ISSN 10637826, Semiconductors, 2011, Vol. 45, No. 13, pp. 1684–1688. © Pleiades Publishing, Ltd., 2011. Original Russian Text © E.A. Denisova, V.V. Uzdovskii, V.I. Khainovskii, 2011, published in Izvestiya vysshikh uchebnykh zavedenii. Elektronika, 2011, Vol. 88, No. 2, pp. 14–21.

MICROELECTRONIC DEVICES AND SYSTEMS

Multichannel Photocells for Image Converters with Color Separation E. A. Denisova, V. V. Uzdovskii^, and V. I. Khainovskii Moscow Institute of Electronic Technology, pr. 4806, str. 5, Zelenograd, Moscow Oblast, 124498 Russia ^email: [email protected] Submitted September 20, 2010

Abstract—The results of a study of photoelectric processes in photosensitive structures based on a multichan nel vertically integrated p–n junction are presented. Optical radiation absorption in the spacecharge region of a multichannel vertically integrated structure is studied. DOI: 10.1134/S1063782611130070

1. INTRODUCTION The development of the photosensitive elements of spectrally selective photoelectric image converters for various radiation ranges, integrated with a reading electronic circuit in one crystal, is an important prob lem of microelectronics. A spectrally selective array photodetector based on three vertically integrated p–n junctions [1] is one of the successful implementations of this type of photodetecting arrays. Its design and photoelectric parameters are highly competitive and often superior to known multiband photoelectric image converters based on charge coupled devices [2–7]. Recently, of particular interest are photodetectors based on multilayer structures providing operation in various bands of the visible spectrum. For example, multilayer photosensitive structures based on amor phous silicon are considered in [8–13]. 2. PHOTODETECTING STRUCTURES AND POTENTIAL DISTRIBUTION IN VERTICALLY INTEGRATED p–n JUNCTIONS The objective of this research is to study the photo electric processes in photocells based on multichannel vertically integrated p–n junctions. The study is based on analytical calculations of a onedimensional thickness model of the photosensi tive cell structure and numerical calculations of its twodimensional thickness model using the ISE TCAD devicetechnological computeraided design system. Upon exposure to optical radiation from above, the cell structures at the depth of the p–n junctions pro vide separation of the photocarriers corresponding to various optical radiation wavelengths. This results from the wavelength dependence of the optical absor bance in silicon [14]. The photocell structure contains n and ptype semiconductor layers on a ptype semi

conductor substrate. By applying various voltages to p and n regions relative to the substrate, “potential wells” with required depths can be created in these lay ers to accumulate and retain sufficient surface con centrations of photoelectrons and photoholes. The electric potential distribution along the substrate depth has minima and maxima in p and nregions, respectively. From a physical point of view, this means the formation of channels for hole and electron accu mulation in p and nlayers, respectively. Upon optical exposure, photocarriers are separated due to the inter nal electric field; thus, photoelectrons and photoholes accumulate in the corresponding channels. The selectivity of “white” light separation into spectral wavelength ranges can be improved using a photocell structure including five vertically connected photodiodes, i.e., five p–n junctions with metallurgi cal boundaries arranged at distances of 0.2, 0.7, 1.2, 1.7, and 2.5 μm from the surface. Hence, the vertical structure includes three nregions, two pregions, and the psubstrate. There is an individual metal contact to each region (and to the substrate) to pick up the corre sponding photosignal. The thicknesses of the semi conductor region were chosen for the reason of sepa rating five individual spectral wavelength regions of optical radiation. Figure 1 shows the vertical section of the photocell structure under consideration. It can be fabricated using standard CMOS technology includ ing ion implantation of the corresponding phosphorus and boron dopants (ions) followed by their “anneal ing” to form n and pregions successively embedded into each other. To fabricate p–n junctions, concen trations of the corresponding dopants should be increased by an order of magnitude (to overcompen sate the previous dopant). As impurity concentrations in the n and pregions increase, the corresponding spacecharge regions (SCRs) of the p–n junctions significantly decrease, hence, the internal electric fields strengthen. To pre

1684

MULTICHANNEL PHOTOCELLS FOR IMAGE CONVERTERS WITH COLOR SEPARATION

1685

y, μm

Optical radiation

1

2

3

4

11 Vsubstr V2

V4

V5 n

0.2

p n p

V3 0.7 1.2

V1

10 9 8

1.7 2.5 µm

7 6

n

5

n p n p n p p+

4 3

ptype substrate

2 Fig. 1. Schematic section of the photosensitive cell con taining five verticallyintegrated p–n junctions; V1–V5 are control voltages.

1 0 0

vent from exceeding the critical value of the SCR elec tric field, lower dopant concentrations were chosen for two surface p–n junctions. To this end, the pregion penultimate to the surface can be formed by etching the previous nregion to a depth of 0.7 μm and filling the etched region with ptype silicon by epitaxial growth. The surface nregion 0.2 μm deep is formed in a conventional way, i.e., by ion implantation. Figure 2 shows the top plan view on the photocell shown on a scale with topological sizes corresponding to the submicrometer sizes of the photocell with three p–n junctions [15–17]. Electronic circuits of the photosignal readout in each n and pregion of the photocell are similar to the readout circuits of the threediode photocell and are formed in the p+ regions 2 μm wide, adjacent to its photosensitive regions. Each readout circuit includes a MOS transistor for setting a corresponding depletion voltage at the n or pregion, then an amplifying MOS transistor, and a third MOS transistor for reading the photosignal to the data bus. The circuit structure of the photocell control con tains five “line buses” for reading the photosignals of five spectral wavelength ranges and one “data bus”. The electric potential distribution in the fivediode photocell can be found by analytically solving the Poisson equation for each n and pregion of its struc ture (see Fig. 1). 3. NUMERICAL SIMULATION RESULTS Analytical calculations resulted in expressions for the electric potential. Calculations were performed for each n and pregion of the structure, based on the solution of the Poisson equation with continuity boundary conditions for the electric field strength and electric potential. SEMICONDUCTORS

Vol. 45

No. 13

2011

1

2

3

4

5

6

7

8

9 10 11 x, μm

Fig. 2. Plane topological view of the arrangement of the basic structural elements of the photocell with five photo diodes: (1) photosensitive surface of the photocell, (2) metallurgical boundaries of the p–n junctions, (3) con tact windows, and (4) p+regions for MOS transistors of photosignal readout circuits.

Numerical simulation results on the electric field distributions in the threediode structure are in agree ment with the results of their analytical studies (Fig. 3). At chosen control voltages, SCR internal fields of p–n junctions are several times weaker than the critical field of an electrical breakdown. One of the main parameters eventually controlling the photocell photosensitivity are the surface concen trations of photocarriers accumulated in the n and pregions. They are determined by the obtained expressions for the electric potential. In the initial (depleted) state, the electric potential distribution set by the control voltages is established between regions. In the calculation, it is taken into account that photo generated electrons and holes should be accumu lated in corresponding “potential wells” of the n and pregions, so that wells would not be overfilled, i.e., an uncontrolled photocarrier spreading between regions would not occur. In the case of limiting the fill ing of regions with photocarriers, potential differences at reverse biased p–n junctions should not be lower than corresponding contact potential differences. Based on the above limitations, the maximum surface concentrations of photocharges accumulated in the n and pregions were calculated. For each region, the following relation was satisfied, ( 1, 2 )

( 1, 2 )

( 1, 2 )

ΔQ n, p photo = Q n, p max – Q n, p min ,

(1)

1686

DENISOVA et al.

n and pregion, respectively, ϕ2 is the contact poten tial difference between the deep nregion and the mid dle pregion, and β is the ratio of the donor concentra tion in the deep nregion to the acceptor concentra tion in the pregion. Furthermore, the expressions for the initial surface concentration of dark holes in the pregion,

ϕn, p(x), B

3 1 2

Q p min =

2ε 0 ε ( V 3 – V 2 + ϕ 3 ) ⎫ +   ⎬, (2) eN a ( γ + 1 ) ⎭

1 2 0 0

1

2

(1) ⎛ ⎛ Q n min⎞ ⎞ – β ⎜ a – ⎜ αw 1 +  ⎟ ⎟ (1) ⎝ ⎝ Nd ⎠ ⎠ ⎩

(2) ⎧ Na ⎨ b

3

4

5 x, μm

Fig. 3. Electric potential distributions in the threediode vertical photocell at T = 300 K: (1) initial steady state, con trol voltages are V1 = V3 = +3 V and V2 = +1 V; (2) equi librium state, control voltages are V1 = V2 = V3 = 0; (•) points of analytical calculation.

and for the highest surface concentration of holes in the pregion upon exposure to light (maximum filling), (1) ⎛ ⎛ Q n max⎞ ⎞ (2) ⎧ Q p max = N a ⎨ b – β ⎜ a – ⎜ αw 2 +  ⎟ ⎟ (1) ⎝ ⎝ Nd ⎠ ⎠ ⎩

⎫ 2ε 0 εϕ 3  ⎬, +  (2) eN a ( γ + 1 ) ⎭

( 1, 2 )

where ΔQ n, p photo is the maximum surface concentra ( 1, 2 )

tion of accumulated photocharges, Q n, p max and ( 1, 2 )

Q n, p min are the maximum and minimum surface con centrations of carriers in filled and empty “potential wells”. For the deep nregion, the expressions for the ini tial surface concentration of dark electrons, ⎛ 2ε 0 ε ( V 1 – V 2 + ϕ 2 )⎞ (1) (1) ⎟ , Q n min = N d a – ⎜ αw 1 +  (1) ⎝ eN d ( β + 1 ) ⎠ and the limiting surface electron concentration for this region, ⎛ 2ε 0 εϕ 2 ⎞ (1) (1) ⎟ , Q n max = N d a – ⎜ αw 2 +  (1) ⎝ eN d ( β + 1 )⎠ (1)

were derived, where N d is the donor concentration in the deep ntype region, a is the depth of the p–n junc tion of the deep nregion and psubstrate, α is the ratio of the acceptor concentration in the psubstrate to the donor concentration in the deep nregion, w1 and w2 are the SCR thicknesses in the psubstrate in the cases of the initial structure state (in the dark) and in the state occupied by photocarriers (upon exposure to light), respectively, ε0 is the permittivity of free space, and ε is the relative permittivity of silicon, e is the elec tron charge, V1 and V2 are the control voltages for the

(2)

were obtained, where N a are the acceptor concen tration in the middle pregion, b is the depth of the p–n junction of the middle pregion and the deep nregion, V3 is the control voltage for the surface nregion, ϕ3 is the contact potential difference between the middle pregion and the surface nregion, and γ is the ratio of the acceptor concentration in the middle pregion to the donor concentration in the sur face nregion. Similarly, in studying the surface nregion filling with electrons, the corresponding expression for the electron concentration was derived, (1) ⎛ ⎛ Qp Qn ⎞ ⎞ ⎫ (2) (2) ⎧ – – Q n = N d ⎨ c – γ b –   β a αw +  ⎟ ⎟ ⎬, ⎜ ⎜ (2) (1) ⎝ ⎝ Na Nd ⎠ ⎠ ⎭ ⎩ (2)

where N d is the donor concentration in the surface nregion, c is the depth of the p–n junction of the sur face nregion and the middle pregion; w corresponds (1) to w1 or w2, Qp corresponds to Qp min or Qp max, Q n cor (1)

(1)

(2)

responds to Q n min and Q n max when determining Q n min (2)

and Q n max , respectively. After substituting the main photocell parameters optimized according to (1), we obtain: (1)

(i) for the deep nregion, ΔQ n photo = 4.36 × 1011 cm–2; SEMICONDUCTORS

Vol. 45

No. 13

2011

MULTICHANNEL PHOTOCELLS FOR IMAGE CONVERTERS WITH COLOR SEPARATION

(ii) for the middle pregion, ΔQp photo = 1.12 × 1012 cm–2; and (2) (iii) for surface nregion, ΔQ n photo = 7.9 × 1011 cm–2. Based on the obtained maximum surface concen trations of photocarriers, the thermal relaxation times in each n and pregion were determined, ( 1, 2 )

ϕn, p(x), B 2

1

eΔQ n, p photo ( n, p ) τ therm =  . ( n, p ) j therm

(2) 1

The maximum thermal generation current densi ties entering relation (2) were numerically calculated ( n1 ) as j therm = 2.1 × 10–6 A/cm2 for the deep nregion, (p)

1687

2 0

( n2 )

j therm = 15 × 10–6 A/cm2 for the pregion, and j therm = 5.2 × 10–6 A/cm2 for the surface nregion. Using expression (2), the current densities of carrier thermal generation, and the maximum accumulated photocar rier concentrations, the corresponding times of the thermal relaxation of “potential wells” were deter ( n1 ) (p) ( n2 ) mined as τ therm = 0.033 s, τ therm = 0.012 s, and τ therm = 0.024 s. Based on the physical meaning, as the time of thermal relaxation of the entire photocell structure, the smallest one was chosen (τterm = 0.012 s). To elim inate the effect of thermal generation on photocharge accumulation, the time of the main cycle of photocell control by electric voltages was chosen ~1000 times shorter than the thermal relaxation time: τcycle = 0.001τtherm = 12 × 10–6 s. In this case, the main cycle clock frequency of photocell control is fcycle = 1/τcycle ≈ 83 kHz. In fact, due to multiple defects in the crystal structure of the photocell n and pregions, the ther mal relaxation time may be several times smaller. Then the corresponding frequency of the photocell control cycle can be ~200–300 kHz. Furthermore, the onedimensional (Fig. 4) and twodimensional distributions of electric potentials in the semiconductor region of the photocell structure were numerically calculated using the ISE TCAD pro gram according to the layer thicknesses (Fig. 1) and chosen dopant concentrations in the layers. In this case, depleting voltages V1 = V3 = V5 = +1.5 V and V2 = V4 = –1.0 V were applied to the n and pregions, respectively. The thermal relaxation time of the structure under consideration was calculated as eΔQ n, p photo n, p τ therm =  . j n, p therm

(3)

In this case, the maximum calculated densities of photocarriers accumulated in each “potential well”: (i) in the deep nregion, ΔQn1 photo = 2.62 × 1011 cm–2; (ii) in the pregion, ΔQp1 photo = 8.2 × 1011 cm–2; (iii) in the middle nregion, ΔQn2 photo = 18.4 × 1011 cm–2; SEMICONDUCTORS

Vol. 45

No. 13

2011

−1 0

2

4

6

8 x, μm

Fig. 4. Electric potential distribution in the fivediode ver tical photocell at T = 300 K: (1) initial stationary depleted state of the n and ptype regions, control voltages are V1 = V3 = V5 = +1.5 V, V2 = V4 = –1.0 V; (2) equilibrium state, control voltages are V1 = V2 = V3 = V4 = V5 = 0.

(iv) in the middle pregion, ΔQp2 photo = 13.9 × 1011 cm–2; and (v) in the surface nregion, ΔQn3 photo = 9.64 × 1011 cm–2. It was found that the corresponding thermoelectric current densities in the n and pregions are 2.3, 6.4, 4.6, 10.0, and 3.3 μA/cm3. Therefore, according to expression (1), the thermal relaxation times of n and n1 p1 n2 pregions are τ therm = 0.018 s, τ therm = 0.021 s, τ therm = p2

n3

0.063 s, τ therm = 0.022 s, and τ therm = 0.047 s. As the total thermal relaxation time of the entire structure, we choose the shortest among the above, i.e., 0.018 s. Then, the time of the photocell control cycle (one period) is τcycle = 0.001τtherm = 18 μs, and the corre sponding frequency of photocell control is fcycle = 1/τcycle = 1/18 μs = 56 kHz. 4. CONCLUSIONS The mathematical analysis and numerical simula tion of the design parameters and variations of the control voltages for the multichannel vertically inte grated photocell showed the following: (I) Dopant concentrations in successively arranged (1) n and pregions should be N d = 1 × 1016 cm–3 in the (2)

deep nregion, N a = 1 × 1017 cm–3 in the pregion, and

1688

DENISOVA et al.

(2)

N d = 1 × 1018 cm–3 in the surface nregion at the (1)

dopant concentration N a = 1 × 1015 cm–3 in the psub strate; the thicknesses of these regions are 1.4, 0.4, and 0.2 μm. (II) The optimum control (depleting) voltages rel ative to the substrates are V1 = V3 = +3 V for nregions and V2 = +1 V for the pregion, which creates a neces sary electric potential relief in the p–n–p–n structure under consideration and causes photocarrier accumu lation in the corresponding “potential wells” of the n (1) and pregions with surface concentrations: ΔQ n photo = 4.36 × 1011 cm–2 in the deep nregion, ΔQp photo = (2)

1.12 × 1012 cm–2 in the middle pregion, and ΔQ n photo = 7.9 × 1011 cm–2 in the surface nregion. (III) The total time of thermal relaxation of the threediode structure, ~12 ms, controls the nominal frequency of the photocell operation control, which is ~80–100 kHz. An analysis of the electric potential and electric field strength distributions in the bulk of the semicon ductor structure under consideration and the study of dark charge accumulation in corresponding “potential wells” of the structure resulted in the determination of the design parameters (thicknesses of semiconductor layers and dopant concentrations in them) and acceptable control electric voltages. An analysis of the optical radiation absorption in SCRs of the photocell multichannel structure made it possible to construct a set of equations, describing photorelaxation processes in the “potential wells” of the structure. The study of photorelaxation by numerically solving this set of equations made it possible to obtain the time depen dences for photocharge accumulation in the photocell n and pregions, and to calculate the spectral charac teristics of the photosensitivity of the n and pregions and corresponding photorelaxation times. ACKNOWLEDGMENTS This study was supported by the Ministry of Educa tion and Science of the Russian Federation within the Federal target program “Scientific and Pedagogical

Personnel of Innovative Russia” for 2009–2013, state contract no. p1470. REFERENCES 1. R. B. Merrill, US Patent No. 5969875, Int.Cl.G01J 3/50, U.S. Cl. 250/226 (1999). 2. R. Barsan, IEEE Trans. Electron. Dev. 26, 123 (1979). 3. V. L. Khainovskii and V. V. Uzdovskii, Opt. Eng. 33, 2352 (1994). 4. V. L. Khainovskii and V. V. Uzdovskii, in Proceedings of the 40th International Symposium on Optical Engineering Institute (San Diego, CA, USA, 1995), Vol. 2551, pp. 189–196. 5. V. I. Khainovskii and V. V. Uzdovskii, Opt. Eng. 36, 1678 (1997). 6. V. I. Khainovskii, V. V. Uzdovskii, and N. M. Gordo, Izv. Vyssh. Uchebn. Zaved., Elektron., No. 3, 45 (1999). 7. V. I. Khainovskii, V. V. Uzdovskii, N. M. Gordo, and R. A. Fedorov, Izv. Vyssh. Uchebn. Zaved., Elektron., No. 1, 28 (2000). 8. K. Eberhardt, T. Neidlinger, and M. B. Schubert, IEEE Trans. Electron. Dev. 42, 1763 (1995). 9. J. Zimmer, D. Knipp, H. Stiebig, and H. Wagner, IEEE Trans. Electron. Dev. 46, 884 (1999). 10. M. Topic, H. Stiebig, D. Knipp, and F. Smole, IEEE Trans. Electron. Dev. 46, 1839 (1999). 11. V. Gradisnik, M. Pavlovic, B. Pivac, and I. Zulim, IEEE Trans. Electron. Dev. 49, 550 (2002). 12. K.D. Cho, H.S. Tae, and S.I. Chien, IEEE Trans. Electron. Dev. 50, 359 (2003). 13. H. I. Kwon, I. M. Kang, B.G. Park, et al., IEEE Trans. Electron. Dev. 51, 178 (2004). 14. W. C. Dash and R. Newman, Phys. Rev. 99, 1151 (1955). 15. E. A. Ignatjeva, V. V. Uzdovskii, and V. I. Khainovskii, Izv. Vyssh. Uchebn. Zaved., Elektron., No. 1, 35 (2008). 16. E. A. Ignatjeva, V. V. Uzdovskii, and V. I. Khainovskii, Proc. Tech. Univ. Russia, Electron., No. 3, 38 (2008). 17. E. A. Ignatjeva, V. V. Uzdovskii, and V. I. Khainovskii, in Proceedings of the 9th Annual IEEE 2008 Interna tional Workshop and Tutorials on Electron Devices and Materials, Erlagol, Altai, July 1–5, 2008 (2008), pp. 62–68.

Translated by A.M. Kazantsev

SEMICONDUCTORS

Vol. 45

No. 13

2011