There are quite a few publications on DEPFETs since their invention in 1985. For your convenience and as an overview, we placed here an excerpt from the paper "Novel pixel detectors for X-ray astronomy and other applications", G. Lutz, R. H. Richter and L. Strüder, NIM-A, Volume 461, 1 April 2001, Pages 393-404. For more information please see the publications page at the MPI Semiconductor Lab,

The Depleted P- Channel Field Effect Transistor (DEPFET)

The DEPFET principle has been proposed in 1985 by Kemmer and Lutz [1]. It has been subsequently verified experimentally [2] confirming its interesting properties.

Operation principle of a DEPFET

The principle of the DEPFET is shown in Fig. 1. It is based on the sideward depletion as used in the semiconductor drift chamber and the field effect transistor which can be of MOS-type (as shown in the figure) or junction type. The transistor is located on top of a low-doped n-type semiconductor substrate. It becomes fully depleted by applying a sufficiently high negative voltage to the backside p+ contact.


Fig. 1. The DEPFET structure and device symbol.

By suitable doping, a potential minimum for electrons is formed below the transistor channel. The fully depleted bulk is the sensitive volume of the detector in which electron–hole pairs created by the incident radiation are separated by the electric field. While the holes are moved to the negatively biased backplane, the electrons are collected in the local potential minimum below the channel of the transistor (the “internal gate”) and thus increase its charge density by induction. As a consequence, the transistor current is increased as long as the signal charge is not removed from the internal gate. Removal of the charge (emptying of the internal gate) can be performed in several ways, some of them will be described below. The device symbol for the DEPFET is shown in Fig. 1. It is derived from the usual transistor symbol, adding the internal gate and the reset contact.

The use of DEPFET arrays as pixel detectors

Arranging many DEPFETs over an extended area leads to a pixel detector array with each single DEPFET providing one pixel. By choosing suitable voltages on sources, drains and gates of the transistors, one can turn on a single transistor (or a group of transistors) and measure the charge collected in the selected pixel (or group of pixels) by comparing the measured transistor current with the value obtained for empty internal gate. As the signal charge is not destroyed by the act of measurement, the signal can be measured repeatedly and also with varying granularity (by choosing different combinations of pixels). Fig. 2 shows an arrangement of DEPFETS in which each pixel (or a group of pixels) can be read out individually. In this arrangement, the drains of the DEPFETs are connected in a columnlike fashion, while sources and gates are connected in rows. Also indicated in this circuit diagram are clear electrodes which are used for emptying the internal gates. They are also connected in a row-like fashion. The methods for clearing internal gates will be discussed below.


Fig. 2. The circuit diagram of a DEPFET matrix with one single output node. Parallel readout by providing each drain column with a separate readout electronics leads to a large increase of readout speed.

Several schemes for readout are possible. The drain currents of each individual pixel in the row may be read out in parallel or, as indicated in the figure, in series through one output node if the multiplexer for the drain current (on the bottom of the figure) is included. For applications in X-ray imaging, continuous scanning of rows and subsequent clearing seems most natural when full-frame readout is desired. Contrary to readout of CCDs, the signal charge is not moved to an output node, but is read out at the location of production. Therefore "out-of-time events" (data collected during the readout cycle of CCDs with wrongly assigned position information) are completely avoided.

Several principal points have to be addressed in order to arrive at a satisfactorily working device: clearing of the signal charge (of the internal gate); the pixel topology; the readout method and the topological layout of the matrix. These points will be discussed below.

DEPFET topologies

Principally, one distinguishes between open and closed geometries. Open geometries are usually linear structures in which source and drain are of rectangular shape connected by the channel which is steered by the gate. For these topologies, care has to be taken to avoid detrimental effects due to the sideward limits of the structure.

Methods for clearing and signal measurement

After each readout the internal gate has to be emptied. Contrary to standard readout where the signal charge is added to a large sea of charge, in the DEPFET the internal gate contains only the signal (neglecting the leakage current). Complete clearing of the internal gate, therefore, avoids the noise due to fluctuations in the left-over charge. In closed geometry clearing can be performed across a potential barrier towards an n-doped clear contact embedded in the source or the drain of the transistor. In open geometries, the clear contact can be located also to the side of the channel. In the latter version, fast clearing of a geometrically small internal gate can be achieved. Common to both geometries is the necessity to prevent signal charge from reaching the clear contact instead of the internal gate during the charge collection phase. This can be accomplished by implanting a buried p-doped layer beneath the clear contact.

The signal is measured by the difference of the transistor currents, with and without signal charge being present in the internal gate. For the empty state, one can take either the measurement before the signal charge deposition or the status immediately after clearing. While the first method works also for incompletely cleared internal gate, it has disadvantages with respect to noise performance since it is sensitive to low-frequency noise. In addition, the currents have to be read out twice and the first readings stored throughout the complete readout cycle in order to be able to compute the difference, while for the second option the difference can be evaluated directly.


[1] J. Kemmer and G. Lutz Nucl. Instr. and Meth. A 253 (1987), p. 365

[2] J. Kemmer, G. Lutz et al.Nucl. Instr. and Meth. A 288 (1990), p. 92

Edit | Attach | Watch | Print version | History: r5 < r4 < r3 < r2 < r1 | Backlinks | Raw View | Raw edit | More topic actions
Topic revision: r5 - 2008-04-28 - LaciAndricek
This site is powered by the TWiki collaboration platform Powered by PerlCopyright © 2008-2020 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding TWiki? Send feedback