Compton Telescope

The Compton Spectrometer and Imager employs a novel Compton telescope design, utilizing sixteen 3-D imaging, high spectral resolution germanium detectors (GeDs). The Compton imaging serves three purposes: imaging the sky, measuring polarization, and very effectively reducing background.

Compton Telescope Basics

A Compton telescope measures the position and deposited energies of a sequence of Compton scatters from a gamma-ray to determine the path through an active detector. By determining the relative order of scatters in addition to the Compton scatter angle and position of the first interaction, one can constrain the origin of the photon to an annulus on the sky, called the event circle.

Schematic of a Compton event showing the relevant parameters measured in a Compton telescope. An incoming gamma ray scatters twice and ends up in a photoabsorption event. The active detector volume is depicted in blue.

If more than one photon comes from the same sky position, then their respective events circles will overlap and reveal the location of the source (see the back projection animation below). Iterative deconvolution techniques are used to develop an image.

Back-projection of Compton circles from simulated 0.662 MeV point source.

10 iterations of LMEM algorithm on simulated 0.662 MeV point source.

The energy, timing and position of each interaction in both layers was recorded to determine the original direction and energy of the gamma-ray to make an image and spectrum. CGRO was a hugely successful mission making the first all-sky maps in the MeV energy range and studies of a multitude of galactic and extragalactic objects [Schoenfelder & COMPTEL 2000].

Compton telescopes, like COSI, are referred to as “compact” Compton telescopes due to their compact design. They use the same principle as COMPTEL using the energy and position of a series of scatters to reconstruct the original source direction. However, there are no longer two separate planes; the whole detector volume acts simultaneously as the scatterer and absorber, allowing for 2+ site events.

The relative order of Compton scatters from a single photon can be determined by using redundant information from the energies and scatter angles.

One complication of compact Compton telescopes is that the time in-between scatters is shorter than the timing resolution of traditional detectors, therefore, all of the scatters from one photon will appear simultaneous. One can use redundant information from the deposited energies and the geometry of the scatters as well as the chance of absorption within the detector medium to determine the most probable sequence. This technique, referred to as Compton Kinematic Discrimination [Boggs & Jean 2000], allows for significant background suppression.

One major benefit of Compton telescopes is that they are inherently sensitive to polarization. A polarized photon will predominantly Compton scatter 90º relative to it’s polarization angle (see Klein-Nishina cross-section). Therefore, if you have polarized gamma-rays incident upon a Compton telescope, counting the number of scatters as a function of the azimuth angle in the detector will show a modulation with the highest number of scatters being at 90º and 270º relative to the polarization direction. From the magnitude of this modulation one can determine the degree of polarization of the source.

Modulation seen in the azimuth angle in Compton telescope resulting for a polarized source.

The Compton Spectrometer and Imager

The heart of COSI is an array of 16 crossed-strip cryogenic germanium detectors (GeDs) with 3-D position resolution, excellent spectroscopy, and high efficiency. Accurate measures of the position and energy deposits within our detector are key for our science goals.

The COSI cryostat shown with a cutaway view to expose the 16 GeDs within. The detectors are maintained at their cold operating point with a mechanical cryocooler. The cryostat is surrounded on all four sides and the bottom by Bismuth Germanate (BGO) scintillator detectors which act as an anti-coincidence shield.

The COSI GeD fabrication process was developed at LBNL using their amorphous germanium contact technology [Luke et al. 1992]. Each crystal, measuring 8cm×8cm×1.5cm, has 64 aluminum strip electrodes deposited on the anode and cathode in an orthogonal orientation. These strips are what give the GeDs their 3D position sensitivity (see below for an explanation).

One of the COSI-APRA (Balloon) GeDs wire bonded in it’s cryostat mount. The mirror reflection shows the orthogonal electrodes. The COSI-APRA GeDs had only 37 strip electrodes per detector face, compared to the 64 per face for COSI-SMEX.

One reason we use germanium detectors is because they have excellent energy resolution, however, one disadvantage is that the crystals need to be cooled to almost -200ºC in order to operate. Traditionally this is done using liquid nitrogen, but to avoid any consumables on our instrument, we use a CryoTel Stirling cryocooler. The detectors are housed in an evacuated aluminum cryostat, which is surrounded on all four sides and the bottom with Bismuth Germanate (BGO) scintillator detectors, which act as an anti-coincidence shield to reduce background and effectively constrain the field of view to 25% of the sky.

GeD Positioning Germanium Detectors

When a gamma-ray interacts within the COSI detectors, it scatters a handful of times within the germanium crystal, much like a billiard ball, before slowing down enough to stop. At each one of these Compton scatters, a huge number of electron-hole pairs are created from ionization by the recoil electron. With a high voltage applied across the detectors, the electrons travel in one direction and the holes travel in the opposite direction along the field lines, and it’s this motion which induces a signal in the COSI detectors. (Read more about how semiconductor detectors work on wiki.)

The number of electron-hole pairs created in a Compton scatter is directly relative to the energy of the recoil electron. And the size of the induced signal in the COSI electronics is a measure of the number of electrons and holes transported through the detector. Therefore, the pulse height of the signal measured from one interaction gives a measure of the energy deposited. Adding up the energies from the successive Compton scatters from one photon gives the total initial energy.

Determining the position of the interactions within COSI is done in two ways. As mentioned above, the germanium crystals each have 37 strip electrodes deposited in an orthogonal orientation on the top and bottom face. The signal from the electrons and holes will only be detected by strips directly above and below the area of the interaction. The intersection of the two activated strips gives the X-Y position of interaction. From the strip width of 2 mm, this defines an X-Y pixel size of 4 mm².


The Z-positioning is obtained from the difference in arrival time of the electrons and holes at their respective electrodes, which we refer to as the Collection Time Difference (CTD). If an interaction occurs near the anode, the electrons have a much shorter distance to travel and will arrive much before the holes drift through the depth of the detector to be collected on the cathode. An interaction that occurs within the center of the detector would lead to an equal arrival time of electrons and holes, resulting in a CTD ∼ 0.


By precisely measuring the position and energy of each Compton scatter within our detectors, we can obtain a modest angular resolution of >4º.

Na-22 (511 keV, 1274 keV) spectrum of the Compton-scatter events. 3 keV FWHM resolutions are currently achieved, however these can be approved upon by further calibrations and corrections.

COSI image of a lab Cs-137 (662 keV) source using our real time analysis program Realta. The image of the source is displayed in galactic coordinate through our complete reconstruction using our differential dGPS.