IEEE ELECTRON DEVICE LETTERS, VOL. 35, NO. 2, FEBRUARY 2014 229
Dependence of the Noise Behavior on the
Drain Current for Thin Film Transistors
Ya-Hsiang Tai, Chun-Yi Chang, Chung-Lun Hsieh, Yung-Hsuan Yang,
Wei-Kuang Chao, and Huan-Ean Chen
Abstract— In this letter, a noise formula is newly proposed
to calculate the low frequency noise for the three kinds of amorphous silicon, low temperature polycrystalline silicon, and amorphous indium-gallium-zinc oxide (a-IGZO) thin film tran-sistors (TFTs). It is found that the noise behavior of the TFT depends on its drain current in a simple manner. Based on the analysis, the ratios of drain current to the noise level for these TFTs are compared. It reveals that a-IGZO TFT is the best candidate to be used in the active pixel sensor.
Index Terms— Thin-film transistor (TFTs), low frequency noise
(LFN), amorphous indium-gallium-zinc oxide (a-IGZO), active pixel sensor (APS).
I. INTRODUCTION
R
ECENTLY, active matrix flat panel technology based on thin-film transistors (TFTs) have gained considerable significance in large area flat panel digital imaging applications in view of their large area readout capability. The pixel archi-tecture progresses from passive pixel sensor (PPS) [1] to active pixel sensor (APS) [2]. In the APS circuits, the TFT is used to amplify the sensing voltage at the gate to the drain current by its transconductance to be read out. In the mean time, the noise current of the TFT is also gathered by the external readout circuit. Thus, many reports studied on the low frequency noise (LFN) properties of TFTs [3]–[7], which include the most popular TFT technologies, namely, amorphous silicon (a-Si), low temperature polycrystalline silicon (LTPS), and amorphous indium gallium zinc oxide (a-IGZO). In these reports, the noise spectrum is expressed in a function of gate voltage VG and drain current ID asS= αHq/ f ∗ W LCO X|VG− VT|I
2
D (1)
where is Hooge parameter [8], q is the elementary electron charge, f is the frequency, W and L are the channel width and length, respectively, CO X is the gate dielectric capacitance per unit area, and VT is the threshold voltage. The Hooge parameter is a good index to characterize and discuss the noise
Manuscript received October 18, 2013; accepted October 30, 2013. Date of publication January 2, 2014; date of current version January 23, 2014. This work was supported by the Industrial Technology Research Institute and National Science Council of China under Grant NSC 100-2628-E-9-21-MY. The review of this letter was arranged by Editor K. Uchida.
Y.-H. Tai is with the Department of Photonics and Display Institute, National Chiao Tung University, Hsinchu 30010, Taiwan.
C.-Y. Chang, C.-L. Hsieh, and Y.-H. Yang are with the Department of Photonics and Institute of Electro-Optical Engineering, National Chiao Tung University, Hsinchu 30010, Taiwan (e-mail: kid78423.eo00g@nctu.edu.tw).
W.-K. Chao and H.-E. Chen are with the Display Technology Center, Industrial Technology Research Institute, Hsinchu 30010, Taiwan.
Color versions of one or more of the figures in this letter are available online at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/LED.2013.2291565
Fig. 1. Drain-current noise spectral densities of the a-IGZO TFTs and background noise measured at different VGfrom−1 to 15 V and a constant
VD of 10 V.
properties for different materials. However, since ID of the TFT depends on VG and the dependence varies with material [9], it would be complicated when the noise in the operation of APS circuit is considered. Thus, we would like to propose a new expression of the noise formula only depending on ID. Thus, we can determine the most suitable TFT technology to be adapted in the view point of signal-to-noise ratio (SNR).
II. EXPERIMENTALPROCEDURE
The fabrication processes of the a-Si, LTPS, and a-IGZO TFTs in this letter are described elsewhere in [10], [11], and [12], accordingly. The channel widths of these TFTs are 20 μm, 15 μm, and 20 μm, respectively, and their channel lengths are all 5μm. The gate and drain electrodes of the TFT are biased with lithium batteries in series and the source is virtually grounded by connecting it to the current preamplifier (Signal Recovery 5182) which converts ID to a measurable voltage signal. The measurement setup, including breadboard that carries the device under test and the current preamplifier, is put in a shielding box to shield out the interference in environment. The voltage signal is linked out of the shield-ing box to a signal source analyzer (Agilent E5052B) with transmission lines.
III. RESULTS ANDDISCUSSION
The LFN spectral densities for the a-IGZO TFT measured at different VG from−1V to 15 V and a constant VD of 10 V are shown in Fig. 1. It is observed that all the power noise spectrums follow the 1/ f dependence. For the a-Si and LTPS TFTs, even though not shown here, the 1/ f dependence of the
0741-3106 © 2014 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
230 IEEE ELECTRON DEVICE LETTERS, VOL. 35, NO. 2, FEBRUARY 2014
Fig. 2. Dependences of IDand N i A on VGfor the a-Si, LTPS, and a-IGZO
TFTs.
power spectrum and is also observed. The Hooge parameters
αH of the a-Si, LTPS, and a-IGZO TFTs are extracted to be 0.511, 0.595, and 0.00258, respectively, which are consistent with the previous reports [5]–[7]. According to these Hooge parameters, a-IGZO TFT is the device with the lowest noise, but how it can perform in APS is yet to be determined.
Here, we try to analyze the LFN behavior in another approach by integrating the noise power spectrum from 10Hz to 1 KHz and then taking the square root, as the following definition:
Ni A =
1K H z 10H z
Sd f (2)
where NiA is a new index to represent the noise in ampere. The noise spectrums at various VG and VD are measured, and the respective NiA are calculated and plotted against VG together with ID, as shown in Fig. 2. As can be seen, even though the graphs are in different scales, the curves of NiA coincide with the curves of ID for all the devices, expect for the low VG region where the measured noise level is limited by the noise background of the current preamplifier. It suggests the strong correlation between NiA and ID.
Follow the Fig. 1, the background is lower the TFT noise. And thus, the measured noise above the background noise is discussed only. The background noise comproses the sig-nal source asig-nalyzer (SSA), sheilding box and breadboad. And then, ignoring the noise background, the NiA and ID are correlated in a simple manner of power-law dependence. In Fig. 3, the curves of IDversus NiA are plotted in logarithmical
Fig. 3. NiA versus ID with both axes in logarithmical scale for the three
the TFTs.
scale for the three the TFTs. The linearity of the curves confirms that their correlation is universal for various TFTs, which is in the form of:
ID= K ∗ Ni Am (3) where m and K are fitting parameters. The fitted values of m for the a-Si, LTPS and a-IGZO TFTs are 1.38, 1.69 and 1.77, respectively, and the fitted K values are 8.73×107, 7.59×109 and 2.61×1011, accordingly.
Even the newly proposed expression describes the relation between NiA and ID very well, we wonder if there is a theoretical ground for it. Thus, we compare the conventional noise expression in Eq. (1) with ours in Eq. (3). Putting the S in Eq. (1) into Eq. (2), the square of NiA can be written as:
Ni A2= (αHq ln(100)/W LCO X)(ID2/|VG− VT|) (4) Taking a close look at Eq. (4), the items in the first bracket are all constants for a device. Thus, as long as the|VG− VT| in the second bracket is in the relation of power-law to ID, NiA can be expressed by the only variable ID on the right hand side of the Eq. (4). In this case, Eq. (4) can be reduced to the form of Eq. (3). Now the question comes to whether ID versus|VG− VT| is in the power-law relation. We assume that current of the TFT follows
ID = 1/2μCO XW
L(VG− VT)β
Vre fβ−2 (5) whereμ is the field effect mobility and Vre fβ−2is a constant and reference value. Substituting Eq. (5) into Eq. (4), we obtain:
Ni A2 = αHq ln(100) W LCO X × 1 2μCO X W L 1/β ID2−1/β (6) Comparing Eq. (6) and the square of Eq. (3), the exponential term gives:
m= 2β2β − 1 (7)
The value of β plays an important role in the derivation above to get the m and K values of Eq. (3). They are extracted
TAI et al.: DEPENDENCE OF THE NOISE BEHAVIOR ON THE DRAIN CURRENT 231
Fig. 4. The curves of SNR in dB scale versus IDin logarithm scale.
from the ID− VG curves of a-Si, LTPS, a-IGZO TFTs to be 1.38, 1.08, and 1.06, respectively, which correspond to the m values of 1.57, 1.86, and 1.89. These values are consistent with those values previously obtained from Fig. 3.
In the APS application, ID of the amplifying TFT is measured to be the signal [2], while the noise current is also incorporated in the measurement. In this case, the best SNR that a TFT can provide is expected as the ratio of signal power to the noise power:
S N R = ID2/Ni A2 (8) Accordingly, we can transform Fig. 3 to Fig. 4, which shows the SNR in dB scale versus log (ID) and provides us a better way to evaluate the noise behavior of the TFTs. As shown in Fig. 4, the IGZO TFT has SNR higher than 100dB, which corresponding to 16-bit of the analog-to-digital converter (ADC) system for the APS application. Furthermore, it is seen that a-IGZO TFT provides higher signal current to be more easily read out. As a result, the IGZO TFT is the best choice to be used in the active pixel sensor [13].
IV. CONCLUSION
In this letter, a new index for the noise behavior of TFTs is proposed to express the noise level in ampere. With this new
index, the noise behavior can be described to be simple function of the drain current for the TFTs made of different materials. Considering both the current signal and noise levels, a-IGZO TFT is the most suitable device to be used in the APS applications.
REFERENCES
[1] A R. A. Street, X. D. Wu, R. Weisfield, et al., “Two dimensional amorphous silicon image sensor arrays,” in Proc. MRS Symp., vol. 377. 1995, pp. 757–766.
[2] N. Faramarzpour, M. J. Deen, and S. Shirani, “An approach to improve the signal-to-noise ratio of active pixel sensor for low-light-level appli-cations,” IEEE Electron Devices Lett., vol. 53, no. 9, pp. 2384–2391, Sep. 2006.
[3] J. M. Lee, W. S. Cheong, C. S. Hwang, et al., “Low-frequency noise in amorphous indium-gallium-zinc-oxide thin-film transis-tors,” IEEE Electron Device Lett., vol. 30, no. 5, pp. 505–507, May 2009.
[4] T. C. Fung, G. Baek, and J. Kanicki, “Low frequency noise in long channel amorphous In-Ga-Zn-O thin film transistors,” J. Appl. Phys., vol. 108, no. 7, pp. 074815-1–074518-10, 2010.
[5] I. T. Cho, W. S. Cheong, C. S. Hwang, et al., “Comparative Study of the Low-frequency-noise behaviors in a-IGZO thin-film transistors with Al2O3and Al2O3/SiNx gate dielectrics,” IEEE Electron Device Lett.,
vol. 30, no. 8, pp. 828–830, Aug. 2009.
[6] L. Y. Su, H. K. Lin, C. C. Hung, et al., “Role of HfO2 /SiO2 gate
dielectric on the reduction of low-frequent noise and the enhancement of a TFT electrical performance,” J. Display Technol., vol. 8, no. 12, pp. 695–698, 2012.
[7] J. Rhayem, D. Rigaud, M. Valenza, et al., “1/f noise modeling in long channel amorphous silicon thin film transistors,” J. Appl. Phys., vol. 87, no. 4, pp. 1983–1989, 2000.
[8] F. N. Hooge, “1/f noise is no surface effect,” Appl. Phys. Lett., vol. 29, no. 3, pp. 139–140, 1969.
[9] J. Rhayem, D. Rigaud, M. Valenza, et al., “SPICE models for amorphous silicon and polysilicon thin film transistors,” J. Electrochem. Soc., vol. 144, no. 8, pp. 2833–2839, 1997.
[10] Y. H. Tai, M. H. Tsai, and S.-C. Huang, “The linear combination model for the degradation of amorphous silicon thin film transistors under drain AC stress,” Jpn. J. Appl. Phys., vol. 47, no. 8, pp. 6228–6235, 2008.
[11] Y. H. Tai, S. C. Huang, and P. T. Chen, “Degradation mechanism of poly-Si TFTs dynamically operated in OFF region,” IEEE Electron Device
Lett., vol. 30, no. 3, pp. 231–233, Mar. 2009.
[12] Y. H. Tai, H. L. Chiu, and L. S. Chou, “The deterioration of a-IGZO TFTs owing to the copper diffusion after the process of the source/drain metal formation,” J. Electrochem. Soc., vol. 159, no. 5, pp. 200–203, 2012.
[13] A. Carbone, C. Pennetta, and L. Reggiani, “Trapping-detrapping fluctu-ations in organic space-charge layers,” Appl. Phys. Lett., vol. 95, no. 23, pp. 233303–233305, 2009.
