Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Powder filters for a dilution fridge scanning tunneling microscope Quentin, Damien Charles 2016

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2016_may_quentin_damien.pdf [ 70.2MB ]
Metadata
JSON: 24-1.0300650.json
JSON-LD: 24-1.0300650-ld.json
RDF/XML (Pretty): 24-1.0300650-rdf.xml
RDF/JSON: 24-1.0300650-rdf.json
Turtle: 24-1.0300650-turtle.txt
N-Triples: 24-1.0300650-rdf-ntriples.txt
Original Record: 24-1.0300650-source.json
Full Text
24-1.0300650-fulltext.txt
Citation
24-1.0300650.ris

Full Text

Powder Filters for a Dilution FridgeScanning Tunneling MicroscopebyDamien Charles QuentinB.A.Sc., The University of British Columbia, 2012A THESIS SUBMITTED IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFMASTER OF APPLIED SCIENCEinThe Faculty of Graduate and Postdoctoral Studies(Physics)THE UNIVERSITY OF BRITISH COLUMBIA(Vancouver)March 2016c© Damien Charles Quentin 2016AbstractIn this thesis I present the design, construction, and characterization ofmetal powder microwave filters for a dilution refrigeration scanning tunnel-ing microscope (STM) in the Laboratory for Atomic Imaging Research atthe University of British Columbia. Scanning tunneling spectroscopy (STS)measurements performed by the STM are able to reveal features in the localdensity of states with energy resolution in the µeV regime if the sampleand tunnel junction are cooled below 100 mK. The filters described in thiswork eliminate thermal noise and electromagnetic interference, which de-crease energy resolution in STS measurements, up to seemingly indefinitefrequency by exploiting the tremendous effective surface area of the metalpowder which dissipates radio-frequency power via eddy currents induced inthe grains.iiPrefaceThe work presented in this thesis is unpublished, original, and independentlyperformed by the author, D. Quentin.iiiTable of ContentsAbstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iiPreface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iiiTable of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . ivList of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viList of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.1 Scanning Tunneling Microscopy . . . . . . . . . . . . . . . . 11.2 UBC LAIR Facility and Instrumentation . . . . . . . . . . . 21.3 The Problem with Noise . . . . . . . . . . . . . . . . . . . . 51.4 Powder Filters . . . . . . . . . . . . . . . . . . . . . . . . . . 71.5 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82 System Details and Design Considerations . . . . . . . . . . 102.1 UHV Wiring . . . . . . . . . . . . . . . . . . . . . . . . . . . 102.2 Design Considerations . . . . . . . . . . . . . . . . . . . . . . 143 Construction and Assembly . . . . . . . . . . . . . . . . . . . 193.1 Powder Filters . . . . . . . . . . . . . . . . . . . . . . . . . . 193.2 Coarse Approach Filters . . . . . . . . . . . . . . . . . . . . 263.3 Cold Bias Filter . . . . . . . . . . . . . . . . . . . . . . . . . 283.4 Tunneling Filter . . . . . . . . . . . . . . . . . . . . . . . . . 323.5 Filter Housing . . . . . . . . . . . . . . . . . . . . . . . . . . 343.6 Wiring Conduit . . . . . . . . . . . . . . . . . . . . . . . . . 364 Attenuation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 394.1 Measurement Setup . . . . . . . . . . . . . . . . . . . . . . . 394.2 Filter Attenuation Characteristics . . . . . . . . . . . . . . . 40ivTable of Contents5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48vList of Tables2.1 Electrical pinouts for UHV wiring port A . . . . . . . . . . . 122.2 Electrical pinouts for UHV wiring port B . . . . . . . . . . . 132.3 Electrical pinouts for UHV wiring port D . . . . . . . . . . . 13viList of Figures1.1 STM inertial slabs . . . . . . . . . . . . . . . . . . . . . . . . 21.2 Nano-G pod . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31.3 Dilution refrigerator insert . . . . . . . . . . . . . . . . . . . . 41.4 Dewar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51.5 Johnson noise circuit . . . . . . . . . . . . . . . . . . . . . . . 61.6 LC-Π circuit diagram . . . . . . . . . . . . . . . . . . . . . . 82.1 UHV wiring diagram . . . . . . . . . . . . . . . . . . . . . . . 112.2 Shear piezo diagram . . . . . . . . . . . . . . . . . . . . . . . 152.3 Sample coarse approach piezo actuator . . . . . . . . . . . . . 152.4 XYZ scanning tube . . . . . . . . . . . . . . . . . . . . . . . . 163.1 De-gassing the epoxy . . . . . . . . . . . . . . . . . . . . . . . 193.2 Use a vacuum pump connected to a vinyl hose to suck theepoxy mixture into a teflon tube. . . . . . . . . . . . . . . . . 203.3 Cure the inductor cores using teflon tubes as molds. . . . . . 203.4 Epoxy core extraction . . . . . . . . . . . . . . . . . . . . . . 213.5 3 cm-long inductor cores. . . . . . . . . . . . . . . . . . . . . 213.6 Winding jig for the inductor cores. . . . . . . . . . . . . . . . 223.7 Inductor with pins installed and one end soldered. . . . . . . 233.8 Feed-through capacitor used in the powder filters . . . . . . . 243.9 First capacitor soldered to inductor . . . . . . . . . . . . . . . 243.10 Open assembly with a single capacitor . . . . . . . . . . . . . 253.11 Settling the powder to the bottom of the inside of the assem-bly with a Dremel . . . . . . . . . . . . . . . . . . . . . . . . 253.12 A fully assembled powder filter . . . . . . . . . . . . . . . . . 263.13 Filter cross section . . . . . . . . . . . . . . . . . . . . . . . . 263.14 Inductor coil used in the coarse approach filters . . . . . . . . 273.15 Teflon endpiece on coarse approach filter . . . . . . . . . . . . 283.16 A fully assembled coarse approach filter . . . . . . . . . . . . 283.17 OFHC material in a lathe . . . . . . . . . . . . . . . . . . . . 293.18 OFHC plate to mount filter to STM head . . . . . . . . . . . 29viiList of Figures3.19 Gold-plated filter tube and mounting plate . . . . . . . . . . 303.20 Jig for curing the epoxy/powder potting mixture in the coldbias filter assembly . . . . . . . . . . . . . . . . . . . . . . . . 313.21 Potted filter tube with inductor. . . . . . . . . . . . . . . . . 313.22 Completed head mount filter . . . . . . . . . . . . . . . . . . 323.23 Instec capacitor . . . . . . . . . . . . . . . . . . . . . . . . . . 323.24 Tunnelling line filter . . . . . . . . . . . . . . . . . . . . . . . 343.25 Filter housing prior to installing filters, wiring, and connectors 343.26 Housing partition populated with filters . . . . . . . . . . . . 353.27 Partially assembled filter housing . . . . . . . . . . . . . . . . 353.28 Fully populated filter housing . . . . . . . . . . . . . . . . . . 363.29 Connector collar with strain relief (left) and strain relief re-moved (right) . . . . . . . . . . . . . . . . . . . . . . . . . . . 373.30 Stainless steel piece to connect the filed collars to the bellows 373.31 Conduit connector, collar, adapter, and bellows epoxied to-gether . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 383.32 Wiring conduit . . . . . . . . . . . . . . . . . . . . . . . . . . 384.1 Measurement housing . . . . . . . . . . . . . . . . . . . . . . 394.2 Housing connected to vector network analyzer . . . . . . . . . 404.3 Attenuation for stainless steel vs. iron powder LC-Π filters . 414.4 Tunnelling line filter attenuation . . . . . . . . . . . . . . . . 424.5 Attenuation for stainless steel vs. iron coarse approach filters 434.6 Attenuation of cold bias filter . . . . . . . . . . . . . . . . . . 444.7 Lukashenko filter attenuation . . . . . . . . . . . . . . . . . . 454.8 H. le Sueur filter attenuation . . . . . . . . . . . . . . . . . . 46viiiChapter 1Introduction1.1 Scanning Tunneling MicroscopyScanning tunneling microscopy (STM) is a tool that can provide atomic to-pographic images and electronic characterization of exciting new materials.The topographic images produced are real-space 2-dimensional images ob-tained by approaching a very fine tip to within a few nanometers of a samplein ultra-high vacuum (UHV), and applying a voltage potential across thejunction. The tip and sample are so close the wavefunctions of the elec-trons in the tip and sample overlap, allowing electrons to tunnel across thejunction when a voltage is applied.Since the tunneling current depends exponentially on gap distance withdecay length h¯(8mφ)−1/2, topographic images can be well-resolved at theatomic scale by two separate techniques: constant-current or constant-heightimaging. In constant-current mode, the feedback control system adjusts thetip-sample separation distance to keep the tunneling current constant asthe tip scans over the surface of the sample. Scanning speed in this mode islimited by the feedback bandwidth. In constant-height mode, the tip-sampleseparation distance is held constant while the current is directly measured.This method is limited to fairly flat sample surfaces to keep the tip fromcrashing into the sample, though since there is no feedback the scanningspeed can be faster.In addition to topographic images, spectroscopic images can be producedusing the same instrumentation with a method called scanning tunnelingspectroscopy (STS), whereby the tip is still scanned over the sample but ateach pixel in the array, the tip is held in position while the bias is swept acrossa range of voltages to extract a tunneling conductance, dI/dV , either bydifferentiation or by adding a small AC modulation. dI/dV is a very usefulquantity to extract since it relates to the local density of states (LDOS),which is useful in exploring many physical phenomena.11.2. UBC LAIR Facility and Instrumentation1.2 UBC LAIR Facility and InstrumentationThe Laboratory for Atomic Imaging Research (LAIR) facility at UBC con-sists of 3 different STM’s housed in separately isolated rooms. Due to theexponential dependence of tunneling current on tip-sample separation, hav-ing an environment with minimal vibrations is essential. The STM roomshave a separate foundation from the rest of the building; they don’t shareany walls, ceilings, or floor space with the rest of the building and are onlycoupled through the earth. This allows for the best possible vibration isola-tion. To further isolate the experiments from vibrations, the microscopes aremounted on massive inertial concrete slabs, which themselves are mountedon pneumatic isolators. The dilution fridge STM, which is the system offocus in this thesis, is mounted on a 70-tonne concrete slab, the largest ofthe three in the LAIR.Figure 1.1: STM inertial slabs (reprinted with permission of Prof. DougBonn) [1]In addition to the slabs being mounted on large pneumatic isolators, thealuminum frames supporting the individual UHV systems are themselvesmounted on smaller pneumatic isolators atop the slabs. Along with acousticdamping pads hanging from the walls, the end result is a low and stabletunneling noise floor, and an eerie silence in the rooms with the doors closedtightly.21.2. UBC LAIR Facility and InstrumentationFigure 1.2: Dilution fridge STM in the Laboratory for Atomic ImagingResearchThe 3He - 4He dilution fridge is built by Janis Research Company, modelJDR-50-UHV-STM. The fridge base temperature is 35 mK. The vector mag-nets are fixed within the liquid helium dewar, which has a belly capacity of65 liters and a boil-off rate of 250-350 cc/hour. The maximum vertical mag-netic field is 7 Tesla at 4.2 K, and the maximum horizontal field is 2 Teslaat 4.2 K. With both coils running simultaneously the maximum field is 1Tesla at 4.2 K.31.2. UBC LAIR Facility and InstrumentationFigure 1.3: Dilution refrigerator (image courtesy of Janis Research Com-pany)41.3. The Problem with NoiseFigure 1.4: Liquid helium dewar for dilution fridge system (image courtesyof Janis Research Systems)1.3 The Problem with NoiseAll wires, to some extent, behave as antennas. A deliberately-designed an-tenna will transmit, and pick up, certain frequencies better than others,but any arrangement of wires will still transmit or receive electromagneticradiation. This pick-up is called electromagnetic interference (EMI), andin general, careful design considerations such as keeping wires short andavoiding loops can greatly reduce this interference and is often sufficient.The dilution fridge STM at the UBC LAIR is located in a low-vibrationfacility, with all instrumentation and control electronics kept in a corridoroutside. As such, all wires leading into the UHV system are 20 to 50 feetlong. Since the wire carrying the tunneling current is connected to an am-plifier with 109 gain, EMI can have a tremendous impact and poses a signif-icant problem. For instance, AC adapters plugged into power outlets nearany wires introduce 60Hz noise peaks several orders of magnitude higherthan the noise floor.In addition to EMI, a significant source of noise is Johnson noise. John-51.3. The Problem with Noiseson noise (sometimes called thermal noise or Nyquist noise) is electronicnoise created by thermal fluctuations of electrons. This noise is present inany dissipative element.Figure 1.5: Equivalent circuit for a resistance at a temperature TThe noise voltage can be expressed as e¯2n = 4kBTR∆f , where kB isBoltzmann’s constant in Joules per Kelvin, T is the temperature in degreesKelvin, R is the resistance in Ohms, and ∆f is the bandwidth in Hertz overwhich the noise is measured. As a rule of thumb to provide some intuition,50 Ω at room temperature produces approximately 1 nV/√Hz. For the vastmajority of electronic systems, this level of noise is not a concern, but in lowtemperature physics experiments where the signals are typically minuscule,Johnson noise can have a relatively large presence and should always beaccounted for. At 300 K, Johnson noise radiation peaks at roughly 6 THz[2].In STM, noise increases electronic temperature. The noisy fluctuationsof the electrons give the illusion the system is at a higher temperature thanit actually is (base temperature). This increase in electronic temperaturecould easily be on the order of a few Kelvin. This can be most easily seenin conductance plots taken by STS, where the thermal broadening causes adecrease in energy resolution. Certain features such as Zeeman or hyperfinesplitting can only be seen in the LDOS at very high energy resolution [3],which is why these experiments are run in the 10’s of milliKelvin regimewith dilution refrigeration systems.Aside from proper shielding and design configuration, the best way tomitigate noise is to install filters. Since the cables and wires are near enoughto each other that they are capacitively coupled, and since the inside of thesteel UHV system behaves to a certain degree like a microwave cavity, it’scrucial to filter each and every wire entering the microscope. Given thatJohnson noise extends into the terahertz regime, special considerations need61.4. Powder Filtersto be made in the design of the filters that allow them to maintain highattenuation at these frequencies.1.4 Powder FiltersThe vast majority of electronic filters used today are active filters. Theseare filters which use an amplifier to keep the filters small, inexpensive, andeasy to manufacture on a large scale. However, amplifiers have very largeresistances, typically on the order of 106 Ω. This poses a problem for sen-sitive low-temperature physics experiments as these filters add considerableJohnson noise. It is therefore necessary to use passive filters (those withoutamplifiers or an external power source) as they can be made with very lowresistance.In 1987, Martinis et al. used metal powder to create a passive filterwhose attenuation increases with frequency seemingly indefinitely [4]. Thefilters consisted of an insulated Manganin coil inside a copper tube filledwith 30 µm grain copper powder. The filter was 0.1 m long, and theymeasured an attenuation of more than 50 dB from 0.5–12 GHz. This filterprovides attenuation via two separate mechanisms. First, it is clear from theimpedance of the inductor, ZL = jωL, that as the frequency ω increases,so does the impedance until eventually it is so high the inductor starts tobehave as an open circuit. However, there is a problem which all inductorshave at very high frequencies. Neighbouring windings are separated by a thinlayer of insulation; this forms a small interwinding capacitance, and beyonda certain frequency the inductor will behave as a short circuit as the higherfrequencies essentially bypass each loop. Thus, at very high frequenciesinductors become ineffective. The Martinis filter’s second mechanism ofattenuation combats this issue, and is what makes this filter special.The effectively massive surface area of the copper powder, whose grainsare insulated by a naturally-grown oxide layer, allows for eddy currents in-duced in the powder to greatly attenuate higher frequencies via the skineffect. The skin effect is the frequency-dependent tendency for current tobe distributed near the edge of a conductor. The AC current density de-cays from the surface exponentially inside the conductor with decay lengthδ =√2ρωµ√√1 + (ρω)2 + ρω, where ρ is the conductor resistivity, ω is theangular frequency of the current, and µ and  are the permeability and per-mittivity of the conductor, respectively. We call δ the skin depth. With thecurrent condensed at the conductor edges, the conductor’s resistance is ef-fectively increased at higher frequencies, allowing greater dissipation of RF71.5. Overviewpower.In 2008, Lukashenko and Ustinov [5] improved upon the design of theMartinis powder filter by adding capacitors to make the filter an LC-Π pow-der filter, so-called because the circuit diagram looks like the Greek letterΠ. LC-Π filters are low-pass filters consisting of a series inductance withcapacitors to ground at both the input and output. They are a convenienttype of filter because they have inherently low impedance, a very flat pass-band response, and their circuit diagram is very simple. Again, their use inpractice is limited due to the cost of implementing inductors.Figure 1.6: Circuit diagram of an LC-Π filterIn addition to introducing capacitors, Lukashenko and Ustinov furtherimproved the Martinis design by using stainless steel powder instead of cop-per powder, since the resistivity of stainless steel is nearly an order of mag-nitude higher than that of copper. As a result, these filters have a cutofffrequency of 1 MHz, a rolloff of -50 dB per decade, and hit the noise floor(-100 dB) around 200 MHz.Where magnetic fields and magnetization are not concerns, stainless steelis a common choice for metal powder. However, if the filters are well awayfrom sensitive systems and magnetic fields, iron powder is sometimes a betteralternative. Due to its high relative magnetic permeability, adding ironpowder to an inductor increases the inductance since inductance is linearlyproportional to permeability.1.5 OverviewIn this thesis I describe the theory, construction, and characterization of anLC-Π metal powder filter system for use in a dilution refrigeration UHVSTM in the Laboratory for Atomic Imaging Research (LAIR) at the Uni-versity of British Columbia (UBC).I begin by discussing the details of the design of the LC-Π powder filters,81.5. Overviewthe housing assemblies, and the conduits. I then discuss special designconsiderations for the different components of the STM system, and how thisaffects the designs of the filtering system. Following that I provide a detailedsummary of the construction and assembly procedure for the different filtersand components in the filtering system. Finally, I characterize performanceand present attenuation data for all filters in the system.9Chapter 2System Details and DesignConsiderationsWhen designing the filtering system it’s crucial to consider the requirementsof each of the lines to be filtered. It’s necessary to know the required band-width, impedances, and capacitances to ensure the filters wouldn’t alter thedata or signals on each of the wires entering the system.2.1 UHV WiringBelow is a wiring diagram for the dilution fridge provided by Janis Re-search Company. The diagram is useful in outlining how and where linesare connected at the various stages of the system. Details on the wiring ofcomponents other than those supplied by Janis are shown in tables in thissection, along with pin assignments for the filter assembly connectors.102.1. UHV WiringFigure 2.1: UHV wiring diagram (image courtesy of Janis Research Systems)There is an adapter on Port A to convert between a 23-pin Delrin sub-C112.1. UHV WiringDevice Port Pin MIL. Pin Bridge PinJanis Thermometer 1: I+ 1 R 3Janis Thermometer 1: I- 2 S 15Janis Thermometer 1: V+ 3 Z 4Janis Thermometer 1: V- 4 Y 16Janis Thermometer 2: I+ 5 T 6Janis Thermometer 2: I- 6 P 18Janis Thermometer 2: V+ 7 G 7Janis Thermometer 2: V- 8 N 19Janis Thermometer 3: I+ 9 U 9Janis Thermometer 3: I- 10 X 21Janis Thermometer 3: V+ 11 V 10Janis Thermometer 3: V- 12 M 22Janis Thermometer 4: I+ 13 H 12Janis Thermometer 4: I- 14 F 24Janis Thermometer 4: V+ 15 A 13Janis Thermometer 4: V- 16 B 25Still Cartridge Heater 17 EStill Cartridge Heater 18 JMC Cartridge Heater 19 LMC Cartridge Heater 20 WNo Pin 21No Pin 22Shutter Sensor 23 CTable 2.1: Electrical pinouts for UHV wiring port Aconnector to a MIL MS3476L14-23S connector. Table 2.1 shows the corre-sponding pins on the MIL connector and the pins on the Lakeshore Resis-tance Bridge.122.1. UHV WiringDevice PinY- Scanner ABias (East) BX- Scanner CX+ Scanner DCapacitive Sensor (Bottom) ERuO2 Sensor FBias (West) GY+ Scanner MRuO2 Sensor NCapacitive Sensor (Inner) PCapacitive Sensor (Top) RRuO2 Sensor SRuO2 Sensor VTable 2.2: Electrical pinouts for UHV wiring port BDevice PinSlip Stick Stack IV- APosition Sensor BSlip Stick Stack VI+ CSlip Stick Stack IV+ DRuO2 Sensor ESlip Stick Stack I- FSlip Stick Stack III- GRuO2 Sensor HSlip Stick Stack V+ JSlip Stick Stack III+ KSlip Stick Stack I+ LSample Input MRuO2 Sensor NSlip Stick Stack VI- PSample Input RSlip Stick Stack II+ SSlip Stick Stack V- TSlip Stick Stack II- UZ Scanner VTable 2.3: Electrical pinouts for UHV wiring port D132.2. Design Considerations2.2 Design ConsiderationsTunneling CurrentThe tunneling signal has a low bandwidth and low current. The onlyspecial consideration to be made in designing the tunneling filter is thelarge impedance of the tunnel junction. The impedance of the tunneljunction is typically on the order of 109 Ω, so it’s important that thefilter impedance to ground be large to avoid losing significant current.Coarse Position ControlThe sample is brought into the STM head by servo-motor control ofa vertical arm which holds the sample and sample plate, but oncethe sample is within about a centimeter of the probe tip, the tip isapproached to the sample by shear stack piezoelectric actuators. Werefer to this as coarse position control, though in this context sincethe actuator resolution is roughly 100 nm, coarse is somewhat of amisnomer—so-called because finer positioning control (less than 0.1nm) is performed by a separate mechanism to be discussed in the nextsection.The coarse approach is performed by (relatively) slowly ramping thevoltage to the piezoelectric stacks and then very rapidly groundingthem. With one side of the stack fixed to the Macor head and theother side simply pressed up against the prism (to which the tip ismounted), shearing the stack has the effect of moving the prism up ordown; however, if the voltage applied to the stack returns to 0 quicklyenough, the stack will slip along the surface of the prism, leaving theprism in its new position. For this reason, this type of actuator issometimes referred to as a stick-slip actuator.142.2. Design ConsiderationsFigure 2.2: The coarse approach piezoelectric actuators shear under an ap-plied bias, then snap back into position to move the STM tip [6].Figure 2.3: The sapphire prism which holds the STM tip is supported bya set of 3 shear stack piezoelectric actuators (image courtesy of RobertDelaney).The shape of the waveform driving the stack of piezoelectric actuatorsis crucial to proper stick-slip action, and if a filter alters this shape toomuch the actuators may not work properly. The max slew rate for thedriver output is 500 MV/s at 2A for 1 µs, 300 times per second. Toavoid altering the shape of the waveform, a filter with a higher cutofffrequency is preferred. Additionally, NiCr wire would be avoided forthis application. NiCr is resistive wire which would lower the cutofffrequency but in addition, the dissipated power would provide roughly1 W of heat. Even though this is a relatively small amount, heatingcan be problematic since iron powder sinters easily. Sintered powderwould form a resistive path to ground within the filter, reducing thevoltage delivered.152.2. Design ConsiderationsFine Position ControlFigure 2.4: The PZT tube allows for lateral and vertical fine position control[6]Fine control of the tip is achieved by a PZT tube with an arrangementof contacts that allows for deflection, elongation, and deformation ofthe tube. Four evenly-spaced contacts on the outside of the tube allowfor lateral control by shearing the tube, and a contact along the insideof the tube allows the entire tube to be elongated and compressed.In order to deform the tube to desired lengths and shapes, the maxi-mum voltage applied to the PZT is 220V. This voltage is high enoughto cause a breakdown of the dielectric in the metal powder grains, sothe wire must be particularly well-insulated.When there is a breakdown of the dielectric and enough current flowsto ground, not only is the piezo tube not able to maintain the desiredposition, but the lost current is picked up by the tunneling amplifierwhich raises the noise floor by orders of magnitude.Cold Bias FilterDue to space constraints, every line into the system is filtered at roomtemperature instead of in the UHV space. However, there is still roomfor some filtering in UHV on the STM head, so a single filter is installedon the bias line.Since the bias line has a high impedance and very low bandwidth,there are no special electronic design constraints; however, making162.2. Design Considerationscomponents for UHV requires special care. Most of the metal used inthe filter should be gold-plated oxygen-free high thermal conductiv-ity (OFHC) copper. OFHC copper is used instead of regular copperbecause of its higher thermal conductivity and low relative volitility(it out-gasses easily), and should be gold-plated to avoid problemswith oxidization. Additionally, to avoid contaminating the vacuum,all internal components should be fully potted in Epo-Tek H77 Epoxy,which is electrically insulating and thermally conductive.ThermometryA number of thermometers are placed throughout the dilution fridgesystem. Temperature measurements are done by 4-wire lock-in mea-surements of RuO2 chips, whose resistance provides an accurate mea-sure of temperature, particularly at low temperatures.Placing an LC-Π filter on these lines adds a capacitance in parallel tothe RuO2 chips, and since the lock-in amplifier drives the measure-ments at 13 Hz, it’s worth considering the impedance of the capacitorto ensure there is no significant current flowing through the capacitor.To provide a reference, consider the case where the impedance of thecapacitor is equal to that of the RuO2 thermometer, which can reacha megaohm when the dilution fridge is running. In this case, equalcurrent would flow through the capacitors as through the thermometerchips. The capacitance required to achieve such an impedance wouldbeC =12pi × 13×106 = 104 pF (2.1)104 pF is a fairly large capacitance, and it is easy to keep the filtercapacitance orders of magnitude below this level.Magnetic FieldThe presence of a strong magnetic field can alter the attenuation char-acteristics of the powder filters by saturating the field within the in-ductors.According to magnetic field profiles provided by the system’s mag-net manufacturer, American Magnetics, Inc., with the superconduct-ing magnets operating at full strength, the room-temperature filtersplaced 1.5–2 m away from the magnets would see a maximum of ap-proximately 100 Gauss. This is a relatively small value, as B-H curves172.2. Design Considerationsof similar iron powder show that 100 Gauss is still well within thelinear regime.18Chapter 3Construction and AssemblyThe following are sets of instructions to make and prepare the filters andassemblies used in this system.3.1 Powder Filters1. Making the inductor cores begins by mixing a batch of Stycast 1266epoxy with -325 mesh iron metal powder, 50/50 by weight. Stycast1266 is a good epoxy to use because it has a fairly low viscosity andhigh resistivity.2. De-gas the epoxy by placing the mixture in a low-pressure chamber.Do this until most of the bubbling has subsided. Placing the epoxy inlow pressure for too long will evaporate some of the epoxy constituentsand will prevent it from curing properly.Figure 3.1: Air violently escaping the epoxy/powder mixture in a low-pressure chamber193.1. Powder Filters3. Use a teflon tube with 1/8” inner diameter connected to a vacuumpump as a mold to cast the epoxy/powder mixture. To contain themixture during the cure, plug the top of the teflon tube with a smallamount of Blu-Tack putty.Figure 3.2: Use a vacuum pump connected to a vinyl hose to suck the epoxymixture into a teflon tube.Figure 3.3: Cure the inductor cores using teflon tubes as molds.4. Teflon is a good material to use because the epoxy doesn’t stick toit, and if the epoxy cores are tough to extract, the tubes will expandwhen heated. The easiest way to extract the cores is to hammer a 1/8”brass rod down the teflon tube and push the cores out the bottom.203.1. Powder FiltersFigure 3.4: Hammer a 1/8” rod through the teflon tubes to extract theinductor cores.Figure 3.5: 3 cm-long inductor cores.5. Once extracted, cut the cores into 3-cm lengths. One by one, placethe cores in a lathe to bore 3-mm-deep holes into the ends using a size55 drill bit. These holes will ultimately hold the inductor pins.6. Use a jig to wind the inductor about the cores. A good wire to use is0.0031” NiCr enameled wire with a resistance of 81.09 Ω/ft manufac-tured by Evanohm. With this wire, each inductor consists of roughly8.5’ of wire for an average resistance around 700 Ω. Once the inductorsare fully wound, apply 5-minute epoxy to the windings to hold themin place.213.1. Powder FiltersFigure 3.6: Winding jig for the inductor cores.7. With the windings fixed in place by epoxy, install the pins on eachinductor. Straighten 2-cm sections of 16 AWG tinned bus wire (thisis very easy to straighten perfectly by clamping one end of the wirein a vise, and then using pliers to pull on the other end until thewire yields and stretches). In order to ensure a tight fit of the pins inthe inductor cores, gently flatten one of the ends of the wire with ahammer to slightly increase its diameter, then coat it with 5-minuteepoxy before finally placing the pins in the inductor core holes.223.1. Powder FiltersFigure 3.7: Inductor with pins installed and one end soldered.8. The next step is to solder the NiCr wire to the inductor pins. Thisstep is quite challenging, and great care must be taken not to meltthe epoxies, which starts to happen around 100 ◦C. Therefore, thesoldering should be done quickly.(a) Strip the enamel from one of the loose wire ends using a razorblade. Be careful not to unwind the inductor or to knick the wire.(b) Ensure a good bond by roughing the pin with sandpaper or arazor to increase surface area.(c) Wrap the free wire around the pin 3 or 4 times to ensure plentyof contact. Increase the strength of the joint by separating con-secutive windings by roughly 0.5 mm. Use a jig to hold the wiretight and apply one or two drops of strong, corrosive acid flux.(d) Apply heat with the soldering iron while simultaneously holdinga small piece of solid-core Pb-Sn solder until it wicks onto the pinwindings.(e) Once all the free wires are soldered to the inductor pins, theassemblies must be thoroughly rinsed and cleaned with warmwater in an ultra-sonic bath to avoid corrosion of the wires bythe acid flux.9. The next step is to solder a feed-through capacitor to one end of theinductors and then to mount the assembly within a tube. The capaci-tor I used is the Tusonix 2482-0120-X5U0-101MLF. It is 100 pF ± 20%with inner diameter 1.22 mm and outer diameter 4.95 ± 0.38 mm. It isvery easy to solder these together as both surfaces are already tinned.233.1. Powder FiltersFigure 3.8: Feed-through capacitor used in the powder filtersFigure 3.9: First capacitor soldered to inductor10. Prepare the filter outer conductor by cutting a 4 cm section of brasstubing, with inner diameter 7/32” and wall thickness of 0.014”. In-sert the inductor/capacitor assembly into the tube, and solder themtogether where the capacitor’s outer contact meets the brass tubing.This capacitor will nest nicely in the tubing if you shave a small amountof brass from the inner wall using a reamer or de-burring tool.11. Once the soldering is complete, all the extra flux has to be washedaway in an unltrasonic bath or the corrosion will cause problems inthe future. Even fairly non-corrosive flux such as rosin flux causesa problem; the fumes from the flux will permeate the powder withinthe filter and strip the oxide layer slowly. After a few months, theresistance parallel to the capacitor will decrease by several orders ofmagnitude, into the kiloohm regime.243.1. Powder FiltersFigure 3.10: Open assembly with a single capacitor12. Next, fill the assembly with -325 mesh iron powder. To completelyfill the space within the assembly, an effective technique is to holdthe assembly upright to fill with powder, and periodically hold theassembly against something that vibrates so all the powder can settleto the bottom of the tube. A Dremel hand drill with a vinyl screwworks quite well without causing any damage.Figure 3.11: Settling the powder to the bottom of the inside of the assemblywith a Dremel13. Since iron powder tends to sinter rather easily and since there wouldbe no way to clean any stray flux inside the filter, the second capaci-tor must not be soldered in place. Use a conductive epoxy that cures253.2. Coarse Approach Filtersat room temperature. Conductive epoxy that cures at higher tem-peratures must be avoided or the powder will sinter when heated. Ifthe powder sinters, the resistance to ground will drop to the kiloohmregime.14. Once the conductive epoxy has cured, all excess solder and epoxy mustbe removed so the outside of the tube is smooth. Otherwise the filterswon’t fit into the filter housing assembly.Figure 3.12: A fully assembled powder filterFigure 3.13: Cross section view of the LC-Π powder filter, to scale. Forclarity, the powder is not shown. Also note that the windings may be toosmall to see.3.2 Coarse Approach FiltersAs discussed in the previous chapter, the filters for the coarse approach piezolines need a higher cutoff frequency than the powder filters. We found thata low cutoff frequency deformed the waveform too much.263.2. Coarse Approach Filters1. To allow for higher currents, the wire used in these inductors is 18AWG enameled copper wire. Since this wire is much more rigid thanthe NiCr wire used in the powder filters in the previous section, theseinductors don’t need to be fixed to an epoxy/powder core. Instead, thewindings are formed around a 1/8” brass rod and maintain their shapewhen slipped off. Once the coils are removed from the rod, carefullybend the free ends to run along the axis of the inductor, as shown inthe figure below.Figure 3.14: Inductor coil used in the coarse approach filters2. Cut a 4 cm-long section of brass tubing with 3/8” outer diameter and5/16” inner diameter to form the filter’s outer conductor.3. Place a section of 5/16” teflon rod in a lathe, and drill a hole severalmillimeters deep along the center with a size 58 drill bit.4. With a razor blade, slice two disks off the rod, roughly 0.5 mm thick.5. Work one of the teflon disks into one of the brass tube ends. Seal itin place with some 5-minute epoxy.6. Insert the inductor in the brass tube and slide the end through thehole in the teflon disk. Use 5-minute epoxy to seal the pin in place.273.3. Cold Bias FilterFigure 3.15: Teflon endpiece on coarse approach filter7. Fill the tube with stainless steel powder.8. Work the second teflon disk in place and seal it with 5-minute epoxyto retain all the powder.Figure 3.16: A fully assembled coarse approach filter3.3 Cold Bias FilterA single filter is installed on the phosphor-bronze stand-off—which connectsthe STM head to the mixing chamber—to filter the bias line at low tem-peratures. At millikelvin temperatures the Johnson noise produced by thisfilter would be significantly smaller than the room temperature equivalent.To fix the head filter to the stand-off, a custom casing had to be ma-chined.1. Place a rod of OFHC copper, at least 1 cm in diamater, in a lathe.Bore a hole with a 6 mm drill bit to a depth of 45 mm.283.3. Cold Bias Filter2. Turn down a 45 mm section to a diameter of 8.5 mm.3. Use a cutting tool to cut off a 40 mm-long section of the OFHC tube.Figure 3.17: OFHC material in a lathe4. Use a milling machine to mill a 3 mm-wide flat edge along the lengthof the tube.5. Cut and drill a thin OFHC plate 60 mm x 12 mm with 4 M3 through-holes. These holes allow the filter assembly to be mounted directlyonto the stand-offFigure 3.18: OFHC plate to mount filter to STM head6. Electroplate gold onto the OFHC pieces.293.3. Cold Bias FilterFigure 3.19: Gold-plated filter tube and mounting plate7. Prepare the inductors in exactly the same was as was described forthe LC-Π powder filters (with NiCr wire and epoxy/powder cores),however, use 316 stainless steel powder instead of iron since the filterwill be exposed to strong magnetic fields.8. Pot the inductor in the gold-plated tubing with a 50-50 mixture ofEpo-Tek H77 epoxy and 316 stainless steel. Begin by preparing themixture.9. Plug one end of the tube with a thin slice of teflon, with a hole in thecentre for the inductor pin to pass through. Do this in a similar wayas was described in the coarse approach filter section, but do not useany epoxy to seal the teflon in place.10. Place the assembly in a vise and a simple jig to keep it upright, andfill the tube with the mixture.303.3. Cold Bias FilterFigure 3.20: Jig for curing the epoxy/powder potting mixture in the coldbias filter assembly11. Place assembly in 150 ◦C oven for a minimum of 1 hour to cure theepoxy.12. Use a pick to gently remove the teflon disk.Figure 3.21: Potted filter tube with inductor.313.4. Tunneling Filter13. Use Epo-Tek H20E silver epoxy to fix the filter tube to the mountingplate.Figure 3.22: Completed head mount filter3.4 Tunneling FilterThe tunneling line filter is very similar to the LC-Π powder filters, exceptthat it has SMA connectors mounted directly to the filter instead of beingmounted in a larger housing.1. Prepare the inductor the same way as for the LC-Π powder filters,again using iron powder.2. Cut the inductor pins so they are 5 mm long.3. Slide the discoidal capacitors over the inductor pins all the way to theNiCr solder joint. Use acid flux and be very quick when soldering thecapacitor to avoid melting the solder on the NiCr joint.Figure 3.23: 100 pF capacitor for tunneling line filters, measuring approxi-mately 3.5 mm in diameter. The part no. is INSTEC 140X101PC0394. Cut a 44 cm-long section of 7/32” outer diameter brass tubing.323.4. Tunneling Filter5. Bore a 1/8” hole in the side of the tubing, large enough to pour powderthrough.6. Cut a small brass sleeve that can just slide over the 44-cm brass tubinglong enough to completely cover the 1/8” hole.7. The discoidal capacitor outer diameter is slightly smaller than thebrass tubing inner diameter, so cut a short section of thin-walled brasstubing so that the capacitor can be connected to the outer conductor’sinside wall.8. Use room-temperature conducting epoxy to attach the small thin-walled sleeve to the discoidal capacitor and the inside of the brassouter conductor.9. Use room-temperature conducting epoxy to attach the SMA connectormale and female pins to the inductor pins.10. Insert one of the pins into the corresponding SMA connector dielectric.11. With the SMA connector held in place, slide the long brass tubing overthe inductor and use room-temperature conducting epoxy to attach itto the SMA connector at the end.12. Slide the brass sleeve over the filter’s outer conductor.13. Attach the other SMA connector in the same way as the first SMAconnector.14. Pour the iron powder through the 1/8” hole in the outer conductor.15. Once the assembly is full of powder, slide the sleeve over the 1/8” holeand seal it in place with room-temperature conducting epoxy.333.5. Filter HousingFigure 3.24: Filter with SMA connectors for tunneling line3.5 Filter HousingTo prevent EMI from re-entering the wires after they’ve been filtered, it’scrucial to hermetically seal all the conduits and assemblies, as well as shieldthe filter outputs from the inputs. In order to achieve this, the filters shouldbe housed in an RF-sealed metal assembly with the filters fixed to a metalpartition within the housing. The box material and thickness should be suchthat fields can not significantly penetrate the walls. Three separate housingswere made for the three multi-wire ports.Figure 3.25: Filter housing prior to installing filters, wiring, and connectorsThe housing assembly consists of 1/4” brass plates machined on a CNC343.5. Filter Housingmilling machine, and fastened together with screws, conductive carbon paste,and silver conductive epoxy.Figure 3.26: Housing partition populated with filtersThe filters were sanded-down to fit snugly in the brass partition, andsealed in place with silver conductive epoxy. Once populated, the partitionwas sealed in the housing with carbon conductive grease.Figure 3.27: Partially assembled filter housingBefore connecting any of the wires to the filters, lossy microwave foamwas installed against the housing partition to help attenuate resonances353.6. Wiring Conduitformed within the partition walls. The foam, manufactured by MAST Tech-nologies (part number MF22-0002-00), is 1/4” thick and has a loss of -21.5dB at 10 GHz. This alone does not completely damp the resonances presentwithin the housing, however. To further attenuate the resonances, vinylpouches filled with approximately 75 mL of water are stuffed in the housing.Since water is quite lossy to microwaves, these pouches are very effective atremoving resonances.Figure 3.28: Fully populated filter housingOnce all the filters are wired and the lossy foam and water pouch areput in place, the housing lid is sealed to the rest of the housing with a layerof conductive carbon paste.3.6 Wiring ConduitIn order for the wires connecting the filter housing to the wiring ports onthe UHV system to not get re-contaminated with noise, their conduit mustbe fully shielded. Since the shielding in coaxial cables typically consistsof a mesh braid, coaxial cables don’t provide adequate shielding at higherfrequencies. In order to account for this, the conduit was made of a sectionof 3/4” rigid bellows with stainless steel adapters to provide the best possibleseal to the connectors.For each conduit end, the strain relief on the collar (Amphenol part no.’sM85049/52-1-16N and M85049/52-1-14N), which allows the conduit connec-tor (Amphenol part no.’s MS3476L14-19P, MS3476L14-19S, MS3476L14-23P and MS3476L14-23S) to be rotated in alignment with the mating con-363.6. Wiring Conduitnector, was cut and filed off. The newly-filed face of the collar was epoxiedto a custom-made adapter which is then slid over and epoxied to the endof the bellows. In this way, the connectors are well-fixed to the bellows andcan withstand enough torque to be used without any special concern forbreaking the conduit in typical use.Figure 3.29: Connector collar with strain relief (left) and strain relief re-moved (right)Figure 3.30: Stainless steel piece to connect the filed collars to the bellows373.6. Wiring ConduitFigure 3.31: Conduit connector, collar, adapter, and bellows epoxied to-getherFigure 3.32: Wiring conduit38Chapter 4Attenuation4.1 Measurement SetupAttenuation data between 10 MHz and 43.5 GHz were collected on an Ag-ilent N5244A PNA-X Network Analyzer. Since the filters don’t have anyconnectors on them, they were mounted in the same assemblies as the onesconnected to the STM connector ports, the only difference being the par-tition within the housing having a single hole for a filter. This housingalso includes the lossy microwave foam and pouches of water to dissipateresonances from within the housing.Figure 4.1: Housing with SMA connectors installed to measure filter atten-uation on the vector network analyzer394.2. Filter Attenuation CharacteristicsFigure 4.2: Housing connected to vector network analyzer4.2 Filter Attenuation CharacteristicsDue to the relatively high resistivity and low magnetic permeability, stainlesssteel powder has been an ideal candidate for some powder filters. Though,as mentioned earlier in this thesis, in choosing between stainless steel powderand iron powder, there is a trade-off: steel is an order of magnitude moreresistive than iron and so is better at dissipating RF power via the skineffect, but iron has a high magnetic permeability and thus increases theinductance of the filter; which has better attenuation depends on the filtergeometry.404.2. Filter Attenuation CharacteristicsFigure 4.3: Attenuation for stainless steel vs. iron powder LC-Π filters414.2. Filter Attenuation CharacteristicsFigure 4.4: Attenuation for tunneling line filterAs shown in Figure 4.3, the roll-off for the iron powder filter attenuationis slightly steeper than that of the stainless steel powder filters, and reachesthe noise floor around 200 MHz instead of 500 MHz. The -3 dB pointof the iron powder filter was measured to be approximately 1 MHz (Thetunneling line filter (Figure 4.4) is essentially the same as the iron powderLC-Π filter, but has SMA connector directly attached). In this geometry,the effect of the increased inductance with iron is stronger than the effectof increased resistance with steel. However, in the geometry of the coarseapproach filters, this does not seem to be the case.Adding iron powder to the filters scales the inductance by the powder’spermeability, and since the inductance in the coarse approach filters haveabout 10 times fewer windings (and thus 10 times lower the inductance)than the LC-Π filters, the absolute value of the increased inductance whenusing iron instead of stainless steel in the coarse approach filters is muchsmaller than the improvement in the LC-Π filters. As seen in Figure 4.4,the increased attenuation resulting from this increased inductance is on thesame order as the increased attenuation due to the increased RF powerdissipated in the more resistive stainless steel powder.424.2. Filter Attenuation CharacteristicsFigure 4.5: Attenuation for stainless steel vs. iron coarse approach filtersThe cold bias line filter behaves very similarly to the coarse approachfilters. This might be surprising because the cold bias line filter has prac-tically the same inductor as the LC-Π stainless steel powder filters, whichhas a much greater inductance (and resistance) than the coarse approachfilters. This suggests that installing capacitors in the Π configuration playsa very significant role in lowing the cutoff frequency. This was also seen byLukashenko and Ustinov where the attenuation reached the noise floor (-100dB) an order of magnitude higher (1 GHz) without capacitors [5].434.2. Filter Attenuation CharacteristicsFigure 4.6: Attenuation of cold bias filter, taken at room temperatureIt’s worth comparing these characteristics to other filters in the liter-ature. The powder filters made by Lukashenko and Ustinov [5] performslightly better than the LC-Π filters in this thesis, even though they usestainless steel powder instead of iron. Whereas the stainless steel LC-Πfilters in this thesis reach -100 dB around 400 MHz, Lukashenko’s stain-less steel filters do this around 200 MHz. The improved performance ofthe Lukashenko and Ustinov filters is likely due to the fact they use 4 nFcapacitors instead of 100 pF.444.2. Filter Attenuation CharacteristicsFigure 4.7: Attenuation of Lukashenko LC-Π powder filters [5]At the time of this writing, it seems this LC-Π powder filter designhas the best performance out of filters used in low temperature physics ex-periments. Other filters have higher cutoff frequencies, but may be moredesirable because they are more easily manufactured. For instance, the mi-crofabricated meander line surrounded with stainless steel powder designedby He´le`ne le Sueur and Philippe Joyez reaches -100 dB around 4.5 GHz, butcan be fabricated much more easily [2].454.2. Filter Attenuation CharacteristicsFigure 4.8: Attenuation of microfabricated le Sueur meander line powderfilters [2]46Chapter 5ConclusionLow temperature physics experiments rely on very low noise systems to meetscientific standards. Scanning tunneling spectroscopy (STS) measurementsaim to resolve features in the local density of states (LDOS), which are onthe order of 10 µeV, requiring experimental temperatures to be extremelylow.The dilution refrigeration scanning tunneling microscope (STM) at theUBC Laboratory for Atomic Imaging Research (LAIR) is capable of reachingtemperatures below 100 mK. However, ambient radio-frequency noise canenter the system and contaminate the tunneling signal. In doing so, energyresolution in STS measurements is decreased which increases the effectivetemperature to levels which may obscure certain features in the LDOS. Inorder to fully exploit the capabilities of the system, a set of microwavefilters was built to eliminate electromagnetic interference (EMI) and reducethe effective temperature of our STS measurements.In this thesis I present a technical description of metal powder filters usedto eliminate EMI up to seemingly indefinite frequencies. A brief overviewof the LAIR STM instrumentation is given, followed by step-by-step in-structions for how to build the different kinds of filters used on this system.Finally, I present attenuation data taken to characterize the filters.The LC-Π iron powder filters have a cutoff frequency of approximately1 MHz and reach the noise floor (approx. -100 dB) at 200 MHz. The filtersused on the coarse approach lines of the STM and the cold bias filter arebuilt without the use of capacitors and reach -80 dB around 5 GHz.47Bibliography[1] http://lair.phas.ubc.ca/facility/, retrieved on March 29, 2016.[2] H. Le Sueur, P. Joyez, “Microfabricated electromagnetic filters for mil-likelvin experiments,” Rev. Sci. Instrum., vol. 77, no. 11, 2006.[3] M. Assig, et al., “A 10 mK scanning tunneling microscope operating inultra high vacuum and high magnetic fields,” Rev. Sci. Instrum., vol. 84,no. 3, 2013.[4] J.M. Martinis, M.H. Devoret, J. Clarke, “Experimental tests for thequantum behavior of a macroscopic degree of freedom: The phase differ-ence across a Josephson junction,” Phys. Rev. B, vol. 35, no. 10, 1987.[5] A. Lukashenko, A.V. Ustinov, “Improved powder filters for qubit mea-surements,” Rev. Sci. Instrum., vol. 79, no. 1, 2008.[6] K. Sapchuk, “Construction and Commissioning of the UBC Ultra HighVacuum Gigahertz 2D Tesla Ultra Low Temperature Scanning TunnelingMicroscope,” April 201248

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0300650/manifest

Comment

Related Items