Chapman University

Chapman University Digital Commons

Computational and Data Sciences (PhD) Dissertations and Theses

Dissertations

Spring 5-2023

Computational Modeling of Superconductivity from the Set of

Time-Dependent Ginzburg-Landau Equations for Advancements in
Theory and Applications
Iris Mowgood
Chapman University, [email protected]

Follow this and additional works at: https://digitalcommons.chapman.edu/cads_dissertations

Part of the Condensed Matter Physics Commons, and the Other Physics Commons

Recommended Citation
I. Mowgood, "Computational modeling of superconductivity from the set of time-dependent Ginzburg-
Landau equations for advancements in theory and applications," Ph.D. dissertation, Chapman University,
Orange, CA, 2023. https://doi.org/10.36837/chapman.000483

This Dissertation is brought to you for free and open access by the Dissertations and Theses at Chapman
University Digital Commons. It has been accepted for inclusion in Computational and Data Sciences (PhD)
Dissertations by an authorized administrator of Chapman University Digital Commons. For more information,
please contact [email protected].
Computational Modeling of Superconductivity from the set of Time-Dependent Ginzburg-Landau

Equations for Advancements in Theory and Applications

A Dissertation by

Iris Kinzie Mowgood

Chapman University

Orange, CA

Schmid College of Science and Technology

Submitted in partial fulfillment of the requirements for the degree of

Doctor of Philosophy in Computational and Data Sciences

May 2023

Committee in charge:

Dr. Armen Gulian

Dr. Domenico Napoletani

Dr. Thomas Piechota

Computational Modeling of Superconductivity from the set of Time-Dependent Ginzburg-Landau

Equations for Advancements in Theory and Applications

Copyright ©2023

by Iris Kinzie Mowgood

III
 ACKNOWLEDGMENTS

I express my gratitude to my advisor, Dr. Armen Gulian, and his lab team, Dr. Sarah Chahid and

Dr. Serafim Teknowijoyo, for their many teachings during my PhD journey. I thank my committee members,

Dr. Domenico Napoletani and Dr. Thomas Piechota for their advice and encouragement.

I am grateful to my many instructors during my coursework in the Computational and Data Science program.

This work is dedicated to my daughters, Georgette Safa and Rose Saba.

IV
 VITA

Education

2016 - M.S. in Physics, California State University, Fullerton

2013 - B.S. in Physics Education, University of California, Santa Cruz

Work

2016-2023 - Adjunct Physics Lecturer, Chapman University

2018-2023 - Research Assistant, Advanced Physics Laboratory, Institute for Quantum Studies, Chapman

University

Summers of 2019 and 2022 - Internships at Advanced Physics Laboratory, Burtonsville, MD

Conferences

APS March Meeting - Las Vegas 2023, Presenter and Session Chair

8th International Workshop on Numerical Modelling of High Temperature Superconductors - Nancy, France

June 2022, Presenter

APS March Meeting – Denver 2020, Accepted Presenter

COMSOL Conference - Boston October 2019, Presenter

V
 LIST OF PUBLICATIONS

Chahid, S., Teknowijoyo, S., Mowgood, I., and Gulian, A. (2023). High-frequency Diode Effect in Supercon-

ducting N b3 Sn microbridges. Physical Review B, 107(5), 054506. doi:10.1103/physrevb.107.054506

Mowgood, I., Teknowijoyo, S., Chahid, S., and Gulian, A. (2022). Novel results obtained by modeling of

dynamic processes in superconductors: phase-slip centers as cooling engines. doi:10.48550/arXiv.2211.01286

Mowgood, I., Melkonyan, G., Dulal, R., Teknowijoyo, S., Chahid, S., and Gulian, A. (2021). Violation of

magnetic flux conservation by superconducting nanorings. Superconductor Science and Technology, 35(4),

045006. doi:10.1088/1361-6668/ac4174

Gulian, A., Foreman, J., Nikoghosyan, V., Sica, L., Abramian-Barco, P., Tollaksen, J., Melkonyan, G.,

Mowgood, I., Burdette, C., Dulal, R., Teknowijoyo, S., Chahid, S., Nussinov, S. (2021). Gravitational wave

sensors based on superconducting transducers. Physical Review Research, 3(4), 043098.

doi:10.1103/physrevresearch.3.043098

VI
 ABSTRACT

Computational Modeling of Superconductivity from the set of Time-Dependent Ginzburg-Landau

Equations for Advancements in Theory and Applications

by Iris Kinzie Mowgood

A full review of the research conducted and published during my PhD studies in Computational and Data

Sciences at Chapman University, under the advisement of Dr. Armen Gulian, are presented. Using the

set of time-dependent Ginzburg-Landau (TDGL) equations with inclusion of the interference current and

the non-equilibrium phonon term, we modeled the dynamics of superconductors in various theory revealing

states and practical purposes. A review of the history and phenomenon of superconductivity, including

modern applications, is introduced. The Josephson effect and associated Josephson junction are discussed

for comparison to our analogous results with the 1-D superconducting wire. The mathematics of microscopic

theory of superconductivity are explored with mention of BCS theory. The time-dependent Ginzburg-Landau

theory is detailed, as well as an interpretation from Dr. Aharonov’s discussion with my advisor on how the

simplest TDGL equation is equivalent to the Schrödinger equation with imaginary Hamiltonian. Methods

of how to transform the TDGL equations into dimensionless form and implement into the modeling software

COMSOL Multiphysics® for a 1-D wire model are revealed in tutorial form. I present my research in the

order of increased dimensions: 1-D, 2-D, and 3-D models, with details of solving each higher dimension with

different gauge choices. The results for the 1-D model are the visualization of the 1-D wire dynamics of

phase slip centers (PSCs), the pitchfork bifurcation of PSC locations at varying energy gaps, the influence of

phonon flux to the PSC, and the proposed application of a 1-D wire cryocooler from these previous results.

The published result of the superconducting diode effect shows vortices entering a 2-D slab unidirectional and

aided in explaining the microscopic theory behind the lab results at the Advanced Physics Laboratory. The

3-D published results were for the use of stacks of superconducting rings as transducers for the gravitational

wave detector, GEFEST, and how best to situate them for highest performance of the detector. The other

3-D model showed how superconducting nanorings, due to their size and closest to the critical temperature,

violate conservation of magnetic flux when of dimensional size of the order of λL .

VII
 TABLE OF CONTENTS

1 Basics of Superconductivity 1

1.1 History . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

1.2 Experimental Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

1.3 Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

2 Mathematics of Superconductivity 12

2.1 Time-Dependent Ginzburg-Landau Set of Equations . . . . . . . . . . . . . . . . . . . . . . . . . 12

2.1.1 The Full Set of TDGL Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

2.2 BCS Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

2.2.1 Connection to BCS’s ∆ function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

2.3 Ahanorov’s Interpretation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

2.4 Dimensionless Form of TDGL Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

2.5 Choice of Gauge . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

3 Computational Superconductivity 20

3.1 Comsol Multiphysics® : Equation-based Physics . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

3.2 COMSOL Implementation: Creating a Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

3.3 Working Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

4 Implementation of 1-D Model 26

5 Dynamics of Phase-Slips Centers (PSCs) 28

5.1 Visualization of PSCs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

5.2 Pitchfork Bifurcation of PSC Locations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30

5.2.1 Automation for Pitchfork Bifurcation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

5.3 Dynamics of PSCs with Phonon Feedback . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34

6 PSC Phonon Cryocooler 37

7 2-D Modeling of Superconducting Diode 41

7.1 Implementation of 2-D Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

7.2 Theory and Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42

7.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

8 3-D Modeling of Superconducting Rings 48

VIII
8.1 Implementation of 3-D Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48

9 Gravitational Wave Sensors based on Superconducting Transducers 53

9.1 Theory and Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53

9.2 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55

10 Violations of Magnetic Flux Conservation in Superconducting Nanorings 58

10.1 Theory and Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58

10.2 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60

11 Conclusion 66

12 Appendices 70

References 78

IX
 LIST OF FIGURES

1 Heike Kamerlingh Onnes (left) and Johannes Diderik van der Waals (right) in front of the helium-

’liquefactor’, Leiden 1908. [6] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

2 Levitating superconductor above magnets. [7] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

3 Illustration of tunneling of Cooper pairs from one superconductor through an insulating barrier

and into the other superconductor. [14] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

4 Illustration of the basis schematic of a SQUID utilizing the Josephson effect[18] . . . . . . . . . . 7

5 Quantum computer from the National Institute of Standards and Technology (NIST). The white

arrow is pointing at fiber optic cable researchers used to control the JJ/transmon qubit. [25] 8

6 Illustration from U.S. Department of Energy website showing use of superconducting magnets for

levitation and propulsion. [27] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

7 Diagram of MRI Machine. [30] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

8 Diagram of Large Hadron Collider. [32] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

9 Comparison of Approximation function to K(x) − E(x) . . . . . . . . . . . . . . . . . . . . . . . 14

10 Space Dimensions from Model Wizard . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

11 Physics module selection in COMSOL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

12 Study selection in COMSOL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

13 Model builder settings for the General PDE used for Equation-based modeling . . . . . . . . . . 23

14 Plot of Cooper pair density along 1-D wire at different moments in time from the model built in

this section. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

15 The lower left corner shows the idealization of the 1-D wire between two massive superconducting

banks. It is set upon a plot of a under dampened phase slippage of the modulus of Ψ vs. x

(dimensional length) or the CP density along the wire. Inset, lower right, is the CP Density

vs. dimensionless time at the center of the wire. . . . . . . . . . . . . . . . . . . . . . . . . . . 26

16 Set of single phase slippage along a 1-D wire versus the modulus of Ψ (square root of density of

pair condensate), evolving over time for increasing τϵ |∆| . . . . . . . . . . . . . . . . . . . . . 28

17 Set of double phase slippage evolving over time for increasing τϵ |∆|. . . . . . . . . . . . . . . . . 29

18 Location of PSCs with increased τϵ |∆| with branching and anti-branching. . . . . . . . . . . . . . 30

X
19 Spatial distribution of Cooper pair density for different intensities of phonon flux action at es-

tablished periodic PSC oscillations. j0 = 0.4 and τϵ |∆| = 0.1. Noticeably, at the beginning

of oscillations period, the sagging starts at the point where the phonon source is located. At

further evolution, the sagging moves towards the geometrical center of the wire. The hotter

the phonon spot is, the closer the location of the phase slip center to it. . . . . . . . . . . . . 34

20 Increase of |p| reduces the period of PSC oscillations. . . . . . . . . . . . . . . . . . . . . . . . . . 35

21 Voltage vs. time across the 1-D wire when PSC oscillations take place with small (p = 0) and

high (p = -0.9, located at the center of the wire) frequencies. Parameters are ∆ = 0.1 and jc

= 0.4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
2
22 (a) Temporal oscillations of Cooper pair density |Ψ(x, t)| in the 1 − D−wire at J = 1.5 (J0 ≈ 1).

Moments of time marked by arrows correspond to those in panel (b). (b) Time oscillations of

|Ψ(x, t)| at the central point of PSC (x = 0). (c) Voltage between the ends of the 1 − D−wire.

(d) Supercurrent, (e) normal and (f ) interference currents for various characteristic moments

of PSC oscillation (color coding is the same in panels a, d, e and f ). . . . . . . . . . . . . . . 37

23 (a) This form of ∂|Ψ(x, t)|/∂t at a PSC (red curve) causes a pulsating phonon flux spectrum.

The flux reverses its sign during the period of PSC oscillation, Cases 1 and 2. (b) Curves

corresponding to phonon term. At a given moment of time, independently of ∂|Ψ|/∂t, the

spectrum of phonon flux has positive and negative segments, separated by the frequency

ωq = 2|Ψ|. Additionally, it depends on variables |Ψ|, T, and γ. The curves shown are for

values of T = 0.9Tc and γ = 0.3. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38

24 Cross-sectional view of the cooler design based on PSC-filament with two surrounding plates.

Acoustic densities should satisfy the relation: ρ1 u1 < ρ2 u2 < ρ3 u3 . . . . . . . . . . . . . . . . 39

25 (a) Surface morphology of the films with inclusion of small areas (nanomountains) with higher

concentration of Sn (zoomed tenfold in the inset). (b) Resistive and magnetic transitions into

superconducting state. (c) A bridge after ion milling of 3D-printed pattern (initial stage of

preparation). (d) Last stage of bridge preparation: the structure in panel (c) was covered by

positive photoresist, two circles were projected via epi- fluorescent microscope and ion milled

to reduce the active part of bridges to micron-scale (the smallest, 2 µm result is shown). The

inset illustrates micro-roughness of the edge line. (e) Effects of ion milling and post annealing

on transition temperature of bridges. (f) Typical voltage-current dependence of bridges above

(T = 20 K) and well-below (T = 1.7 K) the superconducting transition. . . . . . . . . . . . . 43

XI
26 (a) Resistive state at various values of bias temperature for the applied current of opposite

polarities (absolute values are plotted). (b) The SDE in 2 µm bridge at sinusoidal current

amplitude ≈ 10 mA and frequency 0.1 Hz. (c) Same as in panel (b) with 5 µm-wide bridge

at frequency 10 kHz. (d) Same as in panel (c) at frequency 100 kHz. (e) Average of 200

acquisitions with reversed polarity of magnetic field. In all cases, a magnetic field orthogonal

to the surface of the bridge was applied and optimized in the range 50-100 Oe. . . . . . . . . 44

27 a) The entering of vortices from the top of the 2-D slab at the ac current is in its negative polarity.

b) The response voltage of the superconducting diode versus time as related to the ac current

and its polarity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

28 Evolution of Phase Slip Lines with the top of 2-D slab having three dissimilar notches that

encourage the formation of vortices. Panels a) through d) corresponding to increasing time. . 46

29 Comparison of sensitivity of the GEFEST detector with the current (bold curves) and projected

(dotted curves) GW detectors. The characteristic amplitudes of GWs are defined as the

amplitude with a bandwidth equal to the observation frequency. . . . . . . . . . . . . . . . . 53

30 Two orthogonally-oriented and mechanically separate barbells under the influence of an incident

plane periodic GW oscillate around the axis (1-2). The direction of GW is assumed to coincide

with this axis. The GW field-lines (see also Fig. below) are indicated for this case. Both

barbells are rigidly connected with the dedicated superconducting loops (their rectangular

shapes are adopted in the current layout). The magnetic field B is static in space and time.

In the absence of GW, the magnetic-field lines are in parallel to the loop planes, and no current

is flowing through the loop. Upon the action of GW, the barbells’ oscillatory rotation causes

the superconducting loops to rotate in opposite directions, thus generating opposite polarity

currents in the superconducting loops. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54

31 The acceleration field lines of planar wavefront, purely polarized (“+”or“×”) GW are indicated

by the sets of hyperbolas. The direction of accelerations is shown by arrows for a given half-

period of the wave; in the next half-period they will be reversed. A mechanical system of 2

point masses connected by a rigid line is a barbell with nonzero quadrupolar momentum. Two

such barbells will oscillate with respect to each other. Angular amplitude of relative oscillation

δΘ will constitute twice the value of the amplitude δθ of single barbell oscillation relative to

a non-moving observer associated with (x, y) coordinate system and with the magnetic field

B in Fig. 30. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54

32 3D stacking of individual loops in COMSOL at time of modeling for GEFEST. . . . . . . . . . . 56

XII
33 Modeling of increased number of stacked superconducting rings, up to thirteen rings, show that

the induced current will increase linearly. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56

34 (a) A nanoring (the gridline are λL -scaled) in a homogeneous external magnetic field H orthogonal

to the ring’s plane. The field H induces current j (green arrows) which generates its own

magnetic field and magnetization M, thus yielding the induction B = µ0 (H + M) (red

arrows). (b) Violation of magnetic flux conservation in a nanoring with external magnetic field

variation. Here, the dimensionless time and flux are in units t/t0 and Φ/Φext (t), respectively. 60

35 Appearance and disappearance of electric field (shown by arrows) during the evolution of an

external magnetic flux (the amplitude of the external magnetic field is shown in the insets

with a black spot indicating the state of the system). . . . . . . . . . . . . . . . . . . . . . . . 61

36 Magnetic flux and current dynamics in NbN nanorings with different wire diameters. Three curves

in panels (a)-(c) reflect different geometrical definitions of ring diameters (corresponding to

flux density integration): inner, D-d; central, D; outer, D+d. Current dynamics in (d) and

(e) correspond to the case of non-adiabatic on/off switching of the external flux, with γon

= γof f = 10. Here, the dimensionless current and time are in units of 1/(2µ0 ξ(T )) and

πh̄/(8kB (Tc − T )), respectively. Panels (e) and (f) are normalized by their stationary settled

values after the H-field switches on. The time axes in panels (d) through (f) have the same

origin. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63

37 Distribution of the azimuthal component of the electric field in macroscopic (a) and microscopic

(b) ring cross-sections. The two circles correspond to ring cross-sections in a vertical plane.

The sign change in color coding corresponds to directions toward and away from the observer

(the electric field vector encircles the ring). In both panels, color coding corresponds to the

dimensionless electric field in units 1/(2t0 ξ(T )) . . . . . . . . . . . . . . . . . . . . . . . . . . 64

38 Temperature dependence of the non-conservation effect for two nanoring sizes. Each point on

these curves is plotted using the ratio of Φring (t) and Φext (t). Here and below D and d denote

the diameter of the ring and of the ring’s wire correspondingly; λL (0) is the value of λL (0) at

T = 0 (η = 1). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64

XIII
 LIST OF ABBREVIATIONS

Josephson junction (JJ)

Superconducting quantum interference device (SQUID)

Superconductor (SC)

Normal Metal (N)

Cooper pairs (CP)

Insulator (I)

Bardeen–Cooper–Schrieffer (BCS)

Time-dependent Ginzburg-Landau (TDGL)

Phase Slip Centers (PSCs)

Multifrontal massively parallel sparse direct solver (MUMPS)

XIV
 LIST OF SYMBOLS

Tc : Critical Temperature

Hc : Critical Magnetic Field

Jc : Critical Current

Hext : External Magnetic Field

c: the speed of light,

n: the total density of superconducting electrons, and

e: the charge of the electron

nn : population density of the normal electrons

ns : superconducting electrons, ns

vn : velocity of the normal electrons,

vs : velocity of the superconducting electrons,

n: total population of nn and ns , and

ηn : the viscosity coefficient for the normal fluid.

∆: Energy Gap

Ψ: Order parameter of TDGL equation

|Ψ|2 : Cooper pair density

XV
1 Basics of Superconductivity

1.1 History

To begin, I will discuss the history of the discovery of superconductivity. Below is a summary of the history

from several introduction to superconductivity sources[1, 2, 3, 4]. The discovery of superconductivity was

propelled first by the experimental field of liquefying so-called permanent gases, or what is now known as

low temperature physics, starting with Faraday in 1823. His discovery was by accident. Leaving a sealed

U-shaped tube with chloride hydrate gas submerged with one end in an ice bath overnight one particularly

extreme English winter, he returned to discover a condensate in the tube. Therefore, these gases (hydrogen,

nitrogen, and oxygen) were not permanent gases but simply needed much lower temperatures than the

current technology in the laboratory would allow. Other scientists worked to realize methods of reaching

lower temperatures utilizing ideal gas laws through higher pressures during sealing of glass tubes. Louis Paul

Cailletet, influenced by Thomas Andrews of Belfast’s work in the 1860s, developed a technique of cooling

gases through rapid pressure release by allowing a faint mist to leak, in which he accomplished liquefaction

of acetylene, oxygen, nitrogen and carbon monoxide without high pressure. Carl Paul Gottfriend von Linde,

in the following decade, used gas expansion to his advantage with a fine nozzle that released high pressure

gas into a low-pressure region, allowing un-liquefied gas to recirculate back into the high-pressure region

until it was liquefied, producing the main method of refrigeration we use today. All of these developments

in low temperature physics lead to the lightest element, hydrogen, to be liquefied by James Dewar on May

10th, 1898 [1].

It was by the work of Heike Kamerlingh Onnes, photographed with his mentor Johannes Diderick van der

Waals in 1908 in Figure 1, at his cryogenic laboratory at the University of Leiden, that superconductivity

was first discovered in 1911. He has liquefied helium[5] and used it as the cryostat for a sample of mercury

with current passing through it. It was observed that the resistance in the mercury disappeared, and the

current continued without any loss or any additional supply. Believing it may be a malfunction, he and his

lab assistant increased the temperature until they measured a resistance again and then once again lowered

the temperature to see that there exists a critical temperature, Tc , in which resistance is equal to zero at

and below this value. His discovery of superconductivity earned him the Nobel Prize in physics is 1913 ‘for

his investigations on the properties of matter at low temperatures which led, inter alia, to the production of

liquid helium’ [2].

1
Figure 1: Heike Kamerlingh Onnes (left) and Johannes Diderik van der Waals (right) in front of the helium-
’liquefactor’, Leiden 1908. [6]

The discovery of the loss of resistance at low temperatures for superconductors lead to further understanding

of the electromagnetic effects that are a consequence of these types of states of matter. The superconducting

state can be destroyed with a sufficiently large magnetic field, and this limit is known as the critical magnetic

field, Hc . The same goes for the current being supplied through the superconductor; each superconductor

has a critical temperature, Tc , critical current, Jc , and critical magnetic field, Hc , (though for type II

superconductors, there exists two Hc values) uniquely valued. Walter Meissner and Robert Ochsenfeld

discovered the Meissner effect to explain this critical magnetic field and how the interior of the superconductor

is shielded from magnetic fields [2]. The Meissner effect occurs when the magnetic field that was previously

penetrating inside the material is expelled, during the process of cooling the superconductor to Tc and below.

This expulsion is due to the screening current running across the surface, at a depth to later be discussed,

as it enters the superconducting state. As known from electromagnetism[3], the screening current produces

an orthogonal magnetic field opposed to the external magnetic field, Hext , allowing it to levitate above

magnets as seen in Figure 2. This effect occur during steady states and reminds the physicist that the state

2
 Figure 2: Levitating superconductor above magnets. [7]

of superconductivity is an equilibrium state and does not depend on the history of the material, allowing

further understanding of the properties of this phenomenon [3]. The depth at which this screening current

occurs, also known as the penetration depth of Hext , was formalized by Fritz London and Heinz London in

1934 [3]. The London penetration depth, λL , is as follows:

mc2 1/2
λL (T ) = ( ) (1.1)
4πns e2

where λL (T ) is the penetration depth at temperature T , m is the mass of an electron, c is the speed of light,

ns is the total density of superconducting electrons, and e is the charge of the electron. This penetration

depth is temperature dependent and increases with closeness to Tc .

Understanding that this superconducting state is dependent on the population densities of the normal elec-

trons, nn , and the superconducting electrons, ns , in the superconductor, Cornelis Jacobus Gorter and Hendrik

Casimir developed the ‘two-fluid’ model[8] in which both normal and superconducting electrons exist and

exchange with one another as the material is transitioned through its critical quantities (Tc , Hc , Jc ). The

total rate of change of these two densities, nn and ns , respectively are:

Dvn nn ns nn
nn = − ∇n − ns σ∇T − ∇(vn − vs )2 + ηn ∇2 vn (1.2)
Dt n 2n

Dvs ns ns nn
ns = − ∇n + ns σ∇T + ∇(vn − vs )2 (1.3)
Dt n 2n

where vn is the velocity of the normal electrons, vs is the velocity of the superconducting electrons, n is

3
the total population of nn and ns , and ηn is the viscosity coefficient for the normal fluid. The ‘two-fluid’

model is the first developmental towards understanding the supercurrent in the superconductor. As we will

see from the methods of which we implement the time-dependent Ginzburg-Landau equation, the basis of

this model is placed in the current density equation, but the inclusion of interference current is necessary to

understand the full dynamics of the electrons in the superconductor since they also interact with phonons.

1.2 Experimental Results

Josephson junctions

The most important experiment results of superconductivity to application and to many of the models to be

discussed in this thesis is the Josephson junction. It is analogous to our 1-D wire models, and our 1-D wire

model behavior is very similar to the dynamics of the Josephson junction. To put it simply, a Josephson

junction (JJ) is when a weak supercurrent flows through an insulating barrier and the current is j = j0 sin(ωt)

with ω = 2eV /h̄ and V is voltage [9, 4]. The consequences of the effect, known as the Josephson effect,

has led to the development of superconducting quantum interference devices (SQUIDs) that allow detection

of even the faintest magnetic fields and/or variation of them [10, 11]/ We will see in our 1-D wire models

that they are modeled after this effect and considered to be similar to weakly linked JJ, but with no actual

physical weak link.

Shown by Giaever in 1960[12, 13], tunneling through an insulating barrier from one superconductor (SC) to

another or to a normal metal (N), gives a very direct measurement of the energy gap, ∆. The energy gap, ∆,

is the minimum total energy required to create a pair of superconducting electrons, known as Cooper pairs

(CP). The usual experiment arrangement for a N-SC junction are a N strip and a SC strip are evaporated on

a glass substrate in a cross arrangement and current is sent through one end of each strip type with the other

ends attached to potential leads to read their voltages [2]. The junction is the overlapped sections of the two

metal strip types, with an insulator (I) evaporated in between each metal strip layer, making the N-I-SC

configuration. With a N-I-N junction, tunneling of electrons can occur once a voltage is applied between

the metal, depending on the barrier thickness, the height of the potential-energy, typically around 104 kB T,

and the overlap of the electronic wave functions for each metal side of the junction. Though this quantum-

mechanic phenomena gives insight to behavior of electrons at that scale, the N-N tunneling does not give

forth as much information about the electronic states at the Fermi surface or as many details as the SC type

tunneling can. There are two main categories of the superconducting junctions, both depending on whether

the condensate wave functions of the two sides, or banks, overlap and that depends on the relative thickness

of the barrier or insulator. If the barrier is very thin between two superconducting banks, the condensate

4
wave functions overlap and result in a weak supercurrent that flows between the two banks. This type of

weakly linked or weakly coupled junction is known as Josephson tunneling. We will see later in this thesis

that this same principle is applied to the 1-D superconducting wire, as it is analogous to the weakly-linked

Josephson junction of this type, but it does not have a barrier. The center of the wire where a phase slip

center occurs acts like the junction. The second type has a much thicker barrier and allows the tunneling of

quasiparticles that are created when Cooper pairs break up. This type of tunneling of the thermally excited

quasiparticles are observed with SC-SC junctions and N-SC junctions. Tunneling of Cooper pairs (CP) are

illustrated in Figure 3.

Figure 3: Illustration of tunneling of Cooper pairs from one superconductor through an insulating barrier
and into the other superconductor. [14]

The density of states can be elucidated from quasiparticles. Tunneling requires energy of 2∆ to create two

quasiparticles from a Cooper pair and the density of states for quasiparticles is zero +/- ∆ from the Fermi

surface as described from BCS theory [2]. The density of states N (E) is then formalized as :

N (E) = 0 when |E| < |∆| (1.4)

N (0)E
N (E) = when |E| > |∆| (1.5)
(E 2 − |∆|2 )( 1/2)

From experiments, plotting the current versus time for varying low temperatures for these types of junctions

that allow quasiparticle tunneling will provide the energy gap. The I (current, not to be confused with

insulator) vs V (voltage) plots will allow one to deduce the energy gap from the fact that ideally, at T=0 K

5
the voltage will remain zero (no resistance and remaining in the superconductive state) until a voltage V0

is reach through increasing of current. Then one can calculate the energy gap since V0 = 2∆/e. There are

thermal excitations that occupy the states above EF + ∆ when T > 0K, with depletion of the states on the

other bank below EF − ∆, which experimentalists take into account since V < 2∆/e can allow some current

to flow.

There is also the case of SC-N to SC-SC’ tunneling junctions, in which two metals have different critical

temperatures at which they become superconducting. At a certain temperature range, one metal is super-

conducting while the other has a lower Tc and remains in the normal state, becoming a SC-N junction.

Further cooling below the Tc of the previous normal state metal, both become superconducting and it is now

an SC-SC’ junction. At the SC-N junction state, the current will increase rapidly at eV = ∆ due to inversion

of occupied states in the superconductor relative to unoccupied states in the normal metal. However, when

further cooled to the SC-SC’ junction state, there are maximum and minimum values for the current relative

to the two different energy gaps, ∆1 and ∆2 , of the dissimilar metals. At the maximum value for current,

eV = ∆1 − ∆2 and at the minimum, eV = ∆1 + ∆2 . There is a regime between these two values where it

depends on a negative differential resistance, requiring a more complex experimental setup than previously

described [2].

Cohen et al. [15] showed that the tunneling current for the two metals relative to their densities of states,

N1 and N2 , can be defined by the following equation:

Z
I(V ) = I0 N1 (E + eV )N2 (E)[f (E) − f (E + eV )] dE. (1.6)

The derivative of this equation, dI/dV, for the N-SC junction at T << Tc is more useful with experimental

measurements of those quantities using an AC current and phase techniques. This derivative measures the

density of states directly. Since f ′ (E + eV ) is peaked at E + eV = Ef and the density of states for the

normal mental, N1 (E) is constant, it can be shown that:

dI
∝ N2 (EF − eV ) (1.7)
dV

While the simplest derivation of the density of states from the BCS theory replaces the complicated phonon-

mediated interaction with a constant, this derivative above does much more to match real data. McMillian[16]

showed the full details of the phonon interaction due to the effect of N(E) by means of the Eliashberg

formulation, and are best used for weakly coupled superconductors and junctions. For strongly-coupled

6
superconductors and junctions, the BCS derivation for the density of states for phonons matches closest

with the experimental results of the inelastic neutron scattering[17].

The traditional Josephson junctions are considered weakly coupled superconductors, and a voltage Vdc

applied across the junction will produce an oscillating current with the Josephson frequency: ωJ ∝ Vdc . This

oscillation is the basis of the SQUID: the superconducting quantum interference device. Figure 4 shows the

basis schematic for a DC (direct current) SQUID. The SQUID is the Josephson effect, but its geometric

arrangement allows a phase jump between the two JJ opposing each other on the ring portion, as seen from

the schematics.

Figure 4: Illustration of the basis schematic of a SQUID utilizing the Josephson effect[18]

When a magnetic field with flux Φ is introduced through the SQUID opening while there is an initial current,

j0 , flowing through the SQUID, the SQUID will have a maximum current [4]:

jmax = j0 |cos(Φ/2)|. (1.8)

The result’s remarkableness comes from the fact that the SQUID’s sensitivity is of the order of a single

flux quantum. When a flux is produced in the opening of the SQUID, this equation above can determine

the magnetic field, even the very weakest fields, since the SQUID is sensitive to orders of units of flux

quantum and the oscillation of the junctions has been quantified from the Josephson effect. Josephson effect

based interferometers[19, 20, 21] are the most sensitive quantum detectors in modern history and have many

7
practical applications and varying arrangements.

1.3 Applications

The phenomenons of superconductivity have proven useful for modern technologies; in measurements, trans-

portation, medical devices, and even new theories. For measurements, the superconducting quantum inter-

ference device (SQUID) allows detection of the faintest magnetic fields of orders of a single flux quantum,

all aligned with the principles of Josephson junctions and their tunneling properties as described in Section

1.2. However, there are greater applications of the base units of Josephson junctions in quantum computing.

In Figure 6 is a quantum computer that a team of scientists at the National Institute of Standards and

Technology implemented with a transmon qubit that is a Josephson junction in a cavity,[22]. Qubits are the

building blocks of quantum computers because of their extremely short switching times between quantum

states and their scalability, leading to exceptionally fast processing[23]. This team of researchers used a

light-conducting fiber optic (seen in Figure 6 with the white arrow) to control its frequency and in turn

control its quantum state. However, there are limitations to the quantum computer at its current state in

history, as it is an extremely costly to operate it in the proper conditions that superconductors need, which

is at extremely cold temperatures, and the size of the quantum computers at the moment are exceptional

large [24]. More advancements are needed to improve the design of quantum computers.

Figure 5: Quantum computer from the National Institute of Standards and Technology (NIST). The white
arrow is pointing at fiber optic cable researchers used to control the JJ/transmon qubit. [25]

8
Levitation was shown to exist due to the Meissner effect as mentioned in Section 1.1 and in Figure 2. The

Meissner effect proves to be useful for not only train levitation, but the propulsion of the train itself [26].

Figure 6: Illustration from U.S. Department of Energy website showing use of superconducting magnets for
levitation and propulsion. [27]

Recalling that the Meissner effect is when the superconductor transitions below its critical temperature Tc

and expels any external magnetic fields from the bulk of the superconductor. This in turn produces a current

on the surface that produces an opposing magnetic field, causing levitation. We can see from Figure 2 that

there are like-poles between the underside of the train and where the tracks of the trains traditionally are.

The train and the track of magnets will repel each other, producing a non-contact, non-friction interaction

between the two. As known from physics and engineering, friction is the enemy of efficient and any friction

in a system will do work on the system and take away from the energy put into the system, reducing the

usefulness to be extracted from the purpose of the system. For propulsion, the sides of the train contain

alternating magnetic poles that interaction with the sides of the cavity of the train tracks, that also contain

alternating poles. Like poles repel, while opposite poles attract, giving an overall forward push of the train

according to the train conductor’s intention. The strength of magnetic fields can be controlled through

applied currents, and this is how the train conductor controls the propulsion[28].

There exists several operating superconducting levitation trains in the world. In China, the Shanghai Maglev

Train aka Transrapid, Beijing S1 Metro Line, and Changsha Express. In Japan, the Linimo, and in South

Korea, Daejeon Expo and Incheon Airport.

9
Nuclear magnetic resonance is the most common experience people have with applied superconductors.

Typically known to most as magnetic resonance imaging (MRI), the patient is placed lying down in a tunnel

of superconducting magnets, as seen in Figure 7. The superconducting magnets produce a large uniform

magnetic field to align the hydrogen atoms (more exact, the protons) in the water molecules in soft-body

tissues. The polarity of the water molecule are aligned in the same direction at times of large magnetic

fields being present. A radio frequency (since this is a safe wavelength for humans) causes the aligned

molecules to resonates, allowing images with depth and contrast to be processed by the receiver coils. This

has revolutionized medical diagnosis, since this is a noninvasive way to examine the internal issues of the

patient [29].

Figure 7: Diagram of MRI Machine. [30]

Lastly, particle colliders such as the Large Hadron Collider, located on the border of Switzerland and France,

use superconducting magnets to produce extreme magnetic fields that accelerate subatomic particles (elec-

trons, protons) travelling in opposite yet separate tunnels, gaining large momentum, and then opening of

the separate tunnels to record their collisions and observe the splintering into quarks to be discovered [31].

In Figure 8, we see the schematics of a slice of the tunnels themselves, with the superconducting solenoid

(in white hue) at the near center of the tunnel.

10
 Figure 8: Diagram of Large Hadron Collider. [32]

All of these applications rely on finding new superconductors, especially those that operate at higher tem-

peratures, close to room temperature. The less energy that is expended on the cooling of the system, the

more energy there is to be beneficially drawn from it. Therefore, there is great reason to explore and invest

in the research of superconductivity on all levels: theory, experimental and computational wise.

11
2 Mathematics of Superconductivity

2.1 Time-Dependent Ginzburg-Landau Set of Equations

To begin, I would like to introduce the full set of the time-dependent Ginzburg-Landau equations. Even

though all the details will be discussed after the definition of the equation, it is more useful to see the

equations first and then understand the development on a term by term basis. The set of equations involves

one for the order parameter, Ψ, and another for the current density, j. The order parameter, Ψ, has physical

meaning: the square of its modulus gives the density of the Cooper pairs and, in turn, tells us during modeling

which dimensions of the superconducting geometry, with respect to time, are in the superconducting state,

the normal state, and in between the two states. The current density, j, tells us the state of the electrons,

whether they are flowing with a supercurrent, js , a normal current, jn , or with an interference current, jint .

2.1.1 The Full Set of TDGL Equations

The first TDGL equation is for the order parameter Ψ = |Ψ| exp(iθ):

!
2
7ζ(3)|Ψ|2
 
π 1 ∂ ∂ |Ψ| π h 2
i Tc − T
− + 2iφ + 2τϵ2 Ψ+ D (∇ − 2iA) Ψ + − + P (|Ψ|) Ψ = 0,
8(πTc )2
p
8Tc 1 + (2τϵ |Ψ|)2 ∂t ∂t 8Tc Tc

(2.1)

where the theoretical units h̄ = c = e = 1 are used, A is the vector potential, φ is the scalar potential, Tc is

the critical temperature, τϵ is the electron-phonon relaxation time, D is the electronic diffusion coefficient,

ζ(3) is the Reimann zeta function, and P (|Ψ|) is the non-equilibrium phonon term. Ψ is the order parameter,

with |Ψ| equal to the superconducting energy gap and the square of the modulus of Ψ is equal to the density

of Cooper pair condensates. θ(x, y, z, t) is the quantum phase of the order parameter. This is the equation

of superconductors with finite gap, however, one should notice that the gapless case can be recovered by

allowing the electron-phonon relaxation time, τϵ → 0, such that there is a very strong e-ph interaction:

 2 !
∂Ψ 1 ∂
= + (1 − |Ψ|2 ) Ψ (2.2)
∂t 2m ∂r

The equation then is reduced towards the Schrödinger equation with imaginary Hamiltion as will be detailed

in Section 2.3.

12
The original GL theory contained terms of Ψ, as seen above, yet the community prefers ∆, therefore, we will

change the terms of Ψ to ∆:

Order parameter ∆ = |∆| exp(iθ):

!
2
7ζ(3)|∆|2
 
π 1 ∂ ∂ |∆| π h 2
i Tc − T
− + 2iφ + 2τϵ2 ∆+ D (∇ − 2iA) ∆ + − + P (|∆|) ∆ = 0.
8(πTc )2
p
8Tc 1 + (2τϵ |∆|)2 ∂t ∂t 8Tc Tc

(2.3)

The non-equilibrium phonon function has the form:

Z ∞
Γ(ε)
P (|∆|) = −2τε Re dε p , (2.4)
0 (ε + iγ)2 − |∆|2

where γ = (2τε )−1 and Γ(ε) is related with the non-equilibrium population of phonons δNωq via

Z ∞ Z ∞
πλ
Γ(ε) = ωq2 dωq dε′ δ(ε′ + ε − ωq )(uε uε′ + vε vε′ ) × (1 − nε − nε′ )δNωq . (2.5)
2(upF )2 0 |∆|

where λ is the dimensionless electron-phonon interaction constant, u is the speed of sound in the super-

conductor, pF is the electrons Fermi momentum, ωq denotes phonon frequency with the momentum q,
p
uε = |ε|θ(ε2 − |∆|2 )/ ε2 − |∆|2 is the BCS density of states for electrons, vε = uε |∆|/ε, and nε is the

distribution function of non-equilibrium electrons.

Early development utilized the two-fluid model for superconductivity, as mentioned in Section 1.1. However,

during transitions between superconducting and normal states, (e.g. during phase slippage as well as vortices

entering the system in higher dimensions as will be seen in Sections 5.1 and 7.3), there exists a mixed state

of superconducting Cooper pairs and normal electrons,[33]. This mixed state of motion can be quantified as

an interference current that exists during these non-equilibrium superconducting states. Therefore, during

our modeling, we included the interference term for current, jint , together with the superconducting and

normal contributions.

13
Here is the equation for the current density j with all contributions included:

!
2
πσn 2 ∂ |∆|
j = js + jn + jint = Q |∆| −2τϵ
4T ∂t
( p " ! !#) (2.6)
|∆| 1 + (2τϵ |∆|)2 2τϵ |∆| 2τϵ |∆|
+ σn E 1 + K p −E p ,
2T 2τϵ |∆| 1 + (2τϵ |∆|)2 1 + (2τϵ |∆|)2

where jn is the normal component of the current density, js is the superconducting component of the current

density, σn is the conductivity of normal excitations in the superconductor, K(x) and E(x) the complete

elliptic integrals of the first and second type, respectively. δ = 2τϵ ∆0 , Q = −2A + ∇θ, E = −Ȧ − ∇φ.

For acceleration of numerical computations, it is convenient to replace the elliptic integrals in the current

density expression by elementary functions. We will use a very good approximation for K(x) − E(x):

ln(1 + x) − ln(1 − x)
K(x) − E(x) ∼
= + (1 − x) ln(1 − x) (2.7)
2

Figure 9: Comparison of Approximation function to K(x) − E(x)

The figure above shows the close agreement between this approximation function and the difference between

the elliptical integrals. This approximation was used, in parallel with exact Elliptic functions, at COMSOL

modeling.

14
Though useful for steady state, time evolutions of the problems regarding the mixed states of superconductors

are of far more interest and, as seen from applications in Section 1.3, of more use to the scientist and modern

society. During his development of a time-dependent equation of the Ψ-function, Schmid[34] understood

that the Ψ-function for finite gap was not like the typical quantum mechanic Schrödinger wave equation,

but more like that of a diffusion equation.

 2 !
∂Ψ 1 ∂
= + (1 − |Ψ|2 ) Ψ. (2.8)
∂t 2m ∂r

This equation is a parabolic partial differential one that allows for straight forward implementation into the

modeling software. Further confirmation was made by Gor’kov and Eliashberg[35] from the full development

of a model of superconductivity based on the Green’s function.

2.2 BCS Theory

At typical temperatures, electrons are fermions with half-integer spin that are restricted by the Pauli ex-

clusion principle. Unlike bosons, particles with integer spin, they cannot occupy the same energy state.

However, it was realized by Cooper[36] that electrons in superconductors cooled at T= 0 K, or below Tc ,

show a weak attraction that can pair up the electrons known as Cooper pairs, to facilitate the phenomena

of superconductivity, and cause them to behave like bosons into a condensate, operating on the same energy

level. The weak attraction is facilitated by the polarization of the ionic lattice by the first electron traveling

in the pair, allowing the second electron to overcome the repulsive screening of the Couloumb interaction

due to the residual excessive positive lattice charge left by the first electron. The repulsive screening of the

Couloumb interaction is expected of two equal charges, and this is what made Cooper’s idea revolutionary

in explaining the microscopic theory. Phonons must be taken into account due to the fact that the ionic

lattice is elastic in this situation, in which the ionic lattice fluctuates in the present of the traveling Cooper

pairs, despite being at extremely cold temperatures to which most materials are solids and at their most

rigid state.

There are two methods of deriving the BCS Theory with its assumptions. One is the variational method,

used in the original BCS paper[37]. This gives the direct solution for calculating the condensation energy

of the superconducting ground state relative to the normal state. The other method is the more modern

one[38], by canonical transformation using the product of operators based on the pairs of Bloch states

(k ↑, −k ↓): (c−k↓ ,ck↑ ) in a model Hamiltonian in a self-consistent way (bk =< c−k↓ ck↑ >av ) while also

15
introducing a quasi-particle excitation operators, γk , known as Bogoliubons [2]. Both methods come to the

same conclusion, the BCS occupation fraction, vk2 = 1 − u2k :

1 ξk 1 ξk
vk2 = (1 − ) = (1 − ). (2.9)
2 Ek 2 (∆2 + ξk2 )1/2

2.2.1 Connection to BCS’s ∆ function

While Ginzburg and Landau theory was formulated a few years before BCS theory, it was suggested in the

coming years by Bardeen, Cooper and Schrieffer that there may be a resemblance of their energy gap, ∆BCS

to the Ψ-function of Ginzburg and Landau’s theory [37]. This encouraged Gor’kov [39, 40] to formalize this

speculation into the fact that, near Tc the modulus of the Ψ-function is equivalent to the energy gap of

∆BCS , both describing the Cooper pair density in the superconducting state. Gor’kov utilized the Green’s

function for superconductors:

 
Gαβ (x, x′ ) = −i⟨T Ψα (x)Ψ†β (x′ ) ⟩, (2.10)

where T is the tensor product, to transform the BCS theory into:

∇2
 
∂  
i + + ϵF Gαβ (x, x′ ) + iζ⟨T (Ψ†γ (x)Ψγ (x))Ψα (x)Ψ†β (x′ ) ⟩ = δ(x − x′ )δαβ . (2.11)
∂t 2m

As is fully developed in Section 3.4 of Dr. Gulian’s book[4], solutions of the Green’s function and replacement

of variables will lead one to the result Gor’kov and Eliashberg[35] discovered:

"  2  −1  #
1 ∂ 7ζ(3) T − Tc 7ζ(3) 2
− −i − 2eA(r) + ϵF × + |∆(r)| ∆(r) = 0. (2.12)
4m ∂r 6(πTc )2 Tc 8(πTc )2

The previous equation is what connects BCS theory with the TDGL equations and by visual comparison,

the form of this equation matches TDGL. Though limited to the Tc range, the TDGL equations provide

allowance of a spatial variation of the energy gap and ns compared to the generalized London theory. The

16
TDGL equations allow for modeling of type II superconductors as well as type I and provide a clear and

precise modeling method.

2.3 Ahanorov’s Interpretation

Per conversations between Dr. Gulian and Dr. Aharonov, there is an understanding that the quantum

behavior of particles in a cavity has been described by the imaginary Hamiltonian, iĤ and the Schrödinger

equation and the gapless TDGL are equivalent in case of iĤ. Below is the explanation of the statement

above. The Schrödinger equation is

∂Ψ
i = ĤΨ (2.13)
∂t

Replace Ĥ with the kinetic and potential energy:

p̂2
 
∂Ψ
i = +U Ψ (2.14)
∂t 2m

Change the basis:

 2 !
∂Ψ 1 ∂
i = − +U Ψ (2.15)
∂t 2m ∂r

Divide both sides by i:

 2 !
∂Ψ 1 ∂
= −i − +U Ψ = −ĤΨ. (2.16)
∂t 2m ∂r

Now taking the simplest form of TDGL:

 2 !  2 !
∂Ψ 1 ∂ 2 1 ∂ 2
= + (1 − |Ψ| ) Ψ = − − + (|Ψ| − 1) Ψ (2.17)
∂t 2m ∂r 2m ∂r

Replace U = (|Ψ|2 − 1) and we obtain similarities with Equation (2.16):

 2 !
∂Ψ 1 ∂
−ĤΨ = = −i − +U Ψ. (2.18)
∂t 2m ∂r

In quantum measurement theory, the collapse of the wave function is an important topic because this theory

17
is not yet constructed. With the TDGL equations, the Ψ-function oscillates between 0 and its maximum

value during dynamic processes. Thus, analysis of the TDGL set of equations may be of use in understanding

of the disputed topic.

2.4 Dimensionless Form of TDGL Equations

For implementation into the modeling software, one is recommended to change the partial differential equa-

tions to be solved in their dimensionless form, such that the variable |Ψ| is scaled between 0 and 1. This is

in accordance with pi-theorem[41] which states the physics and its dynamics are preserved regardless of the

units used. Below is the dimensionless form of the order parameter:

2
δ2 ∂|Ψ|2 ]

1 ∂ |Ψ| i
Ψ + 1 − |Ψ|2 + p Ψ.


p =− p Ψ− ∇+A (2.19)
1 + δ 2 |Ψ|2 ∂τ 2 1 + δ 2 |Ψ|2 ∂τ κ

The dimensionless replacements are as follows: δ = 2τϵ |∆0 |, η = (Tc −T )/Tc , ψ = ∆/∆0 : dimensionless order

parameter, κ = λL /ξ: Ginzburg-Landau parameter with the London penetration depth λL = [8π 4 σn (Tc −

T )/(7ζ(3))]−1/2 , t0 = ξ 2 /D and coherence length ξ = [πD/8(Tc − T )]1/2 , τ = tD/ξ 2 = t/t0 : dimensionless

time, p = P (|∆|)/η: dimensionless phonon feedback, u = π 4 /[14ζ(3)], σ = σn /t0 ), and gauge choice ϕ = 0.

The same replacements are made for the current density:

η ∂|ψ|2
   r 
˜ ˜ i  ∗˜ ˜ ∗ 2
∇×∇×A=− A+ ψ ∇ψ − ψ ∇ψ |ψ| −2δ
2κ|ψ|2 u ∂τ
( p " ! !#) (2.20)
2√ (|ψ|δ)2 + 1 |ψ|δ |ψ|δ
−σ̃ Ȧ 1 + ηu K p −E p .
π δ |ψ|2 δ 2 + 1 |ψ|2 δ 2 + 1

The dimensionless quantities ∼ will be dropped later on. After the dimensionless equation is formed, Ψ must

be solved for by the separation of parts of the real Ψ terms and imaginary Ψ terms. The full and detailed

solution for 2-D is in the Appendix A (12).

18
2.5 Choice of Gauge

The choice of gauge is related to Maxwell’s equations, allowing 1 degree of calibration freedom. There are

4 potentials: vector and scalar: Ax , Ay , Ax , and φ. We can choose one constraint imposed by the gauge

invariance condition: div(A) = 0, or Ax = 0, or, alternatively, φ = 0. When we reduce dimensionality, we

can impose additional constraints: each dimension - 1 additional constraint.

For example, for 3-D cases, we made the gauge choice of φ = 0. To explain how we can make this choice, let

us talk about the general case. Since H = curl(A), we can add a gradient of any function (χ) to A without

changing the magnetic field:

Anew = A + ∇χ → Hnew = H. (2.21)

Since E = −∂A/∂t − ∇φ:

Enew = −∂Anew /∂t − ∇φnew = −∂(A + ∇χ)/∂t − ∇φnew = E. (2.22)

if φnew = φ − ∂χ/∂t. Therefore, to have φnew = 0 , we make a gauge transformation with ∂χ/∂t, and the

φ-variable can be dropped.

For 2-D cases: We make the Hext homogeneous in space by setting Hext along the z-axis. Then, in addition

to the condition of φ = 0, we have Az = 0 and we consider only Ax and Ay in our computations.

For 1-D cases: We first consider the diameter of the wires to be << λL and ξ. This allows us to treat

the current density simply as the density, since the current density is homogeneous through the wire cross-

section. ∇ (any function) is a longitudinal function, we can eliminate the longitudinal part of A by gauge

transformation. The choice we make is to set Ax along the wire and equate it to zero. If the magnetic

feedback of the current on itself is not important, since the current is kept constant by the external source,

then the curl(Atransverse ) is usually negligible. As a result, Ay and Az can also be dropped, and we are left

with just the φ-potential to be determined.

19
3 Computational Superconductivity

3.1 Comsol Multiphysics® : Equation-based Physics

COMSOL Multiphysics® is a powerful advanced numerical method simulation software that implements

equation-based models as well as fully-coupled modules preloaded with the physics required to model and

simulate dynamic changes to a designed system. The software itself allows for a good deal of customiza-

tion of solvers, but for our applications, we used the default multi-frontal massively parallel sparse direct

solver (MUMPS). MUMPS uses several independent frontal solvers, that are build on LU factorization, to

simultaneously solve the sparse matrices while reserving memory for the frontal solvers and acting on the

whole matrices set to be solved. For our cryocooler 1-D model, we have used Highly Non-linear Newtonian

solver settings to reduce the computational time and to preserve accuracy. For our models, we wanted to use

the set of time-dependent Ginzburg-Landau (TDGL) equations with our inclusion of the phonon term and

interference current term. Therefore, we had to solve for the set of equations, reduce it into a dimensionless

form to be implemented into COMSOL’s Equation-based Physics module. COMSOL recommends reducing

the equation into its dimensionless form for ease of implementations and for the same reason pi-theorem

states: that the physics is still valid, yet the complexity is eliminated for the moment[41]. For some of our

models, especially those with results for NbN, we have restored dimensional values and gained great insight

into the dynamics of the results. There are a few different forms in which to fit the equation into the software

module, and that will be described next.

3.2 COMSOL Implementation: Creating a Model

Below, I will describe how to create a model in COMSOL, beginning with the instructions for a 1-D model.

This will be the basis for understanding our more complex models, in their increased dimensions. The

specifications for each model build will be expanded in greater details in each section respectively.

Select Model Wizard after opening COMSOL. For Space Dimension the selections of geometries are

below in the Fig. 10:

Figure 10: Space Dimensions from Model Wizard

20
For our models, we choose 1-D, 2-D and 3-D. The axisymmetric options were not necessary for our models

but gave a different view point depending on the locations we were calculating from (cross-sections). After

selecting our geometry, we selected the Physics for our model as seen in Fig. 11 Drop down the Mathematics

option and select General Form PDE and add as many PDEs as needed. For our 1-D models, we used

three PDEs: one for the real part of Ψ, one for the imaginary part of Ψ, and one for the vector potential,

Ax . We then select Study at the bottom of the window.

Figure 11: Physics module selection in COMSOL

21
We then need to select the type of study, as seen in Fig 12. All of our models are time-dependent, therefore

we select from the General Studies options: Time Dependent and select Done.

Figure 12: Study selection in COMSOL

This will open the Model Builder settings as seen in Fig 13. First, we will build the 1-D geometry: right-

click on Geometry and select Interval. For the settings of Internal, insert Coordinates: -L0 and L0.

Under the Model Builder options, select Parameters and create a parameter with Name: L0, and set its

Value: 10.

22
 Figure 13: Model builder settings for the General PDE used for Equation-based modeling

The General Form PDE Equation is:

∂2u ∂u
ea 2
+ da +∇·Γ=f (3.1)
∂t ∂t

∂
with ∇ = ∂x . We define General Form PDE 1 for u = u1 (x, t) = Re(Ψ) and under the Settings, insert

the following:

Γ = -ux/kappa^2

23
f = -uxx/kappa^2+((1+delta^2*u2^2)*uxx)/(kappa^2*sqrt(1+delta^2*(u^2+u2^2)))

-(delta^2*u*u2*u2xx)/(kappa^2*sqrt(1+delta^2*(u^2+u2^2)))

+(2*(1+delta^2*u2^2)*((u3+A0)*u2x))/(kappa*sqrt(1+delta^2*(u^2+u2^2)))

+(2*delta^2*u*u2*((u3+A0)*ux))/(kappa*sqrt(1+delta^2*(u^2+u2^2)))

+(u2 *sqrt(1+delta^2*(u^2+u2^2))*u3x/kappa)

-(u*((u3+A0)^2)/sqrt(1+delta^2*(u^2+u2^2)))

+(u*(1-u^2-u2^2+p*(((x-x0)/2)^2<width))/sqrt(1+delta^2*(u^2+u2^2)))

da = 1 and ea = 0

Select the Initial Values 1 and enter in u2=1. Right click on General Form PDE 1 and select Dirichlet

Boundary Condition. Enter r=1 for Prescribed value of u.

We then define General Form PDE 2 for u2 = u2 (x, t) = Im(Ψ) and under the Settings, insert the

following:

Γ = -u2x/kappa^2

f = -u2xx/kappa^2+((1+delta^2*u^2)*u2xx)/(kappa^2*sqrt(1+delta^2*(u^2+u2^2)))

-(delta^2*u*u2*uxx/(kappa^2*sqrt(1+delta^2*(u^2+u2^2))))

-(2*(1+delta^2*u^2)*((u3+A0)*ux)/(kappa*sqrt(1+delta^2*(u^2+u2^2))))-

(2*delta^2*u*u2*((u3+A0)*u2x)/(kappa*sqrt(1+delta^2*(u^2+u2^2))))-

(u*sqrt(1+delta^2*(u^2+u2^2))*(u3x)/kappa)

-(u2*((u3+A0)^2)/sqrt(1+delta^2*(u^2+u2^2)))

+(u2*(1-u^2-u2^2+p*(((x-x0)/2)^2<width))/sqrt(1+delta^2*(u^2+u2^2)))

da = 1 and ea = 0.

Select the Initial Values 2 and enter in u2=0. Right click on General Form PDE 2 and select Dirichlet

Boundary Condition. Enter r=0 for Prescribed value of u2.

Lastly, we define General Form PDE 3 for u3 = u3 (x, t) = Ax and under the Settings, insert the follow-

ing:

Γ = 0 and ea = 0.

f = -(u3+A0)*(u^2+u2^2+2*delta*a11*(2*u*ut+2*u2*u2t))

+((u*u2x-u2*ux)/kappa)*(1+2*(delta/(u^2+u2^2))*a11*(2*u*ut+2*u2*u2t))-j0

da = sigma*(1+(1/pi)*((eta*5.798*(u^2+u2^2))^(1/2))*exact(1/(1+1/((u^2+u2^2)*delta^2))^(1/2)))

Select the Initial Values 3 and enter in u3=-j0-A0. Right click on General Form PDE 3 and select

Dirichlet Boundary Condition. Enter r=-j0-A0 for Prescribed value of u3.

The parameters for the models must be defined under Global Definitions − > Parameters. Insert the

24
following names and values: sigma = 0.172, j0 = 0.4, kappa = 0.3, A0 = 0, p = 0, x0 = 0, width = 0.1,

delta = 0.1, eta = 0.2 and a11 = (eta/5.798)^2.

As mentioned in Section 2.4, for the dimensionless form of TDGL, a function of composite natural logs was

used in good agreement and replacement of the complete elliptical integrals of the first and second type,

K(x) and E(x). To insert that replacement function into our model, right-click on Global Definitions,

select Functions − > Analytic. Enter the following into Expression: ((log(1-x^2))/x)-2*log(1-x).

3.3 Working Model

Now that the model has been fully built in the Equation-based module in COMSOL Multiphysics® , we

can run the simulation and plot results. First, the Mesh settings under the Model Builder options must be

customized to User-controlled mesh with Custom Element Size Parameters: Maximum element size is set

to 0.005, Maximum element growth rate to 1, and Resolution of narrow regions to 1.

Setting the Study 1 under the Model Builder options, select Step 1: Time Dependent and editing

the settings for Output times: range(0,0.1,100). Now we can Compute.

After the solver has completed its tasked, we can create plots under Results to see the solutions. Right-click

on Results and select 1-D Plot Group. Right-click on 1-D Plot Group and select Line Graph. In the

settings, enter sqrt(u^2+u2^2) into Expressions and select Plot. This will produce a plot of modulus of

Ψ(x) also known as the CP density, along the 1-D wire at different moments in time. See Figure 14.

Figure 14: Plot of Cooper pair density along 1-D wire at different moments in time from the model built in
this section.

25
4 Implementation of 1-D Model

The 1-D wire is idealized as being between two massive superconducting banks, as visualized in Figure 15.

The thickness, as mentioned previously, is of dimensional thickness ≈ λL , the London penetration depth

The CP density is sagging at the center due to the nearest of the critical current, jc , but it has not lost

superconductivity yet. Inset is the CP Density vs. dimensionless time at the center of the wire as it holds a

steady value less than 1 but greater than 0.

Figure 15: The lower left corner shows the idealization of the 1-D wire between two massive superconducting
banks. It is set upon a plot of a under dampened phase slippage of the modulus of Ψ vs. x (dimensional
length) or the CP density along the wire. Inset, lower right, is the CP Density vs. dimensionless time at the
center of the wire.

For the visualization of PSCs and phonon feedback results, the above Section 3.3 has the set of the equations

used for the model with gauge choice of φ = 0, leaving us with just Ax to implement.

For the 1-D cryocooler results, the vector potentials are dropped: Ax = Ay = Az = 0 and the influence of

the magnetic induction B is dropped in correspondence. The implementation is similar to those detailed in

Section 3.2 (General Form PDE1 models the Re(Ψ) and General Form PDE2 models the Im(Ψ)) with a few

exceptions:

• Dirichlet boundary conditions: u1 = cos(u3) and u2 = sin(u3). u3 is the phase of the wave function

Ψ.

26
 • The phase u3 is modeled by a Boundary ODES and DAES, since the values of the phases are required

only at the ends of the wire. Therefore, we chose the Distributed ODE of form:

∂ 2 u3 ∂u3
ea 2
+ da =f (4.1)
∂t dt

with f = −φ, ea = 0, da = 1, and initial value: u3 = 0.

This above is for Study 1, to calculate the behavior of Ψ and the scalar potential, φ.

We created a second study: Study 2 to calculate ∇φ. We used the choice of an algebraic equation to create

a function, F (u4) = 0 = 4 − withsol(′ sol1′ , f i, setval(t, t)) and u4 is implemented with another Distributed

ODE with f = −1, ea = 0, da = 1, and initial value: u4 = 0.

During plotting with ∇φ, we used centroid(−u4x) to avoid boundary issues of the plot itself. The mesh was

set to the Extremely fine mesh, with a total of less than 100 degrees of freedom. For comparison, the 3-D

model required up to 14 million DOF , Section 10.2.

27
5 Dynamics of Phase-Slips Centers (PSCs)

5.1 Visualization of PSCs

The order parameter is dependent on the quantum phase: θ(x, y, z, t). The phase difference of the order

parameter between the ends of the superconducting wire provides the dynamics of the phase slippage and

jumps down by periods of 2π. This is analogous to a weakly-linked Josephson junction, as discussed in

Section 1.2. Since the density of pair condensate is governed by the order parameter, it must vanish at

certain times, in most of these cases this occurs at the center of the wire when under no influence of external

sources (Hext , phonon flux, etc.). This results in oscillations of the density of pair condensate and the current

components. Experimental[42] and theoretical work[43, 44] has been done on 1-D PSCs, but few take into

account the interference current and the phonon influence besides the work of Gulian and Zharkov[33]. We

will see from the results for 1-D models all the dynamics that the set of TDGL equations reveal: single and

double phase slip centers, periodic oscillatory behaviors, changes to the PSC locations via increase of the

energy gap revealing a bifurcation, phonon feedback introducing changing in the curvature and response of

PSCs, voltage readings of these oscillatory behaviors, and their usefulness as phonon absorbers and emitters

as the basis for single-particle detectors, radiation detectors, and cryocoolers.

Figure 16: Set of single phase slippage along a 1-D wire versus the modulus of Ψ (square root of density of
pair condensate), evolving over time for increasing τϵ |∆| .

28
I began my PhD research with the simplest model: the 1-D wire and explored how to visualize and create

videos to understand the dynamics of these superconductors using the set of TDGL equations. I first explored

how the increase of the dimensionless energy gap τϵ |∆| (in these models, τϵ |∆| is dimensionless, as mentioned

in the previous Section 2.4) affects the critical current. As seen from Figure 16, as the energy gap (delta)

increases, the critical current lowers. Notice also the time evolution of these PSCs. Not only does the

critical current change with increased energy gap, but the time it takes for the PSC to occur elongates. The

increasing of the 1-D model’s τϵ |∆| parameter shows that the period increases for these oscillatory behaviors.

As will be seen with the phonon feedback results, this period can be increased with introduction of increased

phonon term/hot spots.

Figure 17: Set of double phase slippage evolving over time for increasing τϵ |∆|.

Another interesting computation that came out of this introductory start to visualization of PSCs using

TDGL equations was the double phase slip center, as seen in Figure 17. The locations of the double PSCs

are symmetric about the center of the wire and touch zero, one after the other, but not simultaneously. Note

the value of the delta/energy gap at the top. These values are in between the values of those on the previous

figure, Figure 16. This led to my next result that involved automation to find the locations of these PSCs,

singles and doubles, to reveal a pitchfork bifurcation.

29
5.2 Pitchfork Bifurcation of PSC Locations

Figure 18: Location of PSCs with increased τϵ |∆| with branching and anti-branching.

During an exploration of the effects of increased energy gap τϵ |∆0 | (dimensionless) and the location of the

first PSC with finding the critical current, jc , I created an automated search for these two variables and

stored the locations of the PSCs at their critical currents for each energy gap value. I discovered a pitchfork

bifurcation as seen from Figure 18.

The automation was enabled with the Livelink® which connected MATLAB® with COMSOL® to allow

programming of the study and search. With each τϵ |∆0 |, I had the program increase j, search for a modulus

of Ψ values nearly close to zero; if none, j is increased, if so, the values of location(s), τϵ |∆0 | and jc are stored

and τϵ |∆0 | is increased, and the search starts again. The details of how to automate are in the following

section, with the coding is in Appendix B (12).

This work, along with the visualizations and phonon feedback, was presented at the COMSOL Conference

- Boston October 2019.

30
5.2.1 Automation for Pitchfork Bifurcation

REQUIRED SOFTWARE

1) COMSOL Multiphysics® With LiveLink™ for MATLAB®

2) MATLAB®

This series of automation .m files were applied to superconducting 1-D wire COMSOL model. This particular

model was created by implementing the set of Ginzburg-Landau equations into COMSOL’s time-dependent

Equation-based module with the general coefficient PDE form. Linking Comsol with MATLAB allowed us

to automate several tasks for finding solutions, and each task is described below and in the .m files.

Few tips for getting started

• Launch COMSOL Multiphysics® (Your Version) with MATLAB® Program

Make sure all other programs, even COMSOL and MATLAB standalones, are closed. This singular

program will control both.

• Load model from command line or into .m file

Ensure that you are in the same directory/folder as your COMSOL file.

model = mphload(’ModelName.mph’)

• Access the properties of your model

mphlaunch

mphlaunch will launch your model in the COMSOL user interface. This can be useful as you learn

more about the mphnavigator settings and how to extract m file code for automating.

or

mphnavigator

mphnavigator allows you access to the elements and subproperties of your model. Here is where you

will find the tags for your model and can vary this code according to your model’s tags. There are many

useful lines of code that can be extracted using the ‘Copy Get’ or ‘Copy Set’ options in this window.

31
Other Useful Code Setting values for parameters

L= 10; Set variable value

model.param.set(’L0’, L); Set parameter ’L0’ to value of L

model.study(’std1’).run; Run the study

model.result("pg6").run(); Generate a plot

model.result.export(’anim2’).run; Export an animation/video

model.sol(’sol1’).clearSolutionData(); Clear solution 1

AutomationMinimaCount.m

This file sets a start value for two parameters. A while loop sets an upper limit for one particular parameter.

The AutomationMinimaCount.m file automated the following tasks:

• Setting the new parameters

• Running the study

• Running the plot

• Exporting the plot data (.txt file)

• Loading the plot data into MATLAB

• Finding local minima with a certain prominence. See documentation on: islocalmin

• Count local minima as n and display

• Runs a set of if-else if conditions in which a second parameter is varied, the model solution is cleared

and study is rerun or

• If a particular minima condition is met, the parameter values are stored in a .mat file

• A filename is created from the parameter values

• The plot is created with title and marking of the minima in with red star marker

• The plot is saved as a png and the parameters in the while loop is varied

32
MinimaLocations.m

The MinimaLocations.m file automated the following tasks:

• Loads a .mat data set and stores the columns/parameters

• For loop over all elements

• Load model and clear solution data

• Set parameters

• Run study

• Run plot

• Export plot data

• Import plot txt file

• Checks that plot is not empty and prints error message

• Else finds locations of minima

• Stores unique values of minima locations

• Plots these locations

• Stores each location in list with parameters

The coding for the automation is in Appendix B (12).

33
5.3 Dynamics of PSCs with Phonon Feedback

(a) p = 0

(b) p = 0.2 (c) p = 0.8

Figure 19: Spatial distribution of Cooper pair density for different intensities of phonon flux action at
established periodic PSC oscillations. j0 = 0.4 and τϵ |∆| = 0.1. Noticeably, at the beginning of oscillations
period, the sagging starts at the point where the phonon source is located. At further evolution, the sagging
moves towards the geometrical center of the wire. The hotter the phonon spot is, the closer the location of
the phase slip center to it.

The phonon term in the order parameter of the TDGL equation shows interesting effects on the PSC’s

location when a hot spot (positive phonon fluxes with elevated temperature) is introduced at x = 5 for the

Figure 19. The location of the hot spot is marked on the plot with a vertical dashed red line. As the upper

plot with p = 0 shows, the hot spot is absent and produces no change to the central location of the expected

PSC. The two lower plots in Figure 19 show the main results: the PSC moves closer to the hot spot with

increased |p| value. Notice the depth of the hot spot as p increases, marking an important result to modeling

with the phonon term: there is a “weakness” due to the elevation of the effective temperature. The hot spot

can essentially control the locations of PSC in the manner desired.

34
 Figure 20: Increase of |p| reduces the period of PSC oscillations.

In Figure 20 is another result that the dynamics of the TDGL-equation modeling reveals to us about the

1-D wire in regard to phonon feedback: the PSC oscillations. The frequency of the PSC oscillations are

proportional to the elevated temperature introduced through the p term. As seen from the left plot of

Figure 20, there is a shortening of the period of oscillation as |p| varies from p = −0.1 to p = −0.2, relative

to the change of the CP density versus dimensionless time. The right plot of the same figure shows this

period is exponentially decreasing with increased |p| value.

The periodic behavior of the induced changes in the PSC can be detected by measuring the voltage between

the ends of the wire. The mixed state exchanges supercurrent with normal current and back, with an electric

friend vector E induced in the wire.

35
Figure 21: Voltage vs. time across the 1-D wire when PSC oscillations take place with small (p = 0) and
high (p = -0.9, located at the center of the wire) frequencies. Parameters are ∆ = 0.1 and jc = 0.4.

We can calculate the voltage of the computed results in the software with the integral:

Z
V (t) = E(t)dl. (5.1)

The current supplied is constant, yet the voltage vs. time in Figure 21 shows unidirectional voltage spikes

with an average non-zero value. The higher the frequencies due to the higher p term, the higher the average

voltage value is. For Figure 21, V̄highf requency = 1.34 and V̄lowf requency = 0.17. The critical current plays an

essential role is the oscillatory behavior of the 1-D superconducting wire, as it is causing this non-equilibrium

behavior despite being a dc current.

The practical applications of this type of result is its basis for a highly sensitive single-particle or radiation

detectors[45, 46, 47]. We will see that this basis is useful in the next section regarding the cryocooler as the

substrate, in which the wire is located, can absorb and emit phonons through this kind of mechanism.

This work was presented at the 8th International Workshop on Numerical Modelling of High Temperature

Superconductors - Nancy, France June 2022, APS March Meeting - Las Vegas 2023, and is available on arXiv:

Mowgood, I., Teknowijoyo, S., Chahid, S., and Gulian, A. (2022). Novel results obtained by modeling of dy-

namic processes in superconductors: phase-slip centers as cooling engines. doi:10.48550/arXiv.2211.01286

36
6 PSC Phonon Cryocooler

Utilizing all the knowledge from the previous 1-D results, we turn to the usefulness of these PSCs in super-

conducting wires as a method to transport phonons (heat) from one medium to another. What makes these

PSC results unique is that a critical current, jc , is supplied as a dc current, yet the voltage will oscillate

in time with a frequency: ωP SC ∝ Vdc . This is similar to the Josephson junction discussed in Section 1.2.

However, there is no actual weak link/insulator sandwiched between. Instead, the PSC locations act as the

weak point of the system due to its symmetric location, and we classify our PSCs as “strong” weakly-coupled

systems.

2
Figure 22: (a) Temporal oscillations of Cooper pair density |Ψ(x, t)| in the 1 − D−wire at J = 1.5 (J0 ≈ 1).
Moments of time marked by arrows correspond to those in panel (b). (b) Time oscillations of |Ψ(x, t)| at
the central point of PSC (x = 0). (c) Voltage between the ends of the 1 − D−wire. (d) Supercurrent, (e)
normal and (f ) interference currents for various characteristic moments of PSC oscillation (color coding is
the same in panels a, d, e and f ).

As seen in Figure 22, the PSCs in the wire behave as the previous results of 1-D wires have been presented

in this section: (a) the CP density dips at the center of the wire with time at values of jc , (b) the oscillatory

period is steady with time, and (c) non-zero unidirectional voltage spikes are produced. What is new here

are the panels (d), (e), and (f): the time evolution of the supercurrent, js , the normal current, jn , and the

interference current, jint , respectively. The total current supplied to the system is steady and, though the

components of the current vary, their sum always equals the supplied value. Including jint provides greater

illustrative details of the interaction of the mixed states of superconducting and normal electrons, with js

37
remaining a smooth curve, while jn and jint show dramatic peaks at the time of the phase slippage.

In Figure 23, we see the main results calculated from the modeling: the phonon flux and its reversal of its

sign. For Fig 23a), the red curve corresponds to the unpaired electronic excitations from the general kinetic
2
equation: δnϵ = −α2 ∂|Ψ(x,t)|
∂t . The coefficient alpha is:

α = (1/2T )cosh−2 (ϵ/2T )(R2 /N1 ) (6.1)

where

 p 
N1 (ϵ) = Re [ϵ + iγ]/ (ϵ + iγ)2 − |Ψ|2 (6.2)

and

 p 
R2 (ϵ) = Re |Ψ|/ ϵ + iγ)2 − |Ψ|2 (6.3)

Figure 23: (a) This form of ∂|Ψ(x, t)|/∂t at a PSC (red curve) causes a pulsating phonon flux spectrum.
The flux reverses its sign during the period of PSC oscillation, Cases 1 and 2. (b) Curves corresponding
to phonon term. At a given moment of time, independently of ∂|Ψ|/∂t, the spectrum of phonon flux has
positive and negative segments, separated by the frequency ωq = 2|Ψ|. Additionally, it depends on variables
|Ψ|, T, and γ. The curves shown are for values of T = 0.9Tc and γ = 0.3.

Case 1 demonstrates the flux becoming more negative as the CP density decreases to zero, on its way to the

PSC event. Case 2 illustrates the reversal of flux sign to positive after the PSC event and the CP density is

nearing its max value. The signage of the flux is important to the basis of the crycooler principles: positive

flux is when there is a production of phonons from the superconducting wire, and negative flux is when there

is absorption of phonons. Gulian and Zharkov have predicted this flux behavior in their book[33] but had

38
not suggested its use as a cryocooler.

Fig 23b) shows phonon flux analysis described by Equation (2.5), the distribution of the non-equilibrium

electrons, nϵ . This was calculated from the derivative of ∇φ mentioned in Section 4 due to the choice of

gauge. It shows Cases 1 and 2, negative flux (absorption of phonons, δnϵ < 0) and positive flux (emission of

phonons, δnϵ > 0), respectively. Inset is the result of the difference between the positive and negative flux in

the recombination phonon range with frequency ωq = 2|Ψ| = 1.2. This behavior is governed by the function

of the phonon flux, dWωq , and its dependence on ωq :

dWωq = I(Nω0q )(ph−e) ρ(ωq )dωq (6.4)

where I(Nω0q )(ph−e) is the phonon-electron collision integral [4] and

ρ(ωq ) = Bωq3 /(2π 2 u3 ) (6.5)

with volume B and u, the speed of sound in the superconductor also known as the acoustic phonon propa-

gation speed. Accordingly, u is related to the frequency properties of the phonon flux and a good choice of

materials will yield the effects of the Kapitza resistance.

Figure 24: Cross-sectional view of the cooler design based on PSC-filament with two surrounding plates.
Acoustic densities should satisfy the relation: ρ1 u1 < ρ2 u2 < ρ3 u3 .

In Figure 24, we have the schematics of the design for the cryocooler. Each layer; the top plate, supercon-

ducting filament, and substrate on top of cold finger, have unique mechanical densities, ρ, and u values.

39
The products of their acoustic density, ρu increase from top layer to bottom layer and the increased density

downwards acts like of the principles of a refraction index, similar to those in optics, at the interfaces of the

layers. The phonons emitted from the top layer will be accepted by the lower layers due to the higher acoustic

density, with the 1-D superconducting wire mitigating the transfers during its flux variation states. Recip-

rocally, they will suffer total internal reflection at the interfaces with lower acoustic density. With the use of

a thermal diode, which is modeled in Section 7.3 in 2-D, the design is improved further for phonon-sinkage

to the cold finger.

40
7 2-D Modeling of Superconducting Diode

7.1 Implementation of 2-D Model

We always require two PDEs for Re(Ψ) and Im(Ψ), but as seen in Appendix A (12), not only must Ψ be

separated for the real and imaginary parts, in addition it must be solved for 2-D, x and y. For the 2-D

results, the gauge choice is to have the scalar potential dropped, φx = 0, and we are left to solve Ax (with

the direction of the transport current following in the x-direction), Ay and the electric field, E = −∂A/∂t.

The phonon term is p = p(x, y) = −0.3 along the top of the 2-D slab in order to weaken the order parameter,

thus breaking the inversion symmetry.

The implementation for 2-D:

∂ ∂
The General Form PDE Equation, with ∇ = [ ∂x , ∂y ], is:

∂2u ∂u
ea 2
+ da +∇·Γ=f (7.1)
∂t ∂t

General Form PDE 1 for [ux , uy ] = Re(Ψ):

Γ = [-ux/kappa^2, -uy/kappa^2]

f = 2*(u3*u2x+u4*u2y)/kappa+u2*(u3x+u4y)/kappa-u*(u3^2+u4^2)+u*(1-u^2-u2^2+p*(y>2.5))

da = 1 and ea = 0

Initial value for u = 1

Zero Flux Boundary Condition, −n · Γ = 0, is applied on all four sides.

Periodic Boundary Condition, udestination = usource ; −ndst · Γdst = nsrc · Γsrc , is applied to right and

left sides. Periodic Condition: Continuity and Apply condition on variable u.

General Form PDE 2 for [u2x , u2y ] = Im(Ψ):

Γ = [-u2x/kappa^2, -u2y/kappa^2]

f = -2*(u3*ux+u4*uy)/kappa-u*(u3x+u4y)/kappa-u2*(u3^2+u4^2)+u2*(1-u^2-u2^2+p*(y>2.5))

da = 1 and ea = 0

Initial value for u2 = 0

Zero Flux Boundary Condition, −n · Γ = 0, is applied on all four sides.

Periodic Boundary Condition, u2destination = u2source ; −ndst · Γdst = nsrc · Γsrc , is applied to right and

41
left sides. Periodic Condition: Continuity and Apply condition on variable u2.

The last two General Form PDEs represent the curlA = B:

General Form PDE 3 for [u3] = Ay , Ex = −∂Ax /∂t:

Γ = [0, u4x-u3y]

f = (u*u2x-u2*ux)/kappa-(u^2+u2^2)*u3

da = sigma and ea = 0

Initial value for u3 = 0

Zero Flux Boundary Condition, −n · Γ = 0, is applied on right and left sides.

Flux/Source 1 Boundary Condition, −n · Γ = g − qu3, is applied on top and bottom sides:

g = -Ba*sin(pi+omega*t)-Hext*sign(y)

q=0

Periodic Boundary Condition, u3destination = u3source ; −ndst · Γdst = nsrc · Γsrc , is applied to right and

left sides. Periodic Condition: Continuity and Apply condition on variable u3.

General Form PDE 4 for [u4] = Ay , Ey = −∂Ay /∂t:

Γ = [-u4x+u3y, 0]

f = (u*u2y-u2*uy)/kappa-(u^2+u2^2)*u4

da = sigma and ea = 0

Initial value for u3 = 0

Periodic Boundary Condition, u4destination = u4source ; −ndst · Γdst = nsrc · Γsrc , is applied to right and

left sides. Periodic Condition: Continuity and Apply condition on variable u4.

7.2 Theory and Model

Chahid, Teknowijoyo, and Gulian, at the Advanced Physics Laboratory, constructed superconducting Nb3 Sn

micro-bridges by stacking Nb and Sn layers onto a sapphire substrate, annealing the stacked layers to

evaporate excess Sn, to create the films. The films transformed into bridges through a combination of 3-D-

printing, photo-lithography and ion milling. Contact pads were constructed on top of the sapphire substrates

to obtain results below critical temperature, varying polarity of applied current, varying frequencies, and

polarities of magnetic fields orthogonal to the surface of the bridge up to 50-100 Oe. Below in Figure 25

42
are the different stages of constructing their micro-bridges and plotted data showing that the micro-bridge

is indeed superconducting at T = 1.7 K.

Figure 25: (a) Surface morphology of the films with inclusion of small areas (nanomountains) with higher
concentration of Sn (zoomed tenfold in the inset). (b) Resistive and magnetic transitions into supercon-
ducting state. (c) A bridge after ion milling of 3D-printed pattern (initial stage of preparation). (d) Last
stage of bridge preparation: the structure in panel (c) was covered by positive photoresist, two circles were
projected via epi- fluorescent microscope and ion milled to reduce the active part of bridges to micron-scale
(the smallest, 2 µm result is shown). The inset illustrates micro-roughness of the edge line. (e) Effects of
ion milling and post annealing on transition temperature of bridges. (f) Typical voltage-current dependence
of bridges above (T = 20 K) and well-below (T = 1.7 K) the superconducting transition.

43
Figure 26: (a) Resistive state at various values of bias temperature for the applied current of opposite
polarities (absolute values are plotted). (b) The SDE in 2 µm bridge at sinusoidal current amplitude ≈ 10
mA and frequency 0.1 Hz. (c) Same as in panel (b) with 5 µm-wide bridge at frequency 10 kHz. (d) Same as
in panel (c) at frequency 100 kHz. (e) Average of 200 acquisitions with reversed polarity of magnetic field.
In all cases, a magnetic field orthogonal to the surface of the bridge was applied and optimized in the range
50-100 Oe.

The results of these experiments showed the diode effect, as seen in Figure 26. In Fig b) the superconducting

micro-bridge responses with non-zero resistance, therefore non-zero voltage, in only one direction of the

alternating current and with zero resistance (near zero voltage) in the opposite direction of the current, as

one would expect with the diode effect. In Fig c), a lower frequency is applied and a lower response of

voltage spikes unidirectional are seen. With Fig d) we see that an increase of the frequency increases the

unidirectional voltage response as well.

To further explain the effect, the microscopic behavior was modeled, relating the vortices creation and

annihilation at dissimilar edges of the bridge. To model this behavior, the set of TDGL-equations in 2-D

was implemented in COMSOL as described in Section 7.1, with consideration taken into account of the

weakening of the order parameter at the top edge and the proper geometry constructed as well.

44
7.3 Results

Figure 27: a) The entering of vortices from the top of the 2-D slab at the ac current is in its negative
polarity. b) The response voltage of the superconducting diode versus time as related to the ac current and
its polarity.

Above, Figure 27, is a moment in time of when the TDGL-equations in 2-D models the microscopic behavior

of the superconducting diode in the laboratory. The dynamics are best seen through the video that was

submitted as supplemental material during the time of publication[48]. Here we see the polarity of the ac

current is negative at this moment in time and the diode responses unidirectional to only this polarity with

a voltage spike, as seen from Fig. b. The CP density is plotted in Fig a) and shows the vortices entering in

from the top edge, and they will, with continued time, exit through the both edge, and they are the reason

for the increase in resistance and therefore the non-zero voltage response. When the ac current is in the

positive polarity (t ≈ 900), there is no response of voltage to the diode, though the edges decrease slightly in

their CP density. No vortices exist when the ac current is in the positive polarity. Further past, (t ≈ 1600),

there is again the ac current in the negative polarity and the response of non-zero voltage continues with

general lost or dissipation. This computational result not only explained the macroscopic behavior of the

diode effect based on the microscopic theory from the TDGL-equations, but it also aided in the speed of the

acceptance of the paper, as reviewers appreciated the animation [48] to better understand the theory and

results.

45
Figure 28: Evolution of Phase Slip Lines with the top of 2-D slab having three dissimilar notches that
encourage the formation of vortices. Panels a) through d) corresponding to increasing time.

In Figure 28, an unpublished result of our modeling was related to phase slip centers, which have been

observed [49] and as well as modeled[50][51][52]. In our model, the phase slip lines were encouraged by the

creation of equally distanced notches with slightly varying depth on the top edge of the 2-D slab. The panels

a) to d) show the evolution of the phase slip lines. In Panel a), the vortices start to enter the slab from the

top and there is a decreased halo of CP density around the edges of the notches and the corners produce the

vortices. In Panel b) the vortices have entered across the 2-D slab and begin to leave from the bottom edge.

In Panel c), the decreased halo of CP density around the notches began to stretch downwards and there are

lines, known as phase slip lines, that branch somewhat between the dynamics of the sea of vortices. In the

last panel, we see the phase slip lines connect to the bottom of the slab and there are lines of normal states,

phase slip lines, isolating islands of vortices.

46
Further modeling in 2-D could aid in experiments of other applications of superconducting high-performance

devices of two or more terminals.

This work was published in Physical Review B: Chahid, S., Teknowijoyo, S., Mowgood, I., and Gulian, A.

(2023). High-frequency diode effect in superconducting N b3 Sn microbridges. Physical Review B, 107(5).

doi:10.1103/physrevb.107.054506

47
8 3-D Modeling of Superconducting Rings

8.1 Implementation of 3-D Model

As usual, we still need two PDEs for Re(Ψ) and Im(Ψ), however this time we are now dealing with an

additional three PDEs to compute the vector current density, j via the components of the vector potential

Ax , Ay and Az .

The dimensionless form of Re(Ψ) = ψ1 and Im(Ψ) = ψ2 in 3-D:

1 + δ 2 ψ22 δ 2 ψ1 ψ2
ψ˙1 = p (ψ1.xx + ψ1.yy + ψ1.zz ) − p (ψ2.xx + ψ2.yy + ψ2.zz )
k2 1 + δ 2 (ψ12 + ψ22 ) k2 1 + δ 2 (ψ12 + ψ22 )
2(1 + δ 2 ψ22 ) 2δ 2 ψ1 ψ2
+ p 2 2
(A1 ψ2.x + A2 ψ2.y + A3 ψ2.z ) + p (A1 ψ1.x + A2 ψ1.y + A3 ψ1.z )
k 1 + δ 2 (ψ1 + ψ2 ) k 1 + δ 2 (ψ12 + ψ22 )
p
ψ2 1 + δ 2 (ψ12 + ψ22 ) ψ1
+ (A1.x + A2.y + A3.z ) − p (A21 + A22 + A23 − 1 + ψ12 + ψ22 ) (8.1)
k 1 + δ (ψ12 + ψ22 )
2

1 + δ 2 ψ12 δ 2 ψ1 ψ2
ψ˙2 = p (ψ2.xx + ψ2.yy + ψ2.zz ) − p (ψ1.xx + ψ1.yy + ψ1.zz )
k2 1 + δ 2 (ψ12 + ψ22 ) k2 1 + δ 2 (ψ12 + ψ22 )
2(1 + δ 2 ψ12 ) 2δ 2 ψ1 ψ2
− p (A 1 ψ 1.x + A 2 ψ1.y + A 3 ψ 1.z ) − p (A1 ψ2.x + A2 ψ2.y + A3 ψ2.z )
k 1 + δ 2 (ψ12 + ψ22 ) k 1 + δ 2 (ψ12 + ψ22 )
p
ψ1 1 + δ 2 (ψ12 + ψ22 ) ψ2
− (A1.x + A2.y + A3.z ) − p (A21 + A22 + A23 − 1 + ψ12 + ψ22 ) (8.2)
k 1 + δ (ψ12 + ψ22 )
2

The three dimensionless forms of the current density components are:

  r 
(ψ2 ψ1.x − ψ1 ψ2.x ) 2 2 η
σ [1 + β(|ψ|)] Ȧ1 = − A1 + ψ1 + ψ2 − 4δ (ψ1 ψ1.τ + ψ2 ψ2.τ )
κ (ψ12 + ψ22 ) 5.798

+ (A1.yy − A2.xy − A3.xz + A1.zz ) , (8.3)

48
   r 
(ψ2 ψ1.y − ψ1 ψ2.y ) 2 2 η
σ [1 + β(|ψ|)] Ȧ2 = − A2 + ψ1 + ψ2 − 4δ (ψ1 ψ1.τ + ψ2 ψ2.τ )
κ (ψ12 + ψ22 ) 5.798

+ (A2.xx − A1.xy − A3.yz + A2.zz ) ,

  r 
(ψ2 ψ1.z − ψ1 ψ2.z ) 2 2

η
σ [1 + β(|ψ|)] Ȧ3 = − A3 + ψ 1 + ψ 2 t − 4δ (ψ ψ
1 1.τ + ψ ψ
2 2.τ )
κ (ψ12 + ψ22 ) 5.798

+ (A3.xx − A1.xz + A3yy − A2.yz ) , (8.4)

where

p " ! !#
2p (|ψ|δ)2 + 1 |ψ|δ |ψ|δ
β(|ψ|) = 5.798η|ψ| K p −E p . (8.5)
π |ψ|δ |ψ|2 δ 2 + 1 |ψ|2 δ 2 + 1

We used the Coefficient Form PDE for the Ψ−function:

∂2u ∂u
ea 2
+ da + ∇ · (−c∇u − αu + γ) + β · ∇u + au = f , (8.6)
∂t ∂t

and the General Form Equation for the vector potential A:

∂2u ∂u
ea 2
+ da + ∇ · Γ = f. (8.7)
∂t ∂t

Here, u = (u1, u2)T , which corresponds to u1 → ψ1 and u2 → ψ2 ; and, u = (u3, u4, u5)T , which corresponds

to u3 → Ax , u4 → Ay and u5 → Az . In both equations, ea ≡ 0 and ∇ = [∂/∂x, ∂/∂y, ∂/z]. The coefficient

c is a 2x2−matrix, with the elements:

c11 = ((1 + (deltaˆ2) ∗ u2ˆ2)/((kappaˆ2) ∗ sqrt(1 + (deltaˆ2) ∗ (u1ˆ2 + u2ˆ2)))) ∗ vol (8.8)

c12 = ((deltaˆ2) ∗ u1 ∗ u2/((kappaˆ2) ∗ sqrt(1 + (deltaˆ2) ∗ (u1ˆ2 + u2ˆ2)))) ∗ vol (8.9)

c21 = ((deltaˆ2) ∗ u1 ∗ u2/((kappaˆ2) ∗ sqrt(1 + (deltaˆ2) ∗ (u1ˆ2 + u2ˆ2)))) ∗ vol (8.10)

c22 = ((1 + (deltaˆ2) ∗ u1ˆ2)/((kappaˆ2) ∗ sqrt(1 + (deltaˆ2) ∗ (u1ˆ2 + u2ˆ2)))) ∗ vol. (8.11)

49
The coefficient a is also a 2x2−matrix, with the elements:

a11 = ((u3ˆ2 + u4ˆ2 + u5ˆ2 − 1 + u1ˆ2 + u2ˆ2)/sqrt(1 + (deltaˆ2) ∗ (u1ˆ2 + u2ˆ2))) ∗ vol (8.12)

a12 = (−sqrt(1 + (deltaˆ2) ∗ (u1ˆ2 + u2ˆ2)) ∗ (u3x + u4x + u5z)/kappa) ∗ vol (8.13)

a21 = (sqrt(1 + (deltaˆ2) ∗ (u1ˆ2 + u2ˆ2)) ∗ (u3x + u4x + u5z)/kappa) ∗ vol (8.14)

a22 = ((u3ˆ2 + u4ˆ2 + u5ˆ2 − 1 + u1ˆ2 + u2ˆ2)/sqrt(1 + (deltaˆ2) ∗ (u1ˆ2 + u2ˆ2))) ∗ vol, (8.15)

while f , α, and γ are identically zero. The matrix da is:

 
 1 0 
da =  , (8.16)
0 1

and the vectorial matrix β is:

 
 β11 β12 
β= , (8.17)
β21 β22

where
 
 (−u3 ∗ 2 ∗ (deltaˆ2) ∗ u1 ∗ u2/(kappa ∗ sqrt(1 + (deltaˆ2) ∗ (u1ˆ2 + u2ˆ2)))) ∗ vol 
 
β11 =  (−u4 ∗ 2 ∗ (deltaˆ2) ∗ u1 ∗ u2/(kappa ∗ sqrt(1 + (deltaˆ2) ∗ (u1ˆ2 + u2ˆ2)))) ∗ vol 

 (8.18)
 
(−u5 ∗ 2 ∗ (deltaˆ2) ∗ u1 ∗ u2/(kappa ∗ sqrt(1 + (deltaˆ2) ∗ (u1ˆ2 + u2ˆ2)))) ∗ vol

 
 (−u3 ∗ 2 ∗ (1 + (delta ∗ u2)ˆ2)/(kappaˆ2 ∗ sqrt(1 + (deltaˆ2) ∗ (u1ˆ2 + u2ˆ2)))) ∗ vol 
 
β12 =
 (−u4 ∗ 2 ∗ (1 + (delta ∗ u2)ˆ2)/(kappaˆ2 ∗ sqrt(1 + (deltaˆ2) ∗ (u1ˆ2 + u2ˆ2)))) ∗ vol
 (8.19)

 
(−u5 ∗ 2 ∗ (1 + (delta ∗ u2)ˆ2)/(kappaˆ2 ∗ sqrt(1 + (deltaˆ2) ∗ (u1ˆ2 + u2ˆ2)))) ∗ vol

 
 (u3 ∗ 2 ∗ (1 + (delta ∗ u2)ˆ2)/(kappaˆ2 ∗ sqrt(1 + (deltaˆ2) ∗ (u1ˆ2 + u2ˆ2)))) ∗ vol 
 
β21 =  (u4 ∗ 2 ∗ (1 + (delta ∗ u2)ˆ2)/(kappaˆ2 ∗ sqrt(1 + (deltaˆ2) ∗ (u1ˆ2 + u2ˆ2)))) ∗ vol 

 (8.20)
 
(u5 ∗ 2 ∗ (1 + (delta ∗ u2)ˆ2)/(kappaˆ2 ∗ sqrt(1 + (deltaˆ2) ∗ (u1ˆ2 + u2ˆ2)))) ∗ vol

50
  
 (u3 ∗ 2 ∗ (deltaˆ2) ∗ u1 ∗ u2/(kappa ∗ sqrt(1 + (deltaˆ2) ∗ (u1ˆ2 + u2ˆ2)))) ∗ vol 
 
β22 =  (u4 ∗ 2 ∗ (deltaˆ2) ∗ u1 ∗ u2/(kappa ∗ sqrt(1 + (deltaˆ2) ∗ (u1ˆ2 + u2ˆ2)))) ∗ vol 

 (8.21)
 
(u5 ∗ 2 ∗ (deltaˆ2) ∗ u1 ∗ u2/(kappa ∗ sqrt(1 + (deltaˆ2) ∗ (u1ˆ2 + u2ˆ2)))) ∗ vol

Accordingly, set Γ :

 
 Γ 1 
 
Γ =  Γ2 

 (8.22)
 
Γ3

where
 
 0 
 
Γ1 = 
 −u3y + u4x

 (8.23)
 
−u3z + u5x
 
 −u4x + u3y 
 
Γ2 = 
 0 
 (8.24)
 
−u4z + u5y
 
 −u5x + u3z 
 
Γ3 = 
 −u5y + u4z

 (8.25)
 
0

The matrices f and da are:

 
f
 1 
 
f =
 f2  ,
 (8.26)
 
f2

where

f1 = −(u3 ∗ ((u1ˆ2 + u2ˆ2) − 4 ∗ delta ∗ (sqrt(eta/5.798)) ∗ (u1 ∗ u1t + u2 ∗ u2t))

+((u2∗u1x−u1∗u2x)/kappa)∗(1−4∗delta∗(sqrt(eta/5.798))∗(u1∗u1t+u2∗u2t)/(u1ˆ2+u2ˆ2+epsiˆ2)))∗vol

(8.27)

51
 f2 = −(u4 ∗ ((u1ˆ2 + u2ˆ2) − 4 ∗ delta ∗ (sqrt(eta/5.798)) ∗ (u1 ∗ u1t + u2 ∗ u2t))

+((u2∗u1x−u1∗u2x)/kappa)∗(1−4∗delta∗(sqrt(eta/5.798))∗(u1∗u1t+u2∗u2t)/(u1ˆ2+u2ˆ2+epsiˆ2)))∗vol

(8.28)

f3 = −(u5 ∗ ((u1ˆ2 + u2ˆ2) − 4 ∗ delta ∗ (sqrt(eta/5.798)) ∗ (u1 ∗ u1t + u2 ∗ u2t))

+((u2∗u1x−u1∗u2x)/kappa)∗(1−4∗delta∗(sqrt(eta/5.798))∗(u1∗u1t+u2∗u2t)/(u1ˆ2+u2ˆ2+epsiˆ2)))∗vol

(8.29)

 
 d11 0 0 
 
da = 
 0 d22 0 . (8.30)
 
0 0 d33

d11 = d22 = d33 = sigma ∗ (1 + beta(sqrt(u1ˆ2 + u2ˆ2))) ∗ vol + epsi (8.31)

where the infinitesimal quantity epsi (which stands for the normal conductivity σext of the surrounding

immerser) is added to avoid division by zero during computations of the code. At the time of computation,

we consecutively refined meshes until solutions converged. For demanding cases, such as Figure 36, 4 million

degrees of freedom were required. Similarly, for the case of NbN, 14 million degrees of freedom were involved

for the convergence of solutions to high accuracy.

52
9 Gravitational Wave Sensors based on Superconducting Trans-

ducers

9.1 Theory and Model

GEFEST, (Gravitational wave energy (GE) into magnetic-flux energy (FE) via a superconducting transducer

(ST)) was invented in the Advanced Physics Laboratory and was an international collaboration between

many researchers and labs. GEFEST translates the motion of a system of massive barbells interacting with

gravitational waves into a current signal from superconducting transducers connected to the barbell system.

GEFEST is a non-interferometric GW detector which is compact and can be orientated towards the direction

of the GW source. In Figure 29 below, we can see the wide range of average frequencies that the GEFEST

can detect with minimal amount of strain amplitude to its readings, compared to other GW detectors. The

exceptional and extraordinary details of GEFEST will be summarized here, but the published paper has the

full explanations needed for those interested [53].

Figure 29: Comparison of sensitivity of the GEFEST detector with the current (bold curves) and projected
(dotted curves) GW detectors. The characteristic amplitudes of GWs are defined as the amplitude with a
bandwidth equal to the observation frequency.

53
Figure 30: Two orthogonally-oriented and mechanically separate barbells under the influence of an incident
plane periodic GW oscillate around the axis (1-2). The direction of GW is assumed to coincide with this
axis. The GW field-lines (see also Fig. below) are indicated for this case. Both barbells are rigidly connected
with the dedicated superconducting loops (their rectangular shapes are adopted in the current layout). The
magnetic field B is static in space and time. In the absence of GW, the magnetic-field lines are in parallel to
the loop planes, and no current is flowing through the loop. Upon the action of GW, the barbells’ oscillatory
rotation causes the superconducting loops to rotate in opposite directions, thus generating opposite polarity
currents in the superconducting loops.

In Figure 30, we see the schematics of the design concept of GEFEST. There are two massive barbells, each

connected to a superconducting loop. Under the periodic influence of a GW, the barbells with influence

the loops, which will rotate towards or away from each other and the magnetic fluxes of each with opposite

polarity with be translated to SQUIDs to be detected.

Figure 31: The acceleration field lines of planar wavefront, purely polarized (“+”or“×”) GW are indicated
by the sets of hyperbolas. The direction of accelerations is shown by arrows for a given half-period of the
wave; in the next half-period they will be reversed. A mechanical system of 2 point masses connected by a
rigid line is a barbell with nonzero quadrupolar momentum. Two such barbells will oscillate with respect to
each other. Angular amplitude of relative oscillation δΘ will constitute twice the value of the amplitude δθ
of single barbell oscillation relative to a non-moving observer associated with (x, y) coordinate system and
with the magnetic field B in Fig. 30.

54
Above, in Figure 31, are the GW field-lines. This GW is assumed harmonic with a frequency relative to

cos(ωt). This in turn produces an acceleration in which the GW exerts torque onto the barbells with a

known moment of inertia, I = 2M r2 . It follows from conservation of angular momentum and, with some

integration, that the change in the angle of the barbells due to the GW:

1
δθ(t) = − hGW cos(ωt). (9.1)
2

When the barbell’s sweep an angle, δθ, a magnetic flux is detected by the superconducting transducers/loops,

which will send out of current, J = J(δθ). It is deduced in the full published paper that the maximal current

in the superconducting loop, where Lcircuit is the total inductance in the circuit with loop is:

δΘ Ba2 δθ Ba2 hGW

J= = = (9.2)
Lcircuit Lcircuit 2Lcircuit

9.2 Results

The superconducting transducers are the essential part of GEFEST’s design and I assisted with the modeling

of the impact of increase of loops/rings to induced current, which would result in the increased sensitivity of

detecting GWs. In Figure 32, this is the 3-D geometry used, in conjunction with the implementation of the

TDGL-equations discussed in Section 8.1. The base model was to start with just two rings, which are well

known from induction laws of electromagnetism, to influence and induce field fluxes in one another as that

analogous to a solenoid. The flux that generates an electromotive force, E = −dΦ/dt in a loop is frequency

dependent:

Φ(t) = δθsin(ωt) (9.3)

and in turn results in a voltage, V , between the ends of a loop. If there are N turns/loops, the voltage

becomes VN = N V and the current for superconducting loops are JN = VN /(LN ω), where LN is the

coil’s inductance. This in turn shows us that the maximum flux that can be transferred to the detector is

JN LN = N δϕ. However, this is the same as a single-loop transducer and a coil/solenoid design shows no

advantage. So we worked with the case of parallel loops.

The reasoning behind the use of parallel loops is made best with the modeling and the plotted results in the

figure below. With a coil/solenoid, the separation is typically small and the mutual and self inductances are

the same. However, if we use individual loops with particular spacing in between, each has its own current,

55
which we can sum up the total of each loop. Each loop may be connected to its own individual SQUID

detector, which each loop could be a SQUID in itself and once again, the signals can be summed up to

improve the signal/noise ratio by a factor of N 1/2 .

Figure 32: 3D stacking of individual loops in COMSOL at time of modeling for GEFEST.

Figure 33 shows this case for increased number of loops producing increased induced current amplitude. We

were limited by the computational time of these models due to their millions of degrees of freedom. The

average calculation of a set of loops took a few weeks. However, this shows an even greater number of loops

past thirteen may provide many improvements to GEFEST.

Figure 33: Modeling of increased number of stacked superconducting rings, up to thirteen rings, show that
the induced current will increase linearly.

This work was published in Physical Review Research: Gulian, A., Foreman, J., Nikoghosyan, V., Sica, L.,

56
Abramian-Barco, P., Tollaksen, J., Melkonyan, G., Mowgood, I., Burdette, C., Dulal, R., Teknowijoyo, S.,

Chahid, S., Nussinov, S. (2021). Gravitational wave sensors based on superconducting transducers. Physical

Review Research, 3(4). doi:10.1103/physrevresearch.3.043098

57
10 Violations of Magnetic Flux Conservation in Superconducting

Nanorings

10.1 Theory and Model

Little-Parks experiments[54] as well as the Josephson effects[9] introduced the scientific community to the

phenomenon of superconducting nanorings. SQUIDs[10] are based on these effects and the use of them are

useful for quantum measurement devices as seen in the GEFEST in Section 9.2. It has been introduced and

accepted that superconductors obey the flux conservation theorems of any typical physical system[55], as

intuition should lead one to:

1
LJ + Φext = constant = Φring (10.1)
c

where Φring is the flux through the ring opening, Φext is the externally applied flux, L is the ring inductance

and J is the current induced in the ring by Φext . However, under the guidance and instruction of Dr. Gulian,

I have seen through modeling of the 3-D set of TDGL-equations that there is an exception for finite-gap

superconducting nanorings of the diameter thickness of λL , the penetration depth. The approach taken was

based on the inclusion of the kinetic energy of CP relative to the GL free energy, and there is a majority of

published results[56, 58, 59, 60, 61, 62, 63, 64, 65, 57] that have worked with the time dependent and static

GL equations in the same manner we did.

There were specific conditions to make sure our model was accurate to that of the microscopic superconduct-

ing state of nanorings. The temperature T is near the critical temperature, Tc , which is always necessary

for using the TDGL equations, but it has been shown that the model still remains applicable for description

of the system when farther away from Tc [66]. The other condition is that the fields must vary under the

constraint that, (Dk 2 , ω) << γ where ω describes the speed of variation of the external magnetic field, D is

the diffusion coefficient, k is the characteristic wave vector of non-homogeneous space and is relative to d,

the ring wire diameter, such that k >> 1/d/. γ is the damping caused by inelastic processes in the electron

system, which we have defined before, (Section 2.1.1) as the reciprocal electron–phonon relaxation time τϵ .

Setting the ring up in proper boundary conditions was important, to make sure the computations were just

a result of the superconducting nanoring and not from a connection of conductivity from the environment.

Therefore, we treated the nanoring as if it were immersed in a nonmagnetic dielectric medium with the

constraint that σdiel << σn , and even though the TDGL equations were valid in the totality of the system:

ring and dielectric medium, we modeled the CP condensates to vanish at the medium between the two,

58
through a quantity epsi, as stated in Section 8.1.

The dynamic conditions of the external magnetic field Hext were set by a switch rate. The simulations

were constructed such that the ring is initial in a zero-field condition, then Hext is turned on in a time-

dependent rate, and then kept at a steady. There is the animation [67] of the simulation submitted with

the publication online, and it is recommended that interested readers view it for the full understanding of

the dynamics versus this static moment representation to be presented in the next section. The conditions

on the adiabatic switching of the spatially homogeneous time-dependent Hext = H along the z-axis with

amplitude, Bz0 /µ0 , is:

H(t) = B0z Θ(t)/µ0 (10.2)

where γ1 and γ2 are switching rates. The Θ(t) is dependent on γ1 , γ2 , t = ton , and t = tof f :

Θ(t) = (tanh[γ1 (t − ton )] − tanh[γ2 (t − tof f )]) /2. (10.3)

This results in the vector potential components: Az = 0, Ax = −Θ(t)Bz0 y/2, and Ay = Θ(t)Bz0 x/2. These

values are assigned as a distance significantly far away enough by setting the Dirichlet boundary for the

vector potentials at infinity.

TDGL equations, as well as this method of approach, are acceptable for Type I and Type II superconductors,

as we tested the limited of κ, the GL parameter which governs the type of superconductor, and the results

were the same. We also intentionally kept Hext smaller than the critical field, Hc or Hc1 , so as to make sure

superconductivity was not destroyed. The expectations of flux conservation are that, since the ring is initial

in a zero-state of flux, we expect the flux inside the ring to remain absent as the external magnetic field is

turned on. However, we see from the results this does not hold and makes an important caveat that, for

nanorings, special consideration should be taken when it is in the presence of Hext .

The reasoning for this possible violation comes from the assumption that the electric field is not induced

inside a superconductor and remains zero as a magnetic field is turned on. According to Faraday’s induction
R
law, E = E · dl = −∂Φtot /∂t, due to this flux variation and, as will be shown in results, appears on the

surface (but not the bulk, for sizes of wire thickness > λL ) of the superconductor during the greatest rate of

change of the magnetic field.

Size dependence matters, as there are two characteristic lengths that non-equilibrium superconductor are

associated with: the coherence length ξ(T ) and the magnetic (London) penetration depth λL . However,

59
there exists a penetration length of the longitudinal electric field that needs to be taken into consideration

for nanorings:

s
Dτϵ T
lE = 2 (10.4)
π|∆(T )|

where D is the diffusion coefficient of electrons, τϵ is the energy relaxation time of electronic excitations in

superconductors, and |∆(T )| is the energy gap as a function of temperature. Note the proportionality of

T /|∆(T )| such that as this value increases due to T approaching Tc , lE increases to the range of macroscopic

sizes (millimeters) and the influence of flux cannot be ignored. This is shown in the results with two varying

sizes of rings.

10.2 Results

Figure 34: (a) A nanoring (the gridline are λL -scaled) in a homogeneous external magnetic field H orthogonal
to the ring’s plane. The field H induces current j (green arrows) which generates its own magnetic field and
magnetization M, thus yielding the induction B = µ0 (H + M) (red arrows). (b) Violation of magnetic flux
conservation in a nanoring with external magnetic field variation. Here, the dimensionless time and flux are
in units t/t0 and Φ/Φext (t), respectively.

In Figure 34, we see the geometry of the 3-D ring in Panel a) at the time of the adiabatic switching on

of the external homogeneous magnetic field. The dimensions for all units are made dimensionless before

computations, and the scaling of the ring is of orders of dimensionless λL . It can be seen that there is

external field-induced current j (green arrows) on the surface of the nanoring. This in turn produces its own

magnetic field and magnetization M, and the total induction B = µ0 (H + M) (red arrows). The dielectric

non-magnetic medium occupies the cylinder with boundaries corresponding to infinitely far surfaces on which

the boundary condition B = µ0 H is imposed (only the bottom blue surface is shown). The white dashed

60
line on the blue surface indicates the projection of the ring’s geometrical opening; the surface restricted by

this line is used for the computation of the applied flux Φext and the opening at the equatorial plane—the

flux through the ring, Φring , in both cases, is via surface integration of B = curlA.

Panel b) of Figure 34, we will see the details of the time evolution of this induced electric field: the expecta-

tions of the conservation of flux (green line) and the resulting violation of magnetic flux conservation (blue

line) in the nanoring with external magnetic field applied. Initially, H = 0, so Φring (t) should stay zero in

H(t) not equal to 0. Φext (t) is given by the red line curve. The horizontal green line corresponds to flux

conservation, δΦring (t) = 0, which is not the case as the blue line indicates. We see a ratio of flux violation

to be 20% of the Φext , while this varies for different size rings, as will be shown in Figure 36. The dimensions

of this nanoring are of ring diameter = 12 λL and ring wire diameter = 4λL .

Figure 35: Appearance and disappearance of electric field (shown by arrows) during the evolution of an
external magnetic flux (the amplitude of the external magnetic field is shown in the insets with a black spot
indicating the state of the system).

In Figure 35, we see the evolution of the induced electric field on the surface of the superconducting nanoring.

The upper inset panel is the evolution of Hext applied with respect to time. In Panel a), we see that Hext

is zero and there is no change to the nanoring as expected, and it is in a steady state. In Panel b), Hext

is switching on adiabatically according to the switching parameters set by the computations and during the

greatest flux, we see the induced electric field E (green arrows) appear on the surface and circulate a current

j that will persist even after the flux has stopped. Panel c) shows Hext has ceased an increase or change and

has plateaued to a steady constant value. The energy field is gone, but the current has been started in the

superconducting nanoring and persists. This persistent current is as expected, since once it is set in motion

during the superconducting state, and the superconductivity is not lost because the critical magnetic field

Hc has not been reached, there is no resistance to damping its flow and the computations show this to hold

true.

NbN superconductors are often used in other labs[68, 69, 70] as well as at the Advanced Physics Laboratory,

61
with the results from Section 7.2, the NbN superconducting diode microbridge as a result. Therefore, we

wanted to model our nanorings with the parameters for NbN from a case where Tc = 16.1 [70] which

eventually nearly matched: λL (0) ≈ 167nm and ξ(0) ≈ 6.9nm such that κ ≈ 24. The following are the

parameters used to model the ring as NbN:

Density of electrons n = 1.971029 m−3 , m = m0, vF = 2.09 × 106 ms−1 , kF = 1.80 × 1010 m−1 , mean free path

l = 3.96Å, kF l = 7.15, density of states N (0) = 2.38 × 1028 states m−3 eV −1 , electron–phonon interaction

constant λ = 1.64 and Debye temperature θD = 174K.

Calculations with these parameters gave estimates of the following: elastic scattering time τimp = l/vF =

1.89×10−16 s, diffusion coefficient D = lvF /3 = 2.76×10−4 m2 s−1 and Drude conductivity σ = ne2 τimp /m =

8.6 × 1015s−1 .

We calculated the inelastic electron-phonon scattering time by the function:

2
hθD
τϵ = . (10.5)
kB 4πλ∆3

With regard to temperature, we used an approximation for NbN with its ∆0 = 1.76Tc [71]:

p
∆(T ) = ∆0 1 − T /Tc (0.9663 + 0.7733T /Tc ). (10.6)

During computation, we set κ = 24, δ = 15, D = 50 ≈ 26.4µm, d = 5 ≈ 2.64µm. The degrees of freedom

were in the millions range, with smaller rings requiring up to 14 million DOF and larger rings 1.9 million.

The results are shown in Figure 36 for various sized modeled NbN nanorings. The top three plots (a-c) are

for flux violations for smaller d, ring thickness, from thinnest to thicker (left to right) while the overall D

diameter of the rings are the same for each. The scaling shows that the violation is most significant for

thinner d and especially for the outer surface of the ring (red curve.) The bottom three plots show the

effects of the thickness on the induced current. Panel d) shows that the induced current is greatest for the

large thickness. This makes sense since more mass means more current carriers. However, Panel e) and f)

show interesting spikes in the response for the flux, most pronounced for d = λL . The spikes occur because,

when the increase in the external magnetic field is non-adiabatic, (switching is extremely fast), there is a

higher number of normal electrons being accelerated by induction versus those of Cooper pairs responding

and once the magnetic field is stationary, the Cooper pairs are the only ones left to flow which causes the

sharp increase.

62
Figure 36: Magnetic flux and current dynamics in NbN nanorings with different wire diameters. Three
curves in panels (a)-(c) reflect different geometrical definitions of ring diameters (corresponding to flux
density integration): inner, D-d; central, D; outer, D+d. Current dynamics in (d) and (e) correspond to the
case of non-adiabatic on/off switching of the external flux, with γon = γof f = 10. Here, the dimensionless
current and time are in units of 1/(2µ0 ξ(T )) and πh̄/(8kB (Tc − T )), respectively. Panels (e) and (f) are
normalized by their stationary settled values after the H-field switches on. The time axes in panels (d)
through (f) have the same origin.

In Figure 37, we see the cross-sectional cuts through the rings. Panel a) shows a wire with parameters

d = 40λL and D = 80λL , which is larger than the wire in Panel b) with parameters d = 1λL and D = 10λL .

The color coding is opposite for each cross-section of the individual rings due to its geometry, we are seeing

the circular geometry return the induced electric field to the center of the cross-section. The main difference

between these two computations is that the electric field is only skin thickness for the larger ring, while for

the smaller ring, the electric field fully penetrates the bulk and is nearly homogeneous. The homogeneous

induced electric field on the wire with d = λL illustrates how the violation of flux is greatest with smaller

sized rings, as revealed through this finite element modeling of the TDGL equations.

63
Figure 37: Distribution of the azimuthal component of the electric field in macroscopic (a) and microscopic
(b) ring cross-sections. The two circles correspond to ring cross-sections in a vertical plane. The sign change
in color coding corresponds to directions toward and away from the observer (the electric field vector encircles
the ring). In both panels, color coding corresponds to the dimensionless electric field in units 1/(2t0 ξ(T ))

To better understand the significance of size of the ring versus the flux violation, Figure 38 shows two rings

of different parameter sizes: the blue curve corresponds to with parameters d = 20λL and D = 80λL , and

the red curve corresponds to with parameters d = 4λL and D = 12λL , which is equal to the size of the

results in Figure 34. The x-axis shows the temperature factor, η = (Tc − T )/Tc , such that when T is close

to Tc , η ≈ 0, and when T = 0K, η = 1. Therefore, we see that not only is the violation of flux greater for

smaller ring size, as seen from the scaling of the y-axes, but the closer the temperature of the ring is to the

critical temperature, Tc , the greater the flux violation.

Figure 38: Temperature dependence of the non-conservation effect for two nanoring sizes. Each point on
these curves is plotted using the ratio of Φring (t) and Φext (t). Here and below D and d denote the diameter
of the ring and of the ring’s wire correspondingly; λL (0) is the value of λL (0) at T = 0 (η = 1).

64
This work was published in Superconductor Science and Technology: Mowgood, I., Melkonyan, G., Du-

lal, R., Teknowijoyo, S., Chahid, S., and Gulian, A. (2021). Violation of magnetic flux conservation by

superconducting nanorings. Superconductor Science and Technology. doi:10.1088/1361-6668/ac4174

65
11 Conclusion

In this thesis, I have presented a slice of the profuse amount of knowledge for computationally model-

ing the phenomenon of superconductivity using the set of time-dependent Ginzburg-Landau equations and

COMSOL® Multiphysics. These were no simple tasks, of course. The challenging parts were understanding

all the parameters of the TDGL equations, its development of the theory, the many applications of super-

conductivity and the various methods of modeling. Thankfully I had an advisor who is an expert in the

field and had published books on the phenomenon and mathematics of superconductivity and who I am

eternally grateful for his teachings. The history of superconductivity is just over a century old since Onnes

first discovered it in his lab full of the most essential tool for experimentation in superconductivity: liquefied

helium [5]. Without cooling methods, understanding this quantum behavior would be impossible. Though

there are worldwide searches for high temperature superconductors[72, 73], including at Chapman’s own

Advanced Physics Laboratory, the task is a challenging one indeed. Computational modeling may aid in the

help of finding those subtleties for such discoveries [74] but that is dependent on many aspects regarding the

atomic structure and lattice of the materials. However, we have seen from these results that they can aid to

explain the microscopic theory behind the results found in the laboratory.

The main applications for superconductors[21, 75] using levitation for friction-less trains as well as propulsion

of trains, medical imaging using the large magnetic fields superconductors produce, and advancements to

understand the basic principles of physics through particle colliders and quantum measurement devices will

all benefit from computational physics to the field. The Josephson effect and its corresponding application

as a SQUID provides future encouragement that these devices could advance modern life through their

applications as interferometer detectors[20, 19] as seen from the discussed published findings for GEFEST,

the gravitational wave detector.

The theory of superconductivity and its mathematics were discussed at length, and there is still further

work to be done there. The similarity and distinctions between BCS theory[76, 77], Schrödinger’s equation

and TDGL theory show there is much to connect and dissect since they have shared conclusions of how to

solve for the Cooper pair density, which in turn solves for the superconducting state of matter. There are

conditions towards which materials are BCS materials [78] and others behave with different quantum order.

Our set of TDGL equations were unique in that they included the phonon feedback term, the interference

term of the current density, and a replacement of an elliptical integral difference with an approximation. All

of these inclusions produced novel results in return and aided in the publication of our results.

A tutorial and review of how to solve for the set of TDGL equations was presented. The equations must be

66
solved for a certain gauge choice, that is purely dependent on choice and still resolves to the same results

either way, as has been shown to me by Dr. Gulian. Solving for the gauge choice was presented in full

with examples. The equations were made dimensionless for ease of computation and implementation, as

recommended by the software manufacturer themselves. The physics still hold true due to Buckingham’s

pi-theorem[41]. Once the equation is in dimensionless form of however many dimensions of space are needed,

one needs to separate the imaginary and real parts of the Ψ-function to be implemented in COMSOL® and

separate the parts of the vector potential, A, or the scalar potential, φ, depending on the choice of gauge

and number of dimensions. This solution was provided for 2-D in Appendix A (12).

Once the set of TDGL-equations are made dimensionless and properly separated by parts, the parabolic

partial differential equation (PDE) must be encoded into the COMSOL® Model Wizard as an Equation-

based Model. Using General form PDEs, Coefficient form PDEs, and sometimes Distributed ODE (ordinary).

Each situation was unique according to the dimensions (1, 2 or 3-D), what the external conditions were

(absence or presence of magnetic field, current density distribution, etc.), and certain boundary conditions

were programmed for each. Each section for each dimension is detailed before each set of results. Dr.

Gulian’s book[4] was of most help to detail and describe how to implement and produce models that provide

insight into the dynamics of superconductors.

All the results, most published, some presented during conferences and others currently under consideration,

were described in details. Though the descriptions of the dynamics were included with each figure, the

best method of viewing is with the animations that were published[48] at supplementary material with the

results, and I encourage the reader to download and view the full solutions in video format. Animation from

Violation of magnetic flux conservation by superconducting nanorings

The 1-D models of a wire between two superconducting massive banks prove fruitful for discovering methods

of visualizing the results. We see phase slip centers in their full glory, as single and double phase slips. The

exploration of the locations of the phase slip centers with increased dimensionless energy gap revealed a

pitchfork bifurcation was discussed. The coding of automating the results with Livelink® that connected

programming abilities of COMSOL® through MATLAB® was made into a tutorial in Appendix B (12). The

introduction of a hot spot onto the 1-D wire using the phonon feedback term in the set of TDGL equations

showed interesting results of the oscillatory behavior of decreased period of the PSCs with increased |p|.

This led to the most useful of the 1-D models: the phonon cryocooler. It was shown from modeling that the

1D wire in its oscillatory state, even though it is supplied with a constant j, has moments of positive and

negative phonon flux. These states of positive flux and negative flux allow for phonons to be accepted by the

1-D superconducting filament at certain moments of time and to emit phonons at different moments in time.

67
This facilitated the design of the phonon cryocooler that utilized the Kaptiza resistance and the properties

of increased acoustic densities of the layered cooler design to mitigate phonon propagation outwards, away

from the top plate, to be cooled, and down to the cool sink at the bottom layer. This result was presented

and is still under progress for publication.

For the 2-D model, we aided in explaining the microscopic theory and mechanism behind the superconducting

diode effect for the device built in the Advanced Physics Laboratory[48]. The 2-D model implementation

showed how much more complex the solution of the TDGL equation becomes as we increase the dimensions.

The model proved useful in showing the vortices present during the times of proper polarity of the AC

current applied to the superconducting microbridge. The vortices entered from only one edge of the 2-D

slab, producing a large voltage spike, and left the opposite side of the entry side. When the polarity of

the AC current changed to the other direction, there was no voltage present, illustrating the diode effect as

modeled from the set of TDGL equations.

The 3-D models produced two published results: superconducting transducers/loops for GEFEST, a grav-

itational wave detector[53], and a violation of magnetic flux in nano-sized rings of order of λL [67]. The

superconducting loops for GEFEST were proven to be the best used in combined individual use, each con-

nected to their own SQUID, versus that of a typical solenoid. The model for the superconducting transducers

took into account how the number of individual loops spaced apart would aid in a greater increase in the

induced current that a gravitational wave would produce during the angle sweep of the massive barbells

that are part of the dynamic system of GEFEST. The model of the nano-sized rings took into account

special treatment of the rings as if they are in a dielectric medium, such that the CP density vanishes at

the interface of the medium and the ring, so the computations would show the behavior of the rings in the

presence of a changing magnetic field or flux. The violation model included results of varying ring sizes and

rings with dimensional values of NbN. They all showed the dependency of the flux violation on the thickness

and diameter of the rings and the closeness to Tc . The smaller the ring, the greater the flux violation and

the closer to Tc , the greater the flux violation as well. This showed that consideration should be taken when

conducting experiments of such nano-sized superconducting rings in the presence of a magnetic flux.

The totality of these models and results show that the TDGL-equations are vastly revealing of the knowledge

of superconductivity and quantum behavior. The equations showed methods of advancing science regarding

superconductors, not only in experiment and application but also theory. This was a true joy to study

superconductivity in this way. I started studying superconductivity theory in my Master’s program for

Physics and was very grateful that Chapman had Dr. Gulian to be my next teacher in the subject. I

am very grateful to Chapman University, the CADS program, the many teachers I had during the CADS

68
program, and to my committee for their encouragement. Thank you, this was a blessing.

69
12 Appendices

Appendix A: Separation of Real And Imagination Parts of dimensionless Order

Parameter and Current Density

Separation of Real And Imagination Parts of Dimensionless Order Parameter and Current

Density

 √ 
√
 
|ψ|δ |ψ|δ
|ψ|2 δ 2 +1 ln(1+ √ )−ln(1− √ )

2√ |ψ|2 δ 2 +1 |ψ|2 δ 2 +1 |ψ|2 δ 2 +1 |ψ|δ
σ Ȧ 1 + π ηu|ψ|  2|ψ|δ + |ψ|δ − 1 ln(1 − √ )
|ψ|2 δ 2 +1

η ∂|ψ|2
  r 
i ∗ ∗ 2
=− A+ (ψ ∇ψ − ψ∇ψ ) |ψ| −2δ −∇×∇×A
2κ|ψ|2 u ∂τ

Simplifying Left hand side (LHS):

 √ 
√
 
|ψ|δ |ψ|δ
|ψ|2 δ 2 +1 ln(1+ √ )−ln(1− √ )
2√ |ψ|2 δ 2 +1 |ψ|2 δ 2 +1 |ψ|2 δ 2 +1 |ψ|δ
σ Ȧ 1 + π ηu|ψ|  2|ψ|δ +( |ψ|δ − 1) ln(1 − √ )
|ψ|2 δ 2 +1

  q  
1 |ψ|δ |ψ|δ
 |ψ|δ 1+ |ψ|2 δ 2 ln(1 + √ ) − ln(1 − √ )
 2√ |ψ|2 δ 2 +1 |ψ|2 δ 2 +1
→ σ Ȧ 1 +
 ηu|ψ| 
π  2|ψ|δ

 q 
1
ψ|δ 1+ |ψ|2 δ 2 |ψ|δ
+ − 1 ln(1 − p )
|ψ|δ |ψ|2 δ 2 + 1

s !
2√ 1 1 |ψ|δ |ψ|δ
→ σ Ȧ 1 + ηu|ψ| 1+ ln(1 + p ) − ln(1 − p )
π 2 |ψ|2 δ 2 |ψ|2 δ 2 + 1 |ψ|2 δ 2 + 1
s !
1 |ψ|δ
+ 1+ − 1 ln(1 − p )
|ψ|2 δ 2 |ψ|2 δ 2 + 1

70
Replacements to be made:

2
Re(ψ) = ψ1 , Im(ψ) = ψ2 , ψ = ψ1 + iψ2 , ψ ∗ = ψ1 − iψ2 , |ψ| = ψ12 + ψ22 , A = x
bA1 + ybA2
∂ (ψ12 +ψ22 ) · · · ·
∂ ∂ ∂|ψ|2 ∂ψ ∂ψ ∗
∇=x
b ∂x + yb ∂y , ∂τ = ∂τ = 2(ψ1 ψ1 + ψ2 ψ2 ), ∂τ = ψ1 , ∂τ = ψ2

Handling LHS first:

s !
2√ 1 1 |ψ|δ |ψ|δ
σ Ȧ 1 + ηu|ψ| 1+ ln(1 + p ) − ln(1 − p )
π 2 |ψ|2 δ 2 |ψ|2 δ 2 + 1 |ψ|2 δ 2 + 1
s !
1 |ψ|δ
+ 1+ − 1 ln(1 − p )
|ψ|2 δ 2 |ψ|2 δ 2 + 1

  s !
· · 2√ 1
1 |ψ1 + iψ2 | δ
→σ xbA1 + ybA2 1+ ηu |ψ1 + iψ2 | 1+ 2 ln 1 + p 2
π (ψ1 + ψ22 )δ 2
2 (ψ1 + ψ22 )δ 2 + 1
s ! !
|ψ1 + iψ2 | δ 1 |ψ1 + iψ2 | δ
− ln(1 − p 2 )+ 1+ 2 − 1 ln 1 − p 2
(ψ1 + ψ22 )δ 2 + 1 (ψ1 + ψ22 )δ 2 (ψ1 + ψ22 )δ 2 + 1

  s
· · 2√ 1 1 |ψ1 + ψ2 | δ
→σ xbA1 + ybA2 1+ ηu |ψ1 + ψ2 | 1+ ln(1 + p )
π 2 (ψ12 + ψ22 )δ 2 (ψ12 + ψ22 )δ 2 + 1
s ! !
|ψ1 + ψ2 | δ 1 |ψ1 + ψ2 | δ
− ln(1 − p 2 )+ 1+ 2 − 1 ln 1 − p 2
(ψ1 + ψ22 )δ 2 + 1 (ψ1 + ψ22 )δ 2 (ψ1 + ψ22 )δ 2 + 1

Handling Right hand side (RHS):

h i 2

i ∗ ∗
|ψ|2 −2δ uη ∂|ψ|
p
− A + 2κ|ψ| 2 (ψ ∇ψ − ψ∇ψ ) ∂τ −∇×∇×A

∂ψ ∗ ∂ψ ∗
     
i ∗ ∂ψ ∂ψ
→ − (b
xA1 + ybA2 ) + ψ x + y − ψ x + y
2κ(ψ12 + ψ22 )
b b b b
∂x ∂y ∂x ∂y
η ∂(ψ12 + ψ22 )
 r 
(ψ12 + ψ22 )−2δ − (b
xA1..x + ybA1..y )
u ∂τ

71
 h  ∗
i
i ∗ ∗
→ − (b
xA1 + ybA2 ) + 2κ(ψ12 +ψ22 )
xψ.x + ybψ.y ) − ψ x
ψ (b bψ.x + ybψ.y
 r  · · 
η
(ψ12 + ψ22 )−2δ ψ12 +ψ22 − (b
xA1..x + ybA1..y )
u

 
i  ∗ ∗ ∗ ∗

→ − (b
xA1 + ybA2 ) + bψ ψ.x + ybψ ψ.y − x
x bψψ.x − ybψψ.y
2κ(ψ12 + ψ22 )
 r  · · 
2 2 η
(ψ1 + ψ2 )−2δ ψ12 +ψ22 − (b
xA1..x + ybA1..y )
u

 
i  ∗ ∗ ∗ ∗

→ − (b
xA1 + ybA2 ) + bψ ψ.x − x
x bψψ.x + ybψ ψ.y − ybψψ.y
2κ(ψ12 + ψ22 )
 r  · · 
2 2 η
(ψ1 + ψ2 )−2δ ψ12 +ψ22 − (b
xA1..x + ybA1..y )
u

 
i  ∗ ∗ ∗ ∗


→ − (b
xA1 + ybA2 ) + x (ψ ψ .x − ψψ.x ) + y ψ ψ .y − ψψ.y
2κ(ψ12 + ψ22 )
b b
 r  · · 
η
(ψ12 + ψ22 )−2δ ψ12 +ψ22 − (b
xA1..x + ybA1..y )
u

Subterm replacements: ψ = (ψ1 + iψ2 ) ; ψ ∗ = (ψ1 − iψ2 ) into below

b (ψ ∗ ψ.x − ψψ.x
∗
) + yb ψ ∗ ψ.y − ψψ.y
∗


x

→x
b [(ψ1 − iψ2 ) (ψ1.x + iψ2.x ) − (ψ1 + iψ2 ) (ψ1.x − iψ2.x )] + yb [(ψ1 − iψ2 ) (ψ1.y + iψ2.y )

− (ψ1 + iψ2 ) (ψ1.y − iψ2.y )]

→x
b [ψ1 ψ1.x + iψ1 ψ2.x − iψ2 ψ1.x − iψ2 iψ2.x − ψ1 ψ1.x + iψ1 ψ2.x − iψ2 ψ1.x + iψ2 iψ2.x ] + yb [...]

→x
b [ψ1 ψ1.x − ψ1 ψ1.x + iψ1 ψ2.x + iψ1 ψ2.x − iψ2 ψ1.x − iψ2 ψ1.x + ψ2 ψ2.x − ψ2 ψ2.x ] + yb [...]

→x
b [iψ1 ψ2.x + iψ1 ψ2.x − iψ2 ψ1.x − iψ2 ψ1.x ] + yb [...]

→x
b [2iψ1 ψ2.x − 2iψ2 ψ1.x ] + yb [2iψ1 ψ2.y − 2iψ2 ψ1.y ]

72
→ 2ib
x [ψ1 ψ2.x − ψ2 ψ1.x ] + 2ib
y [ψ1 ψ2.y − ψ2 ψ1.y ]

Replace back into RHS:

h   · · 
i
 ∗ ∗ ∗ ∗

i pη
− (b
xA1 + ybA2 ) + 2κ(ψ2 +ψ2 ) xb (ψ ψ.x − ψψ.x ) + yb ψ ψ.y − ψψ.y (ψ1 + ψ2 )−2δ u ψ12 +ψ22
2 2
1 2

− (b
xA1..x + ybA1..y )
h i  · · 
i
y [ψ1 ψ2.y − ψ2 ψ1.y ]) (ψ12 + ψ22 )−2δ uη ψ12 +ψ22
p
→ − (b xA1 + ybA2 ) + 2κ(ψ12 +ψ22 )
x [ψ1 ψ2.x − ψ2 ψ1.x ] + 2ib
(2ib −

(b
xA1..x + ybA1..y )
h i  · · 
1
pη
→ − (b xA1 + ybA2 ) − κ(ψ12 +ψ22 )
x [ψ1 ψ2.x − ψ2 ψ1.x ] + yb [ψ1 ψ2.y − ψ2 ψ1.y ]) (ψ1 + ψ2 )−2δ u ψ12 +ψ22
(b 2 2
−

(b
xA1..x + ybA1..y )

Combined LHS and RHS:

  s !
· · 2√ 1 1 |ψ1 + ψ2 | δ |ψ1 + ψ2 | δ
σ x
bA1 + ybA2 1+ ηu |ψ1 + ψ2 | 1+ 2 ln(1 + p 2 ) − ln(1 − p 2 )
π 2 (ψ1 + ψ22 )δ 2 (ψ1 + ψ22 )δ 2 + 1 (ψ1 + ψ22 )δ 2 + 1
s !
1 |ψ1 + ψ2 | δ
+ 1+ 2 − 1 ln(1 − p 2 )
(ψ1 + ψ22 )δ 2 (ψ1 + ψ22 )δ 2 + 1

  r  · · 
1 2 2 η
= − (b
xA1 + ybA2 ) − (b
x [ψ 1 ψ 2.x − ψ 2 ψ 1.x ] + y [ψ1 ψ2.y − ψ 2 ψ 1.y ]) (ψ 1 + ψ 2 )−2δ ψ12 +ψ22 −(b
xA1..x + ybA1..y )
κ(ψ12 + ψ22 )
b
u

Isolating j=(b
xA1..x + ybA1..y ):

j = (b
xA1..x + ybA1..y ) =
  r  · · 
1 2 2 η
− (b
xA1 + ybA2 ) − (b
x [ψ1 ψ2.x − ψ 2 ψ 1.x ] + y [ψ1 ψ2.y − ψ2 ψ1.y ]) (ψ1 + ψ 2 )−2δ ψ12 +ψ22
κ(ψ12 + ψ22 )
b
u

  s !
· · 2√ 1 1 |ψ1 + ψ2 | δ |ψ1 + ψ2 | δ
−σ x
bA1 + ybA2 1+ ηu |ψ1 + ψ2 | 1+ 2 ln(1 + p ) − ln(1 − p )
π 2 (ψ1 + ψ22 )δ 2 (ψ12 + ψ22 )δ 2 + 1 (ψ12 + ψ22 )δ 2 + 1
s !
1 |ψ1 + ψ2 | δ
+ 1+ − 1 ln(1 − p 2 )
(ψ12 + ψ22 )δ 2 (ψ1 + ψ22 )δ 2 + 1

Separating j into x and y components:

h i  · · 
1
pη
bj = −b
x x A1 − κ(ψ2 +ψ2 ) [ψ1 ψ2.x − ψ2 ψ1.x ] (ψ1 + ψ2 )−2δ u ψ12 +ψ22
2 2
1 2

73
 s !
· 2√ 1 1 |ψ1 + ψ2 | δ |ψ1 + ψ2 | δ
−σ x
bA1 1+ ηu |ψ1 + ψ2 | 1+ 2 2 2
ln(1 + p 2 ) − ln(1 − p 2 )
π 2 (ψ1 + ψ2 )δ 2 2
(ψ1 + ψ2 )δ + 1 (ψ1 + ψ22 )δ 2 + 1
s ! !
1 |ψ1 + ψ2 | δ
+ 1+ 2 − 1 ln 1 − p 2
(ψ1 + ψ22 )δ 2 (ψ1 + ψ22 )δ 2 + 1

h i  · · 
1
(ψ12 + ψ22 )−2δ uη ψ12 +ψ22
p
ybj = −b
y A2 − κ(ψ12 +ψ22 )
[ψ1 ψ2.y − ψ2 ψ1.y ]

  s !
· 2√ 1 1 |ψ1 + ψ2 | δ |ψ1 + ψ2 | δ
−σ ybA2 1+ ηu |ψ1 + ψ2 | 1+ 2 ln(1 + p 2 ) − ln(1 − p 2 )
π 2 (ψ1 + ψ22 )δ 2 (ψ1 + ψ22 )δ 2 + 1 (ψ1 + ψ22 )δ 2 + 1
s ! !
1 |ψ1 + ψ2 | δ
+ 1+ 2 − 1 ln 1 − p 2 )
(ψ1 + ψ22 )δ 2 (ψ1 + ψ22 δ 2 + 1

These expressions have been used in the computations in the results of Section 7.1

74
Appendix B: Automation code for Pitchfork Bifurcation of PSC Locations

Automatization Code of COMSOL Multiphysics® Model utilizing LiveLink™ for MATLAB®

AutomationMinimaCount.m Code:

MinimaList = []; % Empty list for sorting minima variables

parameter1 = 1; %Starting values of parameter d

parameter2 = 0.001; %Starting values of parameter j

L=10; %Starting values of parameter L for length

model = mphload(’MyModel.mph’);%load model

model.param.set(’L0’, L);

model.sol(’sol1’).clearSolutionData();

while (parameter1 <10.1)

model = mphload(’MyModel.mph’);%load model

model.sol(’sol1’).clearSolutionData();

model.param.set(’d’,parameter1); %set d value

model.param.set(’j’,parameter2);%set j value

model.study(’std1’).run; %run study

model.result(’pg6’).run(); %run plot of Y(X)

model.result.export(’plot4’).run; %export datafile

B = importdata(’YofX.txt’, ’ ’); %import data

x = B(:, 1); %separate column x

Y = B(:, 2);%separate column Y

TF = islocalmin(Y, ’MinProminence’,0.99); %find local min

n = numel(Y(TF)); %count local min

plot(x,Y,x(TF),Y(TF),’r*’) %plot minima for check

axis tight

title([’n=’,num2str(n),’ d= ’,num2str(parameter1),’ j= ’,num2str(parameter2)])

drawnow;

op=[’n=’,num2str(n),’ d= ’,num2str(parameter1),’ j= ’,num2str(parameter2)];

disp(op);

if (n==0)

parameter2=parameter2+0.003;

75
 model.sol("sol1").clearSolutionData();

elseif (n > 1)

parameter2=parameter2-0.004;

model.sol("sol1").clearSolutionData();

elseif (n==1)

d = strrep(num2str(parameter1), ’.’, ’_’);

filename = strcat(’YofXMinima’,d);

plot(x,Y,x(TF),Y(TF),’r*’) %plot minima for check

axis tight

title([’d= ’,num2str(parameter1),’ j= ’,num2str(parameter2)]) %create title with values

fname = ’C:\Users\UserName\Desktop\MinimaResults’;

saveas(gcf,fullfile(fname, filename),’png’);

MinimaList(end+1,:)=[parameter1,parameter2]; %add d and j value to list

parameter1 = parameter1 + 0.01;

parameter2 = 0.001;

model.sol("sol1").clearSolutionData();

end

end

MinimaLocations.m Code:

xMinLocations = []; %Empty list for storing locations along line and parameters d, j

load(’Data.mat’); %Load stored data to reevaluate automatically

A = Data;

d = A(:,1);

j = A(:,2);

for i = 1:90 %length of your data, can use numel(j) or numel(d)

model = mphload(’MyModel.mph’);%load model

model.sol(’sol1’).clearSolutionData();

model.param.set(’j0’,j(i));%set j value

model.param.set(’d’,d(i)); %set d value

model.study(’std1’).run; %run study

model.result(’pg6’).run(); %run plot

model.result.export(’plot4’).run; %export plot data

76
 B = importdata(’PlotData.txt’, ’ ’); %import data

TF = isempty(B); % Ensure plot is not empty

dn = num2str(d(i)); %For printing error message if plot is empty

if (TF == 1)

ms =[’Plot is empty for ’,dn];

disp(ms);

i=i+1;

break

end

x = B(:,1); %separate column x

Y = B(:,2);%separate column Y

[TF1,P] = islocalmin(Y, ’MinProminence’,0.99); %find local min with prominence of 0.99

xall = x(TF1); % Store all minima in xall

xmin = unique(xall); % Store for unique values of x

disp(dn); % display parameter d

disp(xmin); % display unique values of locations x

plot(x,Y,x(TF1),Y(TF1), ’r*’) %plot minima for check

axis tight;

for idx=1:numel(xmin)

xMinLocations(end+1,:)=[d(i),j(i),xmin(idx)]; %add delta and j value to list

end

end

77
References
[1] Stephen Blundell. Superconductivity: A Very Short Introduction. Oxford University Press, 2009.

[2] Michael Tinkham. Introduction to Superconductivity. Courier Corporation, 2004.

[3] David R Tilley and John Tilley. Superfluidity and Superconductivity. Routledge, 2019.

[4] Armen Gulian. Shortcut to Superconductivity. Springer, 2020.

[5] H Kamerlingh Onnes. “The liquefaction of helium.” In: Through Measurement to Knowledge: The

Selected Papers of Heike Kamerlingh Onnes 1853–1926. Springer, 1991, pp. 164–187.

[6] NobelPrize.org. Heike Kamerlingh Onnes – Photo gallery. 2023. url: https://www.nobelprize.org/

prizes/physics/1913/onnes/photo-gallery/.

[7] R Nave. Magnetic Levitation - Hyperphysics. 2023. url: http://hyperphysics.phy-astr.gsu.edu/

hbase/Solids/maglev.html#c1.

[8] Cornelis Jacobus Gorter and Hendrik Casimir. “On supraconductivity I”. In: Physica 1.1-6 (1934),

pp. 306–320.

[9] Brian David Josephson. “Possible new effects in superconductive tunnelling”. In: Physics letters 1.7

(1962), pp. 251–253.

[10] RC Jaklevic et al. “Quantum interference effects in Josephson tunneling”. In: Physical Review Letters

12.7 (1964), p. 159.

[11] JC Taunton and MR Halse. “A Josephson junction analogue incorporating the effects of thermal

noise and phase-dependent conductivity”. In: Journal of Physics E: Scientific Instruments 10.5 (1977),

p. 505.

[12] Ivar Giaever. “Energy gap in superconductors measured by electron tunneling”. In: Physical Review

Letters 5.4 (1960), p. 147.

[13] Ivar Giaever. “Electron tunneling between two superconductors”. In: Physical Review Letters 5.10

(1960), p. 464.

[14] R Nave. Tunneling - Hyperphysics. 2023. url: http://hyperphysics.phy- astr.gsu.edu/hbase/

Solids/Squid.html.

[15] Michael H Cohen, LM Falicov, and JC Phillips. “Superconductive tunneling”. In: Physical Review

Letters 8.8 (1962), p. 316.

[16] WL McMillan and JM Rowell. “Lead phonon spectrum calculated from superconducting density of

states”. In: Physical Review Letters 14.4 (1965), p. 108.

78
[17] DJ Scalapino, JR Schrieffer, and JW Wilkins. “Strong-coupling superconductivity. I”. In: Physical

Review 148.1 (1966), p. 263.

[18] R Nave. Josephson Junction - Hyperphysics. 2023. url: http://hyperphysics.phy-astr.gsu.edu/

hbase/Solids/Squid.html.

[19] Francesco Giazotto and Marıa José Martınez-Pérez. “The Josephson heat interferometer”. In: Nature

492.7429 (2012), pp. 401–405.

[20] Dario Gerace et al. “The quantum-optical Josephson interferometer”. In: Nature Physics 5.4 (2009),

pp. 281–284.

[21] Steven M Anlage. “The physics and applications of superconducting metamaterials”. In: Journal of

Optics 13.2 (2010), p. 024001.

[22] Florent Lecocq et al. “Control and readout of a superconducting qubit using a photonic link”. In:

Nature 591.7851 (2021), pp. 575–579.

[23] John M Martinis. “Superconducting phase qubits”. In: Quantum information processing 8 (2009),

pp. 81–103.

[24] Marco Fellous-Asiani et al. “Limitations in quantum computing from resource constraints”. In: PRX

Quantum 2.4 (2021), p. 040335.

[25] F. Lecocq/NIST. Optical Fiber Could Boost Power of Superconducting Quantum Computers. 2023.

url: https://www.nist.gov/news-events/news/2021/03/optical-fiber-could-boost-power-

superconducting-quantum-computers.

[26] P Bernstein and J Noudem. “Superconducting magnetic levitation: principle, materials, physics and

models”. In: Superconductor Science and Technology 33.3 (2020), p. 033001.

[27] Carly Wilkins. How Maglev Works. 2023. url: https://www.energy.gov/articles/how-maglev-

works.

[28] Hyung-Suk Han and Dong-Sung Kim. “Magnetic levitation”. In: Springer Tracts on Transportation

and Traffic. Springer Netherlands 247 (2016).

[29] National Institute of Biomedical Imaging and Bioengineering (NIBIB). Magnetic Resonance Imaging

(MRI). 2023. url: https://www.nibib.nih.gov/science-education/science-topics/magnetic-

resonance-imaging-mri.

[30] Canon Inc. Structure of MRI Scanner. 2023. url: https : / / global . canon / en / technology /

support28.html.

79
[31] O Brüning, H Burkhardt, and S Myers. “The large hadron collider”. In: Progress in Particle and

Nuclear Physics 67.3 (2012), pp. 705–734.

[32] Raoul Rañoa Programming by Anthony Pesce. GRAPHIC The Large Hadron Collider. 2023. url:

https://graphics.latimes.com/towergraphic-large-hadron-collider/.

[33] Armen M Gulian and Gely F Zharkov. Nonequilibrium electrons and phonons in superconductors.

Springer Science and Business Media, 1999.

[34] Albert Schmid. “A time dependent Ginzburg-Landau equation and its application to the problem of

resistivity in the mixed state”. In: Physik der kondensierten Materie 5.4 (1966), pp. 302–317.

[35] LP Gor’kov and GM Eliashberg. “Dynamical properties of gapless superconductors”. In: Journal of

Low Temperature Physics 2 (1970), pp. 161–172.

[36] Leon N Cooper. “Bound electron pairs in a degenerate Fermi gas”. In: Physical Review 104.4 (1956),

p. 1189.

[37] John Bardeen, Leon N Cooper, and J Robert Schrieffer. “Microscopic Theory of Superconductivity”.

In: Physical Review 106.1 (1957), p. 162.

[38] John Bardeen, Leon N Cooper, and John Robert Schrieffer. “Theory of superconductivity”. In: Physical

review 108.5 (1957), p. 1175.

[39] LP Gor’Kov. “On the energy spectrum of superconductors”. In: Sov. Phys. JETP 7.505 (1958), p. 158.

[40] Lev Petrovich Gor’kov. “Microscopic derivation of the Ginzburg-Landau equations in the theory of

superconductivity”. In: Sov. Phys. JETP 9.6 (1959), pp. 1364–1367.

[41] E. Buckingham. “On Physically Similar Systems; Illustrations of the Use of Dimensional Equations”.

In: Phys. Rev. 4 (4 Oct. 1914), pp. 345–376. doi: 10.1103/PhysRev.4.345. url: https://link.aps.

org/doi/10.1103/PhysRev.4.345.

[42] Sebastien Michotte, Stefan Matefi-Tempfli, and Luc Piraux. “1D-transport properties of single super-

conducting lead nanowires”. In: Physica C: Superconductivity 391.4 (2003), pp. 369–375.

[43] Thomas Eggel, Miguel A Cazalilla, and Masaki Oshikawa. “Dynamical theory of superfluidity in one

dimension”. In: Physical review letters 107.27 (2011), p. 275302.

[44] WJ Skocpol, MR Beasley, and M Tinkham. “Phase-slip centers and nonequilibrium processes in su-

perconducting tin microbridges”. In: Journal of Low Temperature Physics 16.1-2 (1974), pp. 145–167.

80
[45] Cécile Delacour et al. “Quantum and thermal phase slips in superconducting niobium nitride (NbN)

ultrathin crystalline nanowire: Application to single photon detection”. In: Nano letters 12.7 (2012),

pp. 3501–3506.

[46] Anthony Joseph Annunziata. Single-photon detection, kinetic inductance, and non-equilibrium dynam-

ics in niobium and niobium nitride superconducting nanowires. Yale University, 2010.

[47] Lu Zhang et al. “Hotspot relaxation time of NbN superconducting nanowire single-photon detectors

on various substrates”. In: Scientific reports 8.1 (2018), p. 1486.

[48] Sara Chahid et al. “High-frequency diode effect in superconducting Nb3 Sn microbridges”. In: Phys.

Rev. B 107 (5 Feb. 2023), p. 054506. doi: 10.1103/PhysRevB.107.054506. url: https://link.aps.

org/doi/10.1103/PhysRevB.107.054506.

[49] AG Sivakov et al. “Josephson behavior of phase-slip lines in wide superconducting strips”. In: Physical

review letters 91.26 (2003), p. 267001.

[50] D Yu Vodolazov and FM Peeters. “Rearrangement of the vortex lattice due to instabilities of vortex

flow”. In: Physical Review B 76.1 (2007), p. 014521.

[51] GR Berdiyorov et al. “Finite-size effect on the resistive state in a mesoscopic type-II superconducting

stripe”. In: Physical Review B 79.17 (2009), p. 174506.

[52] P Sánchez-Lotero, J Albino Aguiar, and Daniel Domınguez. “Behavior of the flux-flow resistivity

in mesoscopic superconductors”. In: Physica C: Superconductivity and its Applications 503 (2014),

pp. 120–122.

[53] Armen Gulian et al. “Gravitational wave sensors based on superconducting transducers”. In: Phys.

Rev. Res. 3 (4 Nov. 2021), p. 043098. doi: 10 . 1103 / PhysRevResearch . 3 . 043098. url: https :

//link.aps.org/doi/10.1103/PhysRevResearch.3.043098.

[54] WA Little and RD Parks. “Observation of quantum periodicity in the transition temperature of a

superconducting cylinder”. In: Physical Review Letters 9.1 (1962), p. 9.

[55] Lev Davidovich Landau et al. Electrodynamics of continuous media. Vol. 8. elsevier, 2013.

[56] VV Moshchalkov et al. “Effect of sample topology on the critical fields of mesoscopic superconductors”.

In: Nature 373.6512 (1995), pp. 319–322.

[57] GQ Zha. “Superconducting state evolution with applied magnetic flux in mesoscopic rings”. In: The

European Physical Journal B 84 (2011), pp. 459–466.

81
[58] DY Vodolazov and FM Peeters. “Stable and metastable states in a mesoscopic superconducting “eight”

loop in presence of an external magnetic field”. In: Physica C: Superconductivity 400.3-4 (2004),

pp. 165–170.

[59] BJ Baelus, FM Peeters, and VA Schweigert. “Vortex states in superconducting rings”. In: Physical

Review B 61.14 (2000), p. 9734.

[60] BJ Baelus, FM Peeters, and VA Schweigert. “Saddle-point states and energy barriers for vortex entrance

and exit in superconducting disks and rings”. In: Physical Review B 63.14 (2001), p. 144517.

[61] BJ Baelus, SV Yampolskii, and FM Peeters. “Coupled mesoscopic superconductors: Ginzburg-Landau

theory”. In: Physical Review B 66.2 (2002), p. 024517.

[62] LF Chibotaru et al. “Ginzburg–Landau description of confinement and quantization effects in meso-

scopic superconductors”. In: Journal of mathematical physics 46.9 (2005), p. 095108.

[63] D Yu Vodolazov and FM Peeters. “Dynamic transitions between metastable states in a superconducting

ring”. In: Physical Review B 66.5 (2002), p. 054537.

[64] D Yu Vodolazov et al. “Multiple flux jumps and irreversible behavior of thin Al superconducting rings”.

In: Physical Review B 67.5 (2003), p. 054506.

[65] Lin Peng, Zejiang Wei, and Danhua Xu. “Vortex states and magnetization properties in mesoscopic

superconducting ring structures: a finite-element analysis”. In: Journal of Superconductivity and Novel

Magnetism 27 (2014), pp. 1991–1995.

[66] Qiang Du, Max D Gunzburger, and Janet S Peterson. “Analysis and approximation of the Ginzburg–

Landau model of superconductivity”. In: Siam Review 34.1 (1992), pp. 54–81.

[67] Iris Mowgood et al. “Violation of magnetic flux conservation by superconducting nanorings”. In: Su-

perconductor Science and Technology 35.4 (2022), p. 045006.

[68] Jochem JA Baselmans et al. “Doubling of sensitivity and bandwidth in phonon cooled hot electron

bolometer mixers”. In: Applied physics letters 84.11 (2004), pp. 1958–1960.

[69] D Hazra et al. “Superconducting properties of NbTiN thin films deposited by high-temperature chem-

ical vapor deposition”. In: Physical Review B 97.14 (2018), p. 144518.

[70] SP Chockalingam et al. “Superconducting properties and Hall effect of epitaxial NbN thin films”. In:

Physical Review B 77.21 (2008), p. 214503.

[71] DC Carless, HE Hall, and JR Hook. “Vibrating wire measurements in liquid 3 He. I. The normal

state”. In: Journal of low temperature physics 50 (1983), pp. 583–603.

82
[72] Theodore H Geballe. “The never-ending search for high-temperature superconductivity”. In: Journal

of superconductivity and novel magnetism 19 (2006), pp. 261–276.

[73] Xingjiang Zhou et al. “High-temperature superconductivity”. In: Nature Reviews Physics 3.7 (2021),

pp. 462–465.

[74] Ying Sun et al. “Computational discovery of a dynamically stable cubic SH 3-like high-temperature

superconductor at 100 GPa via CH 4 intercalation”. In: Physical Review B 101.17 (2020), p. 174102.

[75] James W Bray. “Superconductors in applications; some practical aspects”. In: IEEE Transactions on

Applied Superconductivity 19.3 (2009), pp. 2533–2539.

[76] CAR Sá De Melo, Mohit Randeria, and Jan R Engelbrecht. “Crossover from BCS to Bose supercon-

ductivity: Transition temperature and time-dependent Ginzburg-Landau theory”. In: Physical review

letters 71.19 (1993), p. 3202.

[77] Jan R Engelbrecht, Mohit Randeria, and CAR Sáde Melo. “BCS to Bose crossover: Broken-symmetry

state”. In: Physical Review B 55.22 (1997), p. 15153.

[78] R Haussmann. “Properties of a Fermi liquid at the superfluid transition in the crossover region between

BCS superconductivity and Bose-Einstein condensation”. In: Physical Review B 49.18 (1994), p. 12975.

83