Radiation Effect on MOS-Structure Q. Wei, L. Andricek, H-G. Moser, R. H. Richter, Max-Planck-Institute for Physics Semiconductor Laboratory DPG-Tagung, Heidelberg, 2007

Outline Motivation Radiation damage on MOS-C and -DEPFET Experimental Conditions and Results Conclusion

Motivation-ILC(DEPFET) Goal: Higgs, SUSY, Identity of dark matter, Existence of extra dimension  precise vertex detector (DEPFET) Up to now so many central topics like higgs particle, SUSY, and existence of extra dimensions should be studied, therefore, ILC become more and more interested, and this project are proposed also together with the LHC at CERN, it will allow us to research the physics topic in the high energy regions. In the interaction point of this huge construction there is a set of cylindrical detectors, so called TESLA TDR design. It give us a basic concept how it consists of. And for the particle tracking we need such a high resolution of detector both for energy but also for position measurement. At the same time the detectors should be radiation hardness, because radiation induced damage will deteriorate the signal noise ratio and finally make the detector off. So my work is to study this radiation effect especially with x-ray on the MOS-structure to get a radiation hardness detector design for ILC, since MOS-structure is simple to produce and has already been a very important component of so many semiconductor devices, in my presentation I will show you three different devices, MOS-C, MOS-DEPFET and gated diode. A set of cylindrical detectors in each ladder around the interaction point each ladder is an array of pixel cell R/O at the end of the ladder outside the sensitive volume Particle tracking  vertex detector at ILC High resolution:~5µm: technology so far possible Pixel size: 20-30 µm Low mass: 0.1 %Xo per layer: thin detectors ≈ 50µm instead of 300µm High occupancy in 1st layer: fast(!) R/O cycle ~ 50MHz per Pixel Small noise: noise ≈ 100 el ENC TESLA TDR Design

Radiation Damage Two types of radiation damage in MOS-Structure: Surface damage due to Ionizing Energy Loss (IEL) accumulation of charge in the oxide (SiO2) and Si/SiO2 interface Oxid charge  shifts of flat band voltage, (depleted  enhancement) Interface traps  leakage current, degradation of transconduction,… Bulk damage due to Non Ionizing Energy Loss (NIEL) displacement damage, built up of crystal defects Increase of leakage current  increase of shot noise,… Change of effective doping concentration  higher depletion voltage,… Increase of charge carrier trapping  singal loss! S/N Ratio deteriorated! Now let us talk about the radiation effect, after radiation there are some radiation damage generated in the MOS-structure. There are two types of radiation damage, bulk damage and surface damage. Due to ionizing energy loss while the x-ray go through the oxide there are some electron hole pairs generated in the oxide, since electron has a fast mobility compared to hole in oxide and go directly to the electrode under the influence of electric field. After the separation of electron and hole pairs the hole drift to interface and are trapped in hole trap near the interface, and finally form oxide charge also interface trap, oxide charge contributes a shift of flat band voltage, which will make MOS-transistor from depletion mode to enhancement mode, and increases the power consumption. Interface trap results in leakage current and make the signal noise ratio worsen. On the other hand due to non inozing energy loss especially for hadron's radiation the bulk damage will be formed in the silicon bulk through displacement with silicon lattice, the macroscopic effect is changing of effective doping concentration and increasing of leakage current and also trapping effect. All of these follows a degradation of Signal noise ratio.

Radiation Effects - Bulk damage displacement due to NIEL: proton & neutron irradiation(simulation): point defects & clusters clusters Vacancy Frenkel Defect Interstitial Silicon atom Different atom in Interstitial site Different atom in Subinterstitial site Let us talk short about bulk damage. After hadron's radiation there are some displacement damage like vacancy, interstitial, or displaced atom with impurities and so on. Depend on type and energy of radiation there are different damage generated, for proton radiation, due to columbic and also nuclear interaction there are not only point defects but also clusters formed. But for neutron radiation, because it is not charged particles, so there is no columbic interaction, so there is normally only cluster. From the macroscopic point of view we could see the leakage current increase with higher radiation fluence, and the effective doping concentration is changed from n-type to p-type at about 2 times 10 to 12 per square cm. increase of leakage current: change of effective dopping concentration ILC ILC

Radiation Effects - Surface damage Ionizing radiation: Damage mechanismus: Gate Electron/Hole Pairs Generated by ionizing radiation SiO2 Si 1.12ev Hopping Transport of Holes through localized states in SiO2 Bulk Long-term Hole trapping near Si/SiO2 Interface Radiation induced Interface trap Si Electric active Interface state Let us give you a comprehensive view of generation of surface damage. Through ionizing radiation the generated electron hole pairs are separated in the oxide bulk and go to corresponding direction according to the applied bias voltage. E.g. positive bias voltage, hole escape the recombination process in the first stage with generated electrons, and through the so called hopping transport drift towards the interface, and form positive oxide charge there. According to Mclean model so called H-model, the radiation induced hole will combine with hydrogen atom or hydrogen complex to create proton, which act with silicon dangling bond to form interface trap and release hydrogen molecule. As mention above, the macroscopic effect is the shift of flat band voltage and stretch-out of cv-curve. Shifts of flat band voltage: ~ Nox Strechout of CV curve: ~ Nit Nox: postive oxide charge and postively charged oxide traps have to be compensated by a more negative gate voltage  negative shift of the theshold voltage (~tox2) Nit: increased density of interface traps  higher 1/f noise and reduced mobility (gm)

Saturation effect for surface damage Saturation mechanism equilibrium Begin Saturation Recombination Recombination Rejection Trapp filling Before I show you the results, I will discuss here about the saturation effect for surface damage. At the beginning some positive oxide charge are formed near the interface, and more and more positive oxide charge are accumulated there, and results in a additional electric field near the interface, that push the new generated and towards interface drifted hole away from the interface to electrode. At the same time, some electrons will go through tunnel effect from silicon into interface and recombine with positive oxide charge there, finally a equilibrium is built up, and saturation of surface damage will come into being. We call the hole traps in total as reservoir. But here the reservoir is still not exhausted, until a very larger electric field applied on the gate contact, to suppress the recombination process and repulsion from accumulated positive oxide charge near the interface. SiO2 Si Reservoir: hole traps are not exhausted, unless a larger bias voltage is applied on the gate! Saturation: equilibrium between trapped filling and recombination Generated holes are pushed away! Recombination of trapped holes with electrons - Recombination of tunneled electrons from silicon into interface with trapped holes!

Experiment Conditions and Methods Irradiation (X-Ray): Co60 (1.17 MeV and 1.33 MeV) GSF – National Research Center for Environment and Health, Munich CaliFa (17.44 KeV) Max-Planck-Institute Semiconductor Labor, Munich Roentgen facility (20 KeV) Research center, Karlsruhe Dose: irradiation up to 1 Mrad with different dose rate (1rad=0.01J/kg) Process: No annealing during irradiation ~ irradiation duration from 1 day to 1 week Radiation levels at the ILC VTX: Dionization≈ 100 .. 200 Krad ≈ 1010 .. 1011 neq(1MeV )/cm2 Here I will introduce you about my experiment conditions and some analyze methods for each component. Irradiation what I used is x-ray at different energy source. Co60, califa setup in our house, also roentgen facility in research center karlsruhe, I irradiated each sample to 1 Mrad with no annealing process in the radiation period. E.g. for radiation environment of ILC for vertex detectors, it is 100krad in 5 years, about 10 to 10, 11 neutron equivalent. For MOS-DEPFEET subthreshold technique is used to calculate the corresponding parameters. For MOS-c CV-measurement, and for gated diode, it is gated diode technique. And some other parameter like transconduction, surface generation velocity, capture cross section also generation lifetime are important for characterizing surface damage. Comparison of different semiconductor devices DEPFET MOS-C Gated diode Nox (method) Δ Vt (IV-Measurement) ΔVFB (CV-Measurement) ∆VFB & ∆Vg (CV-Measurement & gated diode technique) Nit Subthreshold slope (Subthreshold technique) Stretchout (High-low frequency based on the CV) Full width at 2/3 maximal of current (gated diode technique) Other parameters gm S0,

Depleted P-channel FET MOS-DEPFET Drain Source Gate Now i want to show you more something about the MOS-type DEPFET. The MOS-type DEPFET is integrated on a fully depleted silicon substrate. All charge generated by a traversing particles drift to the surface and are collected in the potential minimum for electrons underneath the external gate of DEPFET, here this region is so called internal gate.the charge cloud is stored and modulates the current in the channel of FET which forms the first amplification stage. After the signal readout, the charge will be removed by applying a positive voltage to the clear gate. Charge generation and first amplification stage are combined in one single device, and due to fully depleted sensitive volume in the whole bulk also low capacitance of read out node we could get low noise operation over a large range of temperatures. This is exactly for the application at linear collider experiments. Another important thing is the charge is measured at the place of generation, this means trapping effects due to bulk damage by NIEL is not so relevant. Depleted P-channel FET fully depleted sensitive volume Charge generation and first amplification in a fully depleted pixel cell  good S/N No charge transfer needed R/O at the place where it generated  better rad. tolerance against hadronic irradiation

Bias during irradiaton: Results for MOSDEPFET Bias during irradiaton: 1: empty int. gate, in „off“ state, VGS= 5V, VDrain=-5V  Eox ≈ 0 2: empty int. gate, in „on“ state, VGS=-5V, VDrain=-5V  Eox ≈ -250kV/cm "OFF" "ON" Now let us take a look at the results for MOSDEPFET, if bias status is on, we get more radiation damage compared to off status. The threshold voltage shift is about 4v. And after radiation the induced damage will be healed up in the first tens of hours for on status, but almost no rebound for off status.

Transconductance and Subtreshold slope s=85mV/dec s=155mV/dec Vth=-0.2V Vth=-4.5V Using subthreshold technique we could calculate the interface trap concentration. From our results we could hardly see any degradation of transconduction. No change in the transconductance gm 300 krad  Nit≈2·1011 cm-2 912 krad  Nit≈7·1011 cm-2 Literature: After 1Mrad 200 nm (SiO2): Nit ≈ 1013 cm-2

MOS-Capacitance MOS-C: the upper left – Al1 the upper right – Al2 the lower left –Poly1 the lower right –Poly2 each contact area is about 10 mm2 Al Al2 In this picture you will see the measurement setup exactly. Poly1/2 Aluminium Poly1/Poly2 LTO Substrate Substrate Oxide Ni

Saturation reaches at about 1 Mrad! Results for MOS-C Gate Bias conditions: 0V These are for MOS-C, again the same from the flat band voltage shift we achieve the information about the oxide charge, and using high-low method we get the interface trap. high-low frequency CV  Nit Flat band voltage Shift Nox Saturation reaches at about 1 Mrad!

+ - Bias Effect: DEPFET thickness dependence dn VG SiO2 Si For MOS-C - SiO2 Si VG DEPFET Here I will present you two effects, bias effect and thickness dependence. Under positive bias condition, more hole will be generated and drifted towards interface in comparison with 0 and negative bias, because for the case of 0v the radiation induced hole could recombine with electrons directly after their generation, and this recombination process become dominated. Under negative bias, the recombination process is already avoided, however, holes drift toward electrode, where there are less hole traps. after we irradiated sample with different oxide thickness, we find that there is correlation between flat band voltage shift and radiation dose. This means if we reduce the oxide thickness, we could get much better radiation hardness. Of course this relationship depends on the applied bias condition. For positive electric field, this n is equal to approximately near 2, and for 0v electric field, n is equal to 1. thickness dependence dn

High dose rate Low dose rate For MOS-C Dose rate effect: Radiation hardness by Ni-layer: Substrate n-Typ Poly1/Poly2/LTO Aluminium Oxide Ni Another two effects is dose rate effect and radiation hardness factor. Due to ionizing radiation, holes are generated and drift towards interface and form there positive oxide charge, if the generation time of hole is faster than transport time of the hole to the interface, there are more positive charge formed in the oxide bulk, therefore, electrons generated above this charge sheet drift toward it and holes move to electrode, on the contrary, holes generated in the regions between this charge sheet and interface will drift to interface, all of this results in the reduction of radiation reduced holes. In comparison with low dose rate, there is no such charge sheet in the oxide bulk, so the flat band voltage for low dose rate is higher than for high dose rate, however, for much lower dose rate, in my work the dose rate for the radiation source Fe55 is predominantly small, so the radiation time is even slower than annealing time. So its flat band voltage is shown between these two dose rate. For the topic of radiation hardness we find that if we produce a additional layer of Nitride between metal electrode and silicon, it serves not only as a mask against hydrogen diffusion from outside into oxide but also as a good mask for generation of positive oxide charge. So we get much less flat band voltage shift for the sample with Nitride layer and with the same oxide thickness compared with another sample which is without Nitride layer. High dose rate SiO2 Si Low dose rate

MOS-Gated Diode gated diode wafer the left one – Al the right one –poly each area is 9 mm2 P+ Al/PolyGate A Al Substrate V Edge N Erde Poly-gate Al-gate Al Al P+ P+ grounding Edge

Results for gated diode Shifts of generated current: Increase of maximal current: These are the results for gated diode. Under bias voltage applied on the gate contact, the gated diode goes through accumulation, depletion and inversion. In the depletion stage, you will see this generation current, which shifts and become much higher with increasing dose. Another two parameters generation velocity and generation lifetime also reflect the radiation effect, namely much more reverse current is built up due to the interface trap. Increase of surface generation velosity and decrease of lifetime due to radiation:

Oxide charge decrease with time: (Tunnel annealing @ RT) For bulk damage: Leakage current decrease with time: “Beneficial annealing” & “Reverse annealing” 80 min 60C Another important radiation hardness factor is annealing process, that is not only for bulk damage but also for surface damage. Annealing for bulk damage is more useful at higher temperature, since defect could change to another states through migration or dissociation. For surface damage at RT, tunnel annealing is dominant effect for positive oxide charge, which is proportional with logarithms time, also at higher temperature the surface damage will be recovered in a certain extend. For surface damage: Oxide charge decrease with time: (Tunnel annealing @ RT)

Conclusion Radiation experiment Radiation hardness Study of saturation effect for surface damage(equilibrium between generation and recombination) Study of bias dependence(“+“,“0“,“-“) Study of different semiconductor devices, which are with different oxide thickness - to see the kind of relation between Vfb and oxide thickness(doxn ) Study of dose rate effect(three different radiation sources) Study of devices, with different structure profile(poly-gate with Ni; al-gate without Ni) Radiation hardness Reduced oxide thickness improves radiation hardness Additional Ni layer serve as a good protection against ionizing radiation Annealing process makes not only surface damage but also bulk damage reduced with time Now we come to conclusion.

Thanks for your Attention

Back slides

Prototype Readout System DEPFET Matrix 64x128 pixels, 36 x 28.5µm2 Clear Switcher Gate Switcher let us take a look at a prototype readout system, on the left side, the steering chip for row-wise readout switcher, and on the right hand side clear switcher, in the middle of this picture, DEPFET matrix, under the DEPFET, front end chip curo. Current Readout CUROII PCB with DEPFET matrix r/o board with ADCs etc. Charge collection in "off" state Only selected rows dissipate power low power consumption, expect about 3W for the whole vertex detector! USB based digital interface board

DEPFET Project Status - in Summary thinning technology steering chips Switcher II Technology development tolerance against ion. radition r/o chips Curo II 55Fe beam test

Annealing for MOS-C (PXD4)