A rapid sample-exchange mechanism for cryogen-free dilution refrigerators
compatible with multiple high-frequency signal connections
G. Batey, S. Chappell, M.N. Cuthbert, M. Erfani, A.J. Matthews, G. Teleberg
Reference: JCRY 2295
To appear in:
Received Date: 29 October 2013
Revised Date: 13 January 2014
Accepted Date: 15 January 2014
Please cite this article as: Batey, G., Chappell, S., Cuthbert, M.N., Erfani, M., Matthews, A.J., Teleberg, G., A rapid
sample-exchange mechanism for cryogen-free dilution refrigerators compatible with multiple high-frequency signal
connections, Cryogenics (2014), doi: http://dx.doi.org/10.1016/j.cryogenics.2014.01.007
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
A rapid sample-exchange mechanism for cryogen-free
dilution refrigerators compatible with multiple
high-frequency signal connections
G Batey, S. Chappell, M.N. Cuthbert, M. Erfani, A.J. Matthews
Oxford Instruments Omicron NanoScience, Tubney Woods, Abingdon, Oxfordshire,
OX13 5QX, UK
Researchers attempting to study quantum eﬀects in the solid-state have a need
to characterise samples at very low-temperatures, and frequently in high mag-
netic ﬁelds. Often coupled with this extreme environment is the requirement for
high-frequency signalling to the sample for electrical control or measurements.
Cryogen-free dilution refrigerators allow the necessary wiring to be installed to
the sample more easily than their wet counterparts, but the limited cooling
power of the closed cycle coolers used in these systems means that the experi-
mental turn-around time can be longer. Here we shall describe a sample loading
arrangement that can be coupled with a cryogen-free refrigerator and that al-
lows samples to be loaded from room temperature in a matter of minutes. The
loaded sample is then cooled to temperatures ∼ 10 mK in ∼ 7 hours. This
apparatus is compatible with systems incorporating superconducting magnets
and allows multiple high-frequency lines to be connected to the cold sample.
Keywords: Dilution refrigerator, Sample exchange, Cryogen-free
Over the past century studying condensed matter systems at extremely low
temperatures, and often in extremely high magnetic ﬁelds, has lead to the dis-
covery of several new states of matter, such as: superconductivity in mercury ;
He [2, 3]; superﬂuidity in
He ; the integer quantum Hall
eﬀect in silicon MOSFET devices ; the fractional quantum Hall eﬀect in
GaAs-AlGaAs heterojunctions .
More recently there has been a drive to harness these quantum systems to
realise devices that exploit their quantum nature, for example in the ﬁeld of
Email address: Anthony.Matthews@oxinst.com (A.J. Matthews)
Preprint submitted to Elsevier January 21, 2014
quantum information processing , with the realisation of a general quantum
computer  being the holy grail. Inevitably the development of these quantum
devices requires temperatures < 10 mK, and possibly magnetic ﬁelds > 10 T,
however in addition to these environmental constraints device characterisation
and development also requires the necessary experimental services be installed
at the sample position: most challengingly high-bandwidth, high-ﬁdelity micro-
In the following sections we describe brieﬂy a suitable experimental envi-
ronment for quantum device development (or any other experiments requiring
high-frequency measurements at low-temperatures), then we show that device
characterisation is more convenient with a sample loading mechanism, and de-
scribe its realisation, op eration and performance, before providing a brief con-
2. Experimental Environment
Pulse-tube precooled dilution refrigerators  are becoming increasingly pop-
ular. Initially this popularity stemmed from the fact that they were cryogen-free,
meaning that they could be installed at institutions without the associated low-
temperature research infrastructure, such as a helium liquefaction plant, or in
remote locations. Additionally, there are beneﬁts from an operational point of
view as such systems can be automated to a higher degree than their “wet”
counterparts. It has also been found that these cryogen-free systems have fur-
ther beneﬁts when compared to wet systems with regards to the installation of
experimental services, as will be discussed in the following sections, and this has
driven the recent rise in their uptake.
With the installation of high-frequency wiring these refrigerators have been
developed into measurement systems for circuit quantum electrodynamics 
and superconducting qubits . The integration of superconducting mag-
nets , with the entire system able to be run from a single pulse-tube cooler,
has enabled a wider range of experiments (those requiring magnetic ﬁelds) to
be performed using this cryogen-free technology .
2.1. Low-Temperatures and High Magnetic Fields
Cryogenic systems using liquid helium are usually designed to minimise its
consumption. This is because liquid helium is expensive, reﬁlling the system
can be time consuming, and reﬁlling the system may perturb the experiment to
an unacceptable level. The central neck of a cryostat is often responsible for the
biggest single heat load into the helium bath, and as a result these necks are
usually made as long and as narrow as possible. Dilution refrigerators designed
to be inserted into such a cryostat have to inherit this aspect ratio, which has
tended to limit the experimental real estate available for the installation of
With no boil-oﬀ considerations, cryogen-free systems have evolved to be
much wider than their wet counterparts with experimental plates (to which ser-
vices can be mounted) typically several hundred mm in diameter . This has
enabled more and / or more complex services to be installed on dilution refriger-
ator systems, in particular bulky signal conditioning elements such as cryogenic
ampliﬁers, microwave components (bias-tees, circulators, switches etc.) and ﬁl-
tering (such as metal powder ﬁlters, for example  and the references therein).
Cryogen-free systems can also be designed without the need for a low-
temperature, vacuum-tight vessel, the so called inner vacuum chamber (IVC),
which makes the routing and heat-sinking of the installed services much more
straightforward, see section 2.2.2.
The range of magnets that are able to be produced for cryogen-free operation
is also continually expanding with higher ﬁelds (> 16 T) and vector-rotation
(> 6-1-1 T) available.
For these reasons cryogen-free dilution refrigerators with integrated magnets
have become the workhorse of quantum device development laboratories around
2.2. High-Frequency Wiring
As was noted in section 1 high-ﬁdelity, high-bandwidth wiring is an experi-
mental requirement for quantum device development applications. In addition
to the quality of the signal transmission performance of these cables, they also
need to be thermally anchored adequately to ensure that they do not aﬀect
adversely the base temperature performance of the system onto which they are
installed. In this section we shall: review various options for the coaxial lines and
some of the materials available for the lines themselves, and discuss their rela-
tive merits; describe a convenient method for mounting multiple high-frequency
lines onto a dilution refrigerator; quantify the frequency dependence of signal
transmission of installed lines with S
measurements made with a vector net-
work analyser; comment on the heat load to the mixing chamber likely to result
from the installation of the type of wiring described.
2.2.1. Coaxial cables and materials
To date, most high-frequency cabling installed in dilution refrigerators have
been of “semi-rigid” construction with the UT-85 cable (having an outer diam-
eter of 85/1000 of an inch, approximately 2.16 mm) being commonly used. The
optimal choice of coaxial cable, in terms of both size and material, depends on
its intended application. Typically coaxial cables are used to 1) improve noise
immunity for “small” signals and / or 2) transmit high-frequency signals to /
from the sample.
If using coaxial cables for either of these reasons one should ensure that the
cables themselves are suitable for the intended application. For dilution refrig-
erator based experiments, this suitability is generally determined by two key
parameters: the heat load to the experiment due to the thermal conductivity of
the cable; and its (frequency dependent) attenuation. Both of these parameters
are aﬀected by the choice of the cable geometry (size) and conductor materials.
The heat load conducted to the coldest parts of the dilution refrigerator is al-
ways to be minimised. For a given choice of coaxial cable material and geometry
there is a lower limit to this heat load determined by the bulk thermal conduc-
tivity of the cable materials. This limit is approached as the cable (both the
inner and outer conductor) is p erfectly thermally connected to every available
temperature stage in the refrigerator, of course the conducted heat load can be
much higher than this limit if the thermal connections are inadequate. A con-
venient method of installing semi-rigid coaxial cables into a dilution refrigerator
that gives good thermal performance is discussed in section 2.2.4. The heat load
can only be reduced further by using either cables with a smaller cross sectional
area and / or cables made from materials with a lower thermal conductivity,
however such changes may well have implications for the cable attenuation.
The frequency dependent attenuation of a coaxial cable is determined by
the cable geometry (outer diameter of the inner conductor and inner diame-
ter out of the outer conductor), the (temperature and frequency dependent)
resistivity of the conductor materials and the dielectric losses . In general
smaller diameter cables have higher attenuation at high frequencies than larger
diameter ones, and cables manufactured from materials with higher bulk resis-
tivity have higher attenuation (at a given frequency) than low resistance ones.
Depending on the application, this increase in attenuation can be fortuitous or
problematic. In applications where coaxial cables are used for noise immunity
for small, low-frequency signals, having increased attenuation at high frequen-
cies is advantageous: in fact “lossy” coax cables have been used as microwave
However, for high-bandwidth signals the change in attenuation, α, with fre-
quency, f, is undesirable as it results in the “shape” of signals (in the time-
domain) being modiﬁed as they propagate along the cable and this can cause
problems with, for example, high-ﬁdelity qubit control. Techniques borrowed
from the NMR / MRI world for pulse preshaping using a posteriori knowledge
of the cabling transfer function  can be applied to compensate for this ef-
fect, but it would still be advantageous to keep the frequency response of the
cable as ﬂat as possible. Using (lots of) large-diameter low-resistance cables can
be incompatible with experiments at dilution refrigerator temperatures, as the
thermal and electrical conductivity of a normal metal are closely related .
However, superconducting cables made from Nb, or preferably NbTi (due to its
higher critical ﬁeld and temperature, and lower thermal conductivity), can be
used. Below their superconducting transition temperature these cables provide
very low attenuation and have a small thermal conductivity  so in many
cases are the ideal solution to this problem. However, with cryogen free dilution
refrigerators enabling experiments over extended temperature ranges  some
care needs to be taken, as the electrical performance of these lines will change
(attenuation will increase) dramatically above their transition temperature.
One ﬁnal point is that the desire to keep
≈ 0 is not the same as keeping
α ≈ 0. Indeed, the types of cables described here are very good at transmitting
“thermal noise” from warmer parts of the refrigerator to colder ones, equating
hν ≈ K
T gives a photon frequency of 20 GHz at 1 K and UT-85 cables
operational range can extend to > 60 GHz , and so having some attenuation
in the line is desirable to reduce these thermal perturbations. Attenuators with
a ﬂat frequency response, compatible with cryogenic temperatures , can be
used to increase the attenuation of a line whilst avoiding the complications of
distorting high-bandwidth signals. Details of measurements of such lines will
be given in section 2.2.3.
2.2.2. High-frequency wiring cartridges
As described in section 2.2.1, for some experiments small diameter coaxial
cables with high attenuation at high-frequencies can be appropriate. For ex-
ample, UT-13 cables have an outer diameter of approximately 330 µm and can
be installed and thermally anchored like ﬂexible “DC” wiring. In this section
we focus on semi-rigid cables and describe a convenient method of installing
multiple, conﬁgurable, semi-rigid coaxial lines into a dilution refrigerator in a
way that gives good electrical and thermal performance and allows for the cable
assemblies to be rapidly demounted and modiﬁed if necessary.
Cryogen-free dilution refrigerators typically have several large (40 - 100 mm
diameter) line-of-sight (LoS) ports that allow connections between the room
temperature top-plate and the mixing chamber plate. Whilst traditional wet
dilution refrigerators also often feature LoS ports they tend to be less numerous
and of smaller diameter. Wet systems also require an IVC and so services
need to be installed in vacuum tubes from room temperature to 4 K, making
the thermal anchoring of the installed services more diﬃcult (services can of
course be thermalised by bringing them through the main helium bath, but
then cryogenically compatible, hermetically sealed feed-throughs are required
to bring the services into the IVC).
A typical cryogen-free refrigerator will have experimental plates that can
be used to thermally anchor wiring at temperatures of approximately 50 K,
3 K, 0.8 K, 100 mK and the mixing chamber at around 10 mK. The wiring
cartridge shown in ﬁgure 1 has anchoring plates at each temperature stage.
It has been found that the use of bulkhead connectors at each of these plates
is an eﬀective way to thermalise both the inner and outer conductors of the
cables , and provides a convenient mounting point for any attenuators that
may be added to the lines at any of the stages. The cartridge is designed to be
able to be loaded into the system either completely assembled from the top, or
without the top plate ﬁttings from below. Once the cartridge is installed, split
clamp plates are used to make thermal contact between the cartridge and the
refrigerator. The fact that the entire wiring assembly can be removed in one
piece allows for bench testing of the microwave lines prior to installing them
into the system. It also means, for example, that should there be a desire
to change installed attenuators for ones with a diﬀerent attenuation value the
assembly can be removed from the refrigerator by simply opening one room
temperature o-ring seal and loosening the clamping bolts. With the assembly
removed, the microwave lines or attenuators, between the bulkhead connectors,
can be reconﬁgured and tested before being reﬁtted to the system.
Figure 1: A wiring cartridge for a cryogen-free dilution refrigerator: (a) shows a fully assembled
cartridge with hermetic feed-throughs on the room temperature top plate and additional
attenuators installed above and below some of the thermal stages. (b) shows the detail of
a split clamp used to thermally anchor the cartridge to the refrigerator and the bulkhead
connectors through the cartridge plate. (c) shows how such a section of such a cartridge could
be installed through a line-of-sight port of a dilution refrigerator.
Figure 2: Scattering parameter measurements on coaxial cables installed in wiring cartridges.
The red and black traces show lines with no additional attenuation installed. The green and
blue traces show lines with an additional 28 dB of in-line attenuation. The reduction in the
attenuation between room temperature and the system being cooled is due principally to
sections of superconducting coaxial cable cooling below their transition temperature. The dip
in the attenuation on the blue trace (circled) was due to a lo ose connector in the cartridge
assembly, this was corrected prior to the cartridge being installed into the system and cooled.
2.2.3. Transmission measurements
The microwave performance of installed coaxial cable assemblies has been
measured with an Antitsu model MS2028C/2 vector network analyser  which
recorded the scattering parameters at frequencies up to 12 GHz. Typical curves
between 5 kHz and 8 GHz are shown in ﬁgure 2. The S
be associated with the total attenuation in the line and the measured values
agree well with cable manufactures’ data for expected values of the frequency
dependent attenuation (per unit length) of the cables they produce [19, 23], in
this case the coaxial cable sections themselves were silver-plated stainless steel
inner conductor, stainless steel outer conductor from room temperature to the
4 K plate, and NbTi inner and outer from 4 K to the mixing chamber. Faults
with the coaxial cables, such as loose connectors or cracked solder joints, can
be identiﬁed from scattering parameter measurements , and for the cables
installed on these systems typically result in additional attenuation (reﬂection)
features at frequencies of a few GHz, ﬁgure 2.
2.2.4. Thermal performance
As discussed in section 2.2.1 there is a lower limit to how far the heat load
from installed cabling can be reduced. Here we examine the residual heat load
onto the mixing chamber of a cryogen-free dilution refrigerator when wiring
Figure 3: (a) An image showing how multiple coaxial cable cartridges can be installed onto
a dilution refrigerator system. In-line attenuators are visible below the upper plate, which
is at the position of the dilution refrigerator still. (b) A plot of the typical cooling capacity
available at the mixing chamb er of such a dilution refrigerator. The ﬁt in the plot is of the
form y = ax
cartridges are installed. This heat load can be extracted using knowledge of
the cooling power of the dilution refrigerator at the temperatures of interest.
With knowledge of the cooling power, the base temperature of a refrigerator
with and without wiring installed can be compared and the heat leak extracted
from the diﬀerence in these temperatures. For these measurements three wiring
cartridges, each containing eight UT-85 coaxial lines (24 lines in total) manufac-
tured from cupronickel conductors, were installed onto a Triton200  dilution
refrigerator system, as show in ﬁgure 3. After the addition of the wiring car-
tridges the temperature of the plate mounted at the end of the continuous heat
exchanger, colloquially know as the “100 mK plate”, had increased from 65
mK to 120 mK as measured with a resistive temperature sensor . The base
temperature of the dilution refrigerator, measured using a nuclear orientation
thermometer , was found to have risen to 9.1 mK, corresponding to an in-
creased heat load of ≈ 600 nW. Extrapolating available data for the thermal
conductivity of cupronickel  to 100 mK and calculating the anticipated heat
load conducted through 24 UT-85 coaxial lines with the geometry deﬁned by
manufactures  accounts for ≈ 200 nW of this increase, with a further ad-
ditional ≈ 300 nW expected through the stainless steel refrigerator support
structure, calculated using published values for the thermal conductivity ,
due to the increase in temperature of the 100 mK plate.
3. Rapid Sample Exchange
In section 2 it was shown that cryogen-free dilution refrigerators integrated
with superconducting magnets provide an ideal environment for quantum de-
vice development experiments due to their ease of use and the convenience of
installing experimental services. These systems do, however, have one signiﬁ-
cant drawback compared to their wet counterparts as the integrated supercon-
ducting magnets become larger: the experimental turnaround time. High-ﬁeld
cryogen-free magnets can have masses well in excess of 50 kg and require en-
thalpy changes of several MJ to cool from room temperature to 4 K. The pulse
tube coolers used in these systems typically have cooling powers at the second
stage of ≈ 140 W at room temp erature, falling to ≈ 1 W at 4 K . This
limited cooling power means that the initial cool down from ro om temperature
can take much longer than wet systems (which can be cooled quite quickly if
the cryogen boil-oﬀ can be tolerated).
This drawback can be overcome with a method of exchanging samples that
keeps the rest of the system (and the magnet in particular) cold. To b e useful
for a wide range of experiments, such a mechanism must provide multiple high-
frequency lines to the sample and provide sample temperatures comparable to
the base temperature of the refrigerator. In the following sections we describe
the design, construction and performance of such a sample exchange mechanism.
3.1. The sample exchange concept
Attaching a sample and experimental wiring directly to a probe and loading
the entire assembly into a dilution refrigerator has been attempted, but it was
found that the resulting thermal performance and limited space is incompat-
ible with multiple high-frequency lines and additional microwave components
(ampliﬁes, ﬁlters etc), see section II A of .
An alternative approach to sample loading, as also implemented in , is
to leave the experimental wiring on the refrigerator, where it can be eﬃciently
thermally anchored, in this case by using the wiring cartridge design discussed
in section 2.2, and to load a “sample holder” to connect to this installed wiring.
Additionally, this means that the full sample space of the refrigerator can be
utilised to install other components into the experimental wiring circuits which
may not ﬁt onto a smaller diameter probe. Loading only a sample holder into
the refrigerator introduces the complication of requiring demountable microwave
connectors, but in the following sections we show that this requirement can
be fulﬁlled. It is often also desirable to be able to bias or ground electrical
connections to delicate samples during the cool down process to prevent, for
example, electrostatic sample damage. This is accomplished with a “make-
before-break” arrangement whereby all DC and microwave connections to the
sample holder are individually connected to room temperature connectors on
the loading probe. As will discussed in section 3.1.2 the sample holder can
also be made demountable, allowing the loading probe to be removed after the
sample is attached to the refrigerator.
With a cryogen-free system without an IVC there is no preferred direction for
sample loading. Samples can either be introduced from the top of the system
using a top-loading load-lock (TLLL) or from below using a bottom-loading
load-lock (BLLL). TLLLs require a (central) LoS access port through which the
sample can be introduced, and BLLLs require access through the vacuum and
radiation shields through which the sample can pass. The distance from the
refrigerator top-plate to the magnetic ﬁeld centre-line is normally longer than
that from the bottom of the system to ﬁeld centre, so systems with TLLLs
tend to require more ceiling height for operation than systems with BLLLs.
However, the sample holder and loading arrangement for BLLL systems needs
to be able to pass into the bore of the system magnet whereas in TLLL systems
the thermal and electrical connections can be made above the magnet, oﬀering
more space for these connections.
Both the TLLL and BLLL require move-able baﬄes to allow the sample
holder to be introduced into the system without leaving an unacceptable heat
load from 300 K blackbody radiation in normal operation. For TLLL systems
these can be controlled with a drive rod mechanically connected to the room
temperature top plate. For BLLL systems it is more convenient to make these
baﬄes spring-loaded as the baﬄes themselves are attached to demountable ra-
The choice of connectors is critical to the microwave performance of a cable
assembly. With standard UT-85 type cables the usual room temperate choices
are SMA  connectors for operation up to 18 GHz and SK  connectors for
operation up to 40 GHz. Both of these connectors are screw lock, so unsuitable
for push ﬁt applications, the BMA  connector range is a blind-mate equiva-
lent of the SMA connector, but suﬀers from being rated only to ∼ 20 GHz and
being rather bulky (∼ 10 mm diameter) which limits the density of connections.
SMP  connectors have the advantage of being blind-mate, small diam-
eter (∼ 3 mm) and rated for 40 GHz operation and so were selected to trial
cryogenically. A test piece was made, ﬁgure 4 (upper panel), and mounted onto
a cryogen-free 4 K platform as a test bed. This test bed was equipped with mi-
crowave cabling enabling round-trip attenuation measurements through pairs of
the test connectors to be made from room temperature.
First the test piece was repeatedly mated and unmated to test the reliability
of the connectors with S
measurements made after every cycle, ﬁgure 4 (b).
The connectors were robust to this cycling, so the system was cooled and the
round-trip attenuation monitored as a function of temperature, ﬁgure 4 (c).
There was no degradation in performance of the connectors with temperature,
the overall reduction in the round-trip attenuation with temperature is due to
the temperature dependence of the coaxial cable material (stainless steel) used
between room temperature and the 4 K stage.
Connectors for multiple DC lines were also trialled cryogenically and a nano
d-type connector  was chosen, principally due to its extremely small foot-
3.1.2. The loading probe
The loading probe is essentially identical regardless of whether the system
is top or bottom loading save for the direction of insertion. The loading probe
consists of a vacuum lock which is mounted onto a gate valve on the top /
bottom of the main vacuum chamber and evacuated prior to introducing the
sample holder into the system. Optionally, the loading probe vacuum lock itself
Figure 4: (a) A test piece for SMP connectors. (b) The round-trip attenuation through pairs
of the connectors measured at room temperature as a function of the connector mating cycle
number. The data for the ﬁrst 25 cycles are for one pair of connectors, the last 25 cycles
are for the second pair. (c) The round-trip attenuation through the ﬁrst pair of connectors
measured at room temperature as a function of the temperature of the connectors.
can be ﬁtted with an additional gate valve to allow samples to be stored under
vacuum prior to loading into the system, and after removal.
The sample holder is mechanically connected to drive rods which enter the
vacuum lock via piston seals, and electrically connected to the biasing / ground-
ing wiring on the probe. The drive ro ds can be used to position the sample
holder at the docking station (detailed in section 3.1.3). The sample holder is
pushed into the docking station (making the electrical connections to the wiring
installed on the refrigerator) and then bolted into place using the drive rods,
utilising fasteners captive to the sample holder, to make the thermal connection.
Typically 2, 3 or 4 (depending on the sample holder conﬁguration) M4 threads
are tightened to a torque of 5 Nm. A double thread arrangement allows the
drive rods to then be disengaged from the sample holder and withdrawn from
the system (breaking the electrical connections to the sample on the probe).
Withdrawing the loading probe after the sample holder is loaded removes the
requirement to thermally anchor the loading mechanism, and removes the as-
sociated conducted heat leak; thus the loaded sample holder has no impact on
the base temperature performance of the refrigerator.
3.1.3. The docking station
The docking station provides the mating electrical and thermal connections
for the sample holder. The cabling attached to the refrigerator is routed to
the docking station. For TLLL systems the docking station is a ring around the
(central) LoS port, for BLLL systems it is a stand-oﬀ that brings the connection
ﬂange into the bore of the magnet. Typically 48 DC lines and 14 microwave
cables can be connected to the docking station, however we note that it is
straightforward to scale up the number of connectors, if required, particularly
on TLLL systems as this can be achieved without the need for a larger magnet
A BLLL sample holder attached to its docking station is shown in ﬁgure 5.
Microwave cable links are ﬁtted between the wiring cartridges, running through
the refrigerator, and the docking station.
3.1.4. The sample holder
Examples of BLLL sample holders are shown in ﬁgure 6. In the ﬁgure,
panel (b) is a design for integration with high-ﬁeld magnets with 57 mm cold
bore diameter, giving a clear diameter sample space inside the sample holder
of ∼ 25 mm (reduced from the diameter of the sample holder by the drive rods
internal to the holder required to make the bolted connections to the docking
station, visible in the ﬁgure) by 90 mm long, symmetric about the ﬁeld centre
line of the magnet. Also shown (c) is a larger diameter sample holder for magnets
with a 90 mm cold bore giving a clear sample space diameter of ∼ 50 mm.
Figure 6 (a) shows the mating surface of the BLLL sample holder. In partic-
ular the SMP connector “bullet” adaptors can be seen (the bullet has been re-
moved from the lower left shroud). On the sample holder, “full detent” shrouds
are used to retain the bullet. On the docking station smooth bore so called
Figure 5: A bottom loading sample holder, with its integrated radiation shield, connected to
the mixing chamber docking station. The microwave cable links between the wiring cartridges
and the docking station are visible.
Figure 6: (a) The mating surface of a bottom loading sample holder showing the 14 SMP
connectors and 51-way nano d-connector. Two alignment pins are visible at the left and right
of the holder and two M4 captive fasteners are visible at the top and bottom. (b) A bottom
loading sample holder with the radiation shield removed. The wiring for grounding or biasing
the sample whilst loading can be seen entering the sample holder from the bottom, and the
experimental wiring entering from the top. Also shown is a PCB sample holder that could
be used for mounting a sample into the holder. (c) A larger diameter bottom loading sample
holder connected to the loading probe before being drawn into the vacuum load-lock. The
four drive rods are visible at the base of the sample holder.
“catcher’s mit” shrouds are used which allow for a certain amount of radial and
axial misalignment between the shrouds during loading.
The sample holder features an integrated radiation shield, which also pro-
tects the sample mechanically during the loading and unloading process.
3.2. Sample cool-down
The procedure for loading a sample is straightforward. While the vacuum
load-lock is evacuated the bulk of the mixture is removed from the refrigerator,
for the refrigerators used in this work that are equipped with pre-cool lines 
this takes around 15 minutes, and any superconducting magnet, if ﬁtted, is de-
energised. The sample is then introduced and connected to the mixing chamber,
and the loading probe withdrawn. The refrigerator then re-cools back to its base
temperature. The re-cooling and running to base can be automated, allowing
a sample to be loaded in the evening and be at base temperature ready for
measurements the following morning.
Figure 7: The cool down of a bottom loading sample holder after being loaded onto a dilution
refrigerator system. (left panel) The cool down from room temperature to mK temperatures in
around 7 hours. The high- and low-range mixing temperature sensors are calibrated between
325 K and 1.4 K, and 40 K and 50 mK respectively. At the lowest temperatures the mea-
surements with resistive sensors are replaced with a nuclear orientation thermometer. (right
panel) Cool down and base temperature measurements and (inset) the temperature stability
at the sample position as measured with the nuclear orientation thermometer.
A typical cool down of a bottom loaded sample holder is shown in the left
panel of ﬁgure 7. In this example the sample holder was loaded onto a refriger-
ator equipped with eight of the silver-plated stainless steel upper, NbTi lower
coaxial lines described in section 2.2.3 and a further eight stainless steel lines.
The base temperature attained at the sample position is shown in the right
panel of ﬁgure 7, as measured with a nuclear orientation thermometer over a
period of several days. The mean temperature at the sample position was found
to be 9.85 mK.
3.2.1. Sample turnaround times
Whilst the co ol down time of a loaded sample can be seen to be around
seven hours from ﬁgure 7, the total turnaround time from removing one sample
to having another cold is also of interest. As the loading probes and sample
holders are interchangeable the optimum turnaround can be accomplished by
having a second sample holder set up on a second loading probe whilst the cold
sample is being removed. The removal of the mixture from the refrigerator
and the unloading of the cold sample can be completed in around 20 minutes.
If the loading probe is equipped with the additional gate valve discussed in
section 3.1.2, or if the sample can be vented to atmosphere whilst cold, then
the two loading probes can be exchanged immediately, the vacuum seals can be
demounted and remade in around 10 minutes. Finally the small volume of the
load-lock can be evacuated and leak tested in around 20 minutes. The loading
of the new sample and disconnection of the loading probe then takes a further
Figure 8: The cooling power measured at the sample position on a top-loaded sample holder.
The ﬁt, a second-order polynomial, is included as a guide.
3.3. Cooling power at the sample stage
As all the experimental services are installed onto the refrigerator, and ther-
mally anchored there, the co oling power requirements at the sample position
(inside the sample holder) are less stringent as only heat dissipated in the sam-
ple itself needs to be adsorbed.
Figure 8 shows the measured cooling power at the sample position in a top
loaded sample holder. The temperatures below 50 mK were measured using a
nuclear orientation thermometer and above 50 mK using a calibrated ruthenium-
oxide temperature sensor , both mounted at the sample position. The heat
load was supplied by a resistive heater mounted nearby.
The base temperature at the sample position was found to be just over
9 mK and the cooling capacity available at 100 mK in excess of 120 µW. This is
reduced slightly from the cooling power available at the mixing chamber plate
itself (typically 200 - 400 µW) due to the ﬁnite thermal impedance between the
loaded sample holder and the docking station, but as all experimental services
are anchored directly to the refrigerator (and not to the sample holder) this
reduction poses no problems for experiments.
We have shown that cryogen-free dilution refrigerators equipped with sample
loading mechanisms are a ﬂexible experimental platform, and are the ideal test
bed for quantum device characterisation and development.
Figure 9: (a) Gigahertz quantised charge pumping in a graphene double quantum dot device.
Single electron pumps operating a high frequencies could allow for a new deﬁnition of the
ampere. The straight lines represent I = ±ef . The measured current oscillates between the
quantised values because of a phase diﬀerence between the two drive signals resulting from
unequal lengths of coaxial line. (b) The fractional quantum Hall eﬀect measured in two-
dimensional electron system. Quantised features at fractional values of ν > 2 exemplify the
low electron-temperatures attained in these measurements, from which a value of 15 - 20 mK
can be inferred.
The engineering of such a sample loading arrangement for either top or bot-
tom sample loading has been described, and we have shown that the performance
attained in terms of the base temperatures at the sample (below 10 mK) and the
microwave characteristics of the (up to 14) coaxial lines that are available at the
sample position are as good as can be achieved mounting a sample directly on
the refrigerator, but that the experimental turn-around time has been reduced
from days to a few hours.
4.1. Application Examples
The versatility of the sample loading system is demonstrated by the wide
range of applications to which they have been applied. Figure 9 shows some
speciﬁc examples of experimental data, from diﬀerent laboratories, obtained
using such sample exchange devices.
Figure 9 (a) shows the current generated by a so-called single electron pump
used to realise a quantum standard for electrical current, as measured on a
top-loading system. In this case the pump is made from graphene patterned
into two nanometre size islands separated by tunnel barriers. By applying high
frequency signals (with well deﬁned waveforms) to gates in close proximity to
the islands, individual electrons can be made to jump from island to island
and hence produce a measurable electrical current . For these experiments
low temperatures are important to minimise thermal excitations that would
produce unwanted tunnelling events and the drive frequency needs to be as high
as possible to create a suitably large current. For these reasons well thermalised,
high-ﬁdelity microwave lines to the sample are extremely important.
Figure 9 (b) shows the longitudinal and transverse resistivity measured in a
δ-doped GaAs-AlGaAs quantum well as a function of magnetic ﬁeld as measured
on a bottom loading system . For these experiments high magnetic-ﬁelds
and low electron-temperatures in the wiring to the sample holder are required.
The quasiparticles of the fractional quantum Hall eﬀect state at Landau-level
ﬁlling factor ν =
are predicted to obey non-abelian statistics which could
allow for the development of topologically protected qubits  for quantum
The authors would like to thank C. Wilkinson and R. Brzakalik for their
contributions to the design of the sample loading arrangement and D. Turner-
Cleaver for his assistance developing the mechanical arrangements.
Part of this work is courtesy of the Oﬃce of the Director of National Intelli-
gence, Intelligence Advanced Research Projects Activity (IARPA), through the
Army Research Oﬃce grant W911NF-12-1-0354 and we thank our IARPA col-
laborators for their detailed discussions on various experimental requirements.
The data presented in ﬁgure 9 are reproduced with the kind permission of
(a) T.J.B.M. Janssen of the National Physical Laboratory, London, UK (b) K.
Rasmussen and C.J.S. Olsen of the Center for Quantum Devices, Copenhagen,
 H. Kamerlingh-Onnes, The superconductivity of mercury, Comm. Phys.
Lab. Univ. Leiden, No.s 122 and 124.
 P. Kapitza, Viscosity of liquid helium below the λ-point, Nature 141 (1938)
 J. F. Allen, A. D. Misener, Flow of liquid helium II, Nature 141 (1938) 75.
 D. D. Osheroﬀ, R. C. Richardson, D. M. Lee, Evidence for a new phase of
He, Phys. Rev. Lett. 28 (1972) 885–888.
 K. v. Klitzing, G. Dorda, M. Pepper, New method for high-accuracy deter-
mination of the ﬁne-structure constant based on quantized hall resistance,
Phys. Rev. Lett. 45 (1980) 494–497.
 D. C. Tsui, H. L. Stormer, A. C. Gossard, Two-dimensional magnetotrans-
port in the extreme quantum limit, Phys. Rev. Lett. 48 (1982) 1559–1562.
 C. Nayak, S. H. Simon, A. Stern, M. Freedman, S. Das Sarma, Non-abelian
anyons and topological quantum computation, Rev. Mod. Phys. 80 (2008)
 D. P. DiVincenzo, The physical implementation of quantum computation,
 K. Uhlig, 3He/4He dilution refrigerator with pulse-tube precooling, Cryo-
genics 42 (2002) 73–77.
 M. Goppl, A. Fragner, M. Baur, R. Bianchetti, S. Filipp, J. M. Fink, P. J.
Leek, G. Puebla, L. Steﬀen, A. Wallraﬀ, Coplanar waveguide resonators
for circuit quantum electrodynamics, Journal of Applied Physics 104 (11)
 Y. Tian, H. F. Yu, H. Deng, G. M. Xue, D. T. Liu, Y. F. Ren, G. H.
Chen, D. N. Zheng, X. N. Jing, L. Lu, S. P. Zhao, S. Han, A cryogen-free
dilution refrigerator based josephson qubit measurement system, Review
of Scientiﬁc Instruments 83 (3) (2012) 033907.
 G. Batey, M. Buehler, M. Cuthbert, T. Foster, A. J. Matthews, G. Teleberg,
A. Twin, Integration of superconducting magnets with cryogen-free dilution
refrigerator systems, Cryogenics 49 (12) (2009) 727 – 734.
 C. Wood, D. Mistry, L. Li, J. Cunningham, E. Linﬁeld, A. Davies, On-chip
THz generation and detection at milli-kelvin temperatures for the study
of ultrafast phenomena in conﬁned semiconductor systems, in: Infrared,
Millimeter, and Terahertz Waves (IRMMW-THz), 2012 37th International
Conference on, 2012, pp. 1–3.
 F. Mueller, R. N. Schouten, M. Brauns, T. Gang, W. H. Lim, N. S. Lai,
A. S. Dzurak, W. G. van der Wiel, F. A. Zwanenburg, Printed circuit board
metal powder ﬁlters for low electron temperatures, Review of Scientiﬁc
Instruments 84 (4) (2013) 044706.
 A. Kushino, S. Kasai, S. Kohjiro, S. Shiki, M. Ohkubo, Development of su-
perconducting coaxial cables for cryogenic detectors, Journal of Low Tem-
perature Physics 151 (2008) 650 – 654.
 A. B. Zorin, The thermocoax cable as the microwave frequency ﬁlter for
single electron circuits, Rev. Sci. Instrum. 66 (1995) 4296 – 4300.
 S. Gustavsson, O. Zwier, J. Bylander, F. Yan, F. Yoshihara, Y. Nakamura,
T. P. Orlando, W. D. Oliver, Improving quantum gate ﬁdelities by using a
qubit to measure microwave pulse distortions, Phys. Rev. Lett. 110 (2013)
 R. Franz, G. Wiedemann, Ueber die w¨arme-leitungsf¨ahigkeit der metalle,
Annalen der Physik 165 (1853) 497 – 531.
 See, for example: http://www.micro-coax.com.
 See, for example: http://www.xmacorp.com.
 To attempt to quantify this we have attached a nuclear orientation source
to just the inner conductor of a coax line installed with bulkhead connectors
at each thermal stage (with no additional in-line attenuators). It is rather
diﬃcult to guarantee correct alignment of the source and detector, and to
attain good thermal contact between the source and the coax core in this
arrangement, but this test did put an upper limit on the temperature of
the core of 23 mK with a system base temperature of 10 mK.
 Anritsu: http://www.anritsu.com.
 See, for example: http://www.coax.co.jp.
 M. Kr¨uger, A. Middendorf, I. Ndip, N. Nissen, H. Reichl, Measurement and
analysis of the impact of micrometer scale cracks on the rf performance and
reliability of transmission lines, in: Electronic Components and Technology
Conference, 2009. ECTC 2009. 59th, 2009, pp. 1277–1283.
 Oxford Instruments: http://www.oxford-instruments.com.
 See, for example: http://www.lakeshore.com.
 H. Marshak, Nuclear orientation thermometry, Journal of Research of the
National Bureau of Standards 88 (3) (1983) 175 – 217.
 A. Kushino, M. Ohkubo, K. Fujioka, Thermal conduction measurement of
miniature coaxial cables between 0.3 and 4.5 K for the wiring of supercon-
ducting detectors, Cryogenics 45 (2005) 637 – 640.
 F. Pobell, Matter and Methods at Low Temperatures, 3rd Edition,
 C. Wang, P. E. Giﬀord, Development of 4 K pulse tube cryorefrigerators
at Cryomech, AIP Conference Proceedings 613 (2002) 641 – 648.
 F. R. Ong, J.-L. Orgiazzi, A. de Waard, G. Frossati, A. Lupascu, Insertable
system for fast turnaround time microwave experiments in a dilution re-
frigerator, Review of Scientiﬁc Instruments 83 (9) (2012) 093904.
 See, for example: http://www.hubersuhner.com.
 See, for example: http://www.home.dynawave.com.
 See, for example: http://www.molex.com.
 See, for example: http://www.radiall.com.
 See, for example: http://www.omnetics.com.
 M. R. Connolly, K. L. Chiu, S. P. Giblin, M. Kataoka, J. D. Fletcher,
C. Chua, J. P. Griﬃths, G. A. C. Jones, V. I. Fal’ko, C. G. Smith, T. J.
B. M. Janssen, Gigahertz quantized charge pumping in graphene quantum
dots, Nature Nanotechnology 8 (2013) 417 – 420.
 K. Rasmussen, C. Olsen, Measuring fractional quantum Hall eﬀect, Mas-
ter’s thesis, University of Copenhagen, Niels Bohr Institute, Center for
Quantum Devices (2013).
 S. Das Sarma, M. Freedman, C. Nayak, Topologically protected qubits from
a possible non-abelian fractional quantum hall state, Phys. Rev. Lett. 94
-A rapid sample exchange mechanism for experiments at millikelvin temperatures has been developed.
-This mechanism is compatible with multiple high-frequency microwave links to the sample.
-Samples can be exchanged in minutes and cooled to temperatures ∼ 10 mK in ∼ 7 hours.