COMPASS DC5 Story follow me on Twitter follow me on facebook

What is DC5?
DC5 is a new large-area planar drift chamber for the COMPASS experiment at CERN. It is replacing an aging tracking detector. DC5 was installed in May 2015 and has been collecting valuable data in the 2015 polarized Drell-Yan run. DC5 will continue to be an important tracking detector in the COMPASS 2016, 2017, and 2018 runs.

Who built DC5?
Almost 20 undergraduate and five graduate students have worked on the DC5 project over the years. Together with the technical personnel at our Nuclear Physics Lab (NPL), ODU, and CERN, they were essential in building and testing prototype detectors, in carrying out studies to find the best material and techniques, and in assisting with the construction, assembly, installation, and commissioning of the detector. Currently two UIUC graduate students are extracting physics results from 2015 COMPASS data collected with DC5.

How does a drift chamber work?
A drift chamber is a device that allows the recording of information about charged sub-atomic particles, which are not visible to the human eye. As for every detector, the particle must leave some kind of trace of itself - it must interact with matter, and then this usually tiny signal has to be amplified. In case of a drift chamber, the charged particle passes through a volume of gas and ionizes a gas atom - i.e. it kicks out one of its shell electrons. The negatively charged electron is attracted by the counting wire with positive electrical potential. On its way to the counting wire the electron creates an avalanche - it kicks out more shell electrons, which kick out more electrons, and so forth. When the united cloud of electrons arrives at the counting wire, it is big enough - a typical size is 50,000 - to leave a signal. For a given counting wire, the arrival time of the hit is recorded and this information, together with suitable calibrations, allows later to conclude on the vertical or horizontal coordinate of the hit - depending on the wire orientation. Combining different hits in different layers and the knowledge about the distance of the layers and wires allows to construct three-dimensional space points for each hit. In the computer algorithm to find the best particle trajectories, or tracks, space points from many different detectors are combined.

How is DC5 designed?
DC5 is a stack of alternating layers of cathodes at high negative electrical potential (-1675 Volt) and anodes at positive electrical potential (0 V). The cathodes are thin carbon-painted mylar sheets and each anode consists of 256 parallel counting wires. The counting (sense) wires are made of gold-plated Tungsten and have a diameter of 20 mu - thinner than a hair! The distance between two sense wires (the pitch) is 8mm. Symmetrically between each two sense wires are 100 mu thick field wires made of an Be-Cu alloy and supplied with the same voltage as the cathodes. Together with the two neighboring cathodes, one sense wire and its two neighboring field wires constitute the smallest unit: a drift cell, a cube of 8mm x 8mm x 8mm.

There are four different wire orientations: horizontal (YY'), vertical (XX'), and offset by +(UU')/-(VV')10 with respect to the vertical. Each wire orientation occurs twice in the detector, where the "primed" view refers to the view for that the wires are offset by half a drift cell compared to the non-primed view. This allows to resolve the following ambiguity: for a given hit in DC5 the arrival time at the counting wire is known, but it is not known whether the electron cloud hit it from the left or from the right. (If I tell you that I live 15 minutes away from work, you still do not know whether I live North or South of work.) The second layer of offset counting wires adds the missing piece of information since it will also record a hit. (If in addition I tell you I live close to a subway station, you can combine both pieces of information and figure out where I live because there is no subway station in the North.)

In total there are 20 active layers in DC5 - eight anode planes and twelve cathode planes. In order to hold the wires mechanically in place, they are soldered to Printed Circuit Boards (PCBs). A precisely defined weight is attached to each wire before soldering - if the weight is too heavy, the wire will break, and if the weight is too light, the wire will sag and will start oscillating when voltage is applied. The 21th G10 frame is empty and serves as spacer. The PCBs and the mylar are attached to G10 frames. G10 is a glassfiber-re-inforced epoxy, which is flexible and robust at the same time. The G10 frames are sandwiched between two stainless-steel ("stiffening") frames. They weigh each more half a metric ton and provide the stiffness to keep the G10 frames as planar as possible, also when the chamber is pulled vertical for its operation in COMPASS. Pins through 40 precision holes tightly hold all 21+2 frames together. Sheets of aluminized mylars ("gas windows") cover the stiffening frames, serving two purposes: to create a closed gas volume, and to create a Faraday cage for an undisturbed electrical operation of DC5. Each layer is supplied with high voltage (HV) from the outside. Each PCB is equipped with a set of connectors, which allow to pick up the counting wire signals with Front End Electronic (FEE) cards. Supply and signal cables are routed in the volume surrounding the chamber stack, called the support frame.

How was DC5 constructed?
2014 was the main construction year for DC5. While the two stiffening frames were produced by an outside company, our first task was to construct the G10 frames that hold the wires and the mylar foils. The G10 frames were entirely handmade by our group in the Nuclear Physics Lab: technicians, a faculty, a postdoc, three graduate students and almost a dozen undergraduate students.

To build one G10 frame, four G10 strips were milled by our technical personnel in the department's large CNC machine: two short strips of 2,48m length and two long strips of 2,88m length. The most critical parameter of the strips was their thickness - the requirement was that the strips may not be thinner than the design (4mm, 10mm, or 12.5mm), and thicker by at most 50mu. Students of our group measured the thickness of each strip at dozens of locations using a micrometer and iterated with the technicians whenever necessary. The first and last 20cm of each strip were milled down to about half the thickness. Details such as holes, grooves and cutouts were milled as last step.

Then each four strips were glued together at the corners by creating lap joints. In determining the proper thickness of the lap joint area of individual strips, also the thickness of the glue had been taken into account after various glueing tests performed by the students. The glue was a rather viscous two-component epoxy, which was mixed freshly for every new glueing session with a working time of 30 minutes and a drying time of 16 hours. Before a frame was declared ready, it required polishing and grinding at the lap joints - depending on how much excess glue had leaked out. The glueing and polishing procedures were repeated 21 times. The eight anode frames were then equipped with PCBs, which were first precisely aligned, and then glued on the G10 frame.

Anode Production
DC5 was equipped with wires at the clean room of Old Dominion University [link 1], at the NPL clean room, and at CERN. Prior to the start of the soldering, our students carried out test series on wire tensions and wire quality by investigating samples in an Environmental Scanning Electron Microscope (ESEM). For convenient wiring on the large frames, a practical system to keep wires at the proper distance and a system to transport the wire spool over the length of the frame was developed. Tensions of every single wire were measured at NPL or at CERN by injecting an alternating current of variable frequency into the wire. When a permanent magnet is placed next to the wire, it will start to oscillate at a certain frequency. This frequency is inversely proportional to the mechanical tension and independent of the magnetic field strength. Wires with too low tension were replaced. LINK

Cathode Production
The DC5 cathodes were produced at CERN and its vicinity. To create a cathode plane, a big piece of mylar foil was stretched tightly and glued on the G10 frame. The stretching process required the presence of at least six of us to hold the foil and to attach its edges into dozens of little stamps, which constitute the end pieces of a hydraulically operating stretching machine. Once all stamps were in place, the foil was supplied with mechanical tension and was then brought in contact with the G10 frame spread with glue. After the glue had dried out, the foil was cut in shape and electrical paint was applied in dedicated patterns - the future electrical supplies of the cathode. In bunches of three (one complete view), the cathodes were transported to a local company close to CERN and were painted with a thin layer of carbon (graphite). One of the three cathodes was painted from both sides ("double sided cathode").

Where was DC5 assembled?
After every plane had been successfully tested electrically, it was equipped with an O-ring for gas tightness and stacked on one of the two stiffening frames in the COMPASS clean area at CERN. The stiffening frame was resting horizontally on a well leveled out heavy-duty table. The stack was closed with the second stiffening frame and tightened to the nominal mechanical torque. The deviation of the stack flatness from the design was found to better than 1mm - which means the individual G10 frames combined do not pile up to more thickness than foreseen. Now the drift chamber constitutes a closed volume and the nominal gas flow can start: a mixture of Argon (ionization gas), flammable ethane ("quencher" to confine the electron avalanche), and CF4 (break long-molecular chains and thus to delay wire chamber aging). The surrounding support frame was assembled; HV cables were equipped with connectors and routed; boxes were built with resistor and capacitor networks to protect the sensitive DC5 wires cathodes from over currents; the FEE was installed and tested. A radioactive 90Sr source was placed on top of DC5 and for the first time a physics signal could be seen from DC5: the chamber was amplifying, i.e. the charged particle emitted from the source created an electron avalanche, which was detected by the counting wires. DC5 was installed into COMPASS on May 13, 2015 and then again on April 12, 2016 after some maintenance of the detector in between the COMPASS 2015 and 2016 runs.

Is this the end of the story?
Yes and no. I have not mentioned many other steps involved in the DC5 construction: distance measurements, masking, dremeling, routing, polishing, attaching nylon threads, transferring a G10 frame to a handling frame and flipping it, high-voltage and front-end-electronic tests, and most of all: cleaning. Cleaning the floor, the G10, the wires, the equipment. You have to work in a very clean environment to build a drift chamber. Certain substances in wire chambers may outgas and cause aging or even an early failure of the detector. You may grow tired of it - but the result is rewarding. For many months I could not see the light at the end of the tunnel. There were setbacks and problems. Together we learned how to solve them. We learned about mechanical precisions, how to ship wire detectors and how to use heavy-duty cranes, about connector crimping, high-voltage trips, isolation resistances and small coils in big magnetic fields. Then suddenly DC5 was a ready detector and was installed into COMPASS. A miracle made possible by a collaborative effort of so many people - technicians, physicists and students - working tirelessly over months and years.

And maybe one day, DC5 will get a younger sister.

Visit the COMPASS Research Project webpage for a slideshow.