Our ability to explore the physics of elementary particles depends on the sensors we use to translate flows of energy from particle collisions in our accelerators into electronic pulses in our detectors. The patterns of these pulses are used to reconstruct the underlying particles and their interactions. At the core of the mammoth detector assemblies and snugly surrounding the beam pipes are arrays of silicon sensors. These sensors, derived from integrated circuit technology, provide detailed patterns of interactions to micron-level (40 millionths of an inch) precision, with subnanosecond timing and low mass. The active area of these arrays has increased from a few square centimeters in experiments in the 1980s to 200 square meters in the CMS and ATLAS trackers at the Large Hadron Collider at CERN. The CMS high-granularity calorimeter, or HGCal, will use 600 square meters of silicon. The precision of these detectors enables unique identification of heavy quarks (bottom and charm) that travel a fraction of a millimeter before they decay. The precision was crucial, for example, in the discoveries of the top quark in 1995, CP violation and mixing in the B meson system, and the Higgs boson in 2012.
Research and development to improve the characteristics and develop better silicon detectors with the use of new technologies continue as we upgrade the existing detectors for better performance and develop designs for experiments at future generations of accelerators.
The 3-D integration of pixelated sensors with readout chips was an infant technology when we began R&D in 2006. The 3-D interconnection technique (now called hybrid bonding by the semiconductor industry) can replace the large, costly, solder bump interconnect technology with one that can be directly integrated into semiconductor process lines. It reduces the minimum spacing between pixels from about 50 microns to three, allows multilayer stacked connections through the body of the semiconductor, and dramatically reduces the capacitance of the interconnect, increasing speed and reducing electronic noise. Working with collaborating laboratories and industrial partners, we have developed and demonstrated the first three-layer 3-D bonded devices, with two electronics layers occupying only 35 microns in height, down from the usual hundreds. This hybrid bonding technology is now probably in your smart phone camera.
Future accelerators, including the High-Luminosity LHC, will produce collisions at a rate many times higher than the current LHC. The complexity of these collision events puts a premium on fast timing and recognition of very complex patterns of energy deposited in detectors. A possibility we are exploring is the induced-current detector. 3-D technology allows us to combine small pixels and low electronic noise with sophisticated electronics. The sensitivity and timing capabilities are now so good that we can measure the detailed shape of pulses due to charge movement deep in the silicon. This pattern of pulse shapes can give us much more information than the usual measurement of only the total charge. If this idea works, a single layer of silicon could measure timing to picoseconds, position to microns, as well as track angle, compressing multiple layers of sensor into one. This would greatly increase the power of detectors to select and process interesting events at very high speed. Work is under way on simulations of these effects and collaboration with industry on a 3-D demonstrator.
Another way to address the experimental challenges is to improve the time resolution of silicon detectors. This can be done by designing the silicon to provide internal gain, providing a larger signal with a faster rise time. The low-gain avalanche diode, or LGAD, was designed to accomplish this. The LGAD is a new technology, and improved variants are continually emerging. Fermilab has an extensive program of testing and qualifying these LGAD detectors in bench tests and in the Fermilab Test Beam. The work is a close collaboration with the foundries and with other institutes within CMS and ATLAS. This program has been crucial in the validation and adoption of LGAD technology for the CMS upgrade endcap timing layer.
The current generation of LGADs suffers from dead regions at the edges of each pixel and has only moderate radiation hardness. This limits the pixel size and range of applicability of these devices. By changing the top layers of the sensors (AC coupling) and adding a layer buried below the surface (buried gain layer) we can both eliminate most of the dead region and provide for a more well-defined gain that is also more resistant to radiation. First demonstrators are now being fabricated in collaboration with industry and universities.
Finally, the very large area of the CMS HGCal prompted us to begin the development of large-area sensors, producing the first HEP sensors on 8-inch silicon wafers in collaboration with industry. We developed the process flow with colleagues from other laboratories and integrated designs from contributors all over the world. We have demonstrated high quality 8-inch sensors thinned to 200 microns.
In this work, intense collaboration with the Fermilab ASIC group, support from CMS and DOE, infrastructure at SiDet, strong collaboration with laboratory, university and industrial partners, and the central contributions of summer students, graduate students, and postdocs have all been vital. These are all exciting developments and there is much more to do. As Richard Feynman said: “There is plenty of room at the bottom.”
Ron Lipton is a Fermilab scientist working on R&D and design of detector and sensor assemblies for the CMS HL-LHC tracker and high-granularity calorimeter upgrades. He is co-manager for HGCal sensors in international CMS.
CMS Department communications are coordinated by Fermilab scientist Pushpa Bhat.