SPECTROSCOPY 161
' ' I ' ' ' I ' ' ' I ' ' '-I ~ ' ' I ~ ' ' I ' ' ' Download full-text
I ' ' ~
100 80 60 40 20 0
FIG. 6. 27A1 NMR spectrum recorded for intact needles of white pine
(Pinus strobus L.) with the use of solenoidal coil (133,200 transients,
11,764-Hz spectral width, 10-Hz line broadening). The reference had a
10/~M A1 equivalence to the sample.
single peak resonating at 12 ppm. 4° Several other peaks
have also been described for Al-oxalic acid complexes at
high molarity, one resonating around 12 ppm, one around
16 ppm, and one around 13 ppm, with the resonant fre-
quency dependent upon pH. 41 Other Al-organic acid li-
gands have also been detected in this resonant frequency
range. 41 These reports strongly suggest that the peak de-
tected from the intact pine needles is an Al-organic acid
ligand, and is likely to be Al-oxalate.
In conclusion, the results indicate that the use of low-
aluminum-containing materials allows the design and
construction of NMR probes that offer the NMR inves-
tigation of dilute AI(III) solutions with minimal to no
interference from probe background. The solenoid design
aids in both decreasing pickup of end effects and increas-
ing probe sensitivity for the NMR spectroscopic study of
solutions as dilute as 10 -6 M. It should be noted that,
even with removal of the interfering probe background
and improvements in sensitivity with novel coils, 27A1
NMR spectroscopy at the 10 -6 M level is appropriately
viewed as at the cutting edge of NMR detection. Each
27A1 NMR spectrum at the 10 -6 M level obtained here
with solenoidal probe in a wide-bore magnet represents
some 18-20 h of NMR signal averaging. Further gains in
sensitivity for dilute, natural samples would be realized
by using single-chamber sample containers. The solenoi-
dal probes appear to be valuable new tools for future
studies of environmental and biological aluminum.
This project is funded in part by grants from the Andrew Mellon
Foundation, the Duke University Instrumentation Fund, the Duke NMR
Spectroscopy Center, and the School of the Environment. A.A.R. is
supported in part by NIHP-30-CA-14236. NMR spectra were obtained
at the Duke University NMR Spectroscopy Center established with
funding from the NSF, the NIH, the NC Biotechnology Center, and
1. R. J. Bartlett and D. C. Riego, Soil Sci. 114, 94 (1972).
2. S. Ratkovie, Studia Biophys. 138, 173 (1990).
3. M. Vuletic and K. Kohler, Studia Biophys. 138, 185 (1990).
4. R. D. Handy and F. B. Eddy, J. Fish Biol. 34, 865 (1989).
5. J. P. Baker and C. L. Schofield, Water Air Soil Pollut. 18, 289
6. D.R. Parker, T. B. Kinraide, and L. W. Zelazny, Soil Sci. Soc. Am.
J. 53, 789 (1989).
7. D. Hunter and D. S. Ross, Science 25, 1056 (1991).
8. D. R. Parker and P. M. Bertsch, Environ. Sci. Technol. 26, 908
9. M. S. Shuman, Environ. Sci. Technol. 26, 593 (1992).
10. B. A. Browne, J. G. McColl, and C. T. Driscoll, J. Environ. Qual.
19, 65 (1990).
I I. B. A. Browne, C. T. Driscoll, and J. G. McColl, J. Environ. Qual.
19, 73 (1990).
12. F. Thomas, A. Masion, J. Y. Bottero, J. Rouiller, F. Genevrier,
and D. Boudot, Environ. Sci. Technol. 25, 1553 (1991).
13. J. S. Akitt, Prog. NMR Spectrosc. 21, 1 (1989).
14. R. Benn, A. Rufinska, E. Janssen and H. Lehmkuhl, Organome-
tallies 5, 825 (1986).
15. T. Nagata, M. Hayatsu, and N. Kosuge, Anal. Sci. 7, 213 (1991).
16. A. Ribeiro, unpublished observations.
17. J.J. Delpuech, in NMR ofNewly Accessible Nuclei, Vol. 2, P. Laszlo,
Ed. (Academic Press, New York, 1983), pp. 153-195.
18. J. W. Akitt, Multinuclear NMR, J. Mason, Ed. (Plenum Press, New
York, 1987), p. 259.
19. J. S. Macfall, A. A. Ribeiro, G. P. Cofer, K. H. Dai, B. C. Faust,
and D. D. Richter, paper delivered at the 33rd Experimental NMR
Conference, Asilomar, California (1992).
20. W. H. Kohl, Materials Technology for Electron Tubes (Reinhold
Publishing Corporation, New York, 1951), p. 4.
21. W. Primak, Compacted States of Vitreous Silica (Gordon and Breach,
New York, 1975), p. 38.
22. Heraeus-Amersil Technical Pamphlet "Optical Fused Quartz and
Fused Silica" (Heraeus-Amersil Company, Duluth, Georgia).
23. D. I. Hoult and R. E. Richards, J. Magn. Reson. 24, 71 (1976).
24. D. I. Hoult, Prog. NMR Spectrosc. 12, 45 (1978).
25. K. J. Neurohr, T. Drakenberg, and S. Forsen, in NMR of Newly
Accessible Nuclei, Vol. 2, P. Laszlo, Ed. (Academic Press, New York,
1983), pp. 229-252.
26. E. Oldfield and M. Meadows, J. Magn. Reson. 31, 327 (1978).
27. E. Oldfield and R. P. Skarjune, J. Magn. Reson. 31, 527 (1978).
28. B. Cook and I. J. Lowe, J. Magn. Reson. 49, 346 (1982).
29. S. B. W. Roeder, E. Fukushima, and A. A. V. Gibson, J. Magn.
Reson. 59, 307 (1984).
30. R. A. Venters, J. R. Anderson, J. F. Cline, and B. F Hoffmann, J.
Magn. Reson. 58, 507 (1984).
31. T. M. Grist and J. S. Hyde, J. Magn. Reson. 61, 571 (1985).
32. J. P. Hornak, T. L. Ceekler, and R. G. Bryant, J. Magn. Reson. 68,
33. Belden Electronic Wire and Cable Master Catalog #885 (Belden,
Richmond, Indiana), p. 9.
34. P. R. Bloom, R. M. Weaver, and M. B. McBride, Soil Sci. Soc.
Am. J. 42, 713 (1978).
35. B. R. James, C. J. Clark, and S. J. Riha, Soil Sci. Soc. Am. J. 47,
36. Wilmad Catalog #851 (Wilmad Glass Company, Buena, New Jer-
sey), p. 14.
37. Wilmad Glass and Quartz Tubing Catalog #801 (Wilmad Glass
Company, Buena, New Jersey).
38. L. S. Simeral, T. Zens, and J. Finnegan, Appl. Spectrosc. 47, 1954
39. Voltronics Catalog "Non-magnetic Precision Trimmer Capacitors'"
(Voltronics Corporation, East Hanover, New Jersey).
40. B.C. Faust, W. B. Labiosa, K. H. Dai, J. S. MacFall, B. A. Browne,
A. A. Ribeiro, and D. D. Richter, Geochem. Cosmochim. Acta,
41. F. T. Greenaway, Inorg, Chim. Acta. 116, L21 (1986).
162 Volume 49, Number 2, 1995