2.1.

Focal Plane Design

Due to its physical size of $\sim 14\mathrm{cm}$, the focal plane detector cannot be realized as a single monolithic device. Instead, it will be composed of a combination of four large DEPFET arrays designed to cover the large field of view and one smaller DEPFET optimized for fast timing applications (see Sec. 3.3). Several options for the location of the fast detector are currently being studied. Placing it in the center, as shown in Fig. 3, has the advantage of compactness and permits the observation of very bright sources simultaneously with objects in the large field of view. However, this option requires detailed analysis because of two issues: the complexity in arranging the front-end electronics behind the sensor (i.e., the far side with respect to the mirror system) and as a consequence, the impairment of the graded Z-shield.

Fig. 3

Potential layout of the five focal plane detectors for the WFI: a small but fast DEPFET detector in the center for bright sources (approximately 7 arcmin) and four identical, large-area DEPFET detectors surrounding the central one to span the large field of view of 40 arcmin (indicated as outer circle).

An alternative solution could be to accommodate the fast sensor adjacent on the four large sensors. Depending on whether the observation requires a large field of view for large objects or a small detector with high-time resolution for very bright point sources, the respective detector will be moved in the focal plane to the on-axis position of the mirror system. This arrangement is beneficial for the placement of the front-end electronics including thermal and mechanical interfaces, for the modularity of the focal plane detector and for the implementation of the graded Z-shield.

For a sufficient oversampling of the on-axis PSF of 5 arc sec half energy width of the mirror system, the pixel size has to be matched. With a pixel size of $100\times 100\mu {\mathrm{m}}^2$, the PSF is oversampled by a factor of 2.8. Actually, the oversampling is almost a factor of two higher since the signal charge of an x-ray photon is spread over up to four pixels allowing a more accurate spatial resolution.

3.1.

DEPFET APS Concept

The APS planned for the WFI instrument use DEPFETs as signal amplifying element per pixel (Fig. 4). The signal electrons generated by the interaction of an incident x-ray photon with the fully (sideways) depleted silicon bulk are collected in an internal gate underneath the transistor channel. This increases the conductivity of the MOSFET proportional to the number of stored signal electrons and is, therefore, a measure of the energy of the incident photon. The internal gate persists regardless of the presence of a transistor current. Thus, the pixels are switched on for readout only and are switched off during the remaining time until the next readout occurs. The amount of integrated charge can then be sensed by turning on the transistor current and measuring the difference of the conductivity before and after the charge removal as depicted in Fig. 5.

Fig. 4

Schematic view of a basic circular DEPFET transistor, which is the readout element of each pixel. The signal charge generated by an x-ray photon is collected and stored in the internal gate of the DEPFET located underneath the external gate deep in the pixel bulk volume. The transistor current increases for each electron in the internal gate by a certain amount, e.g., 0.3 nA, depending on the design of the DEPFET.

Fig. 5

Signal evaluation scheme for the readout of an active pixel sensor. After the integration time, the row is switched active by the use of an external gate and the total current is measured. Then, the charge in the internal gate is cleared completely and the current is measured again to determine the baseline offset. The difference of the two measurements gives the effective current that is proportional to the number of electrons collected in the internal gate and thus to the x-ray photon energy.

The pixels are organized in matrices with one global contact for the backside. The time-dependent control voltages for switching on the transistor currents and clearing of the signal charges in the internal gates are applied row-wise, whereas the signal readout is accomplished column-wise. A 64-channel fast and low-noise analog signal processor, called VERITAS-2, will serve for this purpose. This ASIC reads 64 pixels of a row in parallel (i.e., simultaneously) and multiplexes at the same time the signals of the previous row to one output buffer. The number of VERITAS-2 ASICs is chosen according to the number of pixels per row. The readout scheme is the following: one row is switched active and read out while the others are switched off but collect the signal charge generated by incoming x-ray photons. Next, the neighboring row is switched on for readout. This rolling shutter mode is continued until all pixels of the sensor are read out and the cycle starts again with the first row.

This scheme provides a high frame rate because of the simultaneous readout of all pixels of a row. The architecture of the DEPFET matrix and of the VERITAS-2 ASIC permits flexible readout modes, too, since arbitrary sub-areas of the matrix can be selected for readout. No charge transfer to readout nodes is necessary, as would be the case for a CCD. DEPFET APS are not affected by the predominant radiation damage effect of a CCD, the decrease of the charge transfer efficiency. This implies a significantly increased radiation hardness of the sensor.

3.2.

Front-End ASICs for Readout and Control

Two different types of ASICs are needed to operate the DEPFET matrices: the readout ASIC and the switcher ASIC (see Fig. 6).

Fig. 6

Prototype DEPFET detector with $256\times 256\text{pixels}$ developed in the course of the IXO project. The switcher ASICs are placed on the left- and right-hand side, whereas the readout ASICs (here of ASTEROID type) can be seen at the bottom.

The switcher ASIC is needed to switch a selected row to active by applying an appropriate voltage to the external gates of all pixels of the row. Furthermore, the charge clear pulses for the internal gates are applied by these ASICs. Two ASICs control an array of 64 DEPFET rows.

The readout ASIC of VERITAS-2⁸ type will be tailored to the WFI DEPFETs. In addition to the typical source-follower readout, as used in the ASTEROID⁹ ASICs for the MIXS flight detectors, it provides the possibility for drain-readout. For this purpose, a low-noise current-to-voltage converter is placed in front of the preamplifier. The main advantage of the drain-readout is that all the nodes of the DEPFET are at a fixed potential and the readout speed of the system is not limited by the resistor-capacitor (RC) time constant of the input of the readout chain. This capacitance comprises the input capacitance of the preamplifier and the parasitic capacitance associated with the DEPFET source line. Coupling the ASIC in the drain-readout mode with a DEPFET array enables a higher readout speed for the DEPFET matrix compared to the source follower mode. We aim for a readout time of $2.5\mu \mathrm{s}/\text{row}$. The outputs of the analog channels of VERITAS-2 are serialized by a $64:1$ multiplexer with a clocking speed up to 32 MHz and buffered by a fast fully differential output register. While the signals of a DEPFET matrix row are read out, the signals of the previous row are fed via the output buffer to the ADC. One VERITAS-2 ASIC processes 64 DEPFET channels.

The DEPFET and VERITAS-2 architecture allows window-mode readout of the pixel matrices. Selected sub-areas of the matrix can be addressed even with the option to read out different subareas with different speeds.

The switcher ASICs for the WFI will be based on the ASICs that have been developed for the MIXS detector onboard of the BepiColombo satellite. Regarding the VERITAS-2 ASIC, a first prototype version is currently under test. A detailed description is presented in Ref. 9.

3.3.

DEPFET for Large Areas and for Fast Readout

The four large-area DEPFET matrices planned for the WFI essentially provide the field of view of 40 arc min. Each one comprises $448\times 640\text{pixels}$ with a pixel size of $130\times 130\mu {\mathrm{m}}^2$. The pixels can be slightly larger than $130\times 130\mu {\mathrm{m}}^2$ for off-axis photon incidence because the point spread function becomes worse with increasing off-axis angles. Thus a sufficient oversampling of the mirror’s point spread function is still ensured. With the projected readout time per row of $2.5\mu \mathrm{s}$ (and assuming 640 rows and 448 channels per DEPFET sensor), we obtain a whole-DEPFET time resolution of 1.6 ms.

The unprecedented effective area of the Athena mirror system will provide very high photon rates for bright x-ray sources (e.g., $\sim \mathrm{100,000}\text{count}/\mathrm{s}$ for the Crab pulsar). This necessitates very high frame rates and flexible readout modes which can be achieved by operating the $256\times 256\text{pixel}$ central DEFET matrix in a window mode of e.g., 16 rows. Due to the excellent point spread function of 5 arc sec on-axis, a pixel size of $100\times 100\mu {\mathrm{m}}^2$ ($1.8\times 1.8{\mathrm{arcsec}}^2$) was chosen for the central matrix. The matrix will be subdivided into two halves, each read out in parallel by four VERITAS-2 ASICs. With this concept, the WFI count rate capability will be optimized if the focal spot is located and maintained in a relatively precise position on this DEPFET sensor, which spreads the photons between the two readout halves. The feasibility of such a scheme and the impact on the pointing accuracy and stability shall be explored in the study phase. With the option to double the number of readout channels, two rows per half can be read out simultaneously. This allows the full chip to be read out in $160\mu \mathrm{s}$ or a 16-row window (corresponding to five times the on-axis PSF) in $10\mu \mathrm{s}$. Since this option significantly increases the complexity of the detector and its power dissipation, therefore, a detailed trade-off is necessary during the technology development phase.

The more the readout time per row approaches the photon integration time, the higher is the probability for occurrence of so-called energy misfits. They are caused by analog signal processing errors from photons that hit the active row during the signal processing (see Fig. 5). The energy misfits can be prevented by an advanced version of DEPFETs, the gateable DEPFETs. They have an electronic shutter implemented into each active pixel, allowing suppression of x-ray photons that hit the pixel during the readout process. This improves the sensitivity and spectral response of the detector. The improvement by a gateable DEPFET compared to a nongateable DEPFET is demonstrated in Fig. 7.

Fig. 7

Comparison of an ${}^{55}\mathrm{Fe}$ spectrum measured with a standard DEPFET and a gateable DEPFET (i.e., with integrated electronic shutter) for a short integration time of $15\mu \mathrm{s}$ where the occurrence of energy misfits becomes relevant.¹⁰ The gateable DEPFET shows approximately a factor of 10 less “background” because of the suppression of energy misfits caused by signal processing errors. Furthermore, the absence of absurd negative signal energies is demonstrated quite obviously.

In addition, advanced pixel layouts containing an intermediate storage region are in development. Charge generated during the readout is not lost but is collected in a region separated from the internal gate. This signal charge is read out in the next frame. The concept of gateable DEPFETs with an intermediate storage region minimizes dead time together with an increase of the sensitivity of the detector. Prototype devices have been designed, produced, and are currently being studied by testing and simulations. A detailed description of these devices is given in Ref. 10.

3.4.

WFI Detector Characteristics

The expected performance of the WFI detectors is summarized in Table 1. The characteristics are based on the planned design, measurement results of existing detectors, and simulations. The energy spectrum of a prototype gateable DEPFET detector illuminated by an ${}^{55}\mathrm{Fe}$ source is shown in Fig. 8 demonstrating the excellent energy resolution of the DEPFET, which is close to the theoretical limit given by the Fano-noise. This means the energy resolution is primarily limited by the statistical variation in the number of electrons generated in the silicon by the interacting x-ray photon. The spectroscopic performance capability of nongateable prototype DEPFET arrays has been verified in laboratory measurements. The goal is to maintain a spectral resolution with a full width at half maximum below 150 eV for a photon energy of 6 keV throughout the mission lifetime.

Table 1

Wide field imager (WFI) detector characteristics.

Fig. 8

Measurement of an ${}^{55}\mathrm{Fe}$ source spectrum with a DEPFET sensor. A value of 125 eV was obtained for the full width at half maximum (FWHM) of the 5.9 keV line.

Figure 9 shows an example for the resulting quantum efficiency of the WFI instrument that depends on the on-chip and external light-blocking filter (see Table 1). The filters are necessary to attenuate the visual and UV light intensities because the DEPFET is sensitive to these photon energies, too. In the energy band between 1 and 10 keV, the quantum efficiency is typically between 90% and 100%. The reason for this is the back-illumination of an unstructured and ultra-thin photon entrance window in combination with the full depletion of the $450\text{-}\mu \mathrm{m}$-thick DEPFET chip.

Fig. 9

Quantum efficiency of the projected WFI instrument. The black curve shows the values of the DEPFET sensor with on-chip light-blocking filter (70 nm aluminum), while the red curve shows the resulting quantum efficiency of the combination DEPFET with an additional external light-blocking filter (320 nm polypropylene and 40 nm aluminum).

The effect of the quantum efficiency on the field of view averaged effective area is shown in Fig. 10. For energies below 1 keV, the effective area is significantly decreased by the WFI, whereas for energies above 1 keV, the effective area is determined solely by the mirror system. The effective area is at its maximum between 1 and 2 keV.

Fig. 10

Effective area provided by the Athena mirror system (blue curve) and resulting effective area for the WFI instrument (red curve) considering the quantum efficiency of the detector and the absorption by the external light-blocking filter. The presented values are averaged over the field of view, i.e., vignetting is taken into account.