Anal. Chem. 1998, 70, 2454-2458
Design of Solenoidal Microcoils for High-Resolution 13 C NMR Spectroscopy Raju Subramanian and Andrew G. Webb*
Department of Electrical and Computer Engineering & Beckman Institute for Advanced Science and Technology, University of Illinois at UrbanasChampaign, 405 North Mathews Avenue, Urbana, Illinois 61801
Typical high-resolution natural-abundance 13C NMR experiments require milligrams of material. In this study, solenoidal microcoil direct and inverse detection probes have been designed to explore improvements in the limits of detection (LOD) for 13C NMR at 11.7 T. A directdetection radio frequency probe with an observe volume (Vobs) of 1 µL has been developed with carbon observe, deuterium lock, and proton decouple channels. The nonspinning line width of a decoupled [3-13C]-L-alanine (99%) peak achieved is 1.3 Hz. The natural-abundance 13C 3σ LOD for sucrose for a 90-min acquisition time is 18 µg (52 nmol). The 3σ LOD for [3-13C]-L-alanine (99%) is 375 ng (4.2 nmol) for an acquisition time less than 30 s. The inverse detection probe has proton observe and carbon decouple channels and a Vobs of 550 nL. The 3σ LOD for sucrose from a 1D 13C decoupled HMQC spectrum is 1.6 µg (4.5 nmol) for a 14-min acquisition time.
The signal-to-noise ratio (SNR) of an NMR experiment is proportional to the sensitivity of the radio frequency (rf) detection coil, defined as B1/i, where B1 is the transverse magnetic field produced per unit current.1 In general, as the coil diameter decreases the ratio B1/i increases, thereby increasing the coil sensitivity.1,2 The SNR per unit volume has been shown to increase to a size limited by current fabrication techniques. This yields a commensurate improvement in the limit of detection (3σ LOD, defined as SNR ) 3). Initially, the practical use of microcoil NMR was limited by large spectral line widths caused primarily by the magnetic susceptibility mismatch between the coil materials and the surrounding air.2-6 By using a susceptibility matching liquid to surround the rf coil region, Olson et al.7 obtained high(1) Hoult, D. I.; Richards, R. E. J. Magn. Reson. 1976, 24, 71-85. (2) Peck, T. L.; Magin, R. L.; Lauterbur, P. C. J. Magn. Reson. Ser. B 1995, 108, 114-24. (3) Odeblad, E. Micro-NMR in High Permanent Magnetic Fields; Nordisk Forening for Obsterik och Gynekologi, Karolinska Institute: Stockholm, Sweden, 1966. (4) Wu, N.; Peck, T. L.; Webb, A. G.; Magin, R. L.; Sweedler, J. V. Anal. Chem. 1994, 66, 3849-57. (5) Wu, N.; Peck, T. L.; Webb, A. G.; Magin, R. L.; Sweedler, J. V. J. Am. Chem. Soc. 1994, 116, 7929-30. (6) Wu, N.; Peck, T. L.; Webb, A. G.; Magin, R. L.; Sweedler, J. V. Anal. Chem. 1995, 67, 3101-7. (7) Olson, D. L.; Peck, T. L.; Webb, A. G.; Magin, R. L.; Sweedler, J. V. Science 1995, 270, 1967-70.
2454 Analytical Chemistry, Vol. 70, No. 13, July 1, 1998
resolution 1H NMR spectra with nonspinning line widths of 0.6 Hz on a 300-MHz spectrometer. The microcoil employed had an observe volume (Vobs, the sample volume enclosed by the rf coil) of 5 nL, and the 3σ LOD for sucrose with a 10-min acquisition time was 6.44 ng (18.8 pmol). This represented a mass sensitivity (defined as SNR per mol) enhancement of 130 relative to using a conventional 5 mm tube in a 10 mm commercial probe. Since then, developments in microcoil technology have included improving the concentration LODs,8-10 the use of a coil oriented at the magic angle with respect to the main magnetic field,9 incorporation of a lock channel for stable long-term acquisitions,10 and design of microcoils to operate at 500 MHz to exploit the higher chemical shift dispersion.10 Two review articles have been published recently.11,12 So far, microcoils have been designed only for 1H NMR. Heteronuclear techniques are an important component in determining full structural information on unknown compounds. In the present paper, we have designed microcoils to explore improvements in the LOD for 13C NMR. 12C and 13C are the two stable isotopes of carbon with a natural-abundance of 98.9% and 1.1%, respectively. Only the 13C nucleus is NMR detectable with a spin angular momentum quantum number, I ) 1/2. The SNR scales as γ3, where γ is the gyromagnetic ratio, which makes the 13C signal lower by a factor of 64 relative to 1H. Since the naturalabundance of 13C is only 1.1%, the overall 13C sensitivity is ∼6000 times lower than that for 1H. In addition, 13C spin-lattice (T1) relaxation times are typically much longer than those for protons which reduces further the relative sensitivity. Typically, milligram quantities of material are required for natural-abundance 13C NMR spectra with adequate SNR.13 The microcoils used in previous studies have a Vobs between 5 and 130 nL with coil fill factors of up to 57%.4-10 Due to the low relative sensitivity of 13C NMR, the Vobs needs to be increased while maintaining a high filling factor. We present here designs for 13C direct and indirect detection microprobes. The direct detection probe has rf circuitry for carbon observe, deuterium lock, and proton decouple channels with a Vobs of 1 µL. The inverse detection probe has a proton observe and carbon decouple channels with a Vobs volume of 550 (8) (9) (10) (11) (12)
Webb, A. G.; Grant, S. C. J. Magn. Reson., Ser. B 1996, 113, 83-7. Olson, D. L.; Lacey, M. E.; Sweedler, J. V. Anal. Chem. 1998, 70, 645-50. Subramanian, R.; Lam, M. M.; Webb, A. G. J. Magn. Reson., in press. Webb, A. G. Prog. Nucl. Magn. Reson. Spectrosc. 1997, 31, 1-42. Olson, D. L.; Lacey, M. E.; Sweedler, J. V. Anal. Chem. 1998, 70, 257A264A. (13) Gunther, H. NMR Spectroscopy, 2 ed.; John Wiley & Sons: West Sussex, England, 1995; p 469. S0003-2700(98)00299-6 CCC: $15.00
© 1998 American Chemical Society Published on Web 05/15/1998
Figure 1. 13C direct detection probe. (A) 13C-2H double-resonance circuit consists of a carbon observe and deuterium lock channels. The microcoil solenoid (L1 ) 130 nH) has a Vobs of 1 µL and is impedance matched to 50 Ω at 125.7 and 76.7 MHz. Only the AG coil portion of the 1H decouple circuit is shown. (B) The complete proton decouple circuit consists of a modified Alderman-Grant coil impedance matched at 499.89 MHz. The AG coil and the microcoil assembly are encased in a fluid container that contains a susceptibility matching fluid, FC-43. Typical values of the components in the rf microprobe are listed in Table 1, and the electrical characterization parameters are summarized in Table 2.
nL. The coil filling factor for both probes is 69%. EXPERIMENTAL SECTION Chemicals. The [3-13C]-L-alanine (13C, 99%) and D2O (D, 99.9%) were obtained from Cambridge Isotope Laboratories (Andover, MA) and the 200 proof dehydrated ethanol was from McCormick Distilling Co. (Weston, MO). Sucrose was purchased from Fisher Biotech (Fairlawn, NJ). All chemicals were used without further purification. Fluorinert FC-43 (3M Corp., St. Paul, MN), a nonconducting liquid, is a mixture of perfluorinated C12 branched tributylamines with an average molecular weight of 670.9 The volume magnetic susceptibility (χV) of FC-43 is within 15% of χV of copper. Carbon Direct Detection Circuitry. The direct detection microprobe consists of a solenoid double tuned to the 13C (125.7 MHz) and deuterium (76.7 MHz) frequencies and a modified Alderman-Grant (AG)14 coil tuned to the 1H frequency (499.89 MHz). The microcoil solenoid L1 shown in Figure 1A is fabricated manually, as previously described.7 L1 consists of 40 closely wound turns, 20 turns each of two parallel, 50-µm-diameter insulated (6.4 µm thick polyurethane coating) copper wire (California Fine Wire Co., Grover Beach, CA) wrapped on a 850µm-outer diameter (o.d.)/700-µm-inner diameter (i.d.) polyimidecoated fused-silica capillary (Polymicro Technologies Inc., Phoenix, AZ). The measured length of the solenoid (L1) is 2.5 mm giving a Vobs 9 of 1 µL. In the double-resonance circuitry displayed in Figure 1A, a chip capacitor (C1; 700A Series, American Technical Ceramics Corp., Huntington Station, NY) is placed in parallel with the coil and physically close to the microcoil (L1). This placement reduces the effects of inductance contributions from the microcoil leads. A grounded capacitor CT is placed on one side of this L1C1 tank circuit (described later). The other side of the L1C1 tank is split into two arms and the arms are tuned and matched to 13C and 2H frequencies, respectively. The 13C arm of the circuit consists of a 2H trap formed by the L3C3 parallel tank, a tuning capacitor CT,C, (14) Alderman, D. W.; Grant, D. M. J. Magn. Reson. 1979, 36, 447-51.
and a matching capacitor CM,C. Similarly, the 2H arm of the circuit consists of a 13C trap formed by the L2C2 parallel tank, a tuning capacitor CT,D, and a matching capacitor CM,D. The L3C3 and L2C2 parallel tanks present a high impedance at the 2H and 13C frequencies, respectively, and therefore eliminate the signal from the trapped frequency. The capacitors used are a combination of ceramic chip capacitors and/or small-variable capacitors (Gigatrim, 0.6-4.5 pF, Johanson Mfg. Co., Boonton, NJ). The inductors L2 and L3 are multiturned, made by winding 24 AWG insulated copper wire on a 7-mm-diameter screw. The L1C1 tank appears inductive at the 13C and 2H frequencies. For the 13C portion of the circuit in Figure 1A, CT, C is in series with CT. Tuning of the 13C frequency is dominated by CT, C, since CT, C , CT. The lock channel circuit has CT, D in series with CT, and the tuning of the 2H frequency is dominated by CT since CT , CT, D. The 1H decouple circuitry shown in Figure 1B consists of a modified AG coil, a variable tuning capacitor CT,H, and matching capacitors CM,H in a balanced network. The AG coil is a silverplated copper tube (outer conductor of semirigid coax UT 250A-SP; Micro-coax Components, Collegeville, PA), 6.35-mm o.d./ 5.59-mm i.d., 12.75 mm long with a 3.8 mm hole drilled through the center. On the lower ring of this tube, two symmetrical slits are formed just wide enough to accommodate single ceramic chip capacitors (C4). Values of the components used in Figure 1 are listed in Table 1. Probe Assembly. The rf circuitry was assembled on a 3.2 cm × 4.6 cm double-sided printed circuit board (PCB). Teflon flow tubes were attached to the microcoil former by shrink-melt Teflon sleeves (Small Parts Inc., Miami Lakes, FL). The AG coil was first soldered on to the PCB. The microcoil assembly is carefully inserted through the lateral holes of the AG coil until the carbon coil is in the center of the decoupler coil. The capillary is affixed using Quiktite glue (Loctite Corp., Rocky Hill, CT) onto the grooves carved out on posts on either side of the AG coil. The microcoil leads (∼4 mm long) are then soldered on to the PCB. Both coils are encased within a 10-mL polyethylene bottle Analytical Chemistry, Vol. 70, No. 13, July 1, 1998
2455
Table 1. Typical Values of Components Used in the 13C Direct and Inverse Detection rf Microprobes
L1 L2 L3 CM,C CT,C CM,D CT,D
Direct Detection Probe 130 nH C1 18 nH C2 45 nH C3 8.6 pF C4 4.2 pF CT 68 pF CM,H 120 pF CT,H
L4 CM,C CT,C C6
Inverse Detection Probe 45 nH C5 18 pF CM,H 87 pF CT,H 3.9 pF
2.2 pF 82 pF 100 pF 3.3 pF 24.2 pF 5.1 pF 6.2 pF 100 pF 3.5 pF 3 pF
Table 2. Electrical Characterization Parameters for the Direct and Indirect Detection rf Microprobes channel
2H 13C 1H 13C 1H
S11 (dB)
S12 (dB)
Q
Direct Detection Probe -32 -30 -34 -29 -28 -40a
44 71 81
Inverse Detection Probe -39 -41 -40 -21
74 61
a The S 13C frequency with the carbon and proton 12 measured at channels connected to the network analyzer.
which is filled with the FC-43 susceptibility matching fluid. The L2C2 and L3C3 parallel tank traps are tuned independently to 13C (125.7 MHz) and 2H (76.5 MHz) frequencies, respectively, before soldering on to the PCB. The rest of the circuit is completed as described earlier with all the components on one side of the PCB. The other side of the PCB has the copper layer intact thereby providing a good rf ground. The finished rf microprobe is mounted on a custom-built platform, which is housed inside an aluminum outer shell. Electrical characterization was performed using a two-port network analyzer (8751A Hewlett-Packard, Palo Alto,CA). When tuned and impedance matched to 50 Ω, the measured input reflection coefficient (S11) was -34 and -32 dB for the carbon and deuterium channels, respectively. This means that less than 0.1% of the power going into the probe is reflected back at their respective frequencies. The measured reverse transmission coefficients (S12, S21) yielded an isolation of -29 and -30 dB. The measured S11 for the 1H channel was -28 dB. Q values were computed by measuring the frequency points (f1, f2) at which the real and reactive components were equal on a Smith chart.15 The measured Q values were 44, 71, and 81 for 2H,13C, and 1H, respectively. The rf microprobe parameters are summarized in Table 2. Inverse Detection Probe. The inverse detection probe consists of an inner solenoid tuned to the 1H frequency and an outer modified AG coil tuned to the 13C frequency: the circuitry is shown in Figure 2. The solenoid L4 consists of 10 turns of 125 µm × 50 µm rectangular insulated copper wire on a 850-µm-o.d./ (15) Application Note 154; Hewlett-Packard Co., Palo Alto, CA, 1972.
2456 Analytical Chemistry, Vol. 70, No. 13, July 1, 1998
Figure 2. Inverse detection probe consisting of proton observe and carbon decouple channels. The microcoil (L4 ) 45 nH) has a Vobs of 550 nL and impedance matched to 50 Ω at 499.89 MHz. The modified AG coil is tuned and matched to 125.7 MHz. As in the direct detection probe, the AG coil and the microcoil are encased in a container that contains a susceptibility matching fluid (FC-43). Typical values of the components in the rf microprobe are listed in Table 1, and the electrical characterization parameters are summarized in Table 2.
700-µm-i.d. polyimide-coated silica capillary, is 1.4 mm long and has a Vobs of 548 nL. The carbon AG coil is the same as the proton AG coil in Figure 1B but with 100 pF C5 capacitors. Both the 13C and 1H are balanced circuits and are assembled and characterized in the same way as the direct detection probe. Typical values for the components in Figure 2 are listed in Table 1, and electrical measurements are summarized in Table 2. NMR Spectroscopy. NMR experiments were carried out on a Unity-500 spectrometer (Varian NMR Instruments, Palo Alto, CA) with an 11.7 T (500 MHz), 51-mm-bore magnet (Oxford Instruments, Oxford, England). Data were processed on a PC platform using the NUTS software package (Acorn NMR Inc., Fremont, CA). To attain the optimum resolution with a new rf probe, it is critical that the microcoil be at the horizontal and vertical center of the magnet and shim stack. The procedure for microprobe positioning has been previously described.9 RESULTS AND DISCUSSION Direct Detection Probe. Initial experiments to evaluate the 13C observe, 2H lock, and 1H decoupler channels in the direct detection probe were performed using a 230 mM solution of [3-13C]-L-alanine (99%) in D2O. The Vobs was manually shimmed on the lock signal from the solvent. Starting from zero shims, Z1 (coarse), Z2 (coarse), X, and Y shims contribute to over 95% of the increase of the lock level with the remainder contributed by higher-order shims. Typical 90° pulse lengths for the direct detection probe were 1.4 and 3.2 µs for the 13C and 1H channels with 8 and 0 dB transmitter attenuations (amplifier rated outputs were 300 and 50 W, respectively). A 13C spectrum of [3-13C]-Lalanine is shown in Figure 3A. Proton decoupling was performed using WALTZ-1616 and the 1H-decoupled spectrum is shown in Figure 3B. The 13C line width of the methyl carbon in Figure 3B was measured to be 1.3 Hz and has a Lorentzian line shape. After shimming, pulse width, and decoupler strength determinations, distilled water is injected through the flow tubes followed by the solution of interest. (16) Shaka, A. J.; Keeler, J.; Freeman, R. J. Magn. Reson. 1983, 53, 313.
Figure 3. 13C spectra of 230 mM [3-13C]-L-alanine (99%) in D2O acquired with the direct detection microprobe. (A) coupled spectrum (32 scans); (B) 1H-decoupled (8 scans). The line width of the methyl peak is 1.3 Hz. The LOD for a 3-min acquisition time is 375 ng (4.2 nmol). Data parameters: 725 Hz spectral width; matched filter; 1000 complex points; 2.9-s prepulse delay.
Since the susceptibility matching liquid (FC-43) is a mixture of perfluorinated alkanes, it is important to determine whether it gives rise to any contaminating signal. A 13C spectrum with an empty coil was acquired (1024 scans, 30-kHz spectral width; 4.3-s prepulse delay) and no signal above the noise floor was detected. It is well-known that transferring polarization from protons to carbon nuclei can increase the sensitivity of direct 13C detection and there are a number of techniques that have been developed for this purpose.17 One of the versatile schemes that is used is the distortionless enhancement by polarization transfer (DEPT) pulse sequence.18 The pulse sequence is 90X(1H)-(1/2J)180X(1H), 90X(13C)-(1/2J)- θ(Y(1H), 180X(13C)-(1/2J)-decouple(1H), acquire(13C), where J ) 140 Hz is the average 13C-1H scalar coupling constant. The DEPT sequence gives in-phase peaks, and the proton multiplicities of carbon atoms (i.e., CH, CH2 or CH3) can be determined by using different values of θ. The theoretical maximum sensitivity gained from the polarization transfer is given by γ1H/γ13C ()4). Also the spin-lattice relaxation time (T1) of the spin system is now determined by the T1 of the protons, which is typically shorter than that for 13C nuclei. DEPT (θ ) 45° and 90°) spectra of 500 mM sucrose in D2O were acquired by coaddition of 1024 scans with an acquisition time of 1.5 h/spectrum. The spectra are displayed in Figure 4 and agree well with previously published data.19 Figure 4A acquired with θ set to a 45° pulse consists of peaks from all protonated carbons (CH2 and CH groups in sucrose). Figure 4B displays the spectrum acquired with θ set to a 90° pulse and consists of peaks from only CH (17) Rahman, A.; Choudhary, M. I. Solving Problems with NMR Spectroscopy; Academic Press: San Diego, 1996; pp 91-121. (18) Dodrell, D. M.; Pegg, D. T.; Bendall, M. R. J. Magn. Reson. 1982, 48, 3237. (19) Pouchert, C. J.; Behnke, J. Aldrich Catalog of NMR Spectra; Aldrich Chemical Co.: Milwaukee, WI, 1993; pp 1-305C.
Figure 4. Sucrose (500 mM) in D2O. DEPT θ ) 45° spectrum (A) consists of CH and CH2 peaks; DEPT θ ) 90° spectrum (B) consists of CH peaks only. The LOD for a 90-min acquisition time is 18 µg (52 nmol). Data parameters: 1024 scans; 7000 Hz spectral width; matched filter; 8000 complex points; 5.2-s prepulse delay.
protonated carbons. The maximum SNR with a matched filter20 applied before the Fourier transform in Figure 4A is 29:1, which corresponds to a 3σ LOD of 18 µg (52 nmol) for the 1.5-h data acquisition period. To improve LODs, 13C labeling is possible; however, 13C labeling is expensive and often difficult. Given the higher coil sensitivities, a potential advantage of microcoils is that only small quantities of 13C-labeled materials need to be synthesized. As an indication of the LOD for 13C labeled compounds, the SNR for the [3-13C]-L-alanine (99%) peak shown in Figure 3B is 216:1. This corresponds to an LOD of 375 ng (4.2 nmol) for an acquisition time of 24 s. The LOD could be decreased to 1 nmol by application of polarization-transfer schemes such as DEPT. It should be noted that if only a single 13C-labeled resonance is to be detected, the spectral width (SW) can be reduced considerably with an increase in the SNR proportional to 1/(SW)1/2. In Figure 3, the spectral width is only 725 Hz, accounting in part for the reduced LOD. Inverse Detection Probe. An inverse detection probe in NMR is used to perform heteronuclear experiments in which the higher γ nucleus is detected, but the signal encodes information related to the lower γ nucleus. Higher SNR is achieved by detecting protons, but the major challenge is in suppressing signals arising from 12C bound protons. Since this probe does not contain a lock channel, it was manually shimmed on the 1H signal from a 10% H2O/90% D2O (v/v) mixture. The 90° pulse lengths for the 1H and 13C channels were 0.9 and 11 µs with 8 and 3 dB transmitter attenuations, respectively. To calculate typical LODs from an inverse detection experiment, a 1D version of the 1H detected heteronuclear multiple(20) Derome, A. E. Modern NMR Techniques for Chemistry Research; Pergamon Press Ltd.: New York, 1987; p 24.
Analytical Chemistry, Vol. 70, No. 13, July 1, 1998
2457
is fully suppressed. The line width is 1.6 Hz and the multiplets are well resolved. The SNR in the HMQC spectrum of ethanol is 583:1, and this corresponds to a 3σ LOD of 2.2 µg (48.4 nmol) for a 3-min acquisition. A 13C decoupled 1D HMQC spectrum of sucrose in D2O (500 mM) was also acquired (512 scans; 1.6-s prepulse delay; not shown). The SNR was 182:1, yielding a 3σ LOD of 1.6 µg (4.5 nmol) for the 14-min acquisition time.
Figure 5. Neat ethanol spectra acquired with the inverse detection probe. (A) 1D HMQC spectrum (128 scans; 2000 complex points; 1.6-s prepulse delay; 300 ms τ delay). The LOD for the spectrum with a 3-min acquisition time is 2.2 µg (48.4 nmol). A minimum of 128 phase-cycled scans is required to achieve adequate suppression of the signal from 12C attached protons. (B) 1H spectrum (1 scan; 8000 complex points). Data parameters: 3000 Hz spectral width; 1 Hz line-broadening.
quantum coherence (HMQC) pulse sequence was used.21-24 The pulse sequence used is BIRD-τ-90X(1H)-(1/2J) -180Y(1H), 90X(13C), 90-X(13C)-(1/2J)-acquire(1H). The bilinear rotation (BIRD) pulse inverts the magnetization of protons bound to 12C and at the end of the delay period τ () 300 ms) the inverted magnetizations go through a null point thereby suppressing signal from protons not bound to 13C. Appropriate phase cycling24 of 90X(13C), 90-X(13C) pulses causes the satellite signals from 13C attached protons to grow, whereas the 12C attached proton signals cancel by subtraction. Theoretically, the HMQC experiment is 64 times more sensitive than conventional 13C direct detection experiment without polarization transfer. Figure 5A shows the 1D HMQC spectrum of neat ethanol acquired in 3 min, and it represents the resonances arising from 13C directly attached to protons. For comparison, a single-scan standard 1H spectrum is displayed in Figure 5B. A 128-step phase cycle was required in the HMQC experiment to achieve a satisfactory suppression of the signal from 12C attached protons. The 1H signal from 12C attached protons in the methyl and methylene groups was only 9% and 2% that of the 13C attached protons and the hydroxyl peak (21) Freeman, R.; Mareci, T. H.; Morris, G. A. J. Magn. Reson. 1981, 42, 3415. (22) Muller, L. J. Am. Chem. Soc. 1979, 101, 4481-4. (23) Summers, M. F.; Marzilli, L. G.; Bax, A. J. Am. Chem. Soc. 1986, 108, 4285-94. (24) Bax, A.; Subramanian, S. J. Magn. Reson. 1986, 67, 565-70.
2458 Analytical Chemistry, Vol. 70, No. 13, July 1, 1998
CONCLUSIONS This study demonstrates that solenoidal microcoils can be successfully applied to high-resolution natural-abundance 13C NMR. The rf probes used in the present study exploit the enhanced sensitivity of microcoils, which is achieved by scaling the dimensions of the NMR detector to that of the small sample size. The direct detection probe has a high coil filling factor (69%) and a Vobs of 1 µL, which is ∼50 times smaller than a 3-mm commercial probe. An innovative double-parallel-wound microcoil circuitry design optimizes the efficiency of the deuterium lock and 13C observe channels. The line width achieved was 1.3 Hz, which is comparable to nonspinning values in commercial probes. The presence of a decouple channel allows for not only proton decoupling but also sensitivity enhancement by polarization transfer. The inverse detection probe demonstrated here also has a high coil filling factor (69%) and a Vobs of 550 nL. The 3σ LOD for natural-abundance direct carbon detection was calculated to be 18 µg for an acquisition time of 90 min. Therefore, useful natural-abundance 13C NMR could be obtained from a few tens of micrograms of sample compared to the milligram quantities required by conventional NMR. 13C enrichment reduces this LOD to hundreds of nanograms. Further reductions in LOD are possible by using a better susceptibility matching liquid since diffusion in susceptibility gradients caused by the 15% susceptibility mismatch between the χV of copper and FC-43 makes a significant contribution to line widths. The LOD could also be reduced by a factor of ∼2 by performing the experiments at a higher external magnetic field of 17.6 T (750 MHz). ACKNOWLEDGMENT This work was supported by a grant from the National Institutes Health (PHS 1 R01 GM53030-01). Spectra were obtained using facilities provided by the Biomedical Magnetic Resonance Laboratory, Shared Instrumentation Grant 1S10RR06243, and the Biomedical Research Technology Grant PHS5 P41 RR05964. We acknowledge the assistance and technical support of Sam Grant, Michael Lam and Babak Behnia (Magnetic Resonance Engineering Laboratory), and Drs. Tim Peck and James Norcross (Magnetic Resonance Microsensors Corp., Savoy, IL). Received for review March 16, 1998. Accepted April 20, 1998. AC980299S
