Highly Ultraviolet Light Sensitive and Highly Reliable Photodiode ...

R11

Highly Ultraviolet Light Sensitive and Highly Reliable Photodiode with Atomically Flat Si Surface Rihito Kuroda, Taiki Nakazawa, Katsuhiko Hanzawa and Shigetoshi Sugawa Graduate School of Engineering, Tohoku University 6-6-11-811, Aza-Aoba, Aramaki, Aoba-ku, Sendai, Miyagi, Japan 980-8579 TEL: +81-22-795-4833, FAX: +81-22-795-4834, Email address: [email protected] Peak to Valley > 1 nm

ABSTRACT Highly UV-light sensitive and highly reliable FSIphotodiode is demonstrated in this work. Using the atomically flat Si surface to uniformly form the thin surface drift layer, a photodiode exhibiting the almost 100% internal Q.E. to UV-Visible-Near IR light and a high stability to UV-light is obtained. The developed photodiode process is compatible to the current manufacturing processes for FSI-photodiodes, arrayed sensors and image sensors. -

INTRODUCTION AND PROPOSED PHOTODIODE CONCEPT

Depth from Si Surface

Ultraviolet light (UV-light) sensing and imaging are widely required in the fields of biological phenomena analysis, space application, environmental assessment and various types of spectrophotometric analyses [1-4]. Highly UV-light sensitive and highly reliable photodiodes and arrayed sensors as well as, a stable and cost effective manufacturing process suitable for mass production are therefore strongly desired. Fig. 1 shows the depth from Si surface where the amount of incident light decreases to 90 % and 37 % due to the absorption as a function of the wavelength, calculated by the absorption coefficient in Si. Due to its high absorption coefficient, most of the UV-light in Si is absorbed within the top few atomic layers. Because of this, conventional buried photodiodes used in the current image sensors do not have any UV-light sensitivity. To increase the UV-light sensitivity, the atomic scale control of the surface dopant profile to form the surface drift layer is fundamentally required. However, the conventional fabrication process induces an unsuitable large roughness to Si surface, as displayed in Fig.2, where the AFM images are shown for an atomically flat Si(100) surface and a typical Si(100) surface after the RCA cleaning with relatively low ammonia concentration in ammonia-peroxide solution [5]. As compared to the atomically flat surface, the conventional surface has much large roughness with peak to valley (P-V) of larger than 1 nm. 100μm 10μm 1μm 100nm

Integrated amount of light decreases to 90% 37% 0.135 nm

10nm (1/e) 1nm

Si atoms

Atomic layers of Si(100) 1st 2nd 3rd

0.1nm 200 300 400 500 600 700 800 900 1000 1100

Wavelength [nm] Fig. 1 Depth from Si surface where the amount of incident light decreases to 90% and 37% due to the absorption as a function of wavelength.

(a)

(b)

Fig. 2 AFM images of (a) an atomically flat Si(100) surface and (b) a typical Si(100) surface after RCA cleaning.

In addition, due to the high photon energy of UV-light, stability to light exposure is indispensable for the long time use of UV-light sensors [6-7]. Recently, an UVlight sensitive BSI-CCD using highly doped Si molecular beam epitaxy on the back surface was reported [8-9]. In this work, we demonstrate a highly UV-light sensitive and highly reliable FSI-photodiode using the atomically flat Si surface, compatible to the current photodiode and image sensor fabrication processes. Using the technologies to atomically flatten the Si surface and preserve the flatness throughout the fabrication [10-12], buried photodiodes were fabricated with uniformly formed few nanometer-thick surface high concentration layers. This surface high concentration layer induces an electrical field that drifts the photogenerated carriers to the buried photodiode without the recombination at the interface states, leading to the high sensitivity characteristic to UV-Visible-Near IR light. -

EXPERIMENTAL SETUP The atomically flat surface is obtained due to the migration of Si atoms by annealing an off-anglecontrolled Si wafer in pure Ar ambient at 850oC or a higher temperature [10, 12]. The modified process technologies of wafer surface cleaning and insulator film formation were employed as summarized in Fig.3 [1011]. By introducing all of these technologies, the atomically flat SiO2/Si interface is able to be formed. Fig.4 shows the AFM images and cross section profiles of the atomically flat Si surfaces. Surfaces are composed of atomic terraces and atomic steps with the height equals to one atomic layer of Si(100): 0.135nm. The surface step morphology can be controlled by the wafer off angle and direction. As illustrated in Fig.5, it is considered that with the conventional flatness, the thin surface drift layer is not uniformly formed and many spots appear where the carrier recombination and generation occurs with very short relaxation time constants due to the interface states, On the contrary, with the atomically flat surface, the thin surface drift layer is uniformly formed and these sensitivity loss and dark current generation spots do not appear. The n+pn buried photodiodes were fabricated on Cz-n Si(100) wafers (dopant concentration: 1x1015cm-3) with the atomically flat and the conventional flat surfaces.

Cz-n wafer, ND=1x1015 cm-3

Cz or Epi wafers

Flatness Control

Native Oxide Removal

Atomically Flattening Anneal or RCA cleaning (Conventional)

Diluted HF solution

Through Implantation Oxide Formation: 7.0 nm

Si Surface Atomically Flattening

Buried p-Si Layer Implantation

Pure-Ar anneal, 850oC~1100oC H2O residue gas less than 20 ppb

B+, 15 keV, 3.4x1012 cm-2

Boron Activation Anneal

Si surface cleaning

RTA, 950oC, 5 sec

No alkali solution, dark condition

Surface n+ layer implantation As+, various implantation conditions

Insulator Formation

Interlayer Formation: NSG 190 nm

Oxygen radical oxidation, 400oC Microwave excited high density plasma equipment

Contact Hole Formation Contact Hole p+-region Implantation

Fig. 3 Process technologies to form the atomically flat Si surface and to preserve the flatness during the fabrication.

Activation Anneal Spike-RTA, 950oC

Al Metallization

Fig. 6. Photodiode fabrication flow. AL

Photo diode 200nm

SiO2

0~30nm

n+

p+

AL

p 250nm

0.135 nm 2500.0

(a)

(a) = 0.058o 400.0

600.0

(b)

800.0

20

10

(b)

Fig. 4 (a) 3x3 μm2 and (b) 1x1 μm2 AFM images and cross section profiles of Si(100) surfaces after the flattening process.

Interface state

AL (light shield)

Fig. 7 Schematic images of the fabricatd n+pn photodiodes, (a) the cross sectional veiw and (b) the birds-eye view.

position [nm]

position [nm]

Atomically Flat or Conventional Flat Interface

+

As implantation Nissin, medium-current equipment

-3

2000.0

1.20 0.90 0.60 0.30 Off angle 0.00 0.0 200.0

As Conc. [cm ]

0.80 0.60 0.40 0.20 Off angle = 0.008o 0.00 0.0 500.0 1000.0 1500.0

height [nm]

height [nm]

Cz-n

Through Implant. Oxide: 7.0 nm 19

10

13

-2

10 keV, 3.4x10 cm

18

3 keV, 13 -2 6.8x10 cm

10

17

SiO2

Interface state

10

SiO2

0.0

10.0

20.0

30.0

40.0

Depth from Si Surface [nm] As+

As+ p-Si

(a)

Fig. 8 Profiles of As from the Si surface measured by SIMS Sensitivity Loss, for wafer numbers 2 and 4 in table 1. Dark Current Increase Table 1

p-Si

(b)

Fig. 5 Schematic images of the cross section of surface thin n+ layer formed on (a) the atomically flat and (b) the conventional flat Si surfaces, respectively.

Figs.6 and 7 show the fabrication flow and structure of the photodiodes. The buried p-Si layer has dopant concentration of 2x1017 cm-3 and thickness of 250 nm, while the surface n+ layer thickness is varied from 0 to 30 nm for different wafers. In this work, the surface drift layer was formed by the As+ implantation. As+ implantation is considered to be advantageous because of its heavy mass, high solubility and small diffusion constant in Si and tendency to segregate at the Si/SiO2 interface during the activation anneal. The As+ implantation conditions were varied for wafers to examine the impact of the profile and the thickness of surface n+ layer to the characteristics of fabricated photodiodes. Typical depth profiles of As measured by secondary ion mass spectroscopy (SIMS) are shown in Fig.8. Table 1 summarizes the fabrication conditions and the extracted surface n+ layer thicknesses.

Fabicated device condtions No.

Flatness

1 2 3 4

Conventional Flat Atomically Flat Conventional Flat Atomically Flat

As+ implantation Dose Energy [keV] [cm-2] 3 3 10

0 6.8x1013 6.8x1013 3.4x1013

Surface n+ layer thickness [nm] 0 3.5 3.5 30.0

The dark current, photo-current and quantum efficiency (Q.E.) as well as, Q.E. and dark current stability to UV-light were evaluated using the on-wafer characterization system shown in Fig.9. The measured range of wavelength is 200-1000 nm. Fig.10 shows the characteristic of the UV-light source used for the UVlight exposure stress. The super high pressure mercury discharge lump was employed. The typical UV-light intensities are 2.0, 4.4, 8.8 and 17.6 mW/cm2 for λ = 254, 303, 313 and 365 nm, respectively. The stability to UVlight was evaluated up to the exposure time of 1000 min while the photodiodes were either reverse biased at -2.0 V or floating during the exposure. The total amount of the light exposure after 1000 min are 1.2x102, 2.6x102, 5.3x102 and 1.1x103 J/cm2 for λ = 254, 303, 313 and 365 nm, respectively.

2

Current Density [A/cm ]

-2

(a) optical fibers Monochromator Shimadzu SPG-120UV

25.0 2

Light Intensity [μW/cm ]

Light source mirrors ob e

Hamamatsu pr

L10290

I-V measurement

wafer

20.0 15.0

-2

200 300 400 500 600 700 800 900 1000

Wavelength [nm]

0.8

0.8 0.6

8.8 mW/cm at 313nm

2

Ushio SuperHigh-Pressure UV Lump, USH-250SC

2

Junction Area: 1.0x10 cm Vpn = -2.0 V

0.6 0.4

100% Internal Q.E. (Calculation) + 13 -2 Atomic. Flat, As 10keV, 3.4x10 cm + 13 -2 Atomic. Flat, As 3keV, 6.8x10 cm + 13 -2 Conv. Flat, As 3keV, 6.8x10 cm + Conv. Flat, without As implantation

0.2

4.4 mW/cm at 303nm 2

0.2

-2

0.0 200 300 400 500 600 700 800 900 1000

2

0.4

2.0 mW/cm at 254nm

Wavelength [nm] Fig. 12 Measured external Q.E. as a function of the wavelength.

0.0 200 250 300 350 400 450 500 550 600 650

Fig. 10 Characteristic of the UV-light source used for the UV-light exposure stress.

RESULTS AND DISCUSSION Fig.11 shows the measured pn junction J-V characteristics of the fabricated photodiodes with and without light. The absolute value of the dark current is relatively high due to the unoptimized process of the pn junction formation. For the atomically flat device with 10keV As+ implantation, the reverse current density is smaller than without n+ layer. Higher current density for devices with 3keV As+ implantation indicates excess interface states creation during the through oxide As+ implantation. Fig.12 shows the measured external and internal Q.E. as a function of the wavelength. The 100% internal Q.E. is calculated by the n and k values of 200nm-thick SiO2 film formed above the photodiodes. With the atomically flat Si surface, very high Q.E. is obtained for 200-1000nm, while with the conventional flatness or without surface n+ layer, Q.E. in UV-light range is very low. It is considered that the slight decrease of the Q.E. for wavelength longer than 400 nm is due to the shallow p-Si layer (250nm). This will be overcome by the optimization of p-Si layer thickness. Fig.13(a-b) shows the dark current and external Q.E. at 250 nm as a function of the surface n+ layer thickness. The impact of the atomically flatness is clearly confirmed in the average values and the variations when the n+ layer thickness is 3.5 nm, indicating that the atomically flat Si surface can enlarge the process margin for the formation of suitable surface drift layer.

-6

10

0.8

Conventional Flat

2

Wavelength [nm]

Dark Current [A/cm ]

Relative Light Intensity [%]

Fig. 9 (a) picture and (b) schematic images of the on-wafer dark current and photo-current measurement system. 2

13

1.0

5.0

4156B

17.6 mW/cm at 365nm

-2

+

0.0

1.0

-2

13

+

External Q.E.

(b)

13

+

Vpn [V] Fig. 11 Measured pn junction J-V characteristics of the fabricated photodiodes with and without light.

Light Source: Hamamatsu L10290

10.0

Agilent chuck stage

+

-7

10

-8

10

-9

10

Atomically Flat +

n pn diode -2 2 Junction Area: 1.0x10 cm Vpn = -1.0 V

External Q.E.

optical filter

10 -2 2 λ=250nm, -3 Junction Area: 1.0x10 cm Atomic. Flat, As 10keV, 3.4x10 cm 2 10 22.1 μW/cm Atomic. Flat, As 3keV, 6.8x10 cm -4 10 Conv. Flat, As 3keV, 6.8x10 cm with light -5 Conv. Flat, without As implantation 10 -6 10 -7 10 -8 10 -9 10 -10 Dark 10 -11 10 -2.0 -1.5 -1.0 -0.5 0.0 0.5

0.6 0.4

+

n pn diode, -2 2 Junction Area: 1.0x10 cm Vpn = -2.0 V

λ = 250 nm

Atomically Flat

0.2

Conventional Flat 0.0 0 5 10 15 20 25 30 35 0 5 10 15 20 25 30 35 (a) Surface n+ Layer Thickness [nm] (b) Surface n+ Layer Thickness [nm] 10

-10

Fig. 13 (a) dark current and (b) External Q.E. at 250 nm as a function of the surface n+ layer thickness. 5 samples were evaluated for each condition.

Figs.14 and 15 show the external Q.E. as a function of the wavelength measured at various UV-light exposure time and the internal Q.E. at 250 nm as a function of the UV-light exposure time up to 1000 min, respectively. The photodiodes were either reverse biased at -2.0 V or floating during the UV-light exposure. The fabrication conditions are shown in the figures. For the atomically flat device with As+ implantation condition of 10keV, 3.4x1013 cm-2, almost no degradation occurred for the UV-Visible-Near IR light, and the almost 100 % internal Q.E. is maintained throughout the UV-light exposure time. With As+ implantation condition of 3keV, 6.8x1013 cm-2, the atomically flat device initially shows the almost 100 % internal Q.E., however a large degradation of Q.E. to UV-light occurred after 10 min. For the conventional flat device, Q.E. to UV-light is low at the all measured times, and a slight degradation is observed in the UVlight range. For each device, the behaviors of the Q.E. to the UV-light exposure are almost the same between the two bias conditions during the UV-light exposure. -

0.6 0.4

100% Internal Q.E. Initial 10min UV-light 100min exposure 1000min

0.2

0.0 200 300 400 500 600 700 800 900 1000

(a)

Wavelength [nm]

External Q.E.

1.0 0.8

+

13

Atomically Flat, As 10keV, 3.4x10 cm during exposure: Floating

0.2

0.8

1.0

-2

0.6 0.4

100% Internal Q.E. Initial 10min 100min UV-light 1000min exposure

0.2

(b)

Wavelength [nm]

1.0

100% Internal Q.E. Initial 10min UV-light 100min exposure 1000min

13

0.0 200 300 400 500 600 700 800 900 1000

-2

0.6 0.4

+

Atomically Flat, As 3keV, 6.8x10 cm during exposure: Vpn = -2.0 V

0.0 200 300 400 500 600 700 800 900 1000

+

13


0.2

0.0 200 300 400 500 600 700 800 900 1000

13

-2

0.6 0.4


0.2

(c)

1.0

0.6 0.4

0.8

+

Conventional Flat, As 3keV, 6.8x10 cm during exposure: Vpn = -2.0 V

0.0 200 300 400 500 600 700 800 900 1000

-2

Atomically Flat, As 3keV, 6.8x10 cm during exposure: Floating

0.8

External Q.E.

1.0

-2

External Q.E.

13

External Q.E.

0.8

+

Atomically Flat, As 10keV, 3.4x10 cm during exposure: Vpn = -2.0 V

External Q.E.

External Q.E.

1.0

0.8

Wavelength [nm] +

13

-2

Conventional Flat, As 3keV, 6.8x10 cm during exposure: Floating

0.6 0.4 0.2


0.0 200 300 400 500 600 700 800 900 1000

(f) (g) (e) Wavelength [nm] Wavelength [nm] Wavelength [nm] Fig. 14 External Q.E. measured as a function of the wavelength at various UV-light exposure time. The fabrication conditions and the pn-junction bias conditions during UV-light exposure are shown in the figures. The photodiode area is 1.0x10-2 cm2. 2.0

Internal Q.E.

+

(a)

13

-2

Atomically Flat, As 10keV, 3.4x10 cm + 13 -2 Atomically Flat, As 3keV, 6.8x10 cm + 13 -2 Conventional Flat, As 3keV, 6.8x10 cm

1.5

during exposure: Vpn = -2.0 V

λ = 250 nm

1.0 0.5 0.0 0 0 10

1

2

10

3

10

10

UV-light Exposure Time [min] 2.0

Internal Q.E.

+

(b)

13

-2

Atomically Flat, As 10keV, 3.4x10 cm + 13 -2 Atomically Flat, As 3keV, 6.8x10 cm + 13 -2 Conventional Flat, As 3keV, 6.8x10 cm

1.5

λ = 250 nm

during exposure: Floating

1.0 0.5 0.0 0 0 10

1

10

2

10

3

10

UV-light Exposure Time [min]

Fig. 15 Internal Q.E. at 250 nm as a function of the UV-light exposure time, while the photodiodes were (a) reverse biased at -2.0 V and (b) floating. The photodiode area is 1.0x10-2 cm2.

CONCLUSION By introducing the atomically flat Si surface to uniformly form the thin surface drift layer, a photodiode exhibiting the almost 100% internal Q.E. to UV-VisibleNear IR light and a high stability to UV-light is demonstrated. The atomically flatness was found to be effective to enlarge the process margin to form the suitable surface drift layer. The developed FSI-photodiode process is compatible to the current photodiode manufacturing, and it is a promising technology for the development of highly UV-light sensitive and highly reliable Si photodiodes, arrayed sensors and image sensors. -

ACKNOWLEDGEMENT Authors would like to acknowledge Y. Kondo and K. Takubo of Shimadzu Corp. for their support on photodiode evaluation.

REFERENCES [1] T. Haraguchi D-Q. Ding, A. Yamamoto, T. Kaneda, T. Koujin and Y. Hiraoka, “Multiple-color Fluorescence Imaging of Chromosomes and Microtubules in Living Cells,” Cell Structure and Function, Vol.24, pp.291-298, 1999. [2] P. E. Malinowski J. Y. Duboz, P. D. Moor, J. John, K. Minoglou, P. Srivastava, Y. Creten, T. Torfs, J. Putzeys, F. Semond et al., “10 μm Pixel-to-Pixel Pitch Hybrid Backside Illuminated AlGaN-on-Si Imagers for Solar Blind EUV Radiation Detection,” IEDM Tech. Dig., pp.348-351, 2010. [3] Y. Goldberg, “Semiconductor near-ultraviolet photoelectronnics,” Semicond. Sci. Technol. Vol.14, pp.R41-60, 1999. [4] B. E. Lendi and V. R. Meyer, “The UV Detector for HPLC -An Ongoing Success Story-,” Liquid Chromatography - Gas Chromatography (LC-GC) Europe, Vol.18, pp.156-163, 2005. [5] T. Ohmi, M. Miyashita, M. Itano, T. Imaoki and I. Kawanabe, “Dependence of Thin-Oxide Films Quality on Surface Microroughness,” IEEE T-ED., Vol. 39, pp. 537-545, 1992. [6] F. M. Li, N. O and A. Nathan, “Degradation Behavior and Damage Mechanisms of CCD Image Sensor With Deep-UV Laser Radiation,” IEEE T-ED, Vol. 51, pp.2229-2236, 2004. [7] T. Watanabe, J-H. Park, S. Aoyama, K. Isobe and S. Kawahito, “Effects of Negative Bias Operation and Optical Stress on Dark Current,” IISW, Session 05-p8, 2009. [8] S. Nikzad M. E. Hoenk, J. Blacksberg, T. J. Jones and T. J. Cunningham, S. E. Holland, W. Kolbe and C. Bebek, “High Performance Large Format UV/optical/near IR Detector Arrays,” IISW, pp.181-183, 2007. [9] J. Blacksberg, S. Nikzad, M. E. Hoenk, S. E. Holland and W. F. Kolbe, “Near-100% Quantum Efficiency of Delta Doped Large-Format UV-NIR Silicon Imagers,” IEEE T-ED, Vol.55, pp.3402-3406, 2008. [10] R. Kuroda, T. Suwa, A. Teramoto, R. Hasebe, S. Sugawa and T. Ohmi, “Atomically Flat Silicon Surface and Silicon/ Insulator Interface Formation Technologies for (100) Surface Orientation Large-Diameter Wafers Introducing High Performance and Low-Noise Metal-Insulator-Silicon FETs,” IEEE T-ED, Vol.56, pp.291-298, 2009. [11] R. Kuroda, A. Teramoto, Y. Nakao, T. Suwa, M. Konda, R. Hasebe, X. Li, T. Isogai, H. Tanaka, S. Sugawa and T. Ohmi, “Complementary Metal–Oxide–Silicon Field-Effect-Transistors Featuring Atomically Flat Gate Insulator Film/Silicon Interface,” Jpn. J. Appl. Phys., Vol. 48, pp.04C048-1-6, 2009. [12] X. Li, A. Teramoto, T. Suwa, R. Kuroda, S. Sugawa and T. Ohmi, “Atomically Flattening Technology at 850oC for Si(100) Surface,” ECS Trans. Vol.28, pp.299-309, 2010.