Magnetic Resonance Microscopy. Группа авторов
for transverse magnetic (TM) and hybrid (HEM) mode types.
When the resonant structure has dielectric/dielectric or dielectric/air interfaces with the surrounding medium, this quantification with integer numbers is no longer valid since the electromagnetic (EM) field does not vanish at the boundaries. In such situations, the EM field outside the resonator must be explicitly expressed. An approximation consists of writing the field outside with similar variations as inside the resonator with evanescent terms, neglecting the contribution of other modes to the decaying part. This can be made part by part, as developed in [4,7] and as shown in Figure 2.2 for the disk resonator.
Figure 2.2 Resonant mode field estimation: part-by-part decomposition of the computation volume, case of the disk resonator.
For example, the field expression remains unchanged within the dielectric disk, while outside it is written as in Equation 2.6, with f a decaying function depending on the computation volume boundaries, Kn the modified Bessel function of the first kind and order n, and ν the radial wavenumber in the region where the EM field is radially decaying:
Combining the boundary conditions at the interface air/dielectric (continuity of the tangential field component) as well as the wavenumber decomposition in each region provides a set of equations describing the modes arising from the combination of a planar dielectric waveguide with a circular dielectric waveguide.
Examples of electromagnetic field maps obtained with the same method extended to the case of a ring resonator are shown in Figure 2.3. As expected from the mode description, the electric field minimum coincides with the magnetic field maximum.
Figure 2.3 Electromagnetic field distribution of the TE01δ mode of a ring resonator (relative permittivity 500, outer radius 10 mm, height 10 mm) filled with a sample (relative permittivity 50) for varying inner to outer radii ratio: magnetic field (first line) and electric field (second line) field maps and lines (gray arrows), and both fields profiles (third line). The 2D maps are plotted in grayscale with a linear value distribution. From [21].
2.3.2 Power Loss Contributions in a Ceramic Probe
As ceramic probes do not require the use of an electronical circuit to tune and match at the Larmor frequency, the power losses are mainly due to the ceramic material losses and the electric field–sample interactions. Another contribution that is not considered here is that of the metallic feeding loop that is used to induce the mode’s field distribution in the ceramic resonator. As it is a small, nonresonant loop, its contribution is considered insignificant.
While operating in different regimes, the loss phenomenon is the same in the ceramic material and in the sample: it is energy dissipated as heat within complex permittivity materials immersed in an electromagnetic field [19]. In practice, these power losses are expressed as the integral over the object volume V of the power loss density, which involves two local variables: the imaginary part of the material permittivity and the electric field intensity. The power losses in a material of complex permittivity
The imaginary part of the permittivity is equal to
The ceramic probe is modeled as a ceramic ring resonator (inner radius rh, outer radius rd, height L, relative permittivity ϵr, loss tangent tan δ) filled with a cylindrical biological sample (radius rh, height L, and electrical conductivity σsample). With the theoretical insight about the TE01δ mode field distribution provided in Section 2.3.1, it is possible, as detailed in [21], to develop an analytical expression for the power losses in the ceramic probe at the cost of some approximations:
The field distribution used to express the dielectric resonator losses is that of a lossless resonator because losses in the ceramic are considered small (tan δ ≪ 10−1).
The field distribution of the ring resonator is assumed equal to that of the corresponding disk without field leakages at the lateral boundaries.
With these assumptions, the power losses expression reduces to Equation 2.8 with the axial wavenumber ky known from the mode study,