Trapped Ion Quantum Information Novel Atom Sources and Ultra-fast Electronic Switches for Trapped-Ion Quantum-Information Experiments Master Thesis Author: Roland Habl¨ utzel Supervisor: Dr. Joseba Alonso Prof. Dr. Jonathan Home Trapped Ion Quantum Information group (TIQI) Institut f¨ ur Quantenelektronik ETH Z¨ urich
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Trapped Ion Quantum Information
Novel Atom Sources and Ultra-fast ElectronicSwitches for Trapped-Ion Quantum-Information
Experiments
Master Thesis
Author:
Roland Hablutzel
Supervisor:
Dr. Joseba Alonso
Prof. Dr. Jonathan Home
Trapped Ion Quantum Information group (TIQI)
Institut fur Quantenelektronik
ETH Zurich
Abstract
The TIQI group is preparing two Paul traps: a segmented linear one at room temperature and a
planar one in a cryogenic environment.
The first part of this master thesis is about atom ovens, which are used to load the ion traps with
atoms. Two heating methods are presented: the electric one and the novel laser one. While the further
is widely used in the ion-trap community and will be set up in the linear trap, the latter showed to be
a better option for the cryogenic environment due to its lower power dissipation: for the calcium ovens
this power is reduced in a factor 13.
The second part of this work deals with “ultra-fast” electronic switches immersed in liquid helium.
These electronic components are able to switch between voltages in the range 0-10 V in less than 5
ns at 4 K. The ultra-fast transport allows the ion to be displaced in the trap faster than one secular
oscillation of it. This new technique can be used also for motional-state squeezing and entanglement
generation. Switches based on GaAs field-effect transistors (FETs) are known to work at 4 K, while
integrated circuits (ICs) using the CMOS technology are preferred because of its low static-power
consumption and high noise immunity, but there are few references about this type of ICs working at
such low temperatures. The selected switch to be incorporated to the cryogenic environment resulted
to be a digital CMOS with rise and fall time of approximately 4 ns in each case within the mentioned
voltage range.
Keywords: Quantum information; Ion trap; Laser-heated atom ovens; Ultra-fast switches
i
Acknowledgements
I would like to thank Prof. Dr. Jonathan Home for giving me the opportunity to let me do my master
thesis in the TIQI group. I would like to thank also Dr. Joseba Alonso for his time and dedication along
my learning of the experimental techniques in the group and for his advices in the personal ambit. It
is for me an honour to had worked with them, Dr. Home and Dr. Alonso, for sharing with me their
experiences in the research of the quantum information in the ion-trap community, opening me the doors
of this field around the world.
I am glad of all the help that I received from the members of the TIQI group, and of all the people
that made possible this achievement: my master work.
Table 11: Detailed voltage configuration used for testing the switches. Each square ($) control signal
(C1 and C2) oscillated at a frequency of 100 kHz and was driven by a SRS DS345 connected to the switch
and to the oscilloscope. The ±9 V and 4.5 V potentials were driven by batteries, while the 5 V by a
voltage source. The SPDT GaAs switches were driven by complementary control signals using a TTL
NOT gate (denoted by an over-line). The symbol “–” stands for an open terminal.
in liquid nitrogen (according the rise and fall times on the data sheet), but some of them did not respond
to the control signal when cooled with liquid helium; these were tested again at room temperature and
all of them started switching again, meaning that they were not cold-start capable at 4 K (table 12).
The increasing in the switching time of the tested analogue CMOS as the temperature is lowered can
be explained by the decrease in free charge carriers in the semiconductors, while the faster switching in
the tested digital ones as the temperature goes down could be a result from the reduction in the load
impedance of the electric components in cryogenics.
Once the CMOS switches were proven to work, the GaAs candidates were ruled out, even though
46
Oscillosope
Voltage generator
Signal generator
Battery
LN2 Dewar
Figure 43: Cryogenic functionality-test set-up. A hollow tube with the test board and part of the cable
tree in its inner was inserted in the dewar. Devices and measurement configuration are described in table
11.
their switching speed is not supposed to be hindered by low temperatures. The reason is that in these
ICs there is internal circuitry other than that purely dedicated to switching. It is theoretically possible to
overcome this problem by adapting the control pulses to the voltages which one wants to switch, but that
poses many experimental complications. However, minimal tests were carried out with these switches,
although the used NOT gate showed to be too slow for our purpose because it introduced a high delay
between the complementary control signals, thus increasing the rise time.
The ultra-fast switching demands an IC with the shortest rise time (the fall time is supposed to be
nearly the same in each case). The CMOS switch CD 74HC 4066M from Texas Instruments will be
used in the cryogenic trap, since it showed the fastest switching for which both integrated SPST switches
worked. As required, the rise time lies below 10 ns. The rising pulse shape of the first SPST switch
(V1/C1) from this IC is depicted in figure 45. Note that in this picture the rise time of the control signal
can be estimated to be smaller than the one of the switch, i.e., this signal is “slower” than the switching
itself, what entails that the switch reacts only to a determined threshold voltage and does not depend on
the shape of the signal used to control it.
47
0 5 0 1 0 0 1 5 0 2 0 0 2 5 0 3 0 03
4
5
6
7
8
9
1 0 # 3 V 1 / C 1 # 3 V 2 / C 2 # 2 V 1 / C 1 # 2 V 2 / C 2 # 1 V 1 / C 1 # 1 V 2 / C 2
Rise t
ime [
ns]
T e m p e r a t u r e [ K ]
(a) analogue/digital CMOS
0 5 0 1 0 0 1 5 0 2 0 0 2 5 0 3 0 00
2 04 06 08 0
1 0 01 2 01 4 01 6 01 8 02 0 0 # 1 V 1 / C 1
# 1 V 2 / C 1 # 2 V 1 / C 1 # 2 V 2 / C 1 # 3 V 1 / C 1 # 3 V 2 / C 1
Rise t
ime [
ns]
T e m p e r a t u r e [ K ]
(b) analogue CMOS
Figure 44: Rise times of the SPST CMOS switches in dependence on the temperature. The labelling in
the legend is according to tables 10 and 11.
CMOS trise (s): V1/C1 trise (s): V2/C2
74VHC 4066 M 4 -
74HC 4066 D 4.7 4.7
CD 74HC 4066M 4.6 4.6
DG 413 DY - -
ADG 413 BRZ 10.6 10.6
DG 413 HSDY 180 141
GaAs trise (s): V1/C1C2 trise (s): V2/C1C2
MASWSS0179 x x
AS179-92LF x 7.4
Table 12: Rise times of the tested switches. Some switches were unable to cold-start when submerged
in liquid helium (denoted by “-”). Only minimal tests were carried out on GaAs switches and their
measurements were discarded (x), since their performance does not suit our future experimental needs.
8.3 Characterization of the selected Switch
As described in the previous section, the CMOS IC with the fastest switching, the CD 74HC 4066M from
Texas Instruments, was chosen for the cryogenic set-up. This switch was characterized in detail to study
its behaviour in liquid helium. Characterizing a switch refers to measure its power dissipation, as well as
the amplitude, rise and fall time of the output signal and the introduced cable-length noise. The set-up
48
Figure 45: Example of a typical rising pulse. CD 74HC 4066M from Texas Instruments at 4 K.
for the characterization of this electronic switch is represented in figure 46.
As already described in previous sections, the control signals C1 and C2 are complementary, i.e., one
is inverted with respect to the other. Since the two SPST switches are in series (i.e., two input voltages
V1 and V2, and common output), the switch is strictly speaking a SPDT switch (figure 38). The signal
generators are coupled by a signal of frequency 1.25 MHz, being this also the frequency of C1 and C2.
This frequency also corresponds to the one that will be used in the experiments on ion transport in the
cryogenic trap.
The DC voltage generators are used in order to scan the behaviour of the switch over different values
for the switched voltages V1 and V2. These input voltages were chosen in the set of integer values 0 V,
... , 7 V.
Since the voltage at the output of the switch is always defined by one of the two input voltages, the
resistance in parallel at the oscilloscope was removed (it was introduced in the last section, because of a
high-impedance state in the switch).
AC coupling
There was a coupling of the switched voltages (appendix E) in the AC driven mode: the lower voltage
was limited by the higher one. In order to be able to measure the behaviour of the switch at higher
amplitudes, the lower potential was connected either directly (0 V) or over a potentiometer (between 0
V and the minimal coupling voltage) to ground.
8.3.1 Power Dissipation
The power dissipated is measured in order to estimate the repercussion of the circuit on the environment.
This power was found by measuring the introduced (DC or time-averaged) currents and the respective
voltages as depicted in figure 47.
49
4 K / 77 K
Switch
V1 V2
Cou
ple
dcl
ock
s
Oscilloscope
C1
C2
V+
Figure 46: Set-up for the characterization of the electronic switch. The complementary control signals
are generated by coupled signal generators and the DC voltages by their corresponding generators. The
output can be read at the oscilloscope.
V1
C1 =ON
V2
C2 =OFF
Oscilloscope
1 MΩ
15 pF
I2
I1
IC2
IC1
I+
V+
Figure 47: Schema of the current-measurement set-up to estimate the power dissipation. The colours
correspond to the ones of the wires in the pin board (figure 41b).
From the precision of the measuring devices it can be concluded that the currents flowing through the
switch were less than 10−6 A in each case in static mode (C1 and C2 constant signals). The switch will
50
stay also the most of the time in this state when unused.
- 8 - 6 - 4 - 2 0 2 4 6 88 . 1
8 . 2
8 . 3
0 1 2 3 4 5 6 7
Pstat
ic [mW
]
V 1 - V 2 [ V ]
V m i n
(a) Dissipated static power.
- 8 - 6 - 4 - 2 0 2 4 6 8
1 0
2 0
3 0
0 1 2 3 4 5 6 7
Pcon
t [mW]
V 1 - V 2 [ V ]
V m i n
(b) Total power dissipated by the switch.
Figure 48: Dissipated power in dependence on the switching amplitude. We observe a symmetric be-
haviour around ∆V = 0.
In the AC-driven switching the static power consumption Pstatic = I+V+ is plotted in figure 48a in
dependence on the switching amplitude |V2 − V1| for many Vmin = min (V1, V2). Since the measurements
were carried out with the output cable connected to the oscilloscope, despite the fact that in the final
configuration by the trap the output will be an electrode, the current flowing through this cable was not
measured, and, moreover, this and other currents are fictitious (as stated below); then, the static power
consumed by the switch will be its dominating dissipation mechanism in the final design.
It can be seen that this power increases for strictly positive and increasing Vmin. An increase in the
static power is also observed as the switching amplitude is increased. From the symmetry of this plot it
can be stated that this power depends only on the absolute value of the amplitude |V1 − V2|. It can be
estimated to be between 8.1 mW and 8.4 mW.
The continuous power consumed by the switch is found by adding the absolute values of the powers
dissipated by each input Pcont = |I+V+| + |IC1VC1 | + |I1V1| + |IC2VC2 | + |I2V2|. Remember that some
of these currents are fictitious, as they originate due to the capacitance of the oscilloscope, in the same
manner as explained in appendix E.
The continuous power consumption is plotted in figure 48b in dependence on the switched amplitude
for many Vmin. This power shows to be symmetric around ∆V = 0 and increases for higher voltages
amplitudes and voltage minimum. The maximum power up to ∆V = 7 V can be estimated to be 30 mW.
51
8.3.2 Switching Behaviour
The behaviour of the switching describes the quality of the pulse in two terms: the rise and fall time, and
the output amplitude ∆Vsw = V1,sw − V2,sw in dependence on the input amplitude ∆V = V1 − V2 (set up
at the signal generators).
- 8 - 6 - 4 - 2 0 2 4 6 8
- 0 . 0 2
- 0 . 0 1
0 . 0 0
0 . 0 1
0 . 0 2
0 . 0 3
% D i f f
% Dif
f [-]
V 1 - V 2 [ V ]
Figure 49: Switching-amplitude deviation in dependence on the switching amplitude. This deviation is
lower than 3 % in all the cases.
The percentage deviation of the output amplitude ∆Vsw−∆V∆V was measured with respect to ∆V and is
plotted in figure 49. This deviation can be bounded by∣∣∆Vsw−∆V
∆V
∣∣ ≤ 0.03 ≡ 3 %.
The rise time and fall time of the switch pulses is plotted in dependence on the amplitude of the
switched voltages ∆V = V1 − V2 (figure 50). Both times are below 5 ns for |∆V | > 2 V. In the ion-trap
experiments this does not represent any problem, since the goal is to switch large voltage amplitudes.
-8 -6 -4 -2 0 2 4 6 8
4
5
6
7
8
9
10 0123456
tris
e [n
s]
V1 - V2 [V]
Vmin
(a) Rise time
-8 -6 -4 -2 0 2 4 6 8
2
3
0123456
tfall
[ns]
V1 - V2 [V]
Vmin
(b) Fall time
Figure 50: Rise and fall time in dependence on the input switch amplitude.
52
Based on one of the goals of this project, the pulse shapes were measured for a switching amplitude
of 10 V and represented in figure 51.
(a) Rising pulse for V1 = 10 V, V2 = 0 V (b) Rising pulse for V1 = 0 V, V2 = 10 V
Figure 51: Rising pulse of the chosen CMOS IC switch between 0 V and 10 V.
8.3.3 Introduced Cable-Length Noise
Another interesting feature to study is the influence of the cable which connects the output from the
switches to the oscilloscope. In the future use of this switch, the output will be directly connected to
the electrode of the trap, which can be pictured as a small capacitance (approx. 1 pF) to ground. The
mere fact of having a cable at the output and of connecting it to the scope dramatically increases the
capacitance. The effect of this increase is that high-frequency components are filtered partially (washed
out), leading to slower switching times, but also possibly to glitches, noise and ripples in the pulses.
Therefore, the shape of the output pulses was recorded while continuously switching between 0 and 7
V (1.25 MHz) for the cable lengths of 0.5, 1, 5 and 10 m. In addition to this, the approx. 1.6 m coaxial
cable for cryogenics (table 13) has to be considered also into account.
There seems to be no major dependence on the cable length, which suggests that the capacitance
of the oscilloscope (15 pF) and that of the cryocoaxial cable (approx. 50 pF @ 50 Ω) already limit the
behaviour of the switch. Indeed, 50 Ω × 50 pF gives a time constant of 2.5 ns, which is in the range of
what it is measured.
An asymmetry between the right and the left case in table 13 can be observed. This is a result of the
internal circuitry of each of the two CMOS SPST switch.
53
V1 = 7 V, V2 = 0 V1 = 0 V, V2 = 7
0.5 m
1 m
5 m
10 m
Table 13: Pulse shapes for different BNC cable lengths.
54
Part IV
Conclusions
9 Summary and Outlook
The greatest success from this work is that the results obtained both regarding atomic ovens and ultra-fast
switches will be implemented into the trap set-ups at the TIQI group within the next months. The main
achievements are:
The electric-heated atom ovens worked properly in the collimator for the room-temperature trap.
Finished ovens were built and kept in an argon atmosphere in order to be used in the future set-up
trap.
Due to the low power consumption of the laser-heated ovens, they showed to be the best alternative
to be used in the cryogenic trap.
The CMOS switch CD 74HC 4066M from Texas Instruments was chosen for the ultra-fast switching
in the cryogenic environment. Its rise and fall time are below 5 ns between 0 V and 10 V. The
switch can be started after it is inserted in the cryostat (cold-start capable).
There are other (less critical but still important) achievements which have been attained in the scope
of this work:
It was shown that shielding a calcium oven with a stainless-steel tube does not increase its temper-
ature (as function of the power dissipated) with respect to the single one.
The beryllium coil winder was improved such that the beryllium wire hangs down freely without
turning the entire device ahead. The coiling process runs automatically.
The PCB of the collimator was improved in order to connect the ovens to the current sources by
screws.
Several switching candidates have been tested and more than one have been proved to work at 77
K and 4 K.
A few interesting experiments remain, which can be carried out based on this work, like estimating
the amount of calcium and beryllium obtained from the ovens, e.g. by measuring the weight of the glass
plates before and after the runs. But especially important steps to take before setting up the final trap
experiments are:
The mount of the atom ovens in the cryogenic set-up needs to be designed. The laser will be
conducted to the oven via an optical fibre. Pinholes can be used in order to collimate the atom gas
56
from the laser-heated sources. The low power of the laser beam should not represent a big problem,
a simple mechanism in order to dissipate its produced heat to the outside can be installed without
worsen the cold environment inside the cryostat. The sources emit atom gas in both front and back
directions, so the scattering of the laser light can be even absorbed in the walls of the mount.
The PCB for the selected switch in the cryogenic set-up needs to be designed. Essentially low pass
filters will be placed in front of the inputs of the switch and the output will be connected directly
to the electrode of the trap.
57
Part V
Appendix
A Design and Construction of a Base for a Turbo-Molecular Pump
(a) Base for TMP. Screw holes for the Turbopump (centre),
hand pulls (opposite sides) and for the feet (corners) are visible.
Roland Hablützel 10.10.2011
Designer DatumGruppeKontakt
Prof. J. Home
Auftrag
TMP holder
Material Anzahl Einheit
±
BaseTMP.idw
Toleranz
Dateiname
Aluminium 1 mm 1%
J. Alonso (32329)
No
Expressauftrag?
1.0
Version
M
4
x
0
.
7
-
6
H
M6x1 - 6H
12,00
R
4
3
,
0
0
M5x0.8 - 6H
20,00
4
3
,
0
0
°
3
3
,
0
0
°
6,50
270,00
15,00
270,0015,00 15,00
15,00
80,00 140,00
80,00
20,00
260,00
20,00
15,00
(b) Projections of the base on the three main planes, including dimensions.
Figure 52: Inventor drawings of the base for the Turbo-Molecular Pump.
An aluminium base was designed, constructed and adapted to the bottom of the TMP (figure 52).
The need for such a base arose from the fact that the electric-heating set-up (figure 8) was a vertical
structure which required to be stabilized. If the TMP turns and falls while it is on, the rotor at a velocity
of 90000 rpm could break off and cause serious injuries.
The hand pulls were purchased at the D-PHYS-Shop and the rubber (neoprene) feet at Distrelec.
With this design, the tube connecting the TMP with the pre pump goes below one of the hand pulls.
59
B Instructions to Build Current-Heated Atom Ovens
The steps followed to build the calcium (sec. 4.1.1) and beryllium (sec. 4.1.2) atom ovens for the electric
heating are described in this section.
B.1 Calcium Oven
One oven was built for each steel-tube size (table 14).
Materials and tools:
Calcium granulate: -16 mesh, 99.5 %
Stainless-steel tubes
Tantalum wire: 0.5 mm, annealed, 99.95 %
Acetone and isopropanol
Chromel and alumel wires (0.25 mm)
Sharp and (needle-nose) stork pliers
File with a sharp edge
Spot-weld machine
Crimps EK-SUBD-F-CLG10, 15 A, 1 mm, copper alloy - gold plated (BarUvac/VACOM)
Crimping tool
We worked with three kinds of tubes with different inner (ID) and outer (OD) diameters (table 14).
All were purchased from McMaster-Carr.
Gauge ID (mm) ID (in) OD (mm) OD (in)
18XT 1.07 0.042 1.27 0.050
17XT 1.27 0.050 1.47 0.058
15XTS 1.60 0.063 1.83 0.072
Table 14: 316 stainless-steel-tube sizes. This type of steel is non magnetic both at room temperature and
at 4 K.
60
The resistance at each temperature is given by the geometry of the sample [Cre08]. In the case of a
hollow cylinder: R = ρ Lπ(r2ext−r2int)
, where ρ, L, rext and rint (= 0 for a wire) are the resistivity, the length,
and the external and the internal radius respectively.
These tubes are made of stainless steel, since this material is widely used in vacuum systems. It has
a relatively high resistivity, transforming in an efficient way electric current into heat. It has also a good
heat conductivity, giving the oven a homogeneous temperature. Another advantage from stainless steel
is that it does not become superconducting at 4 K, so it is also suitable for the cryogenic set-up.
We use tantalum wires to connect the oven to the electric-current source, because this is a good
electric conductor, it has a relative low outgassing rate (figure 4), and has a small thermal conductivity
(compared to the stainless steel).
The calcium is from Alfa Aesar and comes in granular form of approx. 1 mm thickness (mesh -16, the
finest available).
Chromel and Alumel are standard materials in order to measure the temperature, which is related to
the voltage difference at the junction between them. The thinnest available were chosen, since thicker
wires broke off repeatedly after being spot-welded.
(a) Tube is cut. (b) Hole is filed in the
middle and the ends are
opened.
(c) Tantalum wire is fixed
at one end.
(d) The tube is filled with
calcium and the other end
is closed holding a tanta-
lum wire.
Figure 53: Calcium-oven construction.
Clean the tools with acetone and isopropanol. One may use commercial towels or paper on items
that will not go inside the vacuum5.
Cut gently the tube with sharp pliers in order to get a piece with a length of approx. 1 - 1.5
cm (figure 53a). The thicker the tube is, the longer it should be. This length pattern makes the
diameter difference more visible and has the advantage, that a longer (and thus thicker) oven will
need approximately the same current as a smaller one in order to be heated to the same temperature
(due to the homogeneity of the resistivity). Be aware that this piece will spring away when cut;
therefore, cover the biggest space angle around it with the hand in order to prevent injuries. Note
5For the pieces that are going inside the vacuum use optical paper, as the commercial one could leave shavings.
61
also that if the ovens are too short, one will not get enough calcium inside for the experiment. If
they are too long, calcium will be wasted.
Open both ends with stork pliers, since they will be squeezed after the cut.
Cut the tantalum wire in pieces of approx. 90 mm.
File the tubes at their middle with a (triangular) file (figure 53b). Be sure that the hole is not
too small, otherwise one will not get enough calcium gas, and not too big, such that the calcium
granules cannot escape. Clean from the inside with one of the tantalum wires the zone around the
hole in order to remove shavings. An opening width of approx. 0.5 mm is ideal.
All the components are cleaned with acetone (remove organic impurities) and then isopropanol (to
clean the acetone before it dries) in an ultrasonic bath approx. 3 minutes each. Shake ovens in
order to take out possible trapped drops from the chemicals used.
A tantalum wire is introduced in each oven about 2 mm through an open end, which is then
pressed-closed (figure 53c)6.
Fill the ovens selecting the calcium granules. Do it as fast as possible, since the calcium oxidises.
Start with the smallest ovens, as it is very hard to fill it up, otherwise one could also insert small
granules in bigger ovens, remaining at the end big granules for smaller ovens.
Insert another tantalum wire on the open end of the ovens and press-close (figure 53d).
Spot weld the ends of the ovens with two pulses at 7% and 14% of the maximal energy (230 V)7.
Spot weld near one end (at the opposite side of the hole on the tube) an alumel-chromel junction
in order to be able to measure the temperature.
Insert the ends of each wire (2 from the tantalum wires of the oven, one Chromel and one Alumel)
into the crimps and close them with the respective pliers.
B.2 Beryllium Oven
A beryllium oven is a coil of beryllium-wrapped tungsten. The current flow through the tungsten will
heat up the beryllium.
Materials and tools:
6Do not press too fast or too hard since the lateral ends of the now squeezed hole could break on the edges.7Do not weld with tin, because it melts at temperatures which we are going to reach. Spot welding ensures a better
electric contact between the components.
62
Beryllium wire: 0.05 mm, annealed, 99.7 % purity
Tungsten wire: 0.1 mm, 99.95 % purity
Beryllium coil winder (figure 12)
Voltage source: Blanko DF1730SB3A
Heat gun: WELDY PRO 1800 W from Leister Process Technologies
Small weight (nut M2)
Drill with a 1 mm bit
Sharp pliers
Tungsten wire is used (higher resistivity than tantalum) since beryllium needs more heat in order to
produce an adequate amount of vapour atoms (figure 2). The instructions to wrap the beryllium wire
around the tungsten follows:
Tie the small weight to the beryllium wire (still in the reel)8.
Cut approximately 35 cm of each tungsten and beryllium wire.
Tense the tungsten wire between the two rotors. They rotate at the same velocity and are set in
motion by a motor, which is controlled by a voltage source.
– Pass the wire through the tap A (rotating the wire with the fingers will do it easier) and tie
it to a nut in order to fix the wire. Insert the tap in its respective hole (near the motor) and
screw.
– Pass the wire through the tap B and through a 10 cm long hollow tube. Pass the tube through
the corresponding hole in order to bring the wire to the other side. Take out the tube (holding
the tungsten wire) and then press with one hand the rotor in order to compress the spring
while with the other hand two windings around the outer screw are done. Screw and release
in order to tense the tungsten and then fix tap B.
The screw at tap B is long enough such that part of it will stick out (figure 54). Tie the free end
of the beryllium wire to this screw (a small loop is enough in order to hold the beryllium with the
screw).
8Use the heat gun to get smaller loops, do not pull very hard
63
Figure 54: Loop of the beryllium around the left-out screw.
Figure 55: Weight on the beryllium wire (picture rotated 90).
Add more weight to the nut tied to the beryllium wire by passing a cable through it with three
small metal rings (figure 55). The beryllium wire should hang all the way down, i.e., it should not
pass by any edge9. The mass is built as follows:
– Place two metal rings in the centre of the cable and wind once such that the rings are fixed
– Insert the cable through the hole of the nut with the beryllium and wind once below and once
upon the nut
Tilt the set-up about approx. 1.2 to the left10 by turning the corresponding screw. NOTE: the
coil winder was improved such that it is not necessary to turn it ahead in order the beryllium wire
to hang down (figure 56).
Reset the counter before the motor is started. Start with the lowest voltage needed to drive the
rotor, which is approx. 1.2 V. Heat constantly the winding point of the beryllium wire at the
tungsten. The heat gun was set to the maximum heating (around 300 C) and minimum fan speed.
9If so, you prevent the beryllium wire and the mass from rotate freely as it is coiled, and once the mass passes by the
edge, the wire feels a rapidly torque which will end up breaking it in the weakest point10The maximum angle without inserting something below the base
64
Figure 56: Improved beryllium coil winder. The aluminium plate in the basis was cut and the rotating
set-up was moved forwards with respect to the table.
Speed the motor up until voltage values of 7 V. The optimal value is 5 V, that is, around 7 seconds
per turn.
When the oven is finished, cut the exceeding beryllium, release the screw at the end (that tenses
the tungsten wire), open tap A to relax the tungsten and open tap B and in order to take out the
wire.
Once the beryllium is wrapped around the tungsten wire, wind the beryllium-tungsten wire around a
1 mm drill bit by fixing the exceeding tungsten wire together with the bit into the drill (figure 22a) and
holding the other end with the hand.
In order to calculate the amount of wire needed for an oven, the following calculations were carried
out:
Given the length x and the diameter d of the beryllium wire, and the diameter D of the tungsten
wire, the maximum number (tightest) of possible coils is the length of the wire divided by the length of
one loop x(D+d)π . Then the minimal length L of the beryllium-tungsten wire is the number of wrapped
coils times the thickness of the beryllium wire: L = d·x(D+d)π .
The final wire will be coiled around a cylinder with diameter D′ with a spacing of ∆s (figure 57).
Since the diameter of the beryllium-tungsten wire is D + 2d, the length X of the coiled wire is (after an
analogous calculation):
X ≈ (D + 2d+ ∆s)L
(D′ +D + 2d)π=
(D + 2d+ ∆s) · d · x(D + d)(D′ +D + 2d)π2
The values of the different wire diameters are given: d = 0.05 mm, D = 0.1 mm and D = 1 mm. The
spacing is set to σ = 1 mm and the desired coil length is X = 7 mm. The length of the beryllium wire is
calculated to be x ≈ 207 mm.
65
Be
W
D′
D + 2d
Be
WD
d
∆s
Figure 57: Schema of the transversal cut of the beryllium coiled around the tungsten wire and of the
coiled beryllium-tungsten wire.
66
C Developing and setting up PCBs
Figure 58: Original PCB for the collimator of the room-temperature set-up.
PCB is the acronym for printed circuit board. The manufacturing process is:
Print figure in white paper
Check the scale and print the figure in bond paper
The developing room should have yellow lights only
Heat up 2 minutes the UV lamps of the printing machine
Place the paper bond with the printed side on the photosensitive side of the board and both under
a soft vacuum. Turn on 2 minutes the UV light in order to create the impression of the figure on
the board
Take the board into a developer bath for 1 minute
Clean the board with water and place it into the corresponding acid. The board should be stirred
constantly (this was done by fixing the board to a rotating set-up11). To develop again, clean the
acid and repeat from previous step (developer and acid react, therefore clean always with water)
Once it looks well, clean the board using water and sponge
Drill holes if necessary, using the back side to draw their positions.
My plate did not have any photosensitive layer, so instead of developing it under UV light, scotch
tape was cut in the desired form.
11HINT: place the side to develop towards the rotation axis, as this side will be developed faster. This is so, because the
inner face feels more friction with the acid as the board rotates. The difference in the developing time between the inner and
the outer face is about 30 min.
67
C.1 Impractical Spot Welding
The tantalum and tungsten wires should be fixed to the PCB. For this, the wires were tried to be spot
welded onto it without success (figure 59), since copper is a both good thermal and electrical conductor
and it has also a relatively high melting point (for spot welding a local heating is required in order to
melt the metals at the junction).
Figure 59: PCB after trying to spot-weld on it. The copper is deformed due to the welding.
A possible solution is to spot weld the copper to constatan, and then again to the wires.
A new PCB was developed with holes for screws, such that these screws would fix the wires to it, and
it was the one used in the tests with the collimator of section 3.4.
68
D Resistance in parallel to the Oscilloscope
The oscilloscope is schematically considered as a capacitance and a resistance in parallel. In the func-
tionality tests, the output of the switches was connected to the oscilloscope DPO 2014 from Tektronix,
which has an internal resistance of Rosc = 1 MΩ and capacitance Cosc = 15 pF.
In the CMOS technology, an “OFF” SPST switch represents a high-impedance state and can be
considered as an open circuit (in contrast to a grounded wire). When this switch is turned off, the current
flows internally through the oscilloscope as its capacitance is discharged, in analogy to a RC circuit,
governed by the equation
V (t) ∝ e−t
2·π·R·C ,
thus, the potential seen by this device falls exponentially from the high voltage to ground with time
constant τ = Rosc · Cosc = 11.5 µs.
Hence, when a switching signal frequency of 100 kHz is used, the potential at the oscilloscope drops
only to 1− e−10−5
22·π·11.5·10−6 ≈ 7% at the time, when the switch is turned on again (and the potential at the
oscilloscope returns to its high value). Therefore the measurement of the rising time is experimentally
not possible to perform (figure 60).
Figure 60: Signal (violet) without the discharging resistance drops only about 7 % at the time, when the
switch is turned ON again.
A 1 kΩ resistance (figure 61) was connected in parallel at the input of the oscilloscope, in this way
it became the effective resistance of the new schematic RC circuit. In this case the signal drops a factor
e−10−5
22·π·11.5·10−9 ≈ 10−32 of the high voltage value, from the time the switch is turned off until it is turned
on again. Therefore, the measurement of the rise time can be carried out precisely. By doing this, the
amplitude of the new output signal is smaller, but few tests indicated that the difference with respect to
the prior amplitude can be neglected.
69
Figure 61: Resistance in parallel to the Oscilloscope for its faster discharging.
70
E Coupling of Switched Voltages in AC Mode
A CMOS SPDT switch consists of two independently controlled switches with a common output (figure
40a). Each signal defined the state of each switch (ON or OFF). The signals were coupled such that they
were complementary, i.e., one switch was ON while the other was OFF.
An ON switch establishes a low-impedance line from the input to the output, while in the OFF state
the output is floating (high-impedance state). Since the switches were turned on alternately, the output
was always defined to be the voltage of the switch 1 (V1) or switch 2 (V2), which could be controlled by
voltage generators.
We observed that the higher of the two voltages limited the value of the lower one. This issue is
explained in two steps (figure 62):
There is a capacitance at the output of the switch given by the oscilloscope. Without loss of
generality, we may assume V1 > V2. When switch 1 is ON, the capacitance is charged to V1. At the
point where both switches change their states, the capacitance feels a lower potential, therefore it
discharges and part of the current flows from the output channel to the input.
The voltage source does not allow any current flowing in the negative direction (from the negative
(ground) to the positive output). In such a case the source generates automatically a voltage in
order to counteract the negative current.
The result is that the lower voltage was limited by the higher, reducing the voltage amplitude of the
measurements. In order to be able to measure bigger amplitudes, a potentiometer was introduced in the
set-up (see section 7).
This issue will not be present in the final cryogenic set-up since the DC generators will be able to
accept currents in any direction.
71
V1 V2
C2 =OFFC1 =ON
Oscilloscope
V1
(a) t . 0: the capacitance of the oscilloscope is charged to V1.
V1 V2
C2 =ONC1 =OFF
Oscilloscope
V1
I
(b) t = 0: the capacitance of the oscilloscope discharges since
it feels a lower potential V2.
V1 V2
C2 =ONC1 =OFF
Oscilloscope
V2
(c) t & 0: the capacitance of the oscilloscope is charged to V2.
Figure 62: Description of the voltage coupling due to the oscilloscope capacitance.
72
F Precautions working with some Materials
F.1 Calcium
The calcium granules were stored in a closed glass jar. The calcium ingot was closed in a can with a
nitrogen atmosphere.
Contact with calcium produces skin and eyes irritation, as well as respiratory problems when inhaled.
Since we always wore gloves, there was no direct contact the calcium. We avoided the contact of calcium
with water (and in general, with any liquid) due to the risk of generation of hydrogen otherwise.
Some considerations should be taken into account when cutting the calcium (ingot):
When cutting a metal, it can become hot. Do not cool down with water or other possibly reactive
liquid.
We cut it with a saw (the pliers failed, because the ingot was too thick), which became dirty, because
calcium stayed in between the tooth, as well as in the file, when the planar face was filled.
The calcium block was cleaned with acetone and isopropanol.
Abstract of MSDS
Hazard statements [Calcium MSDS, Sciencelab.com]:
Irritating to eyes and skin
Ingestion or inhalation of dust will produce irritation to gastro-intestinal or respiratory tract
Prevention:
Keep under inert atmosphere and dry container
Never add water to this product
Do not breathe dust
Wear suitable protective clothing
Avoid contact with skin and eyes
Keep away from incompatibles such as acid
73
F.2 Beryllium
Beryllium is very poisonous. Thus, we took more safety measures when handling beryllium than by the
construction of the calcium oven. This metal causes serious injuries when it gets in contact with the skin
or the eyes and could be fatal if inhaled for long periods of time. This material also has to be disposed
in a regulated way.
I used gloves and mask with a P3 filter when handling beryllium, since we manipulate and heated it.
The heat gun was never pointing to a person or their clothes for long periods of time when heating the
beryllium wire, since the increase in temperature implies a higher vapour expulsion.
Abstract of MSDS
Hazard statements [MATERIAL SAFETY DATA SHEET, ELECTRONIC SPACE PRODUCTS INTER-
NATIONAL]:
Toxic if swallowed
Causes skin irritation and serious eye irritations
May cause an allergic skin reaction
May cause respiratory irritation, cancer and death if inhaled
Causes damage to organs through prolonged or repeated exposure
Very toxic to aquatic life
Prevention:
Obtain special instructions before use
Wear protective gloves
Wear eye or face protection
Wear respiratory protection
Avoid release to the environment
Do not breathe dust
74
F.3 Stainless Steel
Abstract of MSDS
Hazard statements [MATERIAL SAFETY DATA SHEET, ThyssenKrupp Materuials NA, Inc.]:
Dust or fume (from welding) may cause irritation to the eyes, nose, or throat and may leave a
metallic taste in the mouth
Prevention:
To avoid contact use appropriate protective gloves or clothing to protect against cutting edges
F.4 Acetone
Acetone was used in order to clean organic impurities in the components, since such rests deteriorate the
quality of the vacuum.
Abstract of MSDS
Hazard statements [SICHERHEITSDATENBLATT, MERCK]:
Hazardous in case of skin contact (irritant), of eye contact (irritant), of ingestion, of inhalation
Slightly hazardous in case of skin contact (permeator)
Prevention:
Provide exhaust ventilation or other engineering controls to keep the airborne concentrations of
vapours below their respective threshold limit value
Splash goggles
Lab coat
Vapour respirator
Gloves
F.5 Isopropanol
This was used in order to clean the residual acetone.