p type bsi image sensor with active deep trench interface passivation for radiation hardened imaging systems

Peer-review under responsibility of the organizing committee of the 30th Eurosensors Conference doi: 10.1016/j.proeng.2016.11.211 ScienceDirect 30th Eurosensors Conference, EUROSENSORS

Procedia Engineering 168 ( 2016 ) 176 – 180

1877-7058 © 2016 The Authors Published by Elsevier Ltd This is an open access article under the CC BY-NC-ND license

( http://creativecommons.org/licenses/by-nc-nd/4.0/ ).

Peer-review under responsibility of the organizing committee of the 30th Eurosensors Conference

doi: 10.1016/j.proeng.2016.11.211

ScienceDirect

30th Eurosensors Conference, EUROSENSORS 2016 P-type BSI image sensor with active deep trench interface

passivation for radiation-hardened imaging systems

Bastien Mamdya,b,*, Guo-Neng Lub, François Roya

a STMicroelectronics, 850 Rue Jean Monnet 38920 Crolles, FRANCE

b Institut des Nanotechnologies de Lyon, CNRS UMR5270, Université Lyon 1, Villeurbanne, FRANCE

Abstract

We propose a P-type back-side-illuminated (BSI) image sensor for applications involving harsh ionizing environment Its pixel structure implements a hole-collection vertical photodiode and 2T PMOS shared-readout circuitry It also integrates capacitive deep trench isolation (CDTI) which allows active surface passivation The proposed pixel structure exhibits intrinsically lower level of dark current than its N-type counterpart and improved radiation hardness After receiving a total ionizing dose up to 1 kGy from 6 MeV gamma radiation, the average dark current of the P-type CDTI sensor increases only by a factor 3.6 compared with a 30-fold increase for the N-type counterpart The proposed P-type exhibits much more effective surface passivation, which prevents dark current degradation due to surface thermal generation mechanism

© 2016 The Authors Published by Elsevier Ltd

Peer-review under responsibility of the organizing committee of the 30th Eurosensors Conference

Keywords: CMOS image sensor ; vertical pinned photodiode ; hole-collection photodiode ; back-side illumination ; capacitive deep trench

isolation ; irradiation ;dark current spectroscopy

1 Introduction

For quality considerations in radiotherapy, imaging instrumentation requires radiation-hardened, small-sized image sensors There have been extensive studies on radiation effects on solid-state electronic devices, and three main degradation mechanisms from ionizing particles have been identified [1-3]:

- Creation of Si/SiO2 interface states through H+ release in the oxide layer;

* Corresponding author Tel.: +33-438923167

E-mail address: bastien.mamdy@st.com

© 2016 The Authors Published by Elsevier Ltd This is an open access article under the CC BY-NC-ND license

( http://creativecommons.org/licenses/by-nc-nd/4.0/ ).

Peer-review under responsibility of the organizing committee of the 30th Eurosensors Conference

- Positive charge trapping inside the oxide layer impacting both the threshold voltages of MOS transistors and the extension of space charge regions;

- Displacement damage and crystalline defect formation in the semiconductor bulk

In the case of CMOS image sensors, all these three mechanisms may induce dark current degradation However, the first two mechanisms affect all the pixels of the sensor array and induce widespread dark current degradation in each pixel’s photodiode, whereas the third mechanism, much more localized and randomly distributed, causes very high level of dark current (>500 e-/s*pixel at 60°C) in the affected pixels These may then become permanently white pixels (also called “hot” pixels)

Inside each pixel, dark current generation can result from different contributors that can be distinguished by temperature-varying measurements to determine their activation energy [4-6]:

- Diffusion current contribution coming from electrically neutral regions has an activation energy close to the band gap of silicon (Ea§ Eg = 1.12 eV at room temperature);

- Thermal generation (SRH) in the surface and bulk depletion regions of the photodiode has a lower activation energy (around mid-band gap or between Eg/2 and Eg);

- Dark current dominated by tunnelling from regions presenting high electric field can have an activation energy lower than half of the band gap

Tunnelling effects can be avoided by careful pixel design with consideration of operating conditions and will not be treated in this paper On the other hand, surface thermal generation is minimized in state-of-the-art image sensors by implementing surface passivation

Most image sensors are of N-type with pixel’s photodiode collecting photo-generated electrons Accordingly surface passivation around the photodiode is implemented by enhancing population of holes at the Si/SiO2 interface However, in radiation conditions, positive charge trapping in oxide will push away holes from the interface, thus making the surface passivation ineffective To counter this effect, we propose here a P-type image sensor with a pixel’s photodiode based on the collection of photo-generated holes In that case, Si/SiO2 interface passivation is carried out by the accumulation of electrons at the interface The positive charge trapping from irradiation is expected to naturally hold and even increase the population of surface electrons Our proposed P-type image sensor

is a back-side illuminated (BSI) one integrating active deep trench interface passivation

2 Pixel architecture

Figures 1 a) and b) show respectively cross-sections from 3D-TCAD simulations and transmission electron microphotographs of the proposed pixel structure The back-side illuminated pixel integrates a hole-collection-based vertical photodiode Electrostatic pinning of the P-type photodiode is ensured by N-type implantations in order to fully deplete the photodiode during its reset phase The vertical doping gradient within the photodiode creates a built-in stair-like electrostatic potential which guaranties an efficient charge transfer towards the floating diffusion node during the transfer phase of the readout sequence Pixel-to-pixel electrical and optical isolation is ensured by the recently-developed capacitive deep trench isolation (CDTI) [7] Unlike oxide-filled deep trench isolation (DTI), CDTI can be electrically controlled to accumulate majority carriers (electrons in our case) and pin the Fermi level close to the conduction band at its Si/SiO2 interface Hence, CDTI allow active electrical passivation of the interface states of the side walls The sensor adopts a 2T shared-readout circuitry composed of PMOS transistors comprising a transfer gate (TG), a source follower transistor (SF) and a reset transistor, as shown in Figure 1c

Fig 1 (a) Cross-section from a 3D-TCAD simulation of the studied pixel architecture (b) Transmission electron microphotograph of the

fabricated pixel (c) Schematics of the 2T architecture

3 Experimental Results and discussion

We have evaluated the radiation hardness of the proposed P-type image sensor, in comparison with its state-of-the-art N-type counterpart The N-type image sensor considered here as reference has similar pixel architecture with

an opposite doping type: BSI, N-type vertical photodiode with NMOS readout circuitry Pixel-to-pixel isolation is in this case ensured by standard oxide-filled DTI

Both types of sensor have been irradiated simultaneously at ambient temperature using a 6 MeV gamma ray beam from a Novalis Truebeam STx Varian linear accelerator from the service of radiotherapy of the Lyon Sud hospital

In order to maximize the delivered dose with accuracy, the sensors were placed at a “build-up” depth (under a 1.5

cm thick PMMA plate) during the irradiation and dose rates were carefully calibrated at 225 mGy/s No voltages biases were applied on the sensors during irradiation and they were stored at ambient temperature after irradiation Fig 2 compares the average dark current degradation between the two sensor types as a function of total ionizing dose

c)

+2.0e20

-2.0e20 0.0e00 Doping conc.

cm -3

Fig 2 Average dark current degradation as a function of the total ionizing dose (TID) absorbed for both sensors

The reference N-type DTI sensor deteriorates much faster than the P-type CDTI one After 1 kGy, the average dark current of the reference array has increased by a factor of 30, while that of the P-type CDTI sensor only rises by a factor of 3.6

Spectral analysis of dark current distributions on both types of sensor has also been performed Figure 3a) represents the dark current distribution measured at 60°C on the P-type CDTI sensor Before irradiation the distribution can be separated into three separate populations: the main population of pixels presenting a low level of dark current (peak #1) and two other distributions around 120 h+/s and 240 h+/s respectively These two smaller distributions are associated with tungsten contamination introduced during the fabrication process of the sensor and are therefore not relevant for the current analysis of irradiation damage Similarly, figure 3b) shows the dark current distribution measured at 60°C for the reference N-type DTI sensor Peak #1 of the distribution is once again attributed to the main population of pixels and the absence of other peaks in the spectrum before irradiation indicates that the sensor is free of metallic contamination In both cases, the activation energy associated to the peak #1 has been extracted and is approximately equal to the band gap (Ea § Eg) This means that both sensor types have effective surface passivation before irradiation, and that the remaining major dark current contribution is attributed

to a diffusion mechanism

After irradiation, the level of the main population of the P-type CDTI sensor is not affected by ionizing particles, while that of the main population of the N-type reference sensor drifts up to 600 e-/s This confirms the radiation hardness of the P-type thanks to its surface passivation remaining effective It can also be seen that for both types of sensors, the tail of the distribution increases with the ionizing dose Such behavior is characteristic of a new degradation mechanism affecting only a small percentage of the total population of pixels like for instance the appearance of crystalline defects within the depletion region By combining the annealing temperature of the defects and their activation energy we were able to precisely identify them

4 Conclusion

We have proposed a P-type BSI image sensor with active deep trench interface passivation for radiation-hardened imaging systems After receiving 1 kGy dose from 6 MeV gamma radiation, its average dark current increases only

by a factor 3.6 compared with a 30-fold increase for the N-type counterpart Dark current spectral analysis has shown that, in contrast with N-type reference, the surface passivation in the P-type sensor remains effective after heavy dose irradiation

Acknowledgements

The authors would like to thank Patrice Jalade and his team from the Lyon Sud hospital for their help during the irradiation of the sensors, the entire OCSI team at STMicroelectronics for their guidance during the analysis of the measurements as well as the process integration and process development teams for their support during the fabrication of the sensors

References

[1] F Barry McLean, T R Oldman, Basic mechanisms of radiation effects in electronic materials and devices, Washington DC, 1987

[2] M Moll, Radiation damages in silicon particle detectors – microscopic defects and macroscopic properties, Hamburg 1999

[3] V A J van Lint, Mechanisms of radiation effects in electronic materials, Vol 1, University of California, 1980

[4] S.M Sze, K K Ng, Physics of semiconductor devices, Wiley, 2006

[5] W Shockley, W T Read, Statistics of the recombinations of holes and electrons, Phys Rev., Vol 87, pp 835-842, 1952

[6] R N Hall, Electron-hole recombination in germanium, Phys Rev., Vol 87, pp 387, 1952

[7] N Ahmed, MOS capacitor deep trench isolation (CDTI) for CMOS image sensors, INL, Lyon, 2015