UBC PHYSICS 502 PROJECT (2017 FALL) 1

Superconducting Quantum Interference Device

(SQUID)
Aaron Kraft, Christoph Rupprecht, Yau-Chuen Yam

Abstract—The superconducting state, a macroscopic quantum which is explained in more detail in reference [5] as well as
phenomena which exhibits resistanceless electric transport, en- the Appendix:
ables many unique measurements and experiments. One applica- √
tion is a Superconducting Quantum Interference Device (SQUID), Ψ(~r) = ns · eiφ(~r) (1)
which can measure magnetic flux with incredible precision. Based
on the Josephson Effect, SQUIDs have become instrumental in Here ns = Ψ · Ψ∗ is the cooper pair density and φ(~r) is the
Condensed Matter Physics for their precision and versatility. phase. This wave function in and of itself is very interesting
and gives a clear example of how superconductivity leads
to quantum behavior on a macroscopic scale. Here the wave
I. I NTRODUCTION function describes a large number of cooper pairs which are
phase coherent, so we can see a quantum mechanical behavior

J osephson first formulated the theory for tunneling between

superconductors in 1962. His result, for which he won the
Nobel prize in 1972, indicated that a tunneling supercurrent
on a classical length scale.
Therefore we can assume that the two superconductors
√
in Fig. 1 have wave functions Ψ(~r) = n1 · eiφ1 (~r) and
would occur between two superconductors separated by a √
Ψ(~r) = n2 · eiφ2 (~r) . Given that the typical width of the
small insulating junction. According to his calculations this
supercurrent would be related to the phase difference between
the two superconductors. In practice this principle enables the
phase difference of two superconducting materials to be mea-
sured. And since the phase of a quantum state is also altered by
magnetic flux, this magnetic flux can be measured precisely.
Since its first construction in 1964, SQUIDs have become
an instrumental part of condensed matter physics. Used most Fig. 1. A Josephson junction: It is made of two superconductors connected
frequently as a highly precise magnetometer, SQUIDs can also through a small insulating layer.
be adapted and applied to other problems and combined with
novel techniques. In this paper we will derive the equations insulator is around 100nm, we can also assume that the two
that govern the behavior of a SQUID, and give a brief overview superconductors are weakly coupled with a coupling constant
of its many interesting applications. K, i.e. cooper pairs can tunnel between the two superconduc-
tors, but they still separate enough that they are described by
II. P HYSICS OF A SQUID different wavefunctions. The Schrodinger equation for the two
superconductors (i, j ∈ {1, 2} and i 6= j) then reads:
A Josephson Junction, as shown in Fig. 1 forms the founda-
tion of a SQUID. There are two types of SQUIDs commonly ∂Ψi (~r)
used in applications: a DC SQUID, which contains two parallel i~ = Ei Ψi + KΨj (2)
∂t
junctions, and an RF SQUID, which contains just one junction. Assuming that we have the same type of superconducting
To begin our treatment of SQUIDs we will therefore turn to the material on each side of the barrier (a reasonable assumption,
dynamics of a Josephson Junction. We will then look into the since in practice we can design the Josephson Junction out of
effects of an external magnetic field on a Josephson Junction whatever material we choose), we know that E1 = E2 . But we
in order to show how a SQUID can measure magnetic flux. will also connect this Josephson Junction to a battery such that
we have a potential difference E1 − E2 = qV . Choosing the
A. The Josephson Junction reference point for energy in the middle between the energies,
the Schrodinger Equation becomes:
As shown in Fig. 1, a Josephson junction consists of two
superconductors separated by an insulating material. Interest- ∂Ψi (~r) qV
i~ = (−1)i+1 Ψi + KΨj
ing physics arises due to the tunneling of cooper pairs across ∂t 2
the insulating barrier. While this is quantum mechanical in We can now plug in our superconducting wavefunction (1)
nature, we do not need to turn to BCS theory in order to to get the following four equations for the real and imaginary
understand this. Instead we can choose a phenomenological parts of Ψ that describe the dynamics of tunneling cooper
and macroscopic ansatz for the superconductors wave function, pairs.
UBC PHYSICS 502 PROJECT (2017 FALL) 2

dn1 2K √
= n1 · n2 · sin(φ2 − φ1 )
dt ~
dn2 2K √
= − n2 · n1 · sin(φ2 − φ1 )
dt ~
dφ1 K p qV
= − n2 /n1 · cos(φ2 − φ1 ) −
dt ~ 2~
dφ2 Kp qV
= − n1 /n2 · cos(φ2 − φ1 ) +
dt ~ 2~
Since the current is I = nAq dx
dt with cross section A, charge
carrier density n, and charge of the carriers q, the first two Fig. 2. A DC SQUID is made up two Josephson junctions a and b connected
equations become: in parallel.

Is = I0 · sin(φ2 − φ1 ) (3)
where we assumed I0 = 2K ~ 2qΩ where n1 = n2 = n, and Ω Since ns is proportional to the charge carrier density of
is the volume of the two superconductors. This is known as cooper pairs, we treat it as a current density.1 Similarly, in
the first Josephson Equation and describes how the tunneling a superconductor the current induced by an external magnetic
current will depend on the phase difference between the two field flows mostly on the surface, so the current density J will
superconductors. But how does the phase difference evolve in be zero at some point in the volume.
time? If we again assume n1 = n2 = n and subtract the third This result is shown through a detailed analysis of (5), as
and fourth equations from each other we get: is done in [3]; it is closely related to the Meissner effect.
d qV Essentially, by exploiting Maxwell’s equations we can obtain
(φ2 − φ1 ) = (4) a differential equation for B using (5). By solving it we
dt ~
see that the magnetic field is exponentially suppressed in
which is the second Jospehson equation and describes the time
the superconductor. By plugging that solution into the third
evolution of the phase difference given an external voltage.
Maxwell equation again, this also leads to an exponentially
These two Josephson equations describe the phase evolution
suppressed current in the superconductor. This, in essence, is
of the two superconductors and give rise to very interesting
the Meissner effect, where the current on the surface creates
and non-intuitive physics. Suppose for example that there is a
a magnetic field which cancels the external one. Intuitively
constant voltage applied to our Josephson Junction. Then (4)
this make sense since we know that the magnetic field is
tells us that the phase difference will evolve linearly in time.
not energetically favorable for a superconducting state. But
Plugging this back into (3), we see that we will then get an
for our purposes this means that there exists a path where
AC current. This is known as the AC Josephson effect.
J = 0 around the entire loop. This allows us to simplify (5)
But what if we apply no external voltage? The phase differ-
significantly by integrating over that path from point 1 to 2
ence, given by (4), now becomes constant. Again plugging this
where J = 0.
back into (3) we see a constant supercurrent flowing despite Z 2
2qe 2
Z
no external voltage. This very unconventional behavior results ∇φds = φ2 − φ1 = Ads (6)
directly from the phase coherence of the cooper pairs, which 1 ~ 1
is also known as the DC Josephson effect. where qe is the charge of the electron. Because the wavefunc-
tion is single valued, the change in the phase between the
B. The DC Squid points P and Q in Fig. 2 is:
2qe Q 2qe Q
Z Z
Now that we have the equations that describe just one
Josephson junction, we can apply these to a loop with two φQ − φP = δa + Ads = δb + Ads
~ P ~ P
Josephson Junctions in a magnetic field, as shown in Fig.2. We
where δa and δb are the additional phase differences of the
will see in this section that we can determine the flux present in
superconductors. By using Stokes theorem we get:
the loop between the junctions with extreme precision because I Z
of the unique properties of the Josephson Junction. 2qe 2qe 2qe
δb − δa = Ads = BdA = Φ
We know now from basic quantum mechanics that the ~ ~ ~
probability current in the electromagnetic field is given by Another common approach that considers the gauge invariant
the following equation [5]: phase change can be found in [4] and also reaches the same
( " # " #∗ ) conclusion. The current Iges is now the sum of the two
1 ∗ P − qA Pb − qA
b
J= Ψ Ψ+Ψ Ψ∗ branches. If we use this, the first Josephson equation with
2 m m δ = φ2 − φ1 , δa = δ0 + q~e Φ and δb = δ0 − q~e Φ we get:
If we use (1) we see that this becomes:
n  qe   qe o
Iges = I0 sin δ0 + Φ + sin δ0 − Φ
~  q  ~ ~
J= ∇Φ − A ns (5)
m ~ 1 This can be easily verified by checking the units.
UBC PHYSICS 502 PROJECT (2017 FALL) 3

We can rewrite that as: Putting this in eq(3), the current in the loop then becomes
q 
Iges = 2I0 sin(δ0 )cos
e
Φ (7) 2qe
~ I = I0 sin(2πn − Φ)
~
This result means that we can determine an unknown mag- 2qe
= −I0 sin( Φ) (9)
netic flux by measuring the current. 2I0 sin(δ0 ) can first be ~
determined by putting the dc SQUID in some known magnetic The last equality exploits the identity that sin(2πn − θ) =
flux Φ=Φknown , e.g. Φknown = 0 or the earth’s magnetic field −sin(θ). Recall from (8) that Φ = Φex − LI. Inserting I
flux for example. Then the only independent variable would found in (9), we find a self-consistent relation for the flux
be Φ in the above equation, and we can measure the unknown through the loop, which will be very useful in discussing the
magnetic flux from the current. Let’s look at an concrete RF SQUID in (C.2):
example to see how sensitive a dc squid can be. A typical 2qe
value for the area of a SQUID used as a magnetometer is Φ = Φex − LI0 sin( Φ)
~
A = 1 cm2 . In this case a change in the magnetic flux density 2π
of 1 · 10−11 T is a half period in the current and can therefore = Φex − LI0 sin( Φ) (10)
Φ0
be measured easily.
h
where Φ0 is defined to be 2qe .

C. RF SQUID (C.2) RF SQUID

As opposed to the DC SQUID, which has two junctions, the
RF SQUID is composed of just one junction. In order to fully
account for the behavior of the RF SQUID, we must look at
the inductance of a Josephson Junction in an external field.
(C.1) One junction superconducting loop

Fig. 4. Basic circuit of RF SQUID: a one-junction superconducting

loop coupled to a LC circuit with a alternating current source of radio
frequency.(Credits: [4])

The RF (Radio Frequency) SQUID is a one-junction SQUID

loop that can be used as a magnetic field detector (Fig. 4).
Although it is less sensitive than the DC SQUID, it is cheaper
and easier to manufacture and is therefore more commonly
Fig. 3. A one-junction superconducting loop in an external magnetic field.
used. The Quantum Design SQUID Magnetometer that is
frequently used by experimenters for example, uses a one loop
Suppose a one-junction superconducting loop is put in an RF SQUID.
external flux Φex as shown in Fig. 3. Since in any real material, In the RF SQUID, the one-junction superconducting loop
there is always a certain inductance L in a circuit we must is coupled to a circuit driven by a RF current source, i.e.
now take that into account in our equations. By including the the loop would experience the flux produced by the current
inductance in our equation for the flux we now find that the (ΦRF ). When the loop is put in some magnetic flux Φq that we
total flux is given by: want to measure, the total external flux is Φex = Φq + ΦRF .
Recall from (10) that the flux through the loop is given by
Φ = Φex − LI (8) Φ = Φex − LI0 sin( Φ 2π
Φ), which can be plotted below.
0

where I is the current flowing in the loop. The RF SQUID has two modes of operation, the hysteretic
and non-hysteretic modes. When 2πLI 0
> 1, we can see from
Let δ be the phase difference across the junction, then the Φ0
the plot of Φ vs Φex in Fig. 5 that some parts of the curve
phase change around the loop is
I I will have positive slope and some will have negative slope.
~ 2qe ~ s This is the ‘hysteretic mode’. When the RF current and thus
δ + ∇φd~s = δ + Ad~ (∵ eq(6))
~ ΦRF oscillates over time, as shown by the orange curve in
2qe Fig. 5, the Φ curve will trace out the blue of the curve. A
=δ+ Φ
~ flux quantum enters and leaves the loop during the upward
This phase difference must be equal to 2πn such that there and downward transitions respectively. The amount of ΦRF
is no physical observable change in our wave function. Rear- required to conserve the energy in the transition depends on
ranging, we get how big Φq is (Recall Φex = Φq + ΦRF ). Since the voltage in
2qe the RF circuit is related to ΦRF , the voltage readout reflects
δ = 2πn − Φ the Φq that we are measuring. In fact, as in DC SQUID, when
~
UBC PHYSICS 502 PROJECT (2017 FALL) 4

2πLI0 Fig. 6. A SQUID magnetometer design used in the commercial Quantum

Fig. 5. Plot of Φ vs Φex in the oscillating ΦRF for Φ0
> 1. (Credits:
[4]) Design Magnetic Property Measurement System as shown in reference [11].
Four loops are used, two with currents going clockwise and two with currents
going counterclockwise, to cancel out the effects of a static magnetic flux. The
sample is moved from below the lowest loop to above the highest loop and the
flux is measured as a function of position via the RF SQUID. The isolation
Φq is changed, there is also a periodic variation of the RF transformer is used to decouple the SQUID loop from the pickup coils when
voltage which is measured to determine Φq . the temperature and magnetic field are changed. Any abrupt changes in the
The other mode of SQUID, the nonhysteretic mode, occurs current, like those due to changes in the external field, can damage the
Josephson Junction. This design allows for the SQUID to be kept at low
when 2πLI
Φ0
0
< 1, and the Φ vs Φex relation is therefore single- magnetic field and temperature while the sample temperature and field are
valued. In this case no more energy is lost in the loop. But as varied.
Φq is varied, the current induced in the SQUID loop is varied.
Through the mutual inductance between the SQUID loop and
the RF circuit, the load on the RF circuit and thus RF voltage novel techniques that use a SQUID as part of a more complex
are similarly varied. In this way measuring the RF voltage experimental setup. Because SQUIDs are so prevalent it is
enables us also to measure Φq . difficult to give a full accounting of all the ways they are used
experimentally.
III. SQUIDS IN THE REAL WORLD The magnetization of a complex magnetic system can often
Based on the principles outlined above, it is clear that be used as an order parameter, so measuring it with a SQUID
SQUIDs enable the precise measurement of magnetic flux is a useful way to characterize complex phase transitions.
and therefore, the magnetization of crystal samples. But many The Meissner effect in superconductors, for example, where
issues arise in application that require more complex exper- the superconductor expels all magnetic fields below Tc , can
imental setups than those. In practice, however, one might be readily observed using a SQUID. As a superconducting
be interested in measuring the magnetization at temperatures sample drops below Tc , the system will react to an external
and magnetic fields far above the Tc or Hc of the SQUID. magnetic field by creating its own equal and opposite field
Therefore designs are used that isolate the SQUID from the as previously outlined. In effect the superconducting system
sample. These systems are very complex and because of this becomes magnetized so that the total internal magnetic field is
most SQUID magnetometers are not built at home but rather zero. This change in magnetization can be used to determine
purchased from Quantum Design. The basic schematic of a the Tc of a superconducting sample with good precision. This
Quantum Design SQUID magnetometer is shown in figure 6. principle is outlined in figure 7 where the the magnetization of
Owing to the commercial nature of this system, the intricate supercondcuting cuprate is shown as a function of temperature
details are not readily available. However the basic operation and clearly shows the transition to a superconducting state
is as follows: the sample is placed in the system such that it [6]. This is just one example of how a SQUID can be used
sits inside the four pickup coils. These coils are electrically to determine the transition temperature of a system into or
connected to a SQUID which is kept isolated from any applied out of a magnetic state. The same principle can be used to
magnetic field and far below Tc , usually at Liquid Helium understand transitions as a function of magnetic field as well.
Temperatures. The sample is then moved vertically inside In order to better understand the spin structure of a system,
the four pickup coils and the voltage across the SQUID is a SQUID can be used to measure the magnetization as a
measured as a function of sample position. The use of four function of magnetic field. In complex magnetic systems,
coils cancels out an externally applied magnetic field and helps determining the correct model to describe the underlying
to isolate the flux from the sample itself. This voltage versus spin physics is often difficult. Magnetization measurements
position graph is then fit to a model using a complex algorithm obtained via SQUID are vital for this process. In Cr1/2 NbS2 ,
designed by Quantum Design that uses the equations outlined for example, by comparing magnetization measurements to
above to determine the magnetization. modeled results, it was shown that the intercalated Cr spins
Data obtained via SQUID typically falls into three cate- are described by a Heisenberg Spin Hamiltonian with an added
gories: crystal characterization, analysis of spin structure, and interaction known as a Dyzaloshinkskii-Moriya interaction and
UBC PHYSICS 502 PROJECT (2017 FALL) 5

mental techniques such as scanning probe microscopy. In a

scanning SQUID Microscope, for example, a DC squid is
attached to the end of an SPM and scanned across a sample.
This allows for the determination of the magnetic properties
of a system as a function of position [8]. The limiting factor
to the spatial resolution is the size of the SQUID itself and
currently SQUID loops are constructed that are on the order of
100nm [8]. With these systems the magnetic structure at the
surface of a material can be mapped with excellent resolution.
These microscopes have been used to map out the edge
currents in quantum wells for example [10]. Because currents
generate a magnetic field, the currents in a material can be
mapped out using a scanning SQUID microscope [8], [10].
This is especially useful as non-trivial topolical phases are
Fig. 7. The magnetization as a function of temperature for a superconducting investiagted in materials as some of these systems lead to edge
sample of YBCO measured via a SQUID from reference [6]. The system
undergoes a clear phase transition to a superconducting state at T ≈ 93K. currents [8]. They have also been used to determine spatial
inhomogeneity of the superconducting phase in 2D Niobium
doped SrTiO3 [9]. Figure 9 shows this inhomogeneity at
330mK [9]. Similarly with figure 7, the lower magnetic flux is
crystaline anisotropy [7]. In this example, the magnetization an indication that the sample is superconducting because of the
was measured as function of magnetic field for three different Meissner effect. This inhomogeneity in the superconducting
field orientations as shown in figure 8. Again this is just one state is very useful for understanding the effects of local
of many possible instances in which magnetization data can structural and doping variations [9].
be compared to models to determine underlying spin physics.
This is such a fundamental part of modern condensed matter
physics that it is hard to find a material where this has not
been done.

Fig. 9. The magnetization as a function of position in Niobium doped SrTiO3

thin films as reported in reference [9]. Here a SQUID is scanned across the
sample to determine the magnetic properties as a function of position.

Ultimately it is difficult to outline just how universal

SQUIDs are in modern condensed matter physics. From basic
crystal characterization to scanning SQUID microscopy, there
are very few experiments that are not at least tangentially
connected to the use of a SQUID.

Fig. 8. (a-d) The magnetization of Cr1/2 NbS2 as a function of magnetic IV. C ONCLUSION
field for four different field orientations from reference [7]. The red line is the
modeled result from a Heisenberg Spin Hamiltonian with a Dyzaloshinkskii- Because of this, the SQUID is one of the most versatile mag-
Moriya interaction while the blue dots are the actual measured data. The units netic flux measurement techniques in modern physics. Here we
are represented such that it is the magnetic moment per Cr ion. (e) The spin
structure of the Cr ions determined from this data at B = 0.174 T
have outlined the basic principles which govern the behavior
of the DC and RF SQUID and given a very brief outline of
the SQUID’s myriad uses. To say that this treatment of the
SQUIDs can also be combined with other novel experi- applications is complete would be a mischaracterization as
UBC PHYSICS 502 PROJECT (2017 FALL) 6

SQUIDs are relevant to everything from quantum computing [6] Ruixing Liang, D. A. Bonn, W. N. Hardy, Growth of high quality YBCO
to detectors at CERN. single crystals using BaZrO3 crucibles Physica C: Superconductivity
Volume 304, Issues 1-2, 1 August 1998, Pages 105-111
[7] Benjamin J. Chapman, Alexander C. Bornstein, Nirmal J. Ghimire, David
V. A PPENDIX Mandrus, and Minhyea Lee, Spin structure of the anisotropic helimagnet
Cr1?3NbS2 in a magnetic field Appl. Phys. Lett. 105, 072405 (2014)
Here we will discuss the wave function (1) and its motiva- [8] Kathryn Ann Moler, Imaging Quantum Materials, Imaging quantum
tion as based on reference [3]. At temperatures near the phase materials, Nature Materials 16, 1049?1052 (2017)
[9] Hilary Noad, Eric M. Spanton, Katja C. Nowack, Hisashi Inoue, Minu
transition TC it can be derived directly from BCS theory, as Kim, Tyler A. Merz, Christopher Bell, Yasuyuki Hikita, Ruqing Xu,
was first done by Gor’kov in 1959 [2]. This means that this Wenjun Liu, Arturas Vailionis, Harold Y. Hwang, and Kathryn A. Moler
wavefunction is not explicitly valid for all temperatures, but is Variation in superconducting transition temperature due to tetragonal
domains in two-dimensionally doped SrTiO3 Phys. Rev. B 94, 174516
still applicable for the analysis here because it preserves the (2016)
macroscopic physics we are dealing with. [10] Eric M. Spanton, Katja C. Nowack, Lingjie Du, Gerard Sullivan, Rui-
To begin this analysis we turn to the cooper pair, a bound Rui Du, and Kathryn A. Moler Images of Edge Current in InAs/GaSb
Quantum Wells Phys. Rev. Lett. 113, 026804 (2014)
state of two electrons, one with spin up and one with spin [11] Magnetic Property Measurement System Hardware
down. The wave function for one cooper pair added to an Reference Manual. Quantum Design, San Diego CA.
inert Fermi See can be written as follows: [1]: http://mmrc.caltech.edu/MPMS/Manuals/QD
~ ~
X
Ψ(~ρ) = ei~qR · A(~k) · eik~ρ (11)
~
k>k~f

where ρ ~ = r~1 − r~2 , the center of mass coordinate is

R~ = 1/2(r~1 + r~2 ) and the center of mass momentum is
~q = 1/2(k~1 + k~2 ) (which is zero here owing to the cooper pair
being composed of two electrons with opposite momentum
i.e. ~k and −k).
~ For simplicity we will only look at the spatial
part of the wavefunction and not worry about the symmetry
requirements.
We will be interested in the electrodynamics of this wave-
function so we can now consider the effects of an electric field,
which will add a finite momentum ~~q to the cooper pair.
We can also assume that the N particle wavefunction
can be expressed as the product of all single cooper pair
wavefunctions. Taking this analysis one step further, since
cooper pairs are bosons they are not restricted by the Pauli
exclusion principle. Therefore, we can imagine a state where
all cooper pairs are condensed in the lowest energy state. Since
the cooper pairs are in the same state any applied field will
give all the cooper pairs the same momentum ~~q. Therefore the
wave function (without antisymmetrisation) can be estimated
as:
~ ~
Φ(R~1 , R~2 , ..., R~N ) = ei~q(R1 +···+RN ) Ψ(ρ~1 ) · · · Ψ(ρ~N )
This state has a precisely defined phase at each point ~r in
the superconductor if there is a large number of cooper pairs.
Similarly, the wave functions for the relative coordinates can
be thought of as a cooper pair density since they are functions
of position. From this we can see where the wave function
outlined in (1) is not completely unjustified and has a physical
basis in the physics of cooper pairs. Clearly this is just an
intuitive picture and not an rigorous derivation.

R EFERENCES
[1] Schrieffer, J. Theory Of Superconductivity Avalon Publishing, 1983
[2] Gorkov LP. Microscopic derivation of the Ginzburg-Landau equations in
the theory of superconductivity. Sov. Phys. JETP. 1959 Jan 1;9(6):1364-7.
[3] Kopitzki, K.,Einfhrung in die Festkrperphysik, Springer Berlin Heidel-
berg, 2017
[4] Duzer T.V., Turner C.W., Principles of Superconductive Devices and
Circuits, 1998
[5] Feynman, R. Leighton, R. Sands, M., The Feynman Lectures on Physics,
Pearson/Addison-Wesley, 1963