Inner Vertex Detector


The Detector will be a tracking detector, sitting close to the target in the inner MWPC. Its purpose is the tracking of charged particles close to the interaction point (IP), and optionally the measurement of recoil proton polarization by measuring the deflection angle off a secondary carbon target.

It consists of multiple double sided silicon strip sensors (DSSD), which allow a very high position resolution for charged particle tracks.


The available space is a cylinder inside the inner MWPC, with a diameter of 84 mm. Currently this space is occupied by the target pipe, which will therefore have to be shortened. The first layer should be as close as possible to the target, while a second layer will be positioned a few centimeters downstream, with the best value currently being calculated from simulations. When using a secondary target, a third layer will be needed after the carbon target, so that the deflection angle can be inferred from the two track points in the silicon layers before the secondary target and at least two behind, one from the silicon detector and one from e.g. MOMO.

The available DSSDs are quadratic with a side length of 35.04 mm. Four of these detectors can be arranged on the circular cross-section so that a hole with a radius of 3.13 mm is left open for the photon beam. By using quadratic sensors, an area of 4083 mm2 is covered, 74 % of the total circular area (excluding the hole). Figure 1 shows the arrangement of the detectors.

Figure 1: Arrangement of the DSSDs. Beam enters from left, the black cylinder is the carbon secondary target.


The DSSDs are quadratic with a side length of 35.04 mm, and have an active area of 33.315 mm squared. With 512 strips per side this corresponds to a strip pitch of 65 μm, although in the planned setup every second strip is read out, so that the effective pitch is 130 μm. The stereo angle is 90°, the thickness 285 μm.


For the read-out of the DSSDs the well-known APV25-S1 frontend chip will be used, which is also used by CMS and BELLE-II, among others. One front-end chip reads out 128 channels, performs signal shaping and buffering, and requires one ADC channel to be digitized. One sensor will need 4 APVs for readout, one layer 16 accordingly. A PCB has been designed to carry the front-end chips and on which the sensors are glued in a way that wire bonding of the strips to the front-ends is possible. These PCBs will add 2 mm of material per sensor layer, with a radiation length of approx. 2.3 %, to the 0.3 % from the silicon.

A readout chain for the APV25 front-end has already been developed and used at various beam tests at COSY, DESY and CERN. The newly developed PCB is compatible to existing cabling and support structures.

The power consumption of the APV25 is approx. 350 mW per chip, which means that a thermal power of around 5.6 W needs to be cooled per layer.


The first test module has been equipped and is currently tested in the lab. Figure 2 shows photos from both sides of the module.

Figure 2: Left part shows the front of the first test module. In the top the sensor is partly visible, the corresponding part in the bottom is an equally thick spacer. Below the sensor there is the pitch adapter and two APV25 front-end chips. Right part shows the back side.

First tests have been conducted with a Sr-90 source. A step in the count rate spectrum can be seen where the front PCB ends, because electrons from the Sr-90 source have a kinetic energy where a fraction of them is stopped in the 0.5 mm thick PCB.

Currently the thermal heating from the front-ends poses a problem, because the high temperature increases the reverse bias current (from 2.5 μA at 20° C to around 25 μA in the current setup), and therefore the noise. Figure 3 shows the ADC distribution, taken from a measurement with Sr-90.

Figure 3: ADC spectrum from Sr-90 source. The landau distribution is clearly visible, with a small exponential background at the left.

It has to be determined if a better temperature isolation between sensor and front-ends and different heat transport in the PCB is sufficient to solve this problem, or an active cooling will be required.