betaFeedback

WorksPhysics and Astronomy

Multilayer model for coatings with arbitrary layers for superconducting radio-frequency applications

Abstract

We extend the multilayer model of \etal{Kubo} for superconductor-insulator-superconductor (SIS) structures in two ways: first, by generalizing it to arbitrary sequences of layers of arbitrary type, i.e. superconducting, normal conducting, and insulating; and second, by accounting for all contributions, including ohmic losses and dielectric effects. We examine the maximum applicable field for (SI)nS(\text{SI})^n\text{S} structures. We find that the optimum configuration corresponds to the n=1n=1 case. However, the thickness of the superconducting coating layers can be reduced to below their penetration depth with minor performance penalty. We discuss the ability to model transitions in SS bilayers by introducing a set of virtual layers that represent the transition region through interpolated parameters. We find degradation of the maximum applicable field with thicker transition layers, and a larger effective penetration depth of the electromagnetic fields. Furthermore, the surface impedance of the multilayer structure is calculated using the Leontovich boundary condition, yielding a formulation suitable for integration into finite element simulations. Additionally, the Poynting theorem is used to determine the loss contributions of individual layers.

Introduction

Current superconducting radio-frequency (SRF) applications, such as cavities used in particle accelerators, are primarily manufactured from bulk niobium (Nb), as it has the highest critical temperature TcT_c and highest lower critical field Hc1H_{c1} among pure metals [20]. However, it has been observed that the advantages of superconducting materials are realized only within a thin surface layer (< 1 µm) [24]. This observation has motivated the concept of coated cavities with thin superconducting layers with improved material parameters. Such approaches have already been successfully implemented at CERN facilities in the form of Nb/Cu cavities [2]. Further developments such as Nb3_3Sn coatings on Cu substrates are under investigation [5]. An extension of this concept was proposed by Gurevich [7] in the form of superconductor-insulator-superconductor (SIS) multilayer structures. In this approach, the bulk superconducting substrate is coated with a thin insulating layer, followed by a thin superconducting layer with higher TcT_c and Hc1H_{c1}. The superconducting thin film shields the bulk material from the electromagnetic fields and attenuates them sufficiently to maintain the superconducting state at accelerating fields exceeding those achievable with the bulk material alone. The insulating layer plays an important role in preventing vortex penetration and suppressing Josephson effects [11].

The multilayer model developed by Kubo et al. [12, 13] describes the electromagnetic fields penetrating the multilayer structure as a function of the layer thicknesses and their material parameters. In particular, the model enables the determination of the maximum applicable magnetic field, which can be used to optimize the choice of layer thicknesses. In the present work, we extend this model in two respects. First, the formulation is generalized to an arbitrary number of layers of arbitrary type, i.e. superconducting, normal conducting and insulating. These formulas allow the analysis of e.g. SS bilayers, superconducting coatings on normal conductors, extended SIS structures of the form (SI)n^{n}S, oxide layers on any of the preceding structures, etc. Second, we retain the contributions of normal electrons, i.e. ohmic losses, and dielectric losses in our formulation. This becomes particularly relevant when determining the surface impedance of the structure for applications in finite element (FE) simulations.

Multilayer model

The multilayer structure is defined as illustrated in Figure 1. We consider a configuration consisting of M>1M > 1 layers, where each layer is denoted LkL_k for k=1,…,Mk = 1, \ldots, M. Each layer LkL_k is characterized by a vector of material parameters pk\boldsymbol{p}_k, and by a thickness dkd_k. The thickness of the final layer is assumed to be effectively infinite. For example, in a SIS structure, the substrate thickness is typically around 2 mm, whereas the penetration depth is only on the order of 100 nm to 1000 nm. Consequently, the fields decay well before reaching the end of the substrate.

Figure 1. Multilayer structure of the cavity boundary. Each layer is labeled LkL_k for k=1,…,Mk = 1, \ldots, M and has material parameters pk\boldsymbol{p}_k and thickness dkd_k. The first layer L1L_1 is at the inside of the cavity. The final layer LML_M is the bulk substrate.

Within each layer, we assume homogeneous, isotropic, linear and frequency-dependent materials. Consequently, the constitutive relations are D=εE\boldsymbol{D} = \varepsilon\boldsymbol{E} and B=μH\boldsymbol{B} = \mu\boldsymbol{H}, where D\boldsymbol{D} is the displacement field, ε\varepsilon the permittivity, E\boldsymbol{E} the electric field strength, B\boldsymbol{B} the magnetic flux density, μ\mu the permeability and H\boldsymbol{H} the magnetic field strength. Furthermore, we consider time-harmonic fields,

E(r,t)=ℜ[E(r)e−jωt],H(r,t)=ℜ[H(r)e−jωt],J(r,t)=ℜ[J(r)e−jωt].(1)\boldsymbol{E}(\boldsymbol{r},t) = \Re\left[\mathcal{E}(\boldsymbol{r})e^{-\mathrm{j}\omega t}\right], \quad\boldsymbol{H}(\boldsymbol{r},t) = \Re\left[\mathcal{H}(\boldsymbol{r})e^{-\mathrm{j}\omega t}\right], \quad\boldsymbol{J}(\boldsymbol{r},t) = \Re\left[\mathcal{J}(\boldsymbol{r})e^{-\mathrm{j}\omega t}\right]. \tag*{(1)}

where ω\omega denotes the angular frequency and J\boldsymbol{J} is the current density. In the interior of each layer, the time-harmonic Maxwell equations must be satisfied. Assuming the absence of external charges, they are given by [9]

∇⋅H=0,∇×E=jωμH,∇⋅E=0,∇×H=J−jωεE(2)\nabla\cdot\mathcal{H} = 0, \quad\nabla\times\mathcal{E} = \mathrm{j}\omega\mu\mathcal{H}, \qquad \nabla\cdot\mathcal{E} = 0, \quad\nabla\times\mathcal{H} = \mathcal{J} - \mathrm{j}\omega\varepsilon\mathcal{E} \tag*{(2)}

Furthermore, the interface conditions must hold at each interface between layer Lk−1L_{k-1} and LkL_k for k=2,…,Mk = 2, \ldots, M

n×(E(k)−E(k−1))=0,n×(H(k)−H(k−1))=Kk\boldsymbol{n} \times\left(\mathcal{E}^{(k)} - \mathcal{E}^{(k-1)}\right) = 0, \qquad\boldsymbol{n} \times\left(\mathcal{H}^{(k)} - \mathcal{H}^{(k-1)}\right) = \boldsymbol{K}_k
n⋅(εkE(k)−εk−1E(k−1))=Σk,n⋅(μkH(k)−μk−1H(k−1))=0(3)\boldsymbol{n} \cdot\left(\varepsilon_k\mathcal{E}^{(k)} - \varepsilon_{k-1}\mathcal{E}^{(k-1)}\right) = \Sigma_k, \qquad\boldsymbol{n} \cdot\left(\mu_k\mathcal{H}^{(k)} - \mu_{k-1}\mathcal{H}^{(k-1)}\right) = 0 \tag*{(3)}

where E(k)\mathcal{E}^{(k)} and H(k)\mathcal{H}^{(k)} denote the fields in layer LkL_k for k=1,…,Mk = 1, \ldots, M, Kk\boldsymbol{K}_k are the surface current densities, Σk\Sigma_k are the surface charge densities, and n\boldsymbol{n} is a normal vector that points from Lk−1L_{k-1} to LkL_k, in the presented setup n=ez\boldsymbol{n} = \boldsymbol{e}_z. All currents are modelled within the layers, even when flowing in a very thin region. Hence the surface current densities Kk\boldsymbol{K}_k are zero. Additionally, we utilize the modified Ohm’s law arising from the two-fluid model [15]

J=σ(ω)E,σ(ω)=σn+1ωμλ2(4)\mathcal{J} = \sigma(\omega)\mathcal{E}, \qquad\sigma(\omega) = \sigma_n + \frac{1}{\omega\mu\lambda^2} \tag*{(4)}

where σn\sigma_n is the normal conductivity and λ\lambda is the London penetration depth. The optical conductivity from the two-fluid model may be replaced by more advanced formulations, for example those derived from BCS theory [16, 19], or more recent development on Dynes superconductors [8]. From the above relations, one obtains

∇2E+α2E=0,∇2H+α2H=0,(5)\nabla^2\mathcal{E} + \alpha^2\mathcal{E} = 0, \qquad\nabla^2\mathcal{H} + \alpha^2\mathcal{H} = 0, \tag*{(5)}

with

α(p,ω)=μεω2−1λ2+jωμσn(6)\alpha(\boldsymbol{p},\omega) = \sqrt{\mu\varepsilon\omega^2 - \frac{1}{\lambda^2} + \mathrm{j}\omega\mu\sigma_n} \tag*{(6)}

For superconducting materials, the parameter vector is given by p=(μ,ε,σn,λ)\boldsymbol{p} = (\mu,\varepsilon,\sigma_n,\lambda). In the case of a normal conductor, the limit λ→∞\lambda\to\infty applies, reducing the number of parameters to three. For insulating materials, the same limit holds; however, losses are typically not described through the normal conductivity σn\sigma_n. Instead, the electric permittivity is treated as a complex quantity ε~=ε+j(σn/ω)\tilde{\varepsilon} = \varepsilon+ \mathrm{j}(\sigma_n/\omega), and the dielectric losses are specified through the loss tangent tan⁡δ=σn/(ωε)\tan\delta= \sigma_n/(\omega\varepsilon).

Assuming that the penetration depth is small compared to the wavelengths of interest, the electromagnetic fields can be considered uniform in the planar directions, i.e. ∂xE=∂yE=0\partial_x\mathcal{E} = \partial_y\mathcal{E} = 0. The problem therefore becomes one-dimensional, i.e. E(r)=E(z)\mathcal{E}(\boldsymbol{r}) = \mathcal{E}(z). Furthermore, since ∇⋅E=0\nabla\cdot\mathcal{E} = 0, it follows that ∂zEz=0\partial_z\mathcal{E}_z = 0, implying that Ez\mathcal{E}_z is constant. Because a non-zero constant would correspond to infinite field energy, Ez=0\mathcal{E}_z = 0. Finally, assuming linear polarization, the coordinate system may be chosen such that Ey=0\mathcal{E}_y = 0, leaving Ex\mathcal{E}_x as the only non-zero component of the electric field. Under these assumptions, the Helmholtz equation Eq. (5) reduces to the ordinary differential equation

d2Exdz2+α2Ex=0⟹Ex(z)=Aejαz+Be−jαz(7)\frac{\mathrm{d}^{2}\mathcal{E}_{x}}{\mathrm{d}z^{2}}+\alpha^{2}\mathcal{E}_{x}=0 \quad\Longrightarrow\quad\mathcal{E}_{x}(z)=A e^{\mathrm{j}\alpha z}+B e^{-\mathrm{j}\alpha z} \tag*{(7)}

From Faraday’s law it follows that the magnetic field has Hy(z)\mathcal{H}_{y}(z) as the only non-zero component which is given by

Hy(z)=αωμ[Aejαz−Be−jαz](8)\mathcal{H}_{y}(z)=\frac{\alpha}{\omega\mu}\left[A e^{\mathrm{j}\alpha z}-B e^{-\mathrm{j}\alpha z}\right] \tag*{(8)}

These solutions apply within each layer. Accordingly, we denote the fields in layer LkL_k by Ex(k)(z)\mathcal{E}_{x}^{(k)}(z) and Hy(k)(z)\mathcal{H}_{y}^{(k)}(z) for k=1,…,Mk=1,\ldots,M, each associated with unknown coefficients AkA_k and BkB_k. We express the equations in vector form1

Ex(k)(z)=CkTuk(z),Hy(k)(z)=αkωμkCkTvk(z)(9)\mathcal{E}_{x}^{(k)}(z)=\mathbf{C}_{k}^{T}\mathbf{u}_{k}(z),\qquad\mathcal{H}_{y}^{(k)}(z)=\frac{\alpha_k}{\omega\mu_k}\mathbf{C}_{k}^{T}\mathbf{v}_{k}(z) \tag*{(9)}

with

Ck=(AkBk),uk(z)=(ejαkze−jαkz),vk(z)=(ejαkz−e−jαkz)(10)\mathbf{C}_{k}=\begin{pmatrix} A_k \\ B_k \end{pmatrix},\qquad\mathbf{u}_{k}(z)=\begin{pmatrix} e^{\mathrm{j}\alpha_k z} \\ e^{-\mathrm{j}\alpha_k z} \end{pmatrix},\qquad\mathbf{v}_{k}(z)=\begin{pmatrix} e^{\mathrm{j}\alpha_k z} \\ -e^{-\mathrm{j}\alpha_k z} \end{pmatrix} \tag*{(10)}

For a structure containing MM layers, there are 2M2M unknown coefficients that must be determined. These coefficients can be obtained by imposing the interface conditions at each of the layer interfaces. The position of the interface between adjacent layers Lk−1L_{k-1} and LkL_k is denoted by ζk\zeta_k for k=2,…,Mk=2,\ldots,M. The special case ζ1\zeta_1 corresponds to the boundary between vacuum and the first layer L1L_1. We finally fix the coordinate system by setting z=ζ1=0z=\zeta_1=0. From the interface conditions Eq. (3) and the solutions Eq. (9) one finds that the fields must be continuous over the interfaces. Considering the magnetic field at the ζ1\zeta_1 interface we obtain the first boundary condition. Denoting the magnetic field magnitude with H0H_0 one obtains

Hy(1)(ζ1)=H0(11)\mathcal{H}_{y}^{(1)}(\zeta_1)=H_0 \tag*{(11)}

Only one independent condition arises here, since the electric field magnitude is fixed by Maxwell’s equations once H0H_0 is specified. At each interface ζk+1\zeta_{k+1} between layers LkL_k and Lk+1L_{k+1} for k=1,…,M−1k=1,\ldots,M-1, continuity of both electric and magnetic fields yields

Ex(k)(ζk+1)=Ex(k+1)(ζk+1),(12)\mathcal{E}_{x}^{(k)}(\zeta_{k+1})=\mathcal{E}_{x}^{(k+1)}(\zeta_{k+1}), \tag*{(12)}
Hy(k)(ζk+1)=Hy(k+1)(ζk+1),(13)\mathcal{H}_{y}^{(k)}(\zeta_{k+1})=\mathcal{H}_{y}^{(k+1)}(\zeta_{k+1}), \tag*{(13)}
Equation not extracted
(14)

Finally, the fields must vanish at infinity. As in the first condition, this yields only one independent constraint

lim⁡z→∞Hy(M)=0(15)\lim_{z\to\infty}\mathcal{H}_{y}^{(M)}=0 \tag*{(15)}

Substituting Eq. (9) into Eq. (11) gives

A1−B1=ωμ1α1H0(16)A_1-B_1=\frac{\omega\mu_1}{\alpha_1}H_0 \tag*{(16)}

Similarly, applying the continuity conditions Eq. (12) for k=1,…,M−1k=1,\ldots,M-1

CkTuk(ζk+1)=Ck+1Tuk+1(ζk+1),(17)\mathbf{C}_{k}^{T}\mathbf{u}_{k}(\zeta_{k+1})=\mathbf{C}_{k+1}^{T}\mathbf{u}_{k+1}(\zeta_{k+1}), \tag*{(17)}
CkTvk(ζk+1)=βkCk+1Tvk+1(ζk+1)(18)\mathbf{C}_{k}^{T}\mathbf{v}_{k}(\zeta_{k+1})=\beta_k\mathbf{C}_{k+1}^{T}\mathbf{v}_{k+1}(\zeta_{k+1}) \tag*{(18)}

where

βk=μkμk+1αk+1αk(19)\beta_k=\frac{\mu_k}{\mu_{k+1}}\frac{\alpha_{k+1}}{\alpha_k} \tag*{(19)}

By adding Eqs. (17) and (18) and multiplying with (1/2)exp⁡(−jαkζk+1)(1/2)\exp(-\mathrm{j}\alpha_k\zeta_{k+1}) one obtains

Ak=12e−jαkζk+1Ck+1T[uk+1(ζk+1)+βkvk+1(ζk+1)](20)A_k=\frac{1}{2}e^{-\mathrm{j}\alpha_k\zeta_{k+1}}\mathbf{C}_{k+1}^{T}\left[\mathbf{u}_{k+1}(\zeta_{k+1})+\beta_k\mathbf{v}_{k+1}(\zeta_{k+1})\right] \tag*{(20)}

Similarly, subtracting Eq. (18) from Eq. (17) and multiplying with (1/2)exp⁡(jαkζk+1)(1/2)\exp(\mathrm{j}\alpha_k\zeta_{k+1}) gives

Bk=12ejαkζk+1Ck+1T[uk+1(ζk+1)−βkvk+1(ζk+1)](21)B_k=\frac{1}{2}e^{\mathrm{j}\alpha_k\zeta_{k+1}}\mathbf{C}_{k+1}^{T}\left[\mathbf{u}_{k+1}(\zeta_{k+1})-\beta_k\mathbf{v}_{k+1}(\zeta_{k+1})\right] \tag*{(21)}

These equations can be written compactly as the matrix equation

Ck=τk,k+1Ck+1(22)\mathbf{C}_k=\boldsymbol{\tau}_{k,k+1}\mathbf{C}_{k+1} \tag*{(22)}

where the characteristic matrix τk,k+1\boldsymbol{\tau}_{k,k+1} is given by

τk,k+1=12((1+βk)exp⁡(ȷζk+1[−αk+αk+1])(1−βk)exp⁡(ȷζk+1[−αk−αk+1])(1−βk)exp⁡(ȷζk+1[αk+αk+1])(1+βk)exp⁡(ȷζk+1[αk−αk+1]))(23)\boldsymbol{\tau}_{k,k+1}=\frac{1}{2}\begin{pmatrix} (1+\beta_k)\exp\left(\jmath\zeta_{k+1}\left[-\alpha_k+\alpha_{k+1}\right]\right) & (1-\beta_k)\exp\left(\jmath\zeta_{k+1}\left[-\alpha_k-\alpha_{k+1}\right]\right) \\ (1-\beta_k)\exp\left(\jmath\zeta_{k+1}\left[\alpha_k+\alpha_{k+1}\right]\right) & (1+\beta_k)\exp\left(\jmath\zeta_{k+1}\left[\alpha_k-\alpha_{k+1}\right]\right) \end{pmatrix} \tag*{(23)}

Eq. (22) can essentially be viewed as a recursion relation, i.e. to know the coefficients of layer LkL_k, one must know the coefficients of layer Lk+1L_{k+1}. The recursion ends at the last layer, whose coefficients we determine by substituting Eq. (9) into Eq. (15)

lim⁡z→∞[AMeȷαMz−BMe−ȷαMz]=0(24)\lim_{z\to\infty}\left[A_M e^{\jmath\alpha_M z}-B_M e^{-\jmath\alpha_M z}\right]=0 \tag*{(24)}

For this limit to hold, the coefficient multiplying the exponentially diverging term must vanish. If ℑ(αM)<0\Im(\alpha_M)<0, then ℜ(ȷαMz)>0\Re(\jmath\alpha_M z)>0 implying AM=0A_M=0. Conversely, if ℑ(αM)>0\Im(\alpha_M)>0, then ℜ(−ȷαMz)>0\Re(-\jmath\alpha_M z)>0 and therefore BM=0B_M=0. For a superconducting substrate with typical parameters ω∼1010 s−1\omega\sim10^{10}\ \mathrm{s}^{-1}, λ∼10−7 m\lambda\sim10^{-7}\ \mathrm{m} and σn∼107 S m−1\sigma_n\sim10^7\ \mathrm{S\,m}^{-1} one finds αM∼(103+ȷ107)m−1\alpha_M\sim(10^3+\jmath10^7)\mathrm{m}^{-1}. Similarly, for a normal conducting substrate, where the London penetration depth does not contribute, one finds αM∼105(1+ȷ)m−1\alpha_M\sim10^5(1+\jmath)\mathrm{m}^{-1}. In both cases ℑ(αM)≫0\Im(\alpha_M)\gg0, leading to the conclusion that BM=0B_M=0 and hence

CM=AM(10)(25)\mathbf{C}_M=A_M\begin{pmatrix}1\\0\end{pmatrix} \tag*{(25)}

The coefficients are then determined by propagating this conditions through the recursion relation Eq. (22). In particular first layer coefficients C1\mathbf{C}_1 is determined from the final coefficients CM\mathbf{C}_M by

C1=[∏k=1M−1τk,k+1]CM:=TCM,(26)\mathbf{C}_1=\left[\prod_{k=1}^{M-1}\boldsymbol{\tau}_{k,k+1}\right]\mathbf{C}_M:=\mathbf{T}\mathbf{C}_M, \tag*{(26)}

where the ordered product of all characteristic matrices is denoted by T\mathbf{T} and referred to as the transfer matrix. Using Eq. (25) one obtains

A1=AMT00,B1=AMT10(27)A_1=A_M T_{00},\qquad B_1=A_M T_{10} \tag*{(27)}

where T00T_{00} and T10T_{10} are components of T\mathbf{T}. Combining these expressions with Eq. (16) yields

AM=ωμ1α1[T00−T10]H0(28)A_M=\frac{\omega\mu_1}{\alpha_1}\left[T_{00}-T_{10}\right]H_0 \tag*{(28)}

This determines all coefficients through the recursion relation Eq. (22), and therefore completely specifies the fields in Eq. (9). The resulting electromagnetic fields and current density for an example (SI)n^nS structure are shown in Figure 2.

Figure 2. Electromagnetic fields and current density inside a NbTiN-AlN-NbTiN-AlN-Nb multilayer structure. The material parameters used are listed in Table 1. The thicknesses used for the layers are, d1=d3=75 nmd_1=d_3=75\ \mathrm{nm} and d2=d4=30 nmd_2=d_4=30\ \mathrm{nm}.

Maximum applicable field

The accelerating gradients achievable in SRF cavities are primarily limited by the intrinsic field limits of Nb. With modern surface-preparation techniques, peak magnetic fields of approximately 150 mT are routinely attained [20]. In type II superconductors, two critical magnetic fields exist: the lower critical field Hc1H_{\mathrm{c1}} and the upper critical field Hc2H_{\mathrm{c2}}. For a magnetic field H<Hc1H < H_{\mathrm{c1}}, the material is in the Meissner state, exhibiting perfect diamagnetism, expelling all magnetic flux lines. When H>Hc2H > H_{\mathrm{c2}}, superconductivity is destroyed and the material transitions to the normal conducting state. In the intermediate regime Hc1<H<Hc2H_{\mathrm{c1}} < H < H_{\mathrm{c2}}, the material enters the vortex state, where magnetic flux penetrates the superconductor in the form of quantized Abrikosov vortices [23].

Although the vortex state minimizes the free energy in this regime, the transition from the Meissner state to the vortex state does not necessarily occur, because the Meissner state is metastable. This metastability arises due to the Bean-Livingston surface barrier, which must be overcome before vortices can penetrate the material. The barrier disappears when the magnetic field reaches the superheating field HshH_{\mathrm{sh}}, which constitutes the theoretical upper limit for practical application. For bulk Nb, theoretical estimates place this limit as high as 240 mT [20], whereas the best experimentally achieved values are around 200 mT [25].

In the work of Kubo et al. [12, 13, 11], the maximum applicable field Bmax⁡B_{\max} is evaluated for both SIS multilayer structures and SS bilayers. For SIS structures, they argue that Bmax⁡B_{\max} is determined by the depairing limit of the superconducting coating and the empirical field limit of the substrate material. The magnetic field which exceeds these limits when reaching their respective interfaces determines the maximum field that the structure can sustain. For SS bilayers, the depairing limit is instead considered for both coating and substrate. Here, we extend these considerations to a general (SI)n^{n}S multilayer structure. The depairing limit for each superconducting thin layer corresponds to the current density at which the induced magnetic field reaches the superheating field, Jdp(k)=Hsh(k)/λkJ_{\mathrm{dp}}^{(k)} = H_{\mathrm{sh}}^{(k)}/\lambda_k, while the empirical field limit of the substrate is denoted by Hemp(M)H_{\mathrm{emp}}^{(M)}. Let LkL_k be a superconducting thin layer that is first encountered by the electromagnetic field at the interface z=ζkz = \zeta_k. We then determine the value of the applied field H0H_0 for which the depairing limit in this layer is reached. Denoting the resulting value with H0,sh(k)H_{0,\mathrm{sh}}^{(k)}, we obtain

H0,sh(k)=R[α1ωμ1σ1(kω)T00−T10T00(k)ejαkζk+T10(k)e−jαkζk]Hsh(k)λk(29)H_{0,\mathrm{sh}}^{(k)} = \mathfrak{R}\left[\frac{\alpha_1}{\omega\mu_1\sigma_1(k\omega)}\frac{T_{00}-T_{10}}{T_{00}^{(k)}e^{j\alpha_k\zeta_k}+T_{10}^{(k)}e^{-j\alpha_k\zeta_k}}\right]\frac{H_{\mathrm{sh}}^{(k)}}{\lambda_k} \tag*{(29)}

where we introduce

T(k)=∏i=kM−1τi,i+1.(30)T^{(k)} = \prod_{i=k}^{M-1}\tau_{i,i+1}. \tag*{(30)}

Similarly, we determine the value H0H_0 at which the empirical field limit is reached once the field enters the substrate, denoting this value with H0,emp(M)H_{0,\mathrm{emp}}^{(M)}, we find

H0,emp(M)=R[μMμ1α1αM(T00−T10)ejαMζM]Hemp(M)(31)H_{0,\mathrm{emp}}^{(M)} = \mathfrak{R}\left[\frac{\mu_M}{\mu_1}\frac{\alpha_1}{\alpha_M}(T_{00}-T_{10})e^{j\alpha_M\zeta_M}\right]H_{\mathrm{emp}}^{(M)} \tag*{(31)}

The maximum applicable field Bmax⁡B_{\max} is then defined by the smallest value of H0H_0 that causes either the depairing limit to be reached in any of the superconducting thin layers, or the empirical limit to be reached in the substrate, i.e.

Bmax⁡=min⁡k{μkH0,sh(k),μMH0,emp(M)}.(32)B_{\max} = \min_k\left\{\mu_k H_{0,\mathrm{sh}}^{(k)}, \mu_M H_{0,\mathrm{emp}}^{(M)}\right\}. \tag*{(32)}

Here, the index kk runs over all superconducting thin layers. When extending the SS bilayer configuration to an arbitrary number of superconducting layers, Eq. (32) can still be used, except that the depairing limit is also applied to the substrate.

Following Ref. [11], we illustrate the behavior of Bmax⁡B_{\max} using contour plots as a function of the layer thicknesses. In Figure 3, we present the results for a NbTiN-AlN-Nb structure. The optimal configuration is found for layer thicknesses dNbTiN=260d_{\mathrm{NbTiN}} = 260 nm and dAlN=10d_{\mathrm{AlN}} = 10 nm. Notably, the optimal thickness of the NbTiN layer is approximately 1.4 times its penetration depth. Ideally, however, this thickness should remain below the penetration depth, so that the coating does not behave as a bulk superconductor, which would reintroduce vortex penetration effects that the insulating layer is intended to suppress. Furthermore, a steep decline in Bmax⁡B_{\max} is observed when dNbTiNd_{\mathrm{NbTiN}} is reduced from its optimal value, indicating that enforcing the condition dNbTiN<λNbTiNd_{\mathrm{NbTiN}} < \lambda_{\mathrm{NbTiN}} would lead to a substantial reduction in the theoretically achievable performance.

Figure 3. Maximum applicable field Bmax⁡B_{\max} of a NbTiN-AlN-Nb multilayer system as a function of the insulating and coating layer thicknesses in units of mT at a frequency of 1.3 GHz. The parameters used are listed in Table 1 and are temperature adjusted to 4.2 K according to the two-fluid model.

We next consider the same system, but with an additional NbTiN-AlN layer pair. The corresponding results are shown in Figure 4, where the thickness parameters not displayed in the plots are projected out by maximizing along those directions. The most influential pair of layers are the first ones encountered by the field, as evidenced by the significantly larger variation in Bmax⁡B_{\max} compared with other combinations (see Figure 4 (a)). In

Figure 4. Maximum applicable field Bmax⁡B_{\max} of a NbTiN-AlN-NbTiN-AlN-Nb multilayer system in units of mT as a function of the four thicknesses d1d_1, d2d_2, d3d_3 and d4d_4 of each respective non-bulk layer at a frequency of 1.3 GHz. They are displayed as six contour plots, for each unique pair, by taking maximum projections along the non-varied directions. The parameters used are listed in Table 1 and are temperature adjusted to 4.2 K according to the two-fluid model.

Figure 4 (d), (e) and (f), similar trends are observed, with the optimal configuration approaching the lower-left corner of the parameter space where corresponding layer thicknesses approach zero. Unsurprisingly then, the optimal configuration for this multilayer system is given by d1=dNbTiNd_1=d_{\mathrm{NbTiN}}, d2=dAlNd_2=d_{\mathrm{AlN}} and d3=d4=0d_3=d_4=0, effectively reducing the structure to the three layer case from before. Nevertheless, Figure 4 (b) reveals a much richer parameter space when considering reductions of individual layer thicknesses to below the penetration depth. For example, choosing d1=d3=150d_1=d_3=150 nm and d2=d4=10d_2=d_4=10 nm yields Bmax⁡=452B_{\max}=452 mT. In this configuration, the thickness of each superconducting layer remains below its respective penetration depth, while the reduction in performance is only 15 mT relative to the optimal configuration. A similar conclusion is found when the two superconducting coatings are composed of different materials, for example, when one has a higher superheating field HshH_{\mathrm{sh}}. Although the optimal configuration again eliminates one of the SI pairs, the additional degrees of freedom provided by the extra layers allow for a broader range of layer thickness combinations with only minor reduction in achievable Bmax⁡B_{\max}.

Modeling of layer transitions

Another multilayer structure currently studied is Nb3_3Sn deposited on Nb or Cu substrates. Several fabrication approaches exist, including vapor tin diffusion on Nb substrates [21], co-sputtering on Nb substrates [22], and DC magnetron sputtering on Cu substrates with Nb buffer layer [5]. In all of these methods, an annealing step is required to form the desired A15 crystal phase of Nb3_3Sn, during which Sn evaporates or diffuses into the Nb. As a result, the interface between the Nb3_3Sn layer and the Nb substrate is not sharp. The extent of the transition layer is highly dependent on the deposition method, being more relevant for vapor diffusion than for sputtering-based approaches.

In Figure 5, we illustrate how the multilayer model can be used to represent such a transition region. The transition layer is modeled as a sequence of virtual thin superconductors, each equipped with individual material parameters. One can determine these parameters through, for example, linear interpolation. The result on BmaxB_{\mathrm{max}} and the magnetic field penetration is shown in Figure 6. The maximum applicable field BmaxB_{\mathrm{max}} decreases by approximately 10 mT when the transition layer increases from 1 nm to 10 nm. Additionally, we observe a shift in the peak, indicating that slightly thicker coatings are required to reach the optimal configuration. The magnetic field distribution within the material shows plateaus whose extent scales with the thickness of the transition layer. This delays the decay of the magnetic field and effectively increases the penetration depth of the coating. Furthermore, we observe a slightly more rapid decay within the coating when the transition layer becomes thicker.

Figure 5. Graphic representation of a SS-bilayer with transition layer. The superconducting coating is denoted by S and has material parameters pS\mathbf{p}_{S} and thickness dSd_{S}. Similarly, S′^{\prime} denotes the superconducting substrate, with material parameters pS′\mathbf{p}_{S^{\prime}}. The transition layer has thickness dSd_{S} and is expanded into NN virtual layers, each with thickness dT/Nd_{T}/N and parameters p1,…,pN\mathbf{p}_{1},\ldots,\mathbf{p}_{N}. The effective structure is then of the form illustrated in Figure 1.

Figure 6. Maximum applicable field Bmax⁡B_{\max} (left) and magnetic field penetration (right) of a Nb3_{3}Sn-Nb bilayer with transition layer thicknesses ranging from 1 nm to 10 nm. The material parameters used are listed in Table 1. The transition layer is discretized into 100 virtual layers of equal thickness, and the parameters of each virtual layer are determine through linear interpolation between the values of the substrate and the coating. For the right figure, the thickness of the coating layer is equal to its penetration depth.

It should be noted that the presented model does not account for surface roughness or grain boundaries, which may significantly impact the behavior and applicability of the model. Again, this also depends on the particular deposition method in consideration, where we expect it to be more applicable for vapor diffusion. Nevertheless, if the transition region can be approximated as an effective intermediate layer, the presented model is still applicable.

Surface impedance

Determining the surface impedance of multilayer structures is particularly important for FE simulations of, for example, the eigenmodes of an SRF cavity, from which the quality factor can be obtained. Because of the large discrepancy between the thicknesses of the layers and the overall dimensions of the computational domain, explicitly resolving the layers within the mesh is generally impractical. This limitation can be overcome by replacing the multilayer structure with an effective boundary representation through the surface impedance boundary condition (SIBC). In this work, we consider only the first-order SIBC, also known as the Leontovich boundary condition [14]. Higher-order SIBCs, such as the Mitzner boundary condition, become relevant only when the curvature is comparable to the field attenuation [18], which is not the case here. For further SIBC formulation and their applications in numerical methods, see Ref. [27].

Leontovich boundary condition

The Leontovich boundary condition reads [27]

n×E=Z(ω)[n×(n×H)](33)\mathbf{n} \times\mathbf{E} = Z(\omega) \left[\mathbf{n} \times\left(\mathbf{n} \times\mathbf{H}\right)\right] \tag*{(33)}

where n\boldsymbol{n} is the normal vector to the boundary surface, and Z(ω)Z(\omega) is the frequency-dependent surface impedance. For the configuration shown in Figure 1, the boundary corresponds to the z=0z=0 plane, with the normal vector n=ez\boldsymbol{n}=\boldsymbol{e}_z pointing into the material. From § 2, we further have E=Ex(z)ex\mathcal{E}=\mathcal{E}_x(z)\boldsymbol{e}_x and H=Hy(z)ey\mathcal{H}=\mathcal{H}_y(z)\boldsymbol{e}_y. Using Eq. (33), the surface impedance can therefore be expressed as

Z(p,ω)=Ex(z)Hy(z)∣z=0(34)Z(\boldsymbol{p},\omega)=\left.\frac{\mathcal{E}_x(z)}{\mathcal{H}_y(z)}\right|_{z=0} \tag*{(34)}

Substituting Eq. (9) and using Eq. (27) yields

Z(p,ω)=ωμ1α1T00+T10T00−T10(35)Z(\boldsymbol{p},\omega)=\frac{\omega\mu_1}{\alpha_1}\frac{T_{00}+T_{10}}{T_{00}-T_{10}} \tag*{(35)}

In the special case M=1M=1, the coefficients of the first and last layer coincide, implying that T=Id⁡2T=\operatorname{Id}_2, the 2×22\times2 identity matrix. Consequently, Z(ω)=(ωμ)/αZ(\omega)=(\omega\mu)/\alpha, which reproduces the well-known expression for semi-infinite normal conductors and semi-infinite superconductors within the two-fluid model. Likewise, if all layers are identical, i.e. pk=p\boldsymbol{p}_k=\boldsymbol{p} for all k=1,…,Mk=1,\ldots,M, then αk=α\alpha_k=\alpha and βk=1\beta_k=1. In this case, Eq. (23) gives τk,k+1=Id⁡2\tau_{k,k+1}=\operatorname{Id}_2, and therefore T=Id⁡2T=\operatorname{Id}_2. As expected, this leads to the same result as for the M=1M=1 case. As an example, consider a NbTiN-AlN-Nb multilayer with material parameters listed in Table 1 and layer thicknesses dNbTiN=260 nmd_{\mathrm{NbTiN}}=260\ \mathrm{nm} and dAlN=10 nmd_{\mathrm{AlN}}=10\ \mathrm{nm}. Then one finds a surface impedance Z=1459 nΩ−ȷ1.741 mΩZ=1459\ \mathrm{n}\Omega-\jmath1.741\ \mathrm{m}\Omega.

ParameterValueDescriptionSource
σNbTiN\sigma_{\mathrm{NbTiN}} [S m−1^{-1}]2.86×1062.86 \times10^{6}Normal conductivity of NbTiN at 4.2 KRefs. [24, 4]
λNbTiN\lambda_{\mathrm{NbTiN}} [nm]180.57London penetration depth of NbTiN at 0 KRef. [1]
Tc,NbTiNT_{\mathrm{c,NbTiN}} [K]15.4Critical temperature of thin layer NbTiNRefs. [6, 1]
Bsh,NbTiNB_{\mathrm{sh,NbTiN}} [mT]439Superheating field of NbTiNRef. [10]
εr,AlN\varepsilon_{\mathrm{r,AlN}}10.4Relative permittivity of AlNRef. [26]
tan⁡δAlN\tan\delta_{\mathrm{AlN}}2.4×10−42.4 \times10^{-4}Loss tangent of AlNRef. [26]
σNb\sigma_{\mathrm{Nb}} [S m−1^{-1}]6.58×1066.58 \times10^{6}Normal conductivity of Nb at 300 KRef. [3]
RRRNb\mathrm{RRR}_{\mathrm{Nb}}300Residual resistivity ratio of Nb between 300 K and 4.2 K...
λNb\lambda_{\mathrm{Nb}} [nm]39.0London penetration depth of Nb at 0 KRef. [17]
Tc,NbT_{\mathrm{c,Nb}} [K]9.23Critical temperature of bulk NbRefs. [24, 3]
Bmax,NbB_{\mathrm{max,Nb}} [mT]170Maximum applicable field to bulk NbRef. [11]
λNb3Sn\lambda_{\mathrm{Nb_3Sn}} [nm]120London penetration depth of Nb3_3SnRef. [11]
Bsh,Nb3SnB_{\mathrm{sh,Nb_3Sn}} [mT]454Superheating field of Nb3_3SnRef. [11]

Table 1. Table of material parameters for discussed multilayer structures.

Complex Poynting theorem

While Eq. (35) is perfect for integration into FE simulations, it only tells us the surface impedance of the entire structure, but does not distinguish the contributions of each of the layer. We therefore consider a second way of determining the surface impedance which considers the contributions of each layer separately. We do so through the complex Poynting theorem [9]

12∫ΩdV (E⋅J∗)+2ȷω∫ΩdV (we−wm)+∫∂ΩdS (n⋅S)=0(36)\frac{1}{2}\int_{\Omega} \mathrm{d}V\,(\mathcal{E}\cdot\mathcal{J}^{*})+2\jmath\omega\int_{\Omega}\mathrm{d}V\,(w_e-w_m)+\int_{\partial\Omega}\mathrm{d}S\,(\boldsymbol{n}\cdot\boldsymbol{S})=0 \tag*{(36)}

where

we=ε4∣E∣2,wm=μ4∣H∣2,S=12(E×H∗).(37)w_e=\frac{\varepsilon}{4}|\mathcal{E}|^2,\qquad w_m=\frac{\mu}{4}|\mathcal{H}|^2,\qquad\boldsymbol{S}=\frac{1}{2}(\mathcal{E}\times\mathcal{H}^{*}). \tag*{(37)}

The quantities wew_e and wmw_m are the energy densities from the electric and magnetic fields respectively, and S\boldsymbol{S} is the Poynting vector. Using the Leontovich boundary condition Eq. (33), one finds that

(n⋅S)∣z=0=−12∣H0∣2Z(ω)(38)\left.(\boldsymbol{n}\cdot\boldsymbol{S})\right|_{z=0}=-\frac{1}{2}|\mathcal{H}_0|^2Z(\omega) \tag*{(38)}

Using that E=Ex(z)ex\mathcal{E}=\mathcal{E}_x(z)\boldsymbol{e}_x and H=Hy(z)ey\mathcal{H}=\mathcal{H}_y(z)\boldsymbol{e}_y, we also find

E⋅J∗=σ∗(ω)∣Ex(z)∣2,we(z)=ε∗4∣Ex(z)∣2,wm(z)=μ4∣Hy(z)∣2(39)\mathcal{E}\cdot\mathcal{J}^{*}=\sigma^{*}(\omega)|\mathcal{E}_x(z)|^2,\qquad w_e(z)=\frac{\varepsilon^{*}}{4}|\mathcal{E}_x(z)|^2,\qquad w_m(z)=\frac{\mu}{4}|\mathcal{H}_y(z)|^2 \tag*{(39)}

Then, Eq. (36) can be simplified and rearranged into an expression for the surface impedance as

Z(p,ω)=1∣H0∣2[(σ∗(ω)+ȷωε∗)∫R+dz ∣Ex(z)∣2−ȷωμ∫R+dz ∣Hy(z)∣2](40)Z(\boldsymbol{p},\omega)=\frac{1}{|\mathcal{H}_0|^2}\left[(\sigma^{*}(\omega)+\jmath\omega\varepsilon^{*})\int_{\mathbb{R}^{+}}\mathrm{d}z\,|\mathcal{E}_x(z)|^2-\jmath\omega\mu\int_{\mathbb{R}^{+}}\mathrm{d}z\,|\mathcal{H}_y(z)|^2\right] \tag*{(40)}

We split up the integrals according to the domains of each layer, i.e.

∫R+dz ∣Ex(z)∣2=∑k=1M∫ζkζk+1dz ∣Ex(k)(z)∣2=:∑k=1MIE(k)(pk,ω)(41)\int_{\mathbb{R}^{+}}\mathrm{d}z\,|\mathcal{E}_x(z)|^2=\sum_{k=1}^{M}\int_{\zeta_k}^{\zeta_{k+1}}\mathrm{d}z\,|\mathcal{E}_x^{(k)}(z)|^2=:\sum_{k=1}^{M}I_{\mathcal{E}}^{(k)}(\boldsymbol{p}_k,\omega) \tag*{(41)}
∫R+dz ∣Hy(z)∣2=∑k=1M∫ζkζk+1dz ∣Hy(k)(z)∣2=:∑k=1MIH(k)(pk,ω)(42)\int_{\mathbb{R}^{+}}\mathrm{d}z\,|\mathcal{H}_y(z)|^2=\sum_{k=1}^{M}\int_{\zeta_k}^{\zeta_{k+1}}\mathrm{d}z\,|\mathcal{H}_y^{(k)}(z)|^2=:\sum_{k=1}^{M}I_{\mathcal{H}}^{(k)}(\boldsymbol{p}_k,\omega) \tag*{(42)}

where we define ζM+1=+∞\zeta_{M+1}=+\infty, and IE(k)I_{\mathcal{E}}^{(k)} denotes the integral over the electric field in layer LkL_k, similarly IH(k)I_{\mathcal{H}}^{(k)} denotes the integral over the magnetic field in layer LkL_k, for k=1,…,Mk=1,\ldots,M. Similarly, we can split the surface impedance into the contributions for each layer, Z(p,ω)=∑k=1MZk(pk,ω)Z(\boldsymbol{p},\omega)=\sum_{k=1}^{M}Z_k(\boldsymbol{p}_k,\omega) where

Zk(pk,ω)=1∣H0∣2[(σk∗(ω)+ȷωεk∗)IE(k)(pk,ω)−ȷωμkIH(k)(pk,ω)](43)Z_k(\boldsymbol{p}_k,\omega)=\frac{1}{|\mathcal{H}_0|^2}\left[(\sigma_k^{*}(\omega)+\jmath\omega\varepsilon_k^{*})I_{\mathcal{E}}^{(k)}(\boldsymbol{p}_k,\omega)-\jmath\omega\mu_k I_{\mathcal{H}}^{(k)}(\boldsymbol{p}_k,\omega)\right] \tag*{(43)}

For k=1,…,M−1k = 1,\ldots,M-1, the integrals Eqs. (41) and (42) are easily solved numerically. But for the final layer LML_M the integration extends to infinity. Solving it analytically we find

IE(M)(pM,ω)=∣AM∣22ℑ(αM)e−2ℑ(αM)ζM,(44)I_{\mathcal{E}}^{(M)}(\mathbf{p}_M,\omega)=\frac{\lvert A_M\rvert^2}{2\Im(\alpha_M)}e^{-2\Im(\alpha_M)\zeta_M}, \tag*{(44)}
IH(M)(pM,ω)=12ℑ(αM)∣αMAMωμM∣2e−2ℑ(αM)ζM(45)I_{\mathcal{H}}^{(M)}(\mathbf{p}_M,\omega)=\frac{1}{2\Im(\alpha_M)}\left\lvert\frac{\alpha_M A_M}{\omega\mu_M}\right\rvert^2e^{-2\Im(\alpha_M)\zeta_M} \tag*{(45)}

We consider the same example as before, and find that that the contributions of each layer are: ZNbTiN=649 nΩ−j1.675 mΩZ_{\mathrm{NbTiN}} = 649\,\mathrm{n}\Omega- \mathrm{j}1.675\,\mathrm{m}\Omega, ZAlN=5×10−11 nΩ−j0.013 mΩZ_{\mathrm{AlN}} = 5 \times10^{-11}\,\mathrm{n}\Omega- \mathrm{j}0.013\,\mathrm{m}\Omega and ZNb=810 nΩ−j0.053 mΩZ_{\mathrm{Nb}} = 810\,\mathrm{n}\Omega- \mathrm{j}0.053\,\mathrm{m}\Omega. As expected, and mentioned in [11], the dielectric losses from the insulating layer are negligible compared to the other contributions. However, its contributions to the reactance are not negligible. We see that a significant portion of the losses occur in the NbTiN thin layer, though most losses occur in the Nb bulk substrate.

Conclusion

We have extended upon the model by Kubo et al. [13, 11] in two ways. We generalized the model to an arbitrary number of layers of arbitrary type, i.e. superconducting, normal conducting and insulating, and accounting for all contributions, including ohmic losses and dielectric effects. We have demonstrated how the maximum applicable field for (SI)n(\mathrm{SI})^n structures and arbitrary number of S layers can be obtained, and shown that the optimum corresponds to the n=1n = 1 case. However, we have also shown that n≠1n \ne1 cases can be useful in order to decrease individual layer thicknesses to below their penetration depths. We have further presented the ability to represent transition layers between SS-bilayers by introducing a set of virtual layers with individually assignable parameters. We found that the maximum applicable field decreases as the transition layer thickness is increased, and that the electromagnetic fields penetrate slightly deeper into the material. Finally, we have determined the surface impedance of the structure constituting the Leontovich boundary condition so that it can be incorporated into FE simulations. Additionally, we repeated this computation using the Poynting theorem such that the losses of each individual layer can be distinguished. The code for performing these calculations is included in the supplementary material.

Acknowledgements

The authors acknowledge financial support from the Federal Ministry of Research, Technology and Space (BMFTR), Germany, under grand number 05H2024.

References

  1. [1]M. Asaduzzaman et al. “Superconducting properties of thin film Nb1-xTixN studied via the NMR of implanted 8Li”. In: Journal of Physics: Condensed Matter 37.39 (Sept. 2025), p. 395701. ISSN: 1361-648X. DOI: 10.1088/1361-648x/ae0287. URL: http://dx.doi.org/10.1088/1361-648x/ae0287.
  2. [2]S. Calatroni. “20 Years of experience with the Nb/Cu technology for superconducting cavities and perspectives for future developments”. In: Physica C: Superconductivity 441.1–2 (July 2006), pp. 95–101. ISSN: 0921-4534. DOI: 10.1016/j.physc.2006.03.044. URL: http://dx.doi.org/10.1016/j.physc.2006.03.044.
  3. [3]R.L. David. CRC handbook of chemistry and physics. Internet Version 2005. Boca Raton, FL.: CRC press, 2005.
  4. [4]R. Di Leo et al. “Niobium-titanium nitride thin films for superconducting rf accelerator cavities”. In: Journal of Low Temperature Physics 78.1–2 (Jan. 1990), pp. 41–50. ISSN: 1573-7357. DOI: 10.1007/bf00682108. URL: http://dx.doi.org/10.1007/bf00682108.
  5. [5]D. Fonnesu et al. “Recipe optimization and SRF test of Cu-compatible Nb3Sn films by DC magnetron sputtering from a stoichiometric target”. In: Scientific Reports 16.1 (Jan. 2026). ISSN: 2045-2322. DOI: 10.1038/s41598-025-33547-w. URL: http://dx.doi.org/10.1038/s41598-025-33547-w.
  6. [6]I. González Díaz-Palacio et al. “Thermal annealing of superconducting niobium titanium nitride thin films deposited by plasma-enhanced atomic layer deposition”. In: Journal of Applied Physics 134.3 (July 2023). ISSN: 1089-7550. DOI: 10.1063/5.0155557. URL: http://dx.doi.org/10.1063/5.0155557.
  7. [7]A. Gurevich. “Enhancement of rf breakdown field of superconductors by multilayer coating”. In: Applied Physics Letters 88.1 (Jan. 2006). ISSN: 1077-3118. DOI: 10.1063/1.2162264. URL: http://dx.doi.org/10.1063/1.2162264.
  8. [8]F. Herman and R. Hlubina. “Microwave response of superconductors that obey local electrodynamics”. In: Physical Review B 104.9 (Sept. 2021). issn: 2469-9969. doi: 10.1103/physrevb.104.094519. URL: http://dx.doi.org/10.1103/PhysRevB.104.094519.
  9. [9]J.D. Jackson. Classical Electrodynamics. 3rd ed. New York: Wiley, 1999. isbn: 978-0471309321.
  10. [10]T. Junginger et al. “Critical fields of SRF materials”. en. In: Proceedings of the 9th Int. Particle Accelerator Conf. IPAC2018 (2018), Canada. doi: 10.18429/JACoW-IPAC2018-THPALL118. URL: http://jacow.org/ipac2018/doi/JACoW-IPAC2018-THPALL118.html.DOI
  11. [11]T. Kubo. “Multilayer coating for higher accelerating fields in superconducting radio-frequency cavities: a review of theoretical aspects”. In: Superconductor Science and Technology 30.2 (Dec. 2016), p. 023001. issn: 1361-6668. doi: 10.1088/1361-6668/30/2/023001. URL: http://dx.doi.org/10.1088/1361-6668/30/2/023001.
  12. [12]T. Kubo, Y. Iwashita, and T. Saeki. “RF field-attenuation formulae for the multilayer coating model”. In: Proceedings of the 4th International Particle Accelerator Conference (IPAC 2013). Shanghai, China, 2013, pp. 2343–2345. isbn: 978-3095450-122-9.arxiv.org/abs/1306.4823
  13. [13]T. Kubo, Yoshihisa Iwashita, and Takayuki Saeki. “Radio-frequency electromagnetic field and vortex penetration in multilayered superconductors”. In: Applied Physics Letters 104.3 (Jan. 2014). issn: 1077-3118. doi: 10.1063/1.4862892. URL: http://dx.doi.org/10.1063/1.4862892.
  14. [14]L.D. Landau and E.M Lifshitz. Electrodynamics of continuous media. Oxford, England: Pergamon Press, 1960.
  15. [15]H. London. “The high-frequency resistance of superconducting tin”. In: Proceedings of the Royal Society of London. Series A. Mathematical and Physical Sciences 176.967 (Nov. 1940), pp. 522–533. issn: 2053-9169. doi: 10.1098/rspa.1940.0105. URL: http://dx.doi.org/10.1098/rspa.1940.0105.
  16. [16]D. C. Mattis and J. Bardeen. “Theory of the anomalous skin effect in normal and superconducting metals”. In: Physical Review 111.2 (July 1958), pp. 412–417. issn: 0031-899X. doi: 10.1103/physrev.111.412. URL: http://dx.doi.org/10.1103/PhysRev.111.412.
  17. [17]B. W. Maxfield and W. L. McLean. “Superconducting penetration depth of niobium”. In: Physical Review 139.5A (Aug. 1965), A1515–A1522. issn: 0031-899X. doi: 10.1103/physrev.139.a1515. URL: http://dx.doi.org/10.1103/PhysRev.139.A1515.
  18. [18]K. Mitzner. “Effective boundary conditions for reflection and transmission by an absorbing shell of arbitrary shape”. In: IEEE Transactions on Antennas and Propagation 16.6 (Nov. 1968), pp. 706–712. issn: 0096-1973. doi: 10.1109/tap.1968.1139283. URL: http://dx.doi.org/10.1109/TAP.1968.1139283.
  19. [19]S.B. Nam. “Theory of electromagnetic properties of superconducting and normal systems. I”. In: Physical Review 156.2 (Apr. 1967), pp. 470–486. issn: 0031-899X. doi: 10.1103/physrev.156.470. URL: http://dx.doi.org/10.1103/PhysRev.156.470.
  20. [20]H. Padamsee. RF superconductivity, science, technology and applications. New York, NY: John Wiley & Sons, Apr. 2009. isbn: 978-3-527-40572-5.DOI
  21. [21]S. Posen et al. “Advances in Nb3Sn superconducting radiofrequency cavities towards first practical accelerator applications”. In: Superconductor Science and Technology 34.2 (Jan. 2021), p. 025007. issn: 1361-6668. doi: 10.1088/1361-6668/abc7f7. URL: http://dx.doi.org/10.1088/1361-6668/abc7f7.
  22. [22]N. Sayeed et al. “Fabrication of superconducting Nb3Sn film by Co-sputtering”. In: Vacuum 212 (June 2023), p. 112019. issn: 0042-207X. doi: 10.1016/j.vacuum.2023.112019. URL: http://dx.doi.org/10.1016/j.vacuum.2023.112019.
  23. [23]M. Tinkham. Introduction to superconductivity. 2nd ed. New York, NY: McGraw-Hill, Nov. 1995. isbn: 978-0-070-64878-4.
  24. [24]A.-M. Valente-Feliciano. “Superconducting RF materials other than bulk niobium: a review”. In: Superconductor Science and Technology 29.11 (Sept. 2016), p. 113002. issn: 1361-6668. doi: 10.1088/0953-2048/29/11/113002. URL: http://dx.doi.org/10.1088/0953-2048/29/11/113002.
  25. [25]K. Watanabe et al. “Development of the superconducting rf 2-cell cavity for cERL injector at KEK”. In: Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment 714 (June 2013), pp. 67–82. issn: 0168-9002. doi: 10.1016/j.nima.2013.02.035. URL: http://dx.doi.org/10.1016/j.nima.2013.02.035.
  26. [26]E. Yarar et al. “Low temperature aluminum nitride thin films for sensory applications”. In: AIP Advances 6.7 (July 2016). ISSN: 2158-3226. DOI: 10.1063/1.4959895. URL: http://dx.doi.org/10.1063/1.4959895.
  27. [27]S.V. Yuferev and N. Ida. Surface and impedance boundary conditions: a comprehensive approach. CRC press, 2018. DOI: 10.1201/9781315219929.

Paper details

vlnc:W_6ubmtj2x
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents