Professional Documents
Culture Documents
Chen 2007
Chen 2007
Abstract—This paper proposes an extended 1-D analysis to heavy-doping substrate and low-doping epitaxial layer [4]. The
derive quantum efficiency of various commonly used CMOS pho- 1-D analysis focuses generally on a single photodiode and ne-
todiodes. The theoretical model of the CMOS photodiode with glects its lateral effects. When the feature size of process tech-
the n−/p−epitaxial/p + substrate (n−/p−epi/p + sub) struc-
ture is established from steady-state continuity equations, where nology is scaled down, the crosstalk among pixels of an image
most existing boundary conditions are applied. In particular, the sensor and the lateral effects from photodiodes greatly affect
minority carrier and current densities are continuous across the quantum efficiency. Hence, multidimensional analyses have
interface between two layers with the same dopant type. Models of been proposed for investigating the crosstalk and lateral effects
the other commonly used CMOS photodiodes are also examined. in an image-sensor array. To extend the 1-D into multidimen-
Three CMOS photodiodes with n−/p−substrate (n−/p−sub),
p+/n−/p−substrate (p+/n−/p−sub), and n−/p−epi/p + sub sional analyses, the effects of the crosstalk and the fill factor on
structures are fabricated and characterized to validate the pro- the quantum efficiency of a pixel array must be well addressed.
posed model. Additionally, the surface recombination velocity is Therefore, a multidimensional analysis requires an extra piece-
adequately determined by fitting the simulated quantum efficiency wise function to define an illuminated area and boundary con-
to the measured value. The simulated quantum efficiency of the ditions of the steady state and ambipolar transport equations for
proposed model for these three photodiodes is quite consistent with
the measured values, revealing the feasibility and effectiveness of deriving the divergence of the distributions to acquire quantum
the proposed model in characterizing various CMOS photodiodes. efficiency [5], [6].
Although analyses of multiple dimensions for an image-
Index Terms—CMOS photodiode, quantum efficiency, surface
recombination velocity, 1-D analysis. sensor array are necessary, some characteristics of photodi-
odes are not well addressed in 1-D analyses. Additionally,
only a few analyses were developed that focus particularly
I. INTRODUCTION
on CMOS photodiodes [3], [4], [7]. Accordingly, this paper
Fig. 1. Simplified structures of the n−/p−epi/p + sub photodiode under (a) proposed boundary conditions and (b) conventional boundary conditions, where
the gray area is the space-charge region (the dimension of each layer in these two structures does not represent the actual dimension).
TABLE I
PHYSICAL PARAMETERS USED IN THE PROPOSED EQUATION DERIVATION
all physical parameters applied in the derivation process, assumptions, the 1-D steady-state continuity equations for n-
where Planck’s constant and the velocity of light are 6.626 × and p-type semiconductors are
10−34 J · s and 2.998 × 108 m/s, respectively, and the empirical
absorption coefficient is referred from [8]. Dp d2 pn (x)/dx2 + Gx − ((pn (x) − pn0 ) /τp ) = 0 (1)
The commonly used 1-D analysis for quantum efficiency of
and
photodiodes derives the excess minority carrier density using
1-D steady-state continuity equations. The excess minority
Dn d2 np (x)/dx2 + Gx − ((np (x) − np0 ) /τn ) = 0 (2)
carrier density is then converted into photocurrent density and
quantum efficiency. Since the p-n junction is normally operated respectively. Equations (1) and (2) use boundary conditions
at a reverse-biased voltage, the electron–hole recombination to derive solutions. According to the structure in Fig. 1(a),
in the space-charge region due to the process of photonic four boundary conditions exist. In the following equations,
generation can be neglected. The variation of majority carriers np−epi (x) and np−sub (x) are electron densities in p−epitaxial
caused by incident light can be also ignored, indicating that a and p+substrate, respectively, and np0−epi and np0−sub are
low-injection condition applies. Additionally, the electric field equilibrium electron densities in p−epitaxial and p+substrate,
is negligible in the neutral region. Based on the aforementioned respectively. The subscripts “n − epi” and “n − sub” refer to
CHEN et al.: EXTENDED ANALYSIS TO DERIVE QUANTUM EFFICIENCY OF VARIOUS CMOS PHOTODIODES 2661
the physical parameters of the minority carriers in p−epitaxial B. Quantum Efficiency of n−/p−epi/p + sub Photodiode
and p+substrate, respectively.
The steady-state continuity equations using the boundary
conditions in (3)–(8) are solved to yield the minority carrier
A. Boundary Conditions densities. To facilitate the excess minority carrier densities
solely excited by incident light, the photo-induced excess mi-
1) Surface of the Device: The current density of the inter- nority carrier densities must fulfill the computation of (9)–(11),
face between the SiO2 and Si substrate is utilized to derive a shown on the next page, where fn−epi (x), fn−sub (x), and σ are
boundary condition as follows: expressed as follows:
Jp (x = 0) = qDp (dpn (x)/dx) = q (pn (x) − pn0 ) Sp . (3)
fn−epi (x) = τn−epi G0 e−αx / L2n−epi α2 − 1 (12)
2) Interface Between Two Layers With Different Dopant
fn−sub (x) = τn−sub G0 e−αx / L2n−sub α2 − 1 (13)
Types: Generally, the dark current generated in the space-
charge region is negligible in comparison to the induced pho- Dn−epi x3 − x2 x4 − x3
tocurrent, and the parasitic access resistance is small enough σ= ch sh
Ln−epi Ln−epi Ln−sub
to be ignored. Hence, the boundary conditions at the interface
between n− and p−epitaxial can be simplified as follows: Dn−sub x3 − x2 x4 − x3
+ sh ch . (14)
Ln−sub Ln−epi Ln−sub
pn (x = x1 ) = pn0 (4)
np−epi (x = x2 ) = np0−epi . (5) In the structure in Fig. 1(a), most of the excess minority carriers
recombine rapidly with the majority carriers in p+substrate
The width of the space-charge region is (x2 − x1 ) and is before reaching p−epitaxial because of heavy doping and thick
affected by the impurity doping concentrations of p- and n-type substrate. Hence, the current generated in p+substrate is negli-
semiconductors and the bias voltage across the p-n junction [9]. gible. Consequently, photocurrent densities generated by n- and
3) Interface Between Two Layers With the Same Dopant p-type semiconductors are (15) and (16), respectively, shown on
Type: In the high–low junction between p+substrate and the next page. Lastly, the photocurrent density inside the space-
p−epitaxial, minority carriers that diffuse from a low-doping charge region is determined as
layer to a high-doping layer due to the electric field resulting
from doping gradient can rapidly recombine with majority x3
carriers in the high-doping layer. This phenomenon is similar Jdrift = −q Gx dx = qφ0 (e−αx3 − e−αx2 ). (17)
to that in a p-n junction [4]. Hence, the excess minority car- x2
rier densities in p−epitaxial and p+substrate can be derived
from the boundary conditions like (4) and (5). However, such Consequently, the summation of (15)–(17) is the total pho-
a derivation enables the excess minority carrier densities in tocurrent density (JT T −photo = Jn−photo + Jp−photo + Jdrift )
p−epitaxial and p+substrate to depend only on the factors as- produced by the photodiode in Fig. 1(a). Under ambient tem-
sociated with p−epitaxial and p+substrate, respectively, which perature, the relationship between spectral response () and the
include process recipes and physical parameters. To address quantum efficiency (η) is represented by [9]
the mutual effects of p−epitaxial and p+substrate, this paper
assumes that the minority carrier density and current density
η = hc/qλ = 1.24/λ. (18)
across the real boundary at the interface between p−epitaxial
and p+substrate are continuous. Accordingly, the proposed
boundary conditions at the interface between p−epitaxial and
C. Analysis for the Proposed Boundary Condition
p+substrate are formulated as follows:
The structure in Fig. 1(a) is replaced by that in Fig. 1(b) if the
Jn (x = x3 ) = − qDn−epi (dnp−epi (x)/dx) x=x3 interface between the p+substrate and p−epitaxial is treated
as a p-n junction. Thus, the equations for np−epi (x)photo and
= − qDn−sub (dnp−sub (x)/dx) x=x3 (6)
np−sub (x)photo are given by (19) and (20), respectively, shown
np−epi (x = x3 ) = np−sub (x = x3 ). (7) on the next page, where fn−epi (0) and fn−sub (0) are defined in
(12) and (13), respectively. From (19) and (20), np−epi (x)photo
4) Bottom of the Device: Generally, the thickness of the and np−sub (x)photo depend only on the physical parameters
substrate in a CMOS photodiode is several hundred microm- of p−epitaxial and p+substrate, respectively. Under the pro-
eters. In considering heavy doping of p+substrate and the posed boundary condition, np−epi (x)photo and np−sub (x)photo ,
effect of absorption length, very few excess minority carriers which is represented by (10) and (11), respectively, depend on
are generated at the bottom of the device. Consequently, the the physical parameters of both p−epitaxial and p+substrate.
minority carrier density at the device bottom is treated as the Hence, the proposed analysis enables a designer to understand
minority carrier density during equilibrium, as in the following: clearly the mutual effects of these two layers.
The diffusion lengths, diffusion coefficients, and lifetime
np−sub (x = x4 ) ∼
= np0−sub . (8) of minority carriers refer to empirical data in [10]–[12]. The
2662 IEEE TRANSACTIONS ON ELECTRON DEVICES, VOL. 54, NO. 10, OCTOBER 2007
pn (x)photo = pn (x) − pn (x)G0 =0
τp G0 / L2p α2 − 1 (e−αx1 ((Dp /Lp )ch(x/Lp ) + Sp sh(x/Lp )) − (αDp + Sp )sh ((x − x1 )/Lp )) τp G0 e−αx
= − 2 2
Sp sh(x1 /Lp ) + (Dp /Lp )ch(x1 /Lp ) Lp α − 1
(9)
np−epi (x)photo = np−epi (x) − np−epi (x)G0 =0
fn−epi (x2 ) Dn−epi x3 − x x4 − x3 Dn−sub x3 − x x4 − x3
= ch sh + sh ch
σ Ln−epi Ln−epi Ln−sub Ln−sub Ln−epi Ln−sub
(fn−epi (x3 ) − fn−sub (x3 )) Dn−sub x − x2 x4 − x3
+ sh ch
σ Ln−sub Ln−epi Ln−sub
α x − x2 x4 − x3
+ (Dn−sub fn−sub (x3 ) −Dn−epi fn−epi (x3 )) sh sh
σ Ln−epi Ln−sub
fn−sub (x4 ) Dn−sub x − x2
+ sh − fn−epi (x) (10)
σ Ln−sub Ln−epi
np−sub (x)photo = np−sub (x) − np−sub (x)G0 =0
Dn−epi x4 − x x3 − x2
= sh fn−epi (x2 ) − (fn−epi (x3 ) − fn−sub (x3 )) ch
σLn−epi Ln−sub Ln−epi
α x3 − x2 x4 − x
+ (Dn−sub fn−sub (x3 ) − Dn−epi fn−epi (x3 )) sh sh
σ Ln−epi Ln−sub
fn−sub (x4 ) Dn−epi x3 − x2 x − x3 Dn−sub x3 − x2 x − x3
+ ch sh + sh ch − fn−sub (x)
σ Ln−epi Ln−epi Ln−sub σLn−sub Ln−epi Ln−sub
(11)
Jn−photo = qDp (dpn (x)photo /dx) x=x1
qDp τp G0 ((e−αx1 /Lp ) ((Dp /Lp )sh(x1 /Lp ) + Sp ch(x1 /Lp )) − (αDp + Sp )/Lp ) qDp τp G0 −αx1
= + 2 2 αe (15)
L2p α2 − 1 (Sp sh(x1 /Lp ) + (Dp /Lp )ch(x1 /Lp )) Lp α − 1
Jp−photo = − qDn−epi (dnp−epi (x)photo /dx) x=x2
−qDn−epi fn−epi (x2 ) Dn−epi x3 − x2 x4 − x3 Dn−sub x3 − x2 x4 − x3
= sh sh + ch ch
σ L2n−epi Ln−epi Ln−sub Ln−epi Ln−sub Ln−epi Ln−sub
qDn−epi (fn−epi (x3 ) − fn−sub (x3 )) Dn−sub x4 − x3
− ch
σ Ln−epi Ln−sub Ln−sub
αqDn−epi (Dn−sub fn−sub (x3 ) − Dn−epi fn−epi (x3 )) 1 x4 − x3
− sh
σ Ln−epi Ln−sub
qDn−epi Dn−sub fn−sub (x4 )
− + αqDn−epi fn−epi (x2 ) (16)
σLn−epi Ln−sub
x3 −x x3 −x2
fn−epi (0) e−αx3 sh x−x2
Ln−epi + e−αx2 sh Ln−epi − e−αx sh Ln−epi
np−epi (x)photo = (19)
sh ((x3 − x2 )/Ln−epi )
x5 −x x5 −x4
fn−sub (0) e−αx4 sh Ln−sub + e−αx5 sh x−x4
Ln−sub − e−αx sh Ln−sub
np−sub (x)photo = (20)
sh ((x5 − x4 )/Ln−sub )
CHEN et al.: EXTENDED ANALYSIS TO DERIVE QUANTUM EFFICIENCY OF VARIOUS CMOS PHOTODIODES 2663
Fig. 3. Five simplified CMOS photodiode structures with gray areas indicating space-charge regions (the dimension of each layer in these structures does not
represent the physical one). (a) n+ /p−/p−sub. (b) p+ /n−/p−epi/p + sub. (c) n−/p−sub. (d) p+ /n−/p−sub. (e) n+ /p−/p−epi/p + sub.
Fig. 4. Cross sections of the three CMOS photodiodes. (a) n−/p−sub. (b) p+ /n−/p−sub. (c) n−/p−epi/p + sub.
TABLE II
RECIPES OF THE THREE PHOTODIODES
A PPENDIX
+ γsub sh(∆epi )ch(∆sub ))
The derivation of quantum efficiency of each photodiode
in Fig. 3 is the same as that in Section II-B and, thus, is B[2] (x − x2 )
not illustrated here. The excess minority carrier densities from + γepi sh
δ Ln−p−
different boundary conditions in two sides of each layer can be
analyzed as follows. × (γepi sh(∆epi )sh(∆sub )
1) Since the structure in Fig. 3(a) is similar to that in + γsub ch(∆epi )ch(∆sub ))
Fig. 1(a), the excess minority carrier densities from the
structure in Fig. 3(a) can refer to (9)–(11).
B[3] (x − x2 )
2) The structure of p−epi/p + sub in Fig. 3(b) is the same + sh (γepi ch(∆epi )sh(∆sub )
δ Ln−p−
as that in Fig. 1(a). Accordingly, the excess minority
carrier densities for np−epi (x)photo in p−epitaxial and + γsub sh(∆epi )ch(∆sub ))
np−sub (x)photo in p+substrate can be built by (10) and B[4] (x − x2 )
+ γepi γsub sh ch(∆sub )
(11), respectively, via x4 , x5 , and x6 , replacing x2 , x3 , δ Ln−p−
and x4 , respectively.
B[5] (x − x2 )
3) For n− in Fig. 3(c) and n+ in Fig. 3(e), the excess + γepi sh sh(∆sub )
δ Ln−p−
minority carrier densities are the same as that in (9).
For p+ in Fig. 3(b) and (d), the excess minority carrier B[6] (x − x2 )
+ γepi γsub sh − fn−p− (x)
densities can be also presented by (9), in which physical δ Ln−p−
parameters and process recipes of n− are replaced by the (A7)
corresponding ones of p+ . Moreover, (19) can interpret np−epi (x)photo
the excess minority carrier densities of n− in Fig. 3(b) = np−epi (x) − np−epi (x)|G0 =0
and (d), via replacing physical parameters and process
recipes of p−epitaxial by the corresponding ones of n−. B[1] (x4 − x)
= γp− γepi ch sh(∆sub )
4) The excess minority carrier density of p-substrate in δ Ln−epi
Fig. 3(d) is the same as that in (20). Additionally, the (x4 − x)
+ γsub sh ch(∆sub )
excess minority carrier density of p-substrate in Fig. 3(c) Ln−epi
can be also presented by (20), via x2 and x3 , replacing x4
B[2] (x4 − x)
and x5 , respectively. − γp− ch(∆p− ) γepi ch sh(∆sub )
δ Ln−epi
5) To deduce np−p− (x)photo , np−epi (x)photo , and
np−sub (x)photo , as illustrated in Fig. 3(e), the following (x4 − x)
+ γsub sh ch(∆sub )
six boundary conditions are adopted: Ln−epi
B[3] (x4 − x)
np−p− (x = x2 ) = npo−p− (A1) + sh(∆p− ) γepi ch sh(∆sub )
δ Ln−epi
np−p− (x = x3 ) = np−epi (x = x3 ) (A2) (x4 − x)
+ γsub sh ch(∆sub )
Jn (x = x3 ) = −qDn−p− (dnp−p− (x)/dx)|x=x3 Ln−epi
= −qDn−epi (dnp−epi (x)/dx)|x=x3 (A3) B[4] (x − x3 )
+ γsub ch(∆sub ) γp− ch(∆p−)sh
np−epi (x = x4 ) = np−sub (x = x4 ) (A4) δ Ln−epi
Jn (x = x4 ) = −qDn−epi (dnp−epi (x)/dx)|x=x4 (x − x3 )
+ γepi sh(∆p− )ch
= −qDn−sub (dnp−sub (x)/dx)|x=x4 (A5) Ln−epi
np−sub (x = x5 ) = npo−sub . (A6) B[5] (x − x3 )
+ sh(∆sub ) γp− ch(∆p− )sh
δ Ln−epi
The excess minority carrier densities in p−, p−epitaxial, (x − x3 )
and p+substrate, as shown in Fig. 3(e), are formulated in + γepi sh(∆p− )ch
Ln−epi
(A7)–(A9).
B[6] (x − x3 )
+ γsub γp− ch(∆p− )sh
δ Ln−epi
np−p− (x)photo
(x − x3 )
= np−p− (x) − np−p− (x)|G0 =0 + γepi sh(∆p− )ch
Ln−epi
B[1] (x3 − x)
= γepi sh − fn−epi (x) (A8)
δ Ln−p−
np−sub (x)photo
× (γepi sh(∆epi )sh(∆sub )
+ γsub ch(∆epi )ch(∆sub )) + γp− = np−sub (x) − np−sub (x)|G0 =0
(x3 − x) (B[1] − B[2]ch(∆p− ))
× ch × (γepi ch(∆epi )sh(∆sub ) = −fn−sub (x) +
Ln−p− δ
CHEN et al.: EXTENDED ANALYSIS TO DERIVE QUANTUM EFFICIENCY OF VARIOUS CMOS PHOTODIODES 2667
(x5 − x) R EFERENCES
× γp− γepi sh
Ln−sub [1] A. El Gamal and H. Eltoukhy, “CMOS image sensors,” IEEE Circuits
Devices Mag., vol. 21, no. 3, pp. 6–20, May/Jun. 2005.
B[3] (x5 − x) [2] E. R. Fossum, “CMOS image sensors: Electronic camera-on-a-chip,”
+ γepi sh(∆p− )sh
δ Ln−sub ) IEEE Trans. Electron Devices, vol. 44, no. 10, pp. 1689–1698, 1997.
[3] F. Van de Wiele, “Photodiode quantum efficiency,” in Solid State Imaging.
B[4] (x5 − x) Gronigen, The Netherlands: Noordhoff, 1976, pp. 47–90.
− γepi sh
δ Ln−sub [4] J. S. Lee, R. I. Hornsey, and D. Renshaw, “Analyses of CMOS photo-
diodes (I)—Quantum efficiency,” IEEE Trans. Electron Devices, vol. 50,
× (γepi sh(∆p− )sh(∆epi ) + γp− ch(∆p− )ch(∆epi ))) no. 5, pp. 1233–1238, May 2003.
B[5] (x5 − x) [5] J. S. Lee, R. I. Hornsey, and D. Renshaw, “Analyses of CMOS photodi-
+ sh odes (II)—Lateral photoresponse,” IEEE Trans. Electron Devices, vol. 50,
δ Ln−sub ) no. 5, pp. 1239–1245, May 2003.
[6] D. Levy, S. E. Schacham, and I. Kidron, “Three dimensional analytical
× (γp− ch(∆p− )sh(∆epi ) + γepi sh(∆p− )ch(∆epi )))
simulation of self- and cross-responsivities of photovoltaic detector ar-
B[6] (x − x4 ) rays,” IEEE Trans. Electron Devices, vol. ED-34, no. 10, pp. 2059–2070,
+ γepi sh Oct. 1987.
δ Ln−sub [7] W. J. Liu, O. T.-C. Chen, L. K. Dai, P. K. Weng, K. H. Huang, and
× (γepi sh(∆p− )sh(∆epi ) + γp− ch(∆p− )ch(∆epi )) F.-W. Jih, “A CMOS photodiode model,” in Proc. IEEE Int. Workshop
Behavioral Modeling Simulation, 2001, pp. 102–105.
(x − x4 ) [8] D. K. Schroder, Semiconductor Material and Device Characterization.
+ γsub ch (γp− ch(∆p− )sh(∆epi ) New York: Wiley, 1998.
Ln−sub
[9] S. M. Sze, Physics of Semiconductor Devices. New York: Wiley, 1980.
[10] L. Passari and E. Susi, “Recombination mechanisms and doping density
+ γepi sh(∆p− )ch(∆epi )) . (A9) in silicon,” J. Appl. Phys., vol. 54, no. 7, pp. 3935–3937, Jul. 1983.
[11] W. E. Beadle, J. C. C. Tsai, and R. D. Plummer, “Physical properties,”
in Quick Reference Manual for Silicon Integrated Circuit Technology.
Here, δ, B[1], B[2], B[3], B[4], B[5], and B[6] in (A7)–(A9) New York: Wiley, 1985, pp. 2-42–2-43.
are expressed as follows: [12] S. E. Swirhun, Y. H. Kwark, and R. M. Swanson, “Measurement of elec-
tron lifetime, electron mobility and band-gap narrowing in heavy doped
2 p-type silicon,” in IEDM Tech. Dig., 1986, vol. 32, pp. 24–27.
δ = γepi sh(∆p− )sh(∆epi )sh(∆sub )
[13] [Online]. Available: http://www.wolfram.com/
+ γp− γepi ch(∆p− )ch(∆epi )sh(∆sub ) [14] E. Hecht, “Interference,” in Optics. San Francisco, CA: Addison-
Wesley, 2002, ch. 9, pp. 385–438.
+ γp− γsub ch(∆p− )sh(∆epi )ch(∆sub ) [15] R. R. King, R. A. Sinton, and R. M. Swanson, “Studies of diffused
phosphorus emitters: Saturation current, surface recombination velocity,
+ γepi γsub sh(∆p− )ch(∆epi )ch(∆sub ) (A10) and quantum efficiency,” IEEE Trans. Electron Devices, vol. 37, no. 2,
pp. 365–371, Feb. 1990.
B[1] = fn−p− (x2 ) (A11) [16] R. R. King and R. M. Swanson, “Studies of diffused boron emitters:
B[2] = fn−p− (x3 ) − fn−epi (x3 ) (A12) Saturation current, bandgap narrowing, and surface recombination ve-
locity,” IEEE Trans. Electron Devices, vol. 38, no. 6, pp. 1399–1409,
B[3] = αDn−epi fn−epi (x3 ) − αDn−p− fn−p− (x3 ) (A13) Jun. 1991.
[17] H. S. Bennett, “Hole and electron mobilities in heavily doped silicon:
B[4] = fn−epi (x4 ) − fn−sub (x4 ) (A14) Comparison of theory and experiment,” Solid State Electron., vol. 26,
no. 12, pp. 1157–1166, 1983.
B[5] = α (Dn−sub fn−sub (x4 ) − Dn−epi fn−epi (x4 )) (A15)
B[6] = fn−sub (x5 ). (A16)
In the above equations, fn−epi (x) and fn−sub (x) are presented
in (12) and (13), respectively. Additionally, fn−p− (x), γp− ,
γepi , γsub , ∆p− , ∆epi , and ∆sub are represented by the follow-
ing equations:
Oscal T.-C. Chen (S’89–M’94) was born in
Taiwan, R.O.C., in 1965. He received the B.S. degree
(τn−p− G0 e−αx )
fn−p− (x) = 2 (A17) in electrical engineering from the National Taiwan
Ln−p− α2 − 1 University, Taipei, Taiwan, in 1987, the M.S. and
Ph.D. degrees in electrical engineering from the Uni-
Dn−p− versity of Southern California, Los Angeles, in 1990
γp− = (A18) and 1994, respectively.
Ln−p−
From 1994 to 1995, he was with the Computer
Dn−epi Processor Architecture Department, Computer Com-
γepi = (A19) munication and Research Laboratories, Industrial
Ln−epi Technology Research Institute, serving as System
Dn−sub Design Engineer, Project Leader, and Section Chief. From September 1995 to
γsub = (A20) August 2003, he was an Associate Professor with the Department of Electrical
Ln−sub Engineering, National Chung Cheng University, Chia-Yi, Taiwan, where since
(x3 − x2 ) August 2003, he has been a Professor. His research interests include photoelec-
∆p− = (A21) tronics, very large-scale integration (VLSI) systems, multimedia computing,
Ln−p− and communication systems.
(x4 − x3 ) Dr. Chen has been an Associate Editor of the IEEE Circuits and Devices
∆epi = (A22) Magazine since August 2003. He is a founding member of the Multimedia
Ln−epi Systems and Applications Technical Committee of the IEEE Circuits and
Systems Society. He participates in the Technical Program Committee of many
(x5 − x4 ) IEEE international conferences and symposiums. He was the corecipient of the
∆sub = . (A23)
Ln−sub Best Paper Award of the IEEE TRANSACTIONS ON VLSI SYSTEMS in 1995.
2668 IEEE TRANSACTIONS ON ELECTRON DEVICES, VOL. 54, NO. 10, OCTOBER 2007
Wei-Jean Liu received the B.S. and M.S. degrees Ping-Kuo Weng received the B.S. degree in nuclear
from the Department of Electrical Engineering, engineering from the National Tsing Hua University,
National Chung Cheng University, Chia-Yi, Taiwan, Hsinchu, Taiwan, R.O.C., in 1985 and the Ph.D.
R.O.C., in 1998 and 2000, respectively, where he is degree from the Institute of Electro-Optical Engi-
currently working toward the Ph.D. degree. neering, National Chiao-Tung University, Hsinchu,
His research interests include optoelectronics, in 1992.
device modeling, and CMOS image sensor. Since then, he has been with the Solid-State
Devices Section, Chung-Shan Institute of Science
and Technology, Tao-Yuan, Taiwan. His current re-
search interests include CMOS-image-sensor de-
sign, camera for medical imaging system, and
system-on-chip architecture.