The AGIPD Detector for the European XFEL Julian Becker (DESY), Roberto Dinapoli (PSI), Peter Goettlicher (DESY), Heinz Graafsma (DESY), Dominic Greiffenberg (PSI), Marcus Gronewald (U Bonn), Beat Henrich (PSI), Helmut Hirsemann (DESY), Stefanie Jack (DESY), Robert Klanner (U Hamburg), Hans Krueger (U Bonn), Alessandro Marras (DESY), Aldo Mozzanica (PSI), Bernd Schmitt (PSI), Xintian Shi (PSI), Ulrich Trunk (DESY), Jiaguo Zhang (U Hamburg)

XFEL - Detector Requirements Unique time structure of the beam: 600 µs long bunch trains at a repeatition rate of 10 Hz Each train consists of 2700 bunches with a separation of 220 ns (SASE) Each bunch consists of ~10 12 photons arriving <100 fs Other specifications: X-ray flashes per sec Wavelength: nm ( keV) Peak Brilliance: ph/(s. mm 2. rad 2. 0,1% bandwidth)

XFEL - Detector Requirements Beam provides Simultaneous deposition of all photons Challenges Single photon counting not possible Approach Charge integration High number of bunches  2700 bunches per train (600 µs) Reading out of single frames impossible Analog storage on >200 storage cells Dynamic range of the detector: 0 … <10 4 ph/pixel Dynamic gain switichng  3 gain stages  Single photon counting capability for highest gain

AGIPD - Design

The AGIPD Detector 64 x 64 pixels 2 chips 8 chips 4 modules Specifications 1 Mpixel 16 modules (4 x 4)  1 module: 8 x 2 chips,  1 chip: 64 x 64 pixel Pixel size: 200 x 200 µm 2 Sensor: 500 µm thick Si Hole for direct beam Active cooling

AGIPD - Prototypes AGIPD 0.1 AGIPD 0.2 AGIPD 0.3 No pixels yet 3 readout blocks consisting of:  Readout chain (Preamp + CDS stage)  3 different kinds of leakage current compensation 16 x 16 pixels 100 storage cells No leakage current compensation Different combinations of preamps and storage cell architechures 16 x 16 pixels 200 storage cells Radiation hard storage cell design High speed serial control logic

AGIPD – Characterization AGIPD 0.1 AGIPD 0.2 AGIPD 0.3 Linearity of the gain Stress-test of the input gate at the preamp Temporal behavior of the preamp and CDS stage Energy calibration Noise determination Pixel-to-Pixel variations Storage cells variations First imaging Radiation hardness of storage cells Test of the high speed serial control logic

AGIPD Linearity of the gain (before stress-test) Linearity test using a pulser applying voltage pulses on a 11 pF input capacitance:  A voltage pulse of 1 V corresponds to an equivalent charge of ~20000 x 12 keV photons  Pulsing with 1 kHz for 1 hour: x 20k 12 keV photons Rise time of the pulse: 5 ns

AGIPD Linearity of the gain (before stress-test) High gainstage: 0.35 % (rms) 0.27 x 12 keV ph. (max) (range: x 12 keV ph.) Medium gainstage: 0.15 % (rms) 2.6 x 12 keV ph. (max) (range: x 12 keV ph.) Low gainstage: 0.13 % (rms) 11.8 x 12 keV ph. (max) (range: x 12 keV ph)

AGIPD Linearity of the gain (before stress-test)  Linearity < 1% for whole dynamic range!

AGIPD Linearity of the gain (after stress-test)

Are the pulses arriving fast enough? How does the switching look like?

Direct measurements (Preamp/CDS) High gainstage Resumee: Risetime of pulser is 5 ns Preamp and CDS are properly working within ~50 ns No switching

Direct measurements (Preamp/CDS) Medium gainstage 1x switching

Direct measurements (Preamp/CDS) Low gainstage 2x switchings

Direct measurements (Preamp/CDS) Low gainstage 2x switchings Resumee: Preamp and CDS are properly working within ~50 ns, also when gain switching 1x and 2x

AGIPD 0.2 – Energy calibration Energy calibration done using X-ray fluorescence from Ge (10 keV), Mo (17.5 keV) and Sn (25 keV) Integration time was 1  s Sensor voltage 120 V frames investigated per photon energy Ge (10 keV) Mo (17.5 keV) Sn (25 keV) Nonlog scale to demonstrate low number of fluorescence photons!

AGIPD 0.2 – Energy calibration / Noise  Noise measurements with an integration time of 100 ns reveal a value of (1.15 ± 0.11) keV, corresponding to an ENC of (318 ± 30) e -

AGIPD 0.2 – Pixel-to-Pixel variations  Pixel-to-pixel variations of the gain: ~ ± 1.9 % (rms)

AGIPD 0.2 – Imaging: the „A“

Summary AGIPD 0.1 Linearity better than 1 % for all gain stages Input gate stress-test revealed no degradation neither in gain nor in linearity after extensive pulsing with an equivalent of up to x keV photons Rise time of the preamp and CDS ~ 50 ns (within expectations) AGIPD 0.2 Noise: (1.15 ± 0.11) keV = ENC of (318 ± 30) e -  Single photon resolution demonstrated for high gain stage Pixel-to-Pixel variations: ± 1.9 % (rms) Storage Cell variations: ± 0.65 % (with simple correction algorithm): ± 0.01 % Imaging capability shown