A time projection chamber (TPC) provides a complete, 3D picture of the ionization deposited in a gas (or liquid) volume. It acts somewhat like a bubble chamber, albeit with a fast, all-electronic read-out. The TPC's 3D localization makes it extremely useful in tracking charged particles in a high-track-density environment, and for identifying particles through their ionization energy loss (dE/dx). To honour the 25th anniversary of the TPC, a symposium was organized at the Lawrence Berkeley National Laboratory on 17 October 2003, with workshops that included presentations on the past, present and future of the TPC.

The TPC was invented by Dave Nygren at the Lawrence Berkeley Laboratory (LBL) in the late 1970s. Its first major application was in the PEP-4 detector, which studied 29 GeV e+e- collisions at the PEP storage ring at SLAC. Since then TPCs have been used to study e+e- collisions at PEP, at the TRISTAN collider, at the KEK laboratory and at the Large Electron Positron (LEP) collider at CERN. A TPC could also be the central detector at future e+e- linear colliders.

The device has also figured in a number of experiments involving heavy-ion collisions at machines such as LBL's Bevalac and the Relativistic Heavy Ion Collider (RHIC) at Brookhaven; and now the ALICE collaboration is building a large TPC to study heavy-ion collisions at the Large Hadron Collider (LHC). TPCs have also been used in a whole host of non-accelerator experiments.

TPCs in particle physics

The PEP-4 TPC (figure 1) was built to combine charged-particle tracking with good particle identification by measuring the specific energy loss (dE/dx) of charged particles. This 2 m long cylindrical TPC had an inner diameter of 40 cm and an outer diameter of 2 m, and had most of the features of newer TPCs.

Charged particles from e+e- collisions in the centre of the TPC ionized molecules in a mixture of 80% argon and 20% methane gas at 8.5 atmosphere. A central membrane (the cathode) that was charged to -75 kV produced a strong electric field (figure 2). Under the influence of this field, ionization electrons drifted to one of the two end caps. A solenoidal magnetic field minimized the transverse diffusion and bent the charged particles to allow momentum measurement.

The end caps were divided into six sectors, each one containing a 183-anode multiwire proportional chamber (MWPC). Drifting electrons were accelerated in the strong electric fields around the wires and acquired enough kinetic energy to ionize the gas and produce an avalanche. A single drift electron produced about 1000 electrons in the wire.

The wire signals were sampled 10 million times per second to a 9-bit accuracy by an analogue storage unit based on a charge-coupled device (CCD). The signals were then digitized at a slower rate using inexpensive ADCs. The wire data were used to measure particle energy loss. Because of the high gas pressure, the ionization could be measured accurately and the dE/dx resolution achieved was an unprecedented 3%. This meant that pions, kaons and protons could be identified over most of the kinematic range.

Charged particles were tracked using data from 15 rows of 7 x 7.5 mm2 metallic pads located under the wires. When an electron produced an avalanche on an anode wire, a cloud of positively charged ions remained in the gas. The image charge that formed on the metallic pads was then measured using a charge-sensitive preamplifier. By measuring the relative charge on several adjacent pads, the ionization could be localized to approximately 250 µm. These pads were also read out by the CCD system.

Later TPCs used many of the techniques pioneered by PEP-4. Some notable examples were the ALEPH and DELPHI TPCs at LEP, the TOPAZ experiment at TRISTAN and the early vertex chambers for the CDF experiment at Fermilab. The ALEPH TPC at LEP was one of the larger examples, measuring 3.6 m in diameter and 4.4 m in length, with twice as many dE/dx measurements as the PEP-4 TPC. Both of the LEP TPCs used flash ADC systems instead of CCDs.

TPCs have also been used in a number of smaller experiments, such as in studies of muon decay and capture. The MuCap experiment at the Paul Scherrer Institute, for example, is building a 10 atmosphere hydrogen-gas TPC to measure muon lifetime.

TPCs for heavy ions

With the growth of research with relativistic heavy-ion collisions in the early 1980s, TPCs found another home. The 3D picture of ionization is ideal for tracking particles in high-density environments - hundreds or thousands of particles from a single collision - in which other detectors are overwhelmed by the huge multiplicity. The first large-acceptance TPC was in the Equation-of-State experiment (EOS), which studied heavy-ion collisions at energies of a few GeV per nucleon at the Bevalac. The rectangular EOS TPC measured 150 x 96 x 75 cm. Electrons drifted downwards in a uniform electric field and were amplified by 3000 times in a MWPC. Data were read out from 15 308 pads, which sensed the image charge from positive ions in the same way as in PEP-4.

One key development in using TPCs in heavy-ion collisions concerns the electronics. In the high-track-density environment many pads are required and each pad detects signals from such a large number of tracks that it must be read out by a waveform digitizer. The CCD analogue storage units used with earlier TPCs required considerable power and difficult calibrations. They were also expensive. So EOS used a new technique, the switched capacitor array (SCA), developed by Stuart Kleinfelder.

The EOS SCA consists of an array of 128 capacitors, each connected to an input by a switch. By rapidly opening and closing the switches, the capacitors can be connected to the input one by one, forming an analogue storage unit. The sampling rate is matched to the drift time of electrons across the TPC, and the capacitors are read out by an inexpensive (but slow) analogue-to-digital converter. This scheme reduced the cost and power consumption of waveform digitizers, making TPCs a practical tool for the study of heavy-ion collisions. Packaging was integral to the success of the electronics. The preamplifiers, SCAs, digitizers and multiplexers were mounted directly on the TPC, and a handful of optical fibres replaced 15,000 cables.

After completing its service at the Bevalac, the EOS TPC was moved to the Brookhaven Alternating Gradient Synchrotron (AGS), where it was used to study heavy-ion collisions in experiment E895 and proton-ion collisions in experiment E910, and then to Fermilab for experiment E907 probing higher energy proton-ion collisions. By the time E907 ends, EOS will have seen more than 15 years of service at three different laboratories.

EOS pioneered techniques that were used in many other experiments. At CERN the NA35, NA36 and NA49 experiments all used TPCs to study heavy-ion collisions. NA49 took the construction of the devices to new volumes with four huge TPCs - the largest pair measuring 3.8 x 3.8 x 1.3 m. The complex was read out by 182,000 SCAs; these SCA chips had integral ADCs.

The next step up in heavy-ion collision energy was the RHIC at Brookhaven, and it was natural that TPCs would play a key role. Two original TPC-based proposals merged into the STAR detector, which has a 4 m diameter, 4 m long TPC at its centre. The TPC for STAR follows the geometry of the PEP-4 and ALEPH TPCs, but relies on 138,000 pads that are read out by SCA digitizers for both dE/dx and tracking information. The system is much faster than previous experiments: it can digitize an event containing 70 million volume elements to 10 bit precision and transmit it to the data-acquisition system in 10 microseconds. Figure 3 shows an event in the STAR TPC.

The ALICE heavy-ion experiment at the LHC is built around a mammoth TPC, 2.5 x 5.5 m with 750,000 pads. Analogue-to-digital-converter technology has matured and ALICE has replaced SCAs with custom integrated circuits, each containing 16 x 10-bit ADCs with digital filters for tail suppression and zero-suppression circuitry. Data can be read out 1000 times per second.

Sometimes longitudinally drifting electrons may not be optimal. The collaborations for STAR and CERES (NA45 at CERN) have built cylindrical TPCs where the electrons drift outwards radially from a cylindrical central cathode towards anodes on a concentric outer cylinder. This geometry is advantageous when tracks are parallel to the cylinder's axis, but it introduces many complications. The electric and magnetic fields are no longer parallel, which leads to complex electron-drift trajectories. Curved pad-planes are then required, or the idealized cylindrical geometry must be complicated. For these reasons, radial-drift TPCs have a more complex structure and poorer resolution than linear-drift devices. However, sometimes these factors can provide a worthwhile trade-off.

Non-accelerator applications

TPCs are also used in many non-accelerator experiments such as double-beta decay and dark-matter searches. Often these experiments use dense media, such as liquids, where the active detector volume also serves as the experimental target (for neutrinos or dark matter) or a radioactive source (for double-beta decay, proton decay, etc.).

The first laboratory observation of double-beta decay, in 1987, by Steve Elliott, Alan Hahn and Michael Moe, used a thin layer of 82Se deposited on the central cathode of a TPC. Though very successful, this technique was limited to relatively small sample volumes. Most current efforts use a single material, such as liquid 136Xe, as both a source and drift medium. One particularly ambitious group, the Enriched Xenon Observatory collaboration, plans to use a liquid-xenon TPC to localize double-beta decay events and then insert a probe into the xenon to extract the 136Ba daughter product for detailed study.

Liquid-xenon TPCs are also being used as imaging detectors to track photons with energies of a few MeV. The photon directions are determined by reconstructing double-Compton interactions. A liquid-xenon imager has already been used to study the galactic centre. The technology might also be used to search for photons from smuggled nuclear material.

Liquid-argon TPCs have been studied for many years under the aegis of the ICARUS (Imaging Cosmic and Rare Underground Signals) project. The current T600 prototype, based on 476 tonnes of liquid argon in a volume of 275 m3, recently completed a 68 day engineering run. The collaboration's goal is a 3000 tonne detector in the Gran Sasso Laboratory in Italy, which will study solar and atmospheric neutrinos, terrestrial neutrinos from CERN's Super Proton Synchrotron (SPS), and also proton decay. The solar neutrino study may be quite challenging in terms of backgrounds.

One interesting idea being pursued by several groups is to use drifting ions rather than drifting electrons in a gas or liquid. Both positively and negatively charged ions have been considered; the latter can be formed when an ionization electron attaches itself to a previously uncharged molecule. The advantage of ion drift is that the diffusion can be much smaller. One big drawback is that positively charged ions cannot induce avalanches, greatly complicating the detection of the signal. The much slower drift velocity seems to offer both advantages and disadvantages. Ion-drift TPCs have been considered for a variety of applications, including double-beta decay and dark-matter and axion searches.

Future directions

The most exciting technological developments in gaseous TPCs concern electron amplification, where two new technologies are replacing wire chambers. Gas Electron Multipliers (GEMs) are plastic foils that are metal coated on both sides, with 50-100 µm diameter holes punched in them.

The metal coatings are charged to a potential difference of a few hundred volts, creating strong electric fields in the holes. Electrons drifting into the holes ionize the gas, creating an avalanche much like that formed around anode wires in conventional chambers. GEMs have several advantages over wire chambers. They are easily supported, eliminating wire sag and instability, and can be placed very near to read-out pads, reducing diffusion after amplification. The high hole density provides an even amplification over a large area. Positive ions generated in the avalanche drift naturally away from the amplification region, eliminating the build-up of space charge in the amplification region.

Micromesh gaseous structure chambers (Micromegas) use a thin metal mesh instead of anode wires. The mesh can be supported a small distance above the pads. A simple wire grid above the Micromegas produces a potential difference with the mesh, so electron avalanches form in the strong electric fields around the mesh elements. Like GEMs, Micromegas can be placed very close to read-out pads, greatly reducing diffusion. They also have the same advantage as GEMs for positive-ion elimination.

Both GEMs and Micromegas have a somewhat lower gain than wire chambers. However, two or three layers of GEMs or Micromegas can be cascaded by placing the foils or meshes on top of each other, thereby multiplying the gains. GEMs and Micromegas are beginning to replace wire chambers in some experiments, most notably in the COMPASS experiment at the SPS at CERN. They are also prominent in R&D for future linear colliders and for upgrades of the detectors at RHIC.

Over the past 25 years, TPCs have grown into a proven, mature and flexible technology. With these new developments the next quarter century looks equally bright.

Further reading

J A MacDonald (ed) 1984 The Time Projection Chamber AIP Conf. Proc. 108 (AIP, New York).
Jay N Marx and Dave Nygren 1978 Physics Today October 46.
For more about the Time Projection Chamber symposium, see www-tpc.lbl.gov/symposium/.
For more about the use of TPCs at future linear colliders, see www-lc.lbl.gov/tpc/meeting/.