Rapid-scan EPR imaging

Abstract

In rapid-scan EPR the magnetic field or frequency is repeatedly scanned through the spectrum at rates that are much faster than in conventional continuous wave EPR. The signal is directly-detected with a mixer at the source frequency. Rapid-scan EPR is particularly advantageous when the scan rate through resonance is fast relative to electron spin relaxation rates. In such scans, there may be oscillations on the trailing edge of the spectrum. These oscillations can be removed by mathematical deconvolution to recover the slow-scan absorption spectrum. In cases of inhomogeneous broadening, the oscillations may interfere destructively to the extent that they are not visible. The deconvolution can be used even when it is not required, so spectra can be obtained in which some portions of the spectrum are in the rapid-scan regime and some are not. The technology developed for rapid-scan EPR can be applied generally so long as spectra are obtained in the linear response region. The detection of the full spectrum in each scan, the ability to use higher microwave power without saturation, and the noise filtering inherent in coherent averaging results in substantial improvement in signal-to-noise relative to conventional continuous wave spectroscopy, which is particularly advantageous for low-frequency EPR imaging. This overview describes the principles of rapid-scan EPR and the hardware used to generate the spectra. Examples are provided of its application to imaging of nitroxide radicals, diradicals, and spin-trapped radicals at a Larmor frequency of ca. 250MHz.

Keywords: Deconvolution; Direct detection; Imaging; In vivo; Nitroxides.

Fig. 1

Schematic comparison of CW and rapid scans for a single-line EPR spectrum. A) At each point in a CW first-derivative spectrum the signal is encoded by phase-sensitive detection at the modulation frequency. B) In a rapid-scan experiment the center field is held constant and an additional field is rapidly varied across the full spectrum. The full amplitude of the spectrum is detected in each sweep. The example shows two full cycles of the rapid scan after deconvolution, which is comprised of two up-scans and two down-scans through the single peak.

Fig. 2

(A) Rapid-scan signals (solid lines) for a LiPc sample, obtained at RF frequency = 248 MHz and B₁= 3.6 mG at various scan frequencies and the signals obtained by deconvolving these signals (dashed lines). The signals were obtained at various scan frequencies and a scan width of 2.16 G. The central segment of each scan is plotted. The corresponding scan rates (a–e) are 4.32, 8.64, 21.6, 38.8, and 43.2 kG/s. (B) Deconvolved spectra from part A superimposed on the deconvolution of the spectrum recorded with 1 kHz scan frequency. Reproduced with permission from Ref. [5].

Fig. 3

Relative intensities for the LiPc signal at the center of the scan as a function of scan rate at constant radiofrequency B₁ of 6.5×10⁻³ G (◆) and at the B₁ that gave the maximum signal amplitude (●). The number of sinusoidal scans was held constant. The relative signal amplitudes were scaled to 1.0 for the signal at constant B₁ of 6.5×10⁻³ G collected at the scan rate of 1.3×10³ G/s (scan width of 0.42 G and scan frequency of 1 kHz). The solid lines connect points that were calculated by numerical integration of the Bloch equations using the parameters for LiPc and the experimental scan widths and frequencies. Reproduced with permission from Ref. [11].

Fig. 4

Rapid-scan signals for two LiPc samples separated by ~ 5mm, obtained in the presence of magnetic field gradients with RF frequency = 248 MHz, B₁ = 2.6 mG, and a frequency of 4 kHz for triangular scans. The solid lines are the experimental data and the dashed lines are deconvolved spectra. Reproduced with permission from Ref. [5].

Fig. 5

Block diagram for a rapid-scan imaging system. The rapid-scan driver triggers the digitizer to acquire signal coherent with the triangular or sinusoidal scans.

Fig. 6

A) Spectrum of an oriented crystal of Mn²⁺ in Mg(acac)₂(H₂O)₂ obtained by field-stepped direct detection in 6201 steps over 6200 G. B) CW spectrum obtained with a modulation amplitude of 1.1 G and field resolution of 1.0 G. The B₁ was 174 mG for both spectra. Reproduced with permission from Ref. [32].

Fig. 7

A) 2D spectral-spatial image of phantom consisting of 0.2 mM ¹⁴N-PDT and 0.53 mM ¹⁵N-PDT in deoxygenated aqueous solution at 250 MHz. The samples were in 6 mm OD tubes with the edges separated by 2 mm. B) Comparison of spectral slices (¹⁴N in green and ¹⁵N in blue) through the image with the zero-gradient spectrum. Reproduced with permission from Ref. [50].

Fig. 8

2D spectral-spatial images of I and Ia in a two-compartment phantom with a 10 mm spacer between compartments. A) Left: both compartments contain 0.5 mM diradical I; right: slices through the upper (blue) and lower (red) compartments of the image. B) Left: the upper compartment contains 0.5 mM I and the lower compartment contains 1 mM Ia, generated by the reaction of 0.5 mM I with 1 mM glutathione; right: slices through the upper (blue) and lower (red) compartments of the image. Reproduced with permission from [51].

Fig. 9

Sagittal slice of overlapped MRI and rapid scan images of tumor-bearing mouse leg. A co-registered MRI image, shown in gray scale, was used to identify the location of the tumor that is outlined in magenta. An image of the cleavage rate for dinitroxide I, k^obs, is shown with the color scale. The cleavage rates were obtained by fitting the time dependence of the peaks for monoradical 2 in a series of images [52].

Fig. 10

2-D spectral-spatial image of BMPO-OH in the phantom sample (A), a spectral slice through the image (B) and the error function (C) that was used in the spectral slice fitting routine to distinguish regions that contain BMPO-OH from noise-containing baseline regions. Reproduced with permission from [39].