H. H. Choi, K. Cho, C. D. Frisbie, H. Sirringhaus and V. Podzorov, Critical assessment of charge mobility extraction in FETs, Nat Mater, 2018, 17(1): 2. (DOI: https://doi.org/10.1038/nmat5035)
要点
迁移率正确的表征方法
Shockley equations
物理机理前提
linear relationship between the carrier density n n n and conductivity σ σ σ: σ = μ e n σ = μen σ=μen.
gradual channel approximation
(a) the transverse gate electric field is much greater than the longitudinal source–drain electric field,
(b) the mobility is carrier density independent
电场强度gate方向>>source-drain方向;载流子浓度无关的迁移率。
提取方法前提
Equations (1) and (2) are only applicable when dependences I S D ( V G ) I_{SD}(V_G) ISD(VG) in the linear regime and ∣ I S D ∣ 1 / 2 ( V G ) |I_{SD}|^{1/2}(V_G) ∣ISD∣1/2(VG) in the saturation regime are linear in an extended range of V G V_G VG.
接触电阻和短沟道效应的影响
provided that other extrinsic sources of error, such as short-channel effects or high contact resistance, are excluded
无序系统
无序系统,迁移率与载流子浓度有关。
In disordered systems, μ could be carrier-density dependent (for instance, due to band-tail filling)
Y. Xia, J. H. Cho, J. Lee, P. P. Ruden and C. D. Frisbie, Comparison of the Mobility-Carrier Density Relation in Polymer and Single-Crystal Organic Transistors Employing Vacuum and Liquid Gate Dielectrics, ADVANCED MATERIALS, 2009, 21(21): 2174. (DOI: https://doi.org/10.1002/adma.200803437)
非线性
Nonlinearities in Fig. 1a–c might have diverse microscopic origins (partially discussed in the recommendations below), including contact effects, carrier density-dependent mobility and nonequilibrium biasing of short-channel OFETs.
the hump originates from gated Schottky contacts, and the intrinsic carrier mobility corresponds to the plateau.
Similarly, double-slope characteristics in donor– acceptor polymer FETs, originating from minority carrier injection and trapping
T. Okachi, T. Kashiki and K. Ohya, Device operation mechanism of field-effect transistors with high mobility donor-acceptor polymer semiconductors, Proc Spie, 2015, 9568((DOI: https://doi.org/10.1117/12.2187572)
increased contact resistance strongly suppresses the current at low V G V_G VG but leads to a more rapid increase in I S D I_{SD} ISD at higher V G V_G VG, hence resulting in overestimation of mobility.
reliability factor
μ
e
f
f
≡
r
×
μ
c
l
a
i
m
e
d
μ_{eff} ≡ r × μ_{claimed}
μeff≡r×μclaimed
注意点
…working with unpatterned semiconductors that lead to fringe currents.
Finally, applying source–drain electric fields exceeding the breakdown field of air ( E b r k d = 3 × 1 0 4 V c m – 1 E_{brkd} = 3×10^4 V cm^{–1} Ebrkd=3×104Vcm–1) in top-contact FETs should be avoided, as this can make measurements unreliable. ( 10 μ m ; 30 V 10 \mu m; 30 V 10μm;30V)
(a) μ should only be extracted from linear regime measurements (equation (1)),
(b) contact resistance must be characterized (see below),
© the observation of a behaviour far from the ideal Shockley equations should be explicitly acknowledged, and
(d) the reliability factor r should be quoted.
We also recommend that
(e) authors show the fit to experimental output characteristics (
I
S
D
(
V
S
D
)
I_{SD}(V_{SD})
ISD(VSD) curves measured at several
V
G
V_G
VG values in the range
0
<
∣
V
G
∣
<
∣
V
S
D
∣
m
a
x
0 < |V_G| < |V_{SD}|_{max}
0<∣VG∣<∣VSD∣max) using the Shockley FET model with the claimed μ: such fits are important as they easily show how well the devices are described by the gradual channel model.
In FETs with n-independent μ and negligible contact resistance, μlin and μsat should match
(a) not contact-limited (that is, Schottky contact resistance is much lower than the channel resistance in the VG range used for mobility extraction), and
(b) maintained at thermal and electrical equilibrium during measurements.
d0 is the accumulation layer thickness of a few molecular layers (for definitiveness, we take d0 = 3 nm).
For instance, only a few volts were applied to the 2.5-mm-long channel of rubrene OFET in Fig. 3, resulting in equilibrium transport with very low j S D m a x < 0.2 k A c m – 2 j_{SD}^{max} < 0.2 kA cm^{–2} jSDmax<0.2kAcm–2 and P m a x ≈ 7 m W c m – 2 Pmax ≈ 7 mW cm^{–2} Pmax≈7mWcm–2 (equations (3) and (4)), yet a high μ ≈ 11 c m 2 V – 1 s – 1 μ ≈ 11 cm^2 V^{–1} s^{–1} μ≈11cm2V–1s–1.
This includes accurate measurements of channel geometry (L and W), the minimization of current spreading by semiconductor patterning and the use of correct gate dielectric capacitance measured at sufficiently low frequencies (f < 100 Hz).
contact check
Besides these measurements, yielding RC and a contact-corrected μ, a simple test for severity of contact effects is a V S D V_{SD} VSD-scaling check: in the linear regime, FETs that are not significantly affected by contacts should exhibit I S D I_{SD} ISD proportional to V S D V_{SD} VSD (that is, V S D V_{SD} VSD-independent σ σ σ and μ 2 p μ_{2p} μ2p), as shown in Fig. 3a.