Professional Documents
Culture Documents
Schottky PDF
Schottky PDF
Marco Nardone
Fabrication and characterization of thin film ZnO Schottky contacts based UV photodetectors: A comparative
study
J. Vac. Sci. Technol. B 30, 031206 (2012); 10.1116/1.3701945
Three-dimensional simulations of a thin film heterojunction solar cell with a point contact/defect passivation
structure at the heterointerface
Appl. Phys. Lett. 95, 122108 (2009); 10.1063/1.3233962
Varying the Schottky barrier of thin film Mg ∕ H : p - Si ( 111 ) contacts: Properties and applications
J. Vac. Sci. Technol. A 25, 950 (2007); 10.1116/1.2484574
Schottky junction properties on high quality boron-doped homoepitaxial diamond thin films
J. Vac. Sci. Technol. B 22, 2084 (2004); 10.1116/1.1768186
Reuse of AIP Publishing content is subject to the terms at: https://publishing.aip.org/authors/rights-and-permissions. IP: 148.224.6.252 On: Wed, 24 Feb 2016 14:38:49
JOURNAL OF APPLIED PHYSICS 119, 084501 (2016)
Reuse of AIP Publishing content is subject to the terms at: https://publishing.aip.org/authors/rights-and-permissions. IP: 148.224.6.252 On: Wed, 24 Feb 2016 14:38:49
084501-2 Marco Nardone J. Appl. Phys. 119, 084501 (2016)
such a contacting scheme can in fact produce a built-in field approach can significantly increase the efficiency while obvi-
that effectively separates charge carriers if the metals are of ating the need for a heterojunction.
sufficiently different work functions. The back contact mate- This paper is structured as follows. The pertinent
rials used in that work were Al, Au, and MoO3-coated Au physics and analytical estimates of characteristic lengths of
with measured work functions of 4.1, 5.1, and 5.6 eV, respec- an ABSC device are provided in Sec. II. Section III describes
tively. That work also demonstrated the feasibility of using the numerical methods and input parameters used to model
UV-lithography to create micron-scale metal lines separated the devices. A comparison of the experimental data with
by channels as narrow as 200 nm. Here, we investigate how simulation results for the demonstrated CdSe/CdTe cell,
such a contacting scheme may be used to fabricate high- along with the predictions for the ABSC cell, is provided in
efficiency TFPV cells with absorber layers comprised poly- Sec. IV. The discussion and conclusions are included in
crystalline materials, such as CdTe or CIGS. Secs. V and VI.
A series of recent works10–14 describes the development
of CdS/CdTe and CdSe/CdTe heterojunction PV devices
with micron-scale IBCs of one type of metal (Pt or Ir). The II. PHYSICAL CHARACTERISTICS OF THE ABSC
n-type CdS or CdSe layer was selectively electrodeposited DEVICE
on one of the electrodes, and then a conformal layer of p- The basic physics of ABSC-TFPV is similar to that of
type CdTe was deposited on both (see Fig. 3). Hence, the p-n metal-semiconductor-metal (MSM) structures, but with
heterojunction was located near the back of the device. The unique features due to both transverse and lateral electric
highest efficiency of 5.9% was achieved for a CdSe/CdTe fields. For example, the electric potential varies in 2D, such
cell.14 We demonstrate below that a major limitation in that that the standard 1D energy band diagram is insufficient and
design is the use of a single type of metal contact with a the 2D bands follow the contours in Fig. 1(b), where it can
large work function (5.5 eV). The bi-metallic ABSC be seen that both transverse and lateral electric fields are
generated, with the magnitudes indicated in Fig. 1(c) (the
method of calculation is described below). The conduction
band (CB) surface is illustrated in Fig. 2 where the same pa-
rameter values were used as in Fig. 1 (see Table I). The steep
CB gradient between the metals (lateral direction) and the
FIG. 2. 2D conduction band relative to the Fermi level (gray plane at 0 eV)
at equilibrium. The back of the device is at h ¼ 0 and the location of the two
metal contacts is as shown. FIG. 3. Schematic layout of the IBC CdTe/CdSe device.
Reuse of AIP Publishing content is subject to the terms at: https://publishing.aip.org/authors/rights-and-permissions. IP: 148.224.6.252 On: Wed, 24 Feb 2016 14:38:49
084501-3 Marco Nardone J. Appl. Phys. 119, 084501 (2016)
TABLE I. Input parameter values. CdSe and CdTe values from Refs. 14 and through Schottky junction establishes the lateral built-in
34. Eg—bandgap, v—electron affinity, r—relative permittivity, Nc and field.
Nv—effective density of states in conduction and valence bands, ln and
lp—electron and hole mobilities, Na and Nd—acceptor and donor doping,
Another important condition ispffiffiffiffiffiffi
related to the diffusion
Nt—defect concentration, Et—defect energy above valence band, and rn and length of minority carriers, Ln ¼ Ds, because they must
rp—electron and hole capture cross-sections. migrate from the quasi-neutral zone above M1 to the high-
field/depleted region above the insulator and, finally, to M2.
Parameter Unit Generic CdTe CdSe
Therefore
Eg eV 1.5 1.41 1.74 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
v eV 4.0 4.28 4.58 wM1 ⱗ 2 lskT=q; (4)
r [0] 10 9.4 9.4
Nc cm3 1018 7:5 1017 7:5 1017 where the Einstein relation was used for the diffusion coeffi-
Nv cm3 1019 1:8 1018 1:8 1018 cient, D ¼ lkT=q with thermal voltage kT/q, minority carrier
ln cm2/(V s) 100 500 500 mobility l, and s is the minority carrier lifetime. The factor
lp cm2/(V s) 10 60 60 of 2 in Eq. (4) accounts for the requirement of carrier migra-
Na cm3 1014 5 1015 … tion for half of the contact width. The conditions on the
Nd cm3 … … 5 1015
width of metal M2 appear to be less stringent since it will
Nt cm3 1014 5 1014 2 1014
depend mostly on the majority carrier diffusion length Lp,
Et eV 0.75 0.75 0.75
rn cm2 1012 1012 1015
hence
rp cm2 1015 1015 1012
wM2 ⱗ 2Lp : (5)
relatively small gradient between the back and front surfaces Front surface conditions (charge accumulation, front surface
(transverse direction) are evident in Fig. 2. fields, etc.) will strongly affect the field above M2 (discussed
Stacked, planar devices are always modeled with the further below) and, therefore, minority carrier collection.
directions of light propagation and carrier currents parallel to Using typical TFPV values of a1 ¼ 0:1 lm,
each other. That is not the case in ABSC devices, however, WD ¼ 1 lm, Ln ¼ 1 lm, and Lp ¼ 10 lm, Eqs. (2)–(5) suggest
and a 2D analysis is required to understand the unique inter- length scales of h 1 lm, wins 1 lm, wM1 2 lm, and
play between drift and diffusion currents relative to the wM2 20 lm.
direction of light propagation. There are several characteris-
tic lengths to consider (in what follows, we assume a p-type III. NUMERICAL METHODS AND MODELS
semiconductor and refer to the spatial variables shown in
Fig. 1(a)). First, the depletion width, WD, of the Schottky The numerical methods consisted of solving the Poisson
contact at M2 (cathode, or low work function metal) should equation coupled with the continuity equations for electron
be within the absorber thickness; h ⱗ WD to establish a suffi- and hole drift-diffusion currents using the finite volume
cient field of E > 103 V/cm throughout the absorber. The method with the Scharfetter-Gummel scheme (typical for
depletion width is given by29 numerically solving the semiconductor equations).30 Since
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi the devices we are considering lack the symmetry to be mod-
20 r wbi R
eled in one dimension, COMSOLV Multiphysics software
WD ¼ ; (1)
qNa was employed to solve for the electric potential, /ðx; yÞ, and
the electron, n(x, y), and hole, p(x, y), concentrations in two
where q is the elementary charge, wbi is the built-in potential dimensions. The stationary (steady state) solver is based on
of the Schottky barrier, 0 is the permittivity of free space, r Newton’s method which solves the fully coupled system of
is the relative permittivity of the semiconductor, and NA is
equations.
the acceptor concentration. In addition, the semiconductor
Schematics for the modeled ABSC and CdSe/CdTe
must be thick enough to absorb the majority of incident pho-
devices are provided in Figs. 1 and 3, respectively. Input pa-
tons: h > a1 , where a is the absorption coefficient of the
rameters for the bulk properties are provided in Table I. The
semiconductor. Thus, the conditions
“generic” values were used for numerical modeling of the
a1 ⱗ h ⱗ WD (2) ABSC device and are typical of p-type TFPV, such as
CdTe,31 CIGS,31 copper zinc tin sulfide,32 and perovskites.33
provide the optimal range for the thickness. For comparison with the data, we first simulated the proven
The depletion width also plays a role in determining the IBC CdSe/CdTe device using the parameter values14,34
lateral field between the metals which are separated by the shown in Table I. Acceptor-type doping (Na) in the CdTe
width wins. If D/M is the difference in metal work functions, layer was slightly varied to account for different fabrication
a field on the order of E D/M =ðqwins Þ will be generated steps, as described in Sec. IV A. A band gap of Eg ¼ 1:41 eV
between the metals when was used for CdTe to correspond with the previous measure-
wins ⱗ WD ; (3) ments14 for this device. Although that value is somewhat
lower than the typical 1.5 eV, interdiffusion of Se may have
such that the electric field between the metals is not signifi- resulted in the formation of CdTe1xSex, which can have a
cantly screened by space charges. In that case, a reach- band gap as low as 1.32 eV for x ¼ 0.38.35
Reuse of AIP Publishing content is subject to the terms at: https://publishing.aip.org/authors/rights-and-permissions. IP: 148.224.6.252 On: Wed, 24 Feb 2016 14:38:49
084501-4 Marco Nardone J. Appl. Phys. 119, 084501 (2016)
IV. RESULTS
A. CdSe/CdTe IBC device
Figs. 4 and 5 provide the experimental JV and EQE data
extracted from Refs. 12 and 13, as well as our simulation FIG. 5. Measured and simulated EQE characteristics of the CdSe/CdTe IBC
results, for the IBC CdSe/CdTe device. The efficiency of device. Legend is the same as Fig. 4. Data for cell (c) were not available.
Reuse of AIP Publishing content is subject to the terms at: https://publishing.aip.org/authors/rights-and-permissions. IP: 148.224.6.252 On: Wed, 24 Feb 2016 14:38:49
084501-5 Marco Nardone J. Appl. Phys. 119, 084501 (2016)
B. ABSC device
The ABSC device is similar to the CdSe/CdTe device
but with the CdSe layer removed and with the use of a shal-
lower work function metal at the former CdSe contact. Fig. 1
FIG. 7. Efficiency contours as a function of absorber layer thickness and
shows the electric potential and field distribution using the contact widths (assuming wM1 ¼ wM2 ) with the same parameters as Fig. 6
parameter values given in Sec. III. except for an increase of minority carrier lifetime from s ¼ 1 ns to s ¼ 10 ns.
Reuse of AIP Publishing content is subject to the terms at: https://publishing.aip.org/authors/rights-and-permissions. IP: 148.224.6.252 On: Wed, 24 Feb 2016 14:38:49
084501-6 Marco Nardone J. Appl. Phys. 119, 084501 (2016)
efficiency spans from 9% to 21% over a range of contact 16% when Sbc ¼ 107 cm/s (this scenario uses Sfs ¼ 0 cm/s). It
widths 2 < wM1 ¼ wM2 < 22 lm and absorber thicknesses is apparent that Sbc affected efficiency mainly through FF
0:25 < h < 1:5 lm. In this case, efficiency remains above and Voc. The latter monotonically decayed from 1.0 V to
18% for up to 20–lm contact width when the absorber thick- 0.9 V while FF varied non-monotonically, in correspondence
ness is optimized at h ⱗ 1. The optimum thickness decreased with g. The non-monotonic behavior of g and FF was due to
slightly below 1 lm because the lower density of occupied the effect of Sbc on the minority carrier distribution and,
defect states affects space charge and slightly lowers the built- therefore, on the electric field (as is the case in a high-
in field which, in turn, lowers WD in Eq. (1). In addition, wider injection scenario, n Na ). Our calculations show that
contacts are afforded in this case due to the longer lifetime, as increasing Sbc decreases the transverse field at the metal con-
suggested by Eq. (4). tacts while increasing the lateral field above the insulator,
It should be noted that Eq. (1) may be an oversimplifica- resulting in an optimum Sbc.
tion of the depletion width because WD will also depend on: Next, we consider front surface effects as governed by
(i) the compensation of Na by charged defect states and (ii) varying the surface recombination velocity in the range of
the insulating boundary condition at the front surface which 102 Sf s 107 cm/s. We also considered both the worse
requires that the normal component of the field satisfies case scenario of Sbc ¼ 107 cm/s and passivated Sbc ¼ 103 cm/s.
n E ¼ Qt , where is the semiconductor permittivity, n is The profound effect of front surface recombination is evident
the surface normal unit vector, and Qt is the surface charge in Fig. 9 as efficiency drops from 15.3% to 5.3% over the
density due to traps there. Hence, the properties of both the range of Sfs for Sbc ¼ 107 cm/s. Improving the back surface
bulk and surface defects play important roles in determining passivation so that Sbc ¼ 103 cm/s increased efficiency to a
the field distribution. maximum of 19.1% but resulted in no performance gain at
Fig. 8 shows the effects of back contact recombination large values of Sfs, which dominated efficiency reduction.
velocity, Sbc, on the performance metrics assuming the Losses were greatest for FF and Jsc, while Voc exhibited a rel-
dimensions of the circled point in Fig. 6 and minority carrier atively small loss. Qualitatively, those variations are a result
lifetimes of s ¼ 1 ns and s ¼ 10 ns. Maximum efficiencies of of the increase in recombination in the area of greatest photo-
19.5% and 21.5% were achieved at Sbc ¼ 103 cm/s, which absorption (at the front), while the Schottky junction field is
then decayed to nearly stable values between 15.5% and strongest near the back.
Reuse of AIP Publishing content is subject to the terms at: https://publishing.aip.org/authors/rights-and-permissions. IP: 148.224.6.252 On: Wed, 24 Feb 2016 14:38:49
084501-7 Marco Nardone J. Appl. Phys. 119, 084501 (2016)
Reuse of AIP Publishing content is subject to the terms at: https://publishing.aip.org/authors/rights-and-permissions. IP: 148.224.6.252 On: Wed, 24 Feb 2016 14:38:49
084501-8 Marco Nardone J. Appl. Phys. 119, 084501 (2016)
20 30
N. Mingirulli, J. Haschke, R. Gogolin, R. Ferre, T. F. Schulze, J. S. Selberherr, Analysis and Simulation of Semiconductor Devices
D€usterh€oft, N.-P. Harder, L. Korte, R. Brendel, and B. Rech, Phys. Status (Springer Verlag, New York, 1984).
31
Solidi RRL 5, 159 (2011). M. Gloeckler, A. L. Fahrenbruch, and J. R. Sites, in Proceedings of 3rd
21
S.-Y. Lee, H. Choi, H. Li, K. Ji, S. Nam, J. Choi, S.-W. Ahn, H.-M. Lee, World Conference on Photovoltaic Energy Conversion, Osaka, Japan
and B. Park, Sol. Energy Mater. Sol. Cells 120, 412 (2014). (2003), Vol. 1, pp. 491–494.
22 32
A. Tomasi, B. Paviet-Salomon, D. Lachenal, S. Martin de Nicolas, A. M. Patel and A. Ray, Physica B 407, 4391 (2012).
33
Descoeudres, J. Geissbuhler, S. De Wolf, and C. Ballif, IEEE J. Photovoltaics F. Liu, J. Zhu, J. Wei, Y. Li, M. Lv, S. Yang, B. Zhang, J. Yao, and S. Dai,
4, 1046 (2014). Appl. Phys. Lett. 104, 253508 (2014).
23 34
M. Lu, S. Bowden, U. Das, and R. Birkmire, Appl. Phys. Lett. 91, 063507 A. Fahrenbruch, “CSU report: Modeling results for CdS/CdTe solar cells,”
(2007). 2000.
24 35
F. J. Castano, D. Morecroft, M. Cascant, H. Yuste, M. Lamers, A. A. P. J. Sebastian, Thin Solid Films 245, 132 (1994).
36
Mewe, I. G. Romijn, E. E. Bende, Y. Komatsu, A. W. Weeber et al., in X. Mathew, J. Drayton, V. Parikh, N. R. Mathews, X. Liu, and A. D.
37th IEEE Photovoltaic Specialists Conference (PVSC) (IEEE, 2011), pp. Compaan, Semicond. Sci. Technol. 24, 015012 (2009).
37
001038–001042. X.-H. Zhao, M. J. DiNezza, S. Liu, C. M. Campbell, Y. Zhao, and Y.-H.
25
R. Woehl, M. R€udiger, D. Biro, and J. Wilde, Prog. Photovoltaics 23, 226 Zhang, Appl. Phys. Lett. 105, 252101 (2014).
38
(2015). C. H. Swartz, M. Edirisooriya, E. G. LeBlanc, O. C. Noriega, P.
26
J. Haschke, D. Amkreutz, L. Korte, F. Ruske, and B. Rech, Sol. Energy Jayathilaka, O. S. Ogedengbe, B. L. Hancock, M. Holtz, T. H. Myers, and
Mater. Sol. Cells 128, 190 (2014). K. N. Zaunbrecher, Appl. Phys. Lett. 105, 222107 (2014).
27 39
M. J. Keevers, T. L. Young, U. Schubert, and M. A. Green, in 22nd European R. Cohen, V. Lyahovitskaya, E. Poles, A. Liu, and Y. Rosenwaks, Appl.
Photovoltaic Solar Energy Conference, Milan (2007), pp. 235–239. Phys. Lett. 73, 1400 (1998).
28 40
S. Jeong, M. D. McGehee, and Y. Cui, Nat. Commun. 4, 2950 (2013). C. K. Kang, S. U. Yuldashev, J. H. Leem, Y. S. Ryu, J. K. Hyun, H. S.
29
S. Sze and K. K. Ng, Physics of Semiconductor Devices, 3rd ed. (Wiley, Jung, H. J. Kim, T. W. Kang, H. I. Lee, Y. D. Woo, and T. W. Kim,
New York, 2007). J. Appl. Phys. 88, 2013 (2000).
Reuse of AIP Publishing content is subject to the terms at: https://publishing.aip.org/authors/rights-and-permissions. IP: 148.224.6.252 On: Wed, 24 Feb 2016 14:38:49