Comparison of 10 mm and 30 mm Diameter Surface Coil Resonators for 1 GHz EPR

Abstract

Two surface coil resonators were tested at 1.0 GHz as alternatives to volume resonators that fully contain the sample; a 10 mm diameter coil with Q of 61 and efficiency of 0.044 mT/√W at ~2.5 mm above the coil and a 30 mm diameter coil with Q of 56 and efficiency of 0.0038 mT/√W at ~5 mm above the coil. The 10 mm diameter coil uses power more efficiently, but the 30 mm diameter surface coil detects signal from a larger volume and the B₁ extends further into the sample. The signal intensity as a function of distance from the coil was measured for small samples of lithium phthalocyanine, a 1 mm thick plate of aqueous nitroxide solution, and a 7 mm internal diameter tube containing aqueous nitroxide solution. For a sample localized at a defined distance from the coil the signal intensity can be increased by increasing incident power to compensate for the decreases in B₁ with distance from the coil. However, when the noise is dominated by the source, increasing power increases noise. If incident power is adjusted to compensate for the decrease in B₁ with increasing distance from the coil the S/N for aqueous nitroxide in a 1 mm thick plate, was better for the 10 mm coil than for the 30 mm coil up to about a distance of about 6 mm but that advantage is lost at greater distances from the coil. Samples of nitroxide were used as a phantom to demonstrate 3D spatial imaging with the 30 mm coil.

1. Introduction

For in vivo EPR (electron paramagnetic resonance), L-band (~1 GHz) provides a reasonable tradeoff between increasing signal-to-noise (S/N) and decreasing penetration depth as RF (radio frequency) increases [1–4]. Rapid-scan EPR provides substantial improvements in S/N relative to CW [5]. We have therefore modified our 700 MHz table-top rapid scan spectrometer [6] to operate at about 1.0 GHz [7]. Although volume resonators that fully contain the sample typically are used in EPR, surface coils have potential advantages for spectroscopy and imaging of objects that are too large to fit in a resonator and/or for targeted study of a localized region of a larger object. An early example of the use of a surface coil for EPR is described in Nishikawa et al. [8]. They used a 7 mm diameter coil operating at 1.8 – 1.9 GHz and detected the signal from a nitroxide solution. Surface coil EPR experiments subsequently were reported by Bacic et al. [9] and Nilges et al. [10]. The distribution of the magnetic field B₁ in the vicinity of a surface coil is nonuniform. For a dielectric surface coil the signal intensity in a 30.7 mm OD surface coil was mapped in the plane and perpendicular to it [11]. They showed that the signal intensity falls off exponentially with distance from the surface of the coil. EPR imaging has been used to map the spatial dependence of B₁ for a 9 mm i.d. surface coil operating at 1.05 GHz [12]. A surface dielectric resonator with high sensitivity for lossy samples has recently been reported [13, 14].

At any location the EPR signal generated by a sample is proportional to B₁. With a surface coil the B₁ decreases with distance from the coil. Also, the EPR signal detected at the surface coil decreases with distance of the sample from the resonator with the same dependence on distance as the B₁, so when the position of the sample is varied, the detected signal intensity is proportional to B₁² [15]. This dependence of signal intensity on B₁² is an example of reciprocity [15]. In a discussion of ¹³C NMR flip angles this reciprocity has been stated as “the MR signal has a dual dependence on the coils magnetic field pattern” which has been confirmed by comparison of experiment and theory [16].

The current in a single-wire loop is a simple model for a surface coil. Equations to describe the magnetic fields generated by a current in the coil are provided in elementary texts under the heading of the Biot-Savart law. Along the axis perpendicular to the plane of the loop, the magnetic field decreases with distance as described by the formula:

$$B_1\propto \frac{R^2}{{\left(d^2+R^2\right)}^{3/2}}$$ (1)

• R = Radius of the coil

• d = distance from the plane of the coil

The relationship in Eq (1) is valid only along the axis of the coil. Above the plane of the resonator the B₁ field lines generated by the current in the coil diverge, which complicates prediction of signal intensity for finite-size samples. To gain a practical understanding of signal intensity for finite-size samples as a function of distance above the 10 mm or 30 mm coils experimental data are reported for three sample geometries – small samples of lithium phthalocyanine (LiPc), a 1 mm layer of aqueous nitroxide solution and a cylindrical tube of aqueous nitroxide. Results are compared with predictions based on Eq. (1). For some applications such as an OxyChip [17, 18] or a localized injection of a redox sensor [19], the experiment might involve adjusting incident power to achieve the desired B₁ for a signal at specified distance from the resonator. However, the signal would still be attenuated due to the decrease in intensity with increasing from the distance from the resonator. Increasing resonator diameter increases penetration depth but decreases resonator efficiency. At microwave magnetic field strengths (B₁) that do not saturate a nitroxide radical, the required power from the rf source may fall within a range where source noise is the dominant contribution. A comparison is made of the S/N for an air-saturated nitroxide solution as a function of distance from a 10 mm or a 30 mm surface coil with incident power adjusted to maintain constant B₁ at the position of the sample.

2. Methods and Materials

2.1. Resonator Construction and Characterization.

The 10 mm inner diameter surface coil was made from 1 mm square copper wire. The coupling loop is similar and spaced 2 to 3 mm away. The 30 mm inner diameter resonator was made with copper ribbon 0.5 mm thick and 4 mm wide. The 27 mm inner diameter coupling loop was made with 1 mm square wire and spaced 2.5 mm away. For both resonators the coupling loop is attached to coaxial cable that is mounted on a slider with an adjusting screw to permit changing the coupling. The resonant frequency ω of a coil is given by ω = 1/√(LC) where L is the inductance and C is the capacitance. For the 10 mm and 30 mm resonators the inductance is about 15 nH or 75 nH, respectively. To avoid variations in B₁ around the coil the circumference needs to be segmented into lengths that are small relative to the wavelength. In practice this means segments no larger than about 1/8 of the wavelength. Capacitors were used to break the coils into acceptably small segments and to adjust the resonant frequency. The 10 mm coil had one 1.6 pF capacitor and the 30 mm coil had four 1.5 pF capacitors bridging gaps of about 0.3 mm (Figure 1). The capacitors are non-magnetic chips made by Voltronics, which was bought out by Knowles. The resonators are made with HIPS plastic that was selected because of its low electrical loss. The resonators were mounted in 3D-printed PLA plastic holders (Figure 2) that fit snugly within the bore of an air core magnet [6]. The center of the 30 mm resonator is filled with HIPS plastic to provide a support for samples. Samples are placed parallel to the plane of the resonators. The plastic support center the resonators in the magnetic field with the plane of the surface coil in the xz plane of the gradient coils. The z axis is along the bore of the magnet.

Figure 1.

Capacitors in the surface coils. 10 mm coil (left) with one capacitor and 30 mm coil (right) with four capacitors.

Figure 2.

Surface coil resonators in mounts - 10 mm coil (left) and 30 mm coil (right). The picture of the 10 mm resonator shows the coupling structure. Each picture shows the support structure for samples placed on the resonator.

2.2. Samples

¹⁵N-tempone-d₁₆ (4-oxo-2,2,6,6-tetramethyl-piperidine-d₁₆-1-¹⁵N-1-oxyl, CDN Isotopes, Pointe-Claire, Quebec, Canada), ¹⁵N-tempol-d₁₇ (4-hydroxy-2,2,6,6-tetramethyl-piperidine-d₁₇-1-¹⁵N-1-oxyl, CDN Isotopes, Pointe-Claire, Quebec, Canada), CTPO (3-carbamoyl-2,2,5,5-tetramethyl-3-pyrrolin-1-yloxy, Aldrich Chemical) and LiPc provided by Prof. Swartz, (Dartmouth University, Hanover, New Hampshire) were used as received. The OxyChip was provided by Prof. Kuppusamy (Dartmouth University). Borosilicate flat cells (VitroCom 4410–050 or 4410–100 www.vitrotubes.com) have a sample region that is 1 mm thick × 10 mm wide. Samples of LiPc were deoxygenated under vacuum and studied in flame-sealed tubes. Solutions of nitroxides were in equilibrium with air. The paramagnetic region of an OxyChip is ~ 0.8 mm thick and extends 0.8 × 4 mm parallel to the surface of the coil.

2.3. Resonator Q and Efficiency.

The Q-values (Table 1) measured with a 20 GHz Vector Network Analyzer (Anritsu MS46122A) are about 60 for both resonators. When the 10 mm coil was in the air core magnet [6] the frequency increased by up to 0.003 GHz, and the frequency of the 30 mm coil increased by about 0.02 GHz due to the shielding effect of the magnet on the unshielded coil. The efficiency ratios of the resonators were determined by comparing power saturation of a deoxygenated LiPc sample (5.4 mm high in a 3.8 mm id tube) with data obtained in an 8 mm L-band volume resonator that had been characterized previously [7].. The dimension of the LiPc sample perpendicular to the plane of the surface coil is about 2 mm, so efficiencies are an effective average over this distance. The efficiency, Λ, is measured at a location that is offset from the center of the coil by the thickness of the resonator support and the tube that contained the LiPc. For the 10 mm coil Λ = 0.049 (mT/sqrt(W)) at 2.5 mm above the center. For the 30 mm coil Λ = 0.0035 at 5 mm above the midpoint of the 4 mm height of the resonator. The distances of closest approach for the LiPc sample are similar to that for the nitroxide samples that were used to study the dependence of signal intensity on distance from the coil.

Table 1.

Resonant Frequencies^a and Q-values of Surface Coils.

Resonator diameter	Sample	Frequency (GHz)	Q-value
10 mm	None	1.006	61
	1 mm flat cellb	1.000	60
	8 mm OD tubeb	1.003	59
30 mm	None	0.994	56
	1 mm flat cellb	0.994	59
	8 mm OD tubeb	0.993	56

^a.outside the magnet

^bcontaining aqueous nitroxide solution

To independently verify the efficiency values, the perturbing spheres method was used [20]. A steel ball with 1/16” or ¼” diameter was placed in the tuned resonator, which shifted the resonant frequency. The efficiency is proportional to the difference in squares of the frequencies, with and without the ball. Using this method an efficiency of 0.039 ± 0.001 mT/sqrt(W) was obtained for the 10 mm coil and a value of 0.0040 ± 0.0003 mT/sqrt(W) was obtained for the 30 mm coil. The center of the ball was ~2 mm above the center of the 10 mm coil and 6 mm above the midpoint of the 4 mm height of the 30 mm coil. These values differ from what was measured by power saturation by −20% for the 10 mm coil and a +14% for the 30 mm coils, which is good agreement given the averaging of B₁ over the finite dimensions of the LiPc or the steel balls. Taking the averages of the results from the two measurement methods, the efficiency of the 30 mm resonator (.0038) is a factor of 12 smaller than the efficiency of the 10 mm resonator (0.044), measured near the surface of the resonator. As shown in Eq. 1 the B₁ in the resonator decreases with increasing R and with increasing distance from the center of the coil. The larger radius of the 30 mm coil and increased vertical distance from the center of the coil at which efficiency was measured predict a factor of about 5 decrease in efficiency. The comparison of resonators using Eq. (1) assumes that the current in the two resonators is the same. The resistance of the 30 mm resonator (4.2 ohm) is about a factor of 5 larger than for the 10 mm resonator (0.78 ohm). The larger resistance in the 30 mm resonator and differences in the coupling to the resonator mean that higher power must be used to create the same current as in the smaller resonator, which further decreases the observed efficiency of the larger resonator.

2.4. L-band (1.0 GHz) EPR Spectroscopy and Imaging.

EPR spectra were acquired with a 1.0 GHz (L-band) modification of the previously reported 700 MHz benchtop spectrometer [6, 7]. The bridge for the 700 MHz spectrometer had circuitry for either pulse or rapid-scan operation. The outputs from the 14-bit Teledyne SP Devices ADQ14 digitizer in arbitrary units (a. u.) were converted to mV using the conversion factor of 0.029 mV/a.u. provided by the vendor. Signal-to-noise (S/N) reported in Figure 2 and Table 2 was calculated as the ratio of signal amplitude to rms noise in the baselines of the spectrum. Rms noise was divided by the square root of the number of sinusoidal half cycles (twice the number of scans averaged) to obtain the average values for a single pass through the spectrum that are summarized in Tables 3 and 4.

Table 2.

S/N for ¹⁵N-d₁₆-tempone (normalized to10¹⁵ spins) in surface coil and volume resonators for 100 k scans

	High concentration	Low concentration	Average spins required for S/N of 3	reference
10 mm surface coil	11a	11a	2.7×1014	This work
30 mm surface coil	0.21a	0.21a	1.4×1016	This work
8 mm LGR	14b	17c	2.0×1014	[7]
25 mm Alderman-Grant	0.66b	0.67c	4.5×1015	[7]

^aThe high and low concentrations for the two surface coil resonators were 1.2 mM and 275 μM. Samples were in 1 × 10 × 38 mm flat cells positioned parallel to the surface of the coil.

^bThe high concentration samples were 275 μM for the 8 mm and 25 mm resonators. Samples were in 7 or 23 mm ID tubes for the 8 mm and 25 mm resonators, respectively.

^cThe low concentration samples were 20 and 15 μM for the 8 mm and 25 mm resonators, respectively. Samples were in 7 or 23 mm ID tubes for the 8 mm and 25 mm resonators.

Table 3.

Baseline noise (mV) in 1.2 mM nitroxide spectra at 1.0×10⁻⁶ W source power with various scan frequencies

Resonator\scan frequency	2.5 kHz	17 kHz	26 kHz	35 kHz
Filter width (G)	0.002	0.002	0.002	0.002
10 mm	0.63	0.64	0.69	0.62
30 mm	0.86	0.72	0.97	0.69

Table 4.

Baseline noise (mV) in 1.2 mM nitroxide spectra at non-saturating powers with various scan frequencies

Resonator/scan frequency	2.5 kHz		17 kHz		26 kHz		35 kHz
Filter width (G)	0.002	0.4	0.002	0.4	0.002	0.4	0.002	0.4
10 mma	0.78	0.41	0.86	0.65	0.92	0.82	0.90	0.85
30 mma	2.63	1.82	2.19	1.94	3.28	2.99	2.88	2.52

^aThe incident power was 0.29 mW for the 10 mm resonator and 64 mW for the 30 mm resonator.

To decrease losses in the signal detection path the bridge that was used for the L-band experiments was simplified to perform only rapid-scan experiments. The system uses a Rohde & Schwarz SMA100B signal generator, which exhibits the lowest source noise at 1.0 GHz among the variable frequency sources that were tested. The data acquisition and manipulation software is written in MATLAB. For each sample the microwave power was selected to be within the linear response region of a power saturation curve. The rapid scan frequency was 2.55, 17, 26, or 35 kHz. Deconvolution of the sinusoidal scans and correction for the rapid-scan background signal was performed as previously reported [7, 21]. To vary the distance between the surface coil and the sample, flat pyrex plates 1 mm thick were inserted. At 1 GHz the pyrex plates have low loss and do not impact resonator Q. Signal intensities and S/N shown in the figures are the averages of three replicates for 100 k scans.

3D spatial imaging was performed with 142 gradients equally spaced on a polar grid [22] and maximum gradient of 0.4 mT/cm in each dimension. This is the highest gradient that can be used without causing overlap of signals from the adjacent ¹⁵N hyperfine line. The scan frequency was 26 kHz and the scan width was 2.0 mT centered at the field position for the low field nitroxide line. For each projection 100 k scans were averaged. Images were reconstructed using locally written software. The images were not corrected for the variation in B₁ over the sample. This correction will be implemented in future software versions.

2.5. Calculations

Samples are positioned parallel to the plane of the surface coil. Calculation of the dependence of signal intensity on distance between the sample and the surface coil were performed in MATLAB based on B₁ calculated with Eq. (1), which expresses the signal variation along an axis perpendicular to the coil. The calculations do not account for the increasing divergence of the magnetic field flux lines with increasing distance above the plane of the coil. The signal intensity was the sum of 5 slices through each sample, parallel to the surface of the coil. This distance includes the finite thickness of the surface coil mounts and the thickness of the glass that enclosed the samples. The distance from the surface of the 10 mm coil to the closest edge of the sample was: LiPc in flat cell, 1.1 mm; OxyChip, 0.9 mm; glass flat slide, 1.12 mm; 7 mm ID tube, 1.0 mm. For the 30 mm coil the distances between the midpoint of the 4 mm high coil and the closest edge of the sample was: OxyChip, 3.4 mm; glass flat slide, 3.6 mm; 7 mm ID tube, 3.5 mm. Calculations were performed for two cases: (i) constant incident power or (ii) incident power adjusted for each position relative to the coil to produce the same B₁ at the center of the sample.

3. Results

3.1. Signal to Noise for aqueous ¹⁵N-d₁₆ tempone

The performance of the two resonators for sample positioned parallel to the coil was evaluated using 1.2 mM and 275 μM ¹⁵N-d₁₆ tempone in a 1 mm thick flat cell (Figure 3). The sample-containing region is ~10 mm × 35 mm. The sample extends beyond the diameter of the 10 mm coil, but is fully contained within the 30 mm coil. Table 2 gives S/N for the spins within the active volume of the resonator for each concentration based on averages of 100 k scans. To compensate for the factor of about 12 difference in resonator efficiency and achieve the same B1 at the sample, an incident power of 0.29 mW was used for the 10 mm coil whereas 64 mW was used for the 30 mm coil. For both coils the signal amplitude scales approximately with sample concentration. For the same sample the signal is weaker for the larger coil because of the lower filling factor. There is greater uncertainty in S/N than in signal intensity because of uncertainties in calculation of noise. . Imperfect corrections for the background signal that is induced by the rapid magnetic field scan results in low frequency variations in the baseline. Estimates of noise are based on regions of the baseline with lower contributions from background corrections. The higher power required to achieve the same B₁ for the larger resonator results in increased source noise and a consequent reduction in S/N for the samples on the larger coil (Figure 3). The S/N performance for the two surface coil resonators was compared with that reportedly previously for two volume resonators (8 mm and 25 mm) [7] (Table 2). The highest concentration used in the volume resonators is the lowest for the surface coils. Since the volume of sample in the resonators is different, the S/N is normalized to 10¹⁵ spins. Comparison values are also shown as the number of spins required to achieve a S/N of 3 (Table 2). For a sample directly above the coil the performance of the 10 mm surface coil is similar to that of the 8 mm volume resonator and the performance of the 30 mm surface coil is about a factor of 3 poorer than for the 25 mm volume resonator. These results demonstrate the excellent potential for the use of surface coils for study of radicals near the surface of an animal that is too large to fit into a volume resonator.

Figure 3.

Spectra of 1.2 mM (top) and 275 μM (bottom) ¹⁵N-d₁₆ tempone in a 1 mm flat cell on the surface of the 10 mm or 30 mm coil. The y-axes are arbitrary values with the same scale. Data were acquired with 100 k scans at a scan frequency of 2.55 kHz.

3.2. Dependence of signal intensity on distance from coil with constant power

Signal intensity as a function of distance from the center of the coil to the lower edge of three samples is shown in Figure 4 for the 10 mm and 30 mm coils. For each sample the rf power was selected to be in the linear response region when the sample is positioned as close as possible to the coil. The distance shown on the × axis is between the center of the coil and the nearest edge of the paramagnetic sample. The calculated line is based on Eq. (1), summation of five equally spaced slices through the sample, and the reciprocity-defined decrease of detected signal with increasing distance from the coil. The experimental data for the samples were scaled to match the calculation at the closest position for which data were acquired. The dimensions of the OxyChip (0.8 mm thick × 0.8 × 4 mm) or the LiPc sample (2.6 mm thick × 3 × 4 mm) are small relative to the diameter of both coils. For these samples the experimental dependence of signal intensity on distance from the center of the coil agrees well with the calculation based on Eq. (1). Although the length of sample in the 1 mm flat cell is greater than the diameter of the 10 mm coil, the observed variation of signal with distance from the coil still agrees relatively well with the predictions from Eq. (1). Discrepancy between calculated and observed dependence on position is much greater for the large diameter (7 mm ID) and longer (40 mm) sample of CTPO.

Figure 4.

Dependence of signal intensity on distance from the surface of the 10 mm or 30 mm coil to the edge of sample with constant incident power. Calculated (black); experimental (blue). The flat cell contained 0.78 mM tempol and the 7 mm ID tube contained 1.0 mM CTPO.

3.3. Dependence of signal intensity on distance from the coil with power adjusted

For some experiments the sample of interest is at a known distance from the coil. For those cases the power can be adjusted to compensate for the decrease in B₁ with distance from the coil. When the power is adjusted, the signal intensity is predicted to decrease less rapidly with increasing distance from the coil than if power is not adjusted. Signal intensities as a function of distance from the center of the coil for the samples with modest spatial extents perpendicular to the coil (LiPc, OxyChip and nitroxide in the flat cell) are shown in Figure 5 for the 10 mm and 30 mm coils. For the thin OxyChip or LiPc samples the data are consistent with the predicted improvement in signal intensity when power is adjusted to compensate for the decrease in B₁ with distance from the coils. Improvement is not as good for the nitroxide in the thicker 1 mm flat cell, especially with the 30 mm coil because incident power can only be optimized for a small portion of the larger sample.

Figure 5.

Dependence of signal intensity on distance from the surface of the 10 mm or 30 mm coil to the edge of sample with incident power adjusted for constant B₁ at the center of the sample. Calculated with Eq (1) (black); experimental (blue). The flat cell contained 0.78 mM tempol.

One might expect that the use of a larger diameter coil would always be preferred because of the greater depth of penetration of the microwave (rf) magnetic field. However, the lower efficiency of the larger coil requires higher source power to achieve the same B₁ at the sample. Even with the lowest phase-noise source that was found, the high source power required for a nitroxide radical and the 30 mm resonator results in larger contributions from source noise than in the 10 mm coil. For the available hardware the S/N at a given distance from the coil, with incident power adjusted for constant B₁ at the sample, was compared for the two resonators (Fig. 6) for 0.78 mM nitroxide in the flat cell. At distances less than about 6 mm the S/N is better for the 10 mm coil, but for longer distances the S/N is better for the 30 mm coil.

Figure 6.

Comparison of S/N for 0.78 mM nitroxide in flat cell with incident power adjusted for constant B₁ at the edge of the sample obtained with (○) 10 mm or (◊) 30 mm coil.

The Rohde & Shwarz SMA100B source that was used for these experiments has the lowest phase noise that we could find for a variable frequency source [7]. Fixed frequency sources are available with lower source noise, but would require use of a resonator with adjustable resonant frequency.

3.4. Comparison of baseline (system) noise

Rapid scan EPR generates a background (BG) signal that is induced by the rapidly-varying sinusoidal magnetic field scan. The data for 1.2 mM ¹⁵N-d₁₆ tempone in a 1 mm flat cell were collected at different scan frequencies. Each data set is deconvolved to recover the signal with linear magnetic field axis as described in section 2.4. The noise in the baseline of the spectrum after deconvolution is referred to as the baseline noise. In the final step of data workup and BG correction, a Gaussian filter with a bandwidth equal to the full-width at half height of the EPR signal is applied.

Baseline noise was compared for the two resonators at the lowest source power of 1×10⁻⁶ W and variable scan frequency (Table 3). Rms noise was divided by the square root of number of averages to find the value for a single pass through resonance. In a rapid scan experiment a scan includes both up-field and down-field traverse of resonance. In the data workup the up- and down-field scans are combined so the number of averages is twice the number of scans. At low source power the filter width was set to 0.002 G, which is equivalent to omitting the filter. Baseline noise was also evaluated at the higher power used to record the nitroxide spectra, which was 0.29 mW for the 10 mm resonator and 64 mW for the 30 mm resonator. Noise was compared for data workup with two filters: 0.002 G and 0.4 G. Approximately, 12% of the scan, mostly shorter visually flat regions excluding ¹³C hyperfine lines, were sampled for noise calculation. The estimated uncertainty in the noise values is about 15%. For the 10 mm resonator at the low power there is minimal dependence of noise on scan frequency (Table 3). These values are similar to what was observed at low power for volume resonators and is attributed primarily to thermal noise [7]. Even at low power the noise in the 30 mm resonator is higher than in the 10 mm resonator. At higher powers (Table 4) the final stage filter reduces noise, which is attributed to contributions from incomplete removal of rapid-scan background contributions. Higher noise at 26 kHz than at other scan frequencies is attributed to microphonics in the resonator. The higher noise in the 30 mm resonator than in the 10 mm resonator (Table 4) is due to source noise that dominates at incident power of 64 mW. The dependence of noise on scan frequency suggests that resonator performance could be improved by changing construction to decrease microphonics in the resonator.

3.5. 3D spatial images with the 30 mm surface coil

A long-term goal for the L-band system is to study nitroxides in vivo. A surface coil permits imaging of larger samples than can be handled with volume resonators that fully contain the sample. The performance of the 30 mm surface coil resonator was tested with a phantom (Figure 7) constructed from two 8 mm OD (7 mm ID) tubes separated by a rectangular tube with 6 mm outer (4 mm internal) thickness and 8 mm width. The incident power was selected to be in the linear response regime for the portion of the sample closest to the coil. Each tube contained 1 mM ¹⁵N-d₁₇-tempol. 3D images of the phantom were obtained based on the low-field nitroxide line. Displays are not corrected for the variation in B₁ over the sample region. Slices through the image in three perpendicular planes are shown in Fig. 8. The dimensions of the signal-containing regions are in good agreement with the shapes of the tubes.

Figure 7.

The long axis of the phantom was positioned along the bore of the magnet, which is defined as the z-axis, and parallel to the surface coil, which is the xz plane.

Figure 8.

Slices through 3D image of the phantom based on one of the nitroxide lines, obtained with a gradient of 0.4 mT/cm and 142 projections. Data acquisition parameters were 26 kHz scan frequency, 100 k averages and power of 19 mW.

4. Conclusions

Surface coil resonators can be used to measure EPR signals from regions of an object placed above the plane of the resonators, which permits study of larger objects than can be studied with volume resonators. Resonators with 10 mm and 30 mm coils were compared to understand performance tradeoffs. Resonator efficiency near the surface of the coil is higher for a smaller coil, but signal intensity falls off more quickly with distance from the coil for the smaller coil. For samples with small dimensions along the axis perpendicular to the plane of the coil, signal intensity decreases approximately as B₁². For sample near the coil S/N for the 10 mm surface coil is similar to that for an 8 mm volume resonator. Near the surface of the coil the S/N is higher for the 10 mm coil, but that advantage is lost at distances greater than about 6 mm. For localized samples, the decrease of intensity with distance from the coil can be partially compensated by increasing incident power. At the higher incident powers that are needed to compensate for the lower efficiency of the larger coil, source noise dominates even for the best performing variable frequency source that is currently available at L-band. Baseline noise at the lowest source power was similar across the different scan frequencies, except at 26 kHz, where microphonics are greater, impacting the noise.