Electrical Characteristics of Charge Trap Flash Memory with a Composition Modulated (ZrO<sub>2</sub>)<sub>x</sub>(Al<sub>2</sub>O<sub>3</sub>)<sub>1-x</sub> Film
Electrical Characteristics of Charge Trap Flash Memory with a Composition Modulated (ZrO2)x(Al2O3)1-x Film
Transactions on Electrical and Electronic Materials. 2015. Jun, 16(3): 130-134
Copyright © 2015, The Korean Institute of Electrical and Electronic Material Engineers
This is an open-access article distributed under the terms of the Creative Commons Attribution Non-Commercial License ( which permits unrestricted noncommercial use, distribution, and reproduction in any medium, provided the original work is properly cited.
  • Published : June 25, 2015
Export by style
Cited by
About the Authors
Zhenjie, Tang
Jing, Zhang
Yunhong, Jiang
Guixia, Wang
Rong, Li
Xinhua, Zhu

This research proposes the use of a composition modulated (ZrO 2 ) x (Al 2 O 3 ) 1-x film as a charge trapping layer for charge trap flash memory; this is possible when the Zr (Al) atomic percent is controlled to form a variable bandgap as identified by the valence band offsets and electron energy loss spectrum measurements. Compared to memory devices with uniform compositional (ZrO 2 ) 0.1 (Al 2 O 3 ) 0.9 or a (ZrO 2 ) 0.92 (Al 2 O 3 ) 0.08 trapping layer, the memory device using the composition modulated (ZrO 2 ) x (Al 2 O 3 ) 1-x as the charge trapping layer exhibits a larger memory window (6.0 V) at the gate sweeping voltage of ±8 V, improved data retention, and significantly faster program/erase speed. Improvements of the memory characteristics are attributed to the special energy band alignments resulting from non-uniform distribution of elemental composition. These results indicate that the composition modulated (ZrO 2 ) x (Al 2 O 3 ) 1-x film is a promising candidate for future nonvolatile memory device applications.
With conventional nonvolatile flash memory devices approaching their scaling limitation, tremendous effort has been made to explore the next generation nonvolatile solid state memory devices for use in mobile electronics such as digital cameras, MP3 players, and mobile phones [1 - 5] . Among the family of nonvolatile memories, the charge trap flash memory (CTFM) is one of the most promising candidates due to its advantages such as better scalability, low power, and process simplicity [6] . Nevertheless, one problem in a conventional CTFM device with an Si 3 N 4 trapping layer is the small potential barrier, which leads to poor retention characteristics. Since the Si 3 N 4 based CTFM device faces the challenge of the trade-off between data retention and program/erase speed [7] . In order to further improve memory characteristics and achieve the trade-off among the memory window, program/erase speed and data retention, extensive research has been carried out in recent years. This research has involved the use of nanocrystals (e.g., Ge [8] , Si [9] , Ni [10] , and NiSi [11] ), stacked structure [12 - 14] , binary oxide dielectric film (e.g.,HfO 2 [15] , GdO [16] , and ZrO 2 [17] ), and multiple oxide dielectric film [18 , 19] as a charge trapping layer. This work proposes a bandgap engineering technique using a composition modulated (ZrO 2 ) x (Al 2 O 3 ) 1-x film as the charge trapping layer in the CTFM device to improve the memory characteristics. This involves the formation of a variable bandgap due to the non-uniform distribution of elemental composition in the charge trapping layer causing different energy band offsets.
P-type Si (100) substrates with a resistivity of 3~20 Ω cm were cleaned by the standard Radio Corporation of America process to remove native oxide. Then a 3 nm Al 2 O 3 film (30 deposition cycles) as the tunneling oxide (TO) was deposited by atomic layer deposition (ALD) using trimethylaluminium Al(CH 3 ) 3 precursors at a substrate temperature of 300℃. Following that, a 10 nm composition modulated (ZrO 2 ) x (Al 2 O 3 ) 1-x film (denoted as CM-ZAO) was deposited as the charge trapping layer by ALD using ZrCl 4 and Al(CH 3 ) 3 precursors, and O 3 vapor served as the oxygen source. The ratio of Zr (Al) in the charge- trapping layer first increased (decreased) and then decreased (increased) by controlling the number of ALD deposition cycles. In the process, we deposited 10 units, and every 10 cycles including Al 2 O 3 /ZrO 2 /Al 2 O 3 were a unit. We successively deposited Al 2 O 3 /ZrO 2 /Al 2 O 3 with different cycles in every unit, and the numbers are 5/2/3, 4/3/3, 3/5/2, 2/6/2, 1/8/1, 2/6/2, 2/5/3, 3/3/4, 3/2/5, and 7/2/1 to realize nonuniform distribution of elemental composition. Subsequently, a 12 nm Al 2 O 3 (120 deposition cycles) was deposited by ALD as the blocking oxide (BO). A post deposition annealing process was carried out at a temperature of 900℃ for 30 s in N 2 ambience. Finally, Pt gate electrodes with an area of 7.85 × 10 -5 cm 2 were formed by a magnetron sputtering technique, and the silver glue was then spread on the side of the Si substrates to act as the bottom electrodes. The above fabricated memory device was labeled as S1, and the structural schematic diagram was shown in Fig. 1(a) . To confirm the role of the CM-ZAO film, control samples with single uniform 10 nm (ZrO 2 )0.1(Al 2 O 3 )0.9 or (ZrO 2 )0.92(Al 2 O 3 )0.08 as the charge trapping layer were simultaneously fabricated, which were named S2 and S3, respectively. In the process, ZrCl 4 and Al(CH 3 ) 3 precursors, and O 3 vapor served as an oxygen source were used to deposit the (ZrO 2 ) 0.1 (Al 2 O 3 ) 0.9 and (ZrO 2 ) 0.92 (Al 2 O 3 ) 0.08 films. We deposited 10 Al 2 O 3 /ZrO 2 /Al 2 O 3 units, and the number of cycles in every unit was 5/1/4 and 1/4/1, respectively. The precursor temperatures were kept at 180℃. The X-ray photoelectron spectroscopy (XPS) was used to investigate the elemental composition and energy band alignment of the memory devices. The selected area electron diffraction (SAED) pattern and cross-sectional microstructures of the device were examined by transmission electron microscopy (TEM). The electrical characteristics of the memory devices including memory windows, program/erase speed and data retention characteristics were analyzed by the Keithely 4,200 semiconductor characterization system.
PPT Slide
Lager Image
(a) Schematic diagram and (b) cross-sectional TEM image of S1 with the CM-ZAO charge-trapping layer. The inset is a selected area electron diffraction (SAED) pattern of the CM-ZAO film.
Figure 1(b) is the cross sectional TEM image of the S1, which shows distinct interfaces between the TO, CM-ZAO and BO. Meanwhile, the SAED was performed in the region of the CMZAO, as shown in the inset of Fig. 1(b) , and there was no diffraction spot. The results demonstrated that the CM-ZAO was still in an amorphous state after rapid thermal annealing treatment at 900℃ for 30 s in N 2 ambient.
Since the elemental compositions play a key role in energy band alignment of memory devices [20] , it is important to analyze the composition’s distribution in the charge-trapping layer for S1, S2, and S3. During the process, the memory structures were etched from BO to TO layer by using Ar+ beam sputtering to detect the XPS signals, and the etch rate is 1 nm/min. The position of the CM-ZAO/BO (TO/CM-ZAO) interface was confirmed until the signal of Zr element appear (disappear). Table 1 shows the atomic percent (at.%) distribution of Zr and Al as well as an x value in the charge-trapping layer for the three samples. The Zr (Al) at.% first increases (decreases) and then decreases (increases), and the maximum (minimum) value appears at the middle of the CM-ZAO for S1. It is worth noting that the elemental compositions show almost symmetrical distribution on both sides of the middle position of CM-ZAO. Nevertheless, for S2 and S3, the Zr (Al) at.%, which respectively corresponds to TO/CMZAO and the middle of CM-ZAO in S1, is almost constant in the charge-trapping layer. Based on the above results, we measured the valence band spectra and O1s electron energy loss spectra by XPS at four positions (TO/Si interface, CM-ZAO/TO interface, from CM-ZAO/TO 3 nm and middle of CM-ZAO) in order to obtain the energy band alignment of the S1.
Elemental compositions in the distribution in the charge trapping layer for S1, S2, and S3 samples (measured from the TO/charge trapping layer to the charge trapping layer/BO interface) estimated by XPS. The ZrO2mole fraction value x are also given in the table, where The Zr at.%=x/(5-2x) and the Al at.%=2(1-x)/(5-2x).
PPT Slide
Lager Image
Elemental compositions in the distribution in the charge trapping layer for S1, S2, and S3 samples (measured from the TO/charge trapping layer to the charge trapping layer/BO interface) estimated by XPS. The ZrO2 mole fraction value x are also given in the table, where The Zr at.%=x/(5-2x) and the Al at.%=2(1-x)/(5-2x).
The valence band maximum (VBM) of the cleaned Si surface is determined to be
PPT Slide
Lager Image
by using the linear extrapolation method [21] as shown in Fig. 2(a) . Using the same method the VBM at the above four positions were 3.8 eV, 3.3 eV, 2.9 eV, and 2.3 eV, respectively. Therefore, the valence band offset (VBO) that is relative to the Si substrate was calculated as 3.4 eV, 2.9 eV, 2.5 eV, and 1.9 eV by using formula
PPT Slide
Lager Image
for the above measured positions, respectively. The band gap (E g ) of TO was determined to be 7.3 eV by the onsets of O1s electron energy loss spectra collected at the interfaces of the TO/Si interface, as shown in Fig. 2(b) . Similarly, the bandgaps at CM-ZAO/ TO, from CM-ZAO/TO 3 nm and the middle of CM-ZAO were 6.0 eV, 5.2 eV, and 4.2 eV, respectively. Therefore, using the formula of
PPT Slide
Lager Image
the calculated value of the conduction band offset (CBO) at TO/Si, CM-ZAO/TO, from CM-ZAO/TO 3 nm and the middle of CM-ZAO were 2.8 eV, 2.0 eV,1.6 eV, and 1.2 eV, respectively, where is 1.1 eV. According to the composition of the symmetric distribution around the middle of the CM-ZAO film, the energy band alignment of the S1 was schematically depicted in Fig. 3(a) . The bandgap of CM-ZAO is determined by the Zr and Al content in the film, which monotonously decreases with increasing Zr content, or increases with decreasing Al content. Therefore, the potential barrier E B between TO and CM-ZAO (BO and CM-ZAO) is variable and in the range of 0.8 to 1.6 eV. By comparison, the charge-trapping layer has a uniform elemental composition distribution resulting in a changeless potential barrier for S2 and S3, as shown in Fig. 3(b) .
PPT Slide
Lager Image
(a) Valence band spectra and (b) O 1s electron energy loss spectra at the Si surface, TO/Si interface, CM-ZAO/TO interface, from CM-ZAO/TO 3nm and middle of CM-ZAO.
PPT Slide
Lager Image
Schematic energy band alignment of (a) S1, (b) S2 (dotted line), and S3 (dash dotted lines).
Figure 4(a) shows high frequency (1 MHz) C-V hysteresis of S1 with different gate sweeping voltages to observe its charge storage effect. The C-V curves exhibit an obvious memory window (1 V) in a small sweeping range of ±2 V. When the gate sweeping voltage range is increased to ±8 V, a significant memory window of 6.0 V can be observed. In addition, the flatband voltage shift (ΔV FB ) of the S1, S2, and S3 are shown in Fig. 4(b) , in which relatively large ΔV FB are observed under the same gate sweeping voltages for S1, indicating that a tremendous amount of electron traps exist in the charge-trapping layer of S1 compared to the charge-trapping layer of S2 and S3.
PPT Slide
Lager Image
(a) High frequency capacitance-voltage (C-V) curves corresponding to different gate sweeping voltages and (b) ΔVFB extracted from the C-V curves under different gate sweeping voltages. The inset of Fig. 4(a) is the schematic diagram of S1 with Pt patterns to obtain the C-V curves.
We programmed S1, S2, and S3 with an identical initial flatband voltage (6 V) to compare their data retention characteristics at room temperature, as shown in Fig. 5(a) . S1 exhibited a slight decrease in flatband voltage, whereas S2 and S3 show obvious charge loss after 1.0 × 10 5 s, especially for S3. In order to predict the data retention characteristics for a long time, we extrapolated the experimental data for up to 10 years. The results indicate that the decrease in the memory window from the initial 6.0 V to 5.2 V, 4.5 V, and 2.8 V, and the window losses are about 13.3%, 25%, and 53.3% for S1, S2, and S3, respectively. It is known that T-B tunneling and T-E are the important charge loss mechanisms, as shown in inset of Fig. 5(a) . At room temperature, the T-B tunneling process is the dominant charge loss mechanism, and can be reduced by increasing the potential barrier E B [22 , 23] . Though the chargetrapping layer of S2 is an amorphous state ( Fig. 5(b) and (c) ), the smaller E B (0.8 eV) cannot effectively maintain the trapped electrons, which are easily lost via T-B tunneling. Instead, S3 with a maximum E B (1.6 eV) shows the worst retention performance due to the crystallization of (ZrO 2 ) 0.92 (Al 2 O 3 ) 0.08 charge-trapping layer after post annealing treatment, as shown in Fig. 5(d) and (e) . The crystallization of the film will generate grain boundaries, which can act as current leakage paths [18] . The excellent data retention characteristics of S1 should be attributed to the amorphous charge trapping layer [24] and variable potential barrier E B (0.8 to 1.6 eV), which suppress the leakage of trapped electrons toward the tunneling and blocking oxide.
PPT Slide
Lager Image
(a) Comparison of the data retention characteristics of S1, S2 and S3 at room temperature. The inset is the energy band alignment of S1 (solid line), S2 (dotted line) and S3 (dash dotted lines), showing the charge loss mechanisms: trapped electrons tunnel from traps to the silicon conduction band (T-B) and thermal excited trapped electrons from the traps to the conduction band followed by tunneling through the tunnel oxide (T-E). The (b) and (d) is the cross sectional TEM images of the S2 and S3, respectively. The SAED patterns (c) and (e) indicate that the (ZrO2)0.1(Al2O3)0.9 trapping layer of S2 still maintains an amorphous state, while the (ZrO2)0.1(Al2O3)0.9 trapping layer of S3 crystallize, and the diffraction spots correspond to (111), (002), (202), and (113) planes of the cubic ZrO2 phase, respectively.
Figure 6(a) and (b) show the transient program/erase characteristics of S1, S2, and S3 memory devices. The samples are charged at ±8 V by a single electrical pulse with different width. It is observed that S1 exhibits a faster program and erase speed than S2 and S3. For example, the program times at a memory window of 1 V for S1, S2, and S3 are 2 × 10 -6 s, 2.2 × 10 -5 s and 6.8 × 10 -4 s, respectively. The results can be explained by the energy band diagram shown in Fig. 3 . In a program state, the CBO of CM-ZAO with respect to Si is variable (1.2 to 2.0 eV) in S1, compared to identical CBO 2.0 eV for (ZrO 2 ) 0.1 (Al 2 O 3 ) 0.9 in S2. Hence, at the same gate bias voltages, the electron tunneling distance from the Si substrate to the conduction band of the charge-trapping layer is shorter in CM-ZAO and longer in (ZrO 2 ) 0.1 (Al 2 O 3 ) 0.9 . Moreover, the injected electrons are also reduced through blocking oxide into the gate due to the large potential barrier E B , resulting in a faster program speed for S1 [25] . For the erase case, similar to the program state, the variable VBO (2.9 to 1.9 eV) and large potential barrier result in a faster erase speed for S1 compared to S2. As mentioned above, the poor program and erase speed of S3 should be attributed to the crystallization of the charge-trapping layer.
PPT Slide
Lager Image
Transient (a) program and erase (b) characteristics of S1, S2, and S3.
In summary, a charge trap flash memory with the composition modulated (ZrO 2 ) x (Al 2 O 3 ) 1-x charge trapping layer, Al 2 O 3 tunneling oxide, and Al 2 O 3 blocking oxide has been fabricated and investigated. We have demonstrated that the composition modulated (ZrO 2 ) x (Al 2 O 3 ) 1-x charge trapping layer can simultaneously satisfy a larger memory window, faster program/erase speed, and excellent data retention. The results are attributed to the special energy band structure of the trapping layer. Based on this work, the charge trap memory device with composition modulated (ZrO 2 ) x(Al 2 O 3 )1-x trapping layer may be a potential candidate for future nonvolatile flash memory device applications.
This work was financially supported by the National Natural Science Foundation of China (Grant No.51402004), the science and technology development project of Henan province (Grant No. 142102210380), a key science and technology research project in the Henan Ministry of Education (15A140003, 14A140008 and 13A140021), the innovation fund of Anyang Normal University (Grant No. ASCX/2014-Z45) and Project of National Laboratory of Solid State Microstructures (Grant No. M28026).
Burkhardt M. , Jedaa A. , Novak M. , Ebel A. , Voitchovsky K. , Stellacci F. , Hirsch A. , Halik M. 2010 Adv. Mater. 22 2525 -    DOI : 10.1002/adma.201000030
Huang X. D. , Sin J. K. O. , Lai P. T. 2013 IEEE Electron Device Lett. 34 499 -    DOI : 10.1109/LED.2013.2244557
Tseng R. J. , Huang J. X. , Ouyang J. , Kaner R. B. , Yang Y. 2005 Nano Lett. 5 1077 -    DOI : 10.1021/nl050587l
Lee S. H. , Jung Y. , Agarwal R. 2007 Nat. Nanotechnol. 2 626 -    DOI : 10.1038/nnano.2007.291
Kozicki M. N. , Gopalan C. , Balakrishnan M. , Mitkova M. 2006 IEEE Trans. Nanotechnol. 5 535 -    DOI : 10.1109/TNANO.2006.880407
Lee K. H. , Lin H. C. , Huang T. Y. 2013 Electron Device Letters 34 393 -    DOI : 10.1109/LED.2013.2237748
Park G. H. , Cho W. J. 2010 Appl. Phys. Lett. 96 043503 -    DOI : 10.1063/1.3293291
Hong S. H. , Kim M. C. , Jeong P. S. , Choi S. H. , Kim Y. S. , Kim K. J. 2008 Appl. Phys. Lett. 92 093124 -    DOI : 10.1063/1.2892494
Tiwari S. , Rana F. , Hanafi H. , Hartstein A. , Crabbe' E. F. , Chan K. 1996 Appl. Phys. Lett. 68 1376 -    DOI : 10.1063/1.116085
Tan Z. , Samanta S. K. , Yoo W. J. , Lee S. J. 2005 Appl. Phys. Lett. 86 013107 -    DOI : 10.1063/1.1846952
Ren J. J. , Yan D. , Chu S. , Liu J. L. 2013 Appl. Phys. A 111 719 -    DOI : 10.1007/s00339-013-7641-3
Huang X. D. , Lai P. T. , Sin J. K. O. 2012 Appl. Phys. A 108 229 -    DOI : 10.1007/s00339-012-6881-y
Tang Z. J. , Zhu X. H. , Xu H. N. , Xia Y. D. , Yin J. , Liu Z. G. , Li A. D. , Yan F. 2013 Materials Lett. 92 21 -    DOI : 10.1016/j.matlet.2012.10.024
Zhu C. X. , Huo Z. L. , Xu Z. G. , Zhang M. H. , Wang Q. , Liu J. , Long S. B. , Liu M. 2010 Appl. Phys. Lett. 97 253503 -    DOI : 10.1063/1.3531559
Zhang G. , Wang X. P. , Yoo W. J. , Li M. F. 2007 IEEE Trans. Electron Devices 54 3317 -    DOI : 10.1109/TED.2007.908888
Liu L. , Xu J. P. , Chen J. X. , Lai P. T. 2014 Appl. Phys. A 115 1317 -    DOI : 10.1063/1.4737158
Huang X. D. , Shi R. P. , Lai P. T. 2014 Appl. Phys. Lett. 104 162905 -    DOI : 10.1063/1.4873388
Tan Y. N. , Chim W. K. , Choi W. K. , Joo M. S. , Cho B. J. 2006 IEEE Trans. Electron Devices 53 654 -    DOI : 10.1109/TED.2006.870273
Tang Z. J. , Zhu X. H. , Xu H. N. , Xia Y. D. , Yin J. , Li A. D. , Yan F. , Liu Z. G. 2011 Appl. Phys. A 108 217 -    DOI : 10.1007/s00339-012-6877-7
Tahir D. , Lee E. , Oh S. K. , Tham T. T. , Kang H. J. , Jin H. , Heo S. , Park J. C. , Chung J. G. , Lee J. C. 2009 Appl. Phys. Lett. 94 212902 -    DOI : 10.1063/1.3143223
Chambers S. A. , Liang Y. , Yu Z. , Droopad R. , Ramdani J. , Eisenbeiser K. 2000 Appl. Phys. Lett. 77 1662 -    DOI : 10.1063/1.1310209
Yang Y. , White M. H. 2000 Solid-State Electron 44 949 -    DOI : 10.1016/S0038-1101(00)00012-5
Wang Y. , White M. H. 2005 Solid-State Electronics 49 97 -    DOI : 10.1016/j.sse.2004.06.009
Cha S. Y. , Kim H. J. , Choi D. J. 2010 J. Mater. Sci. 45 5223 -    DOI : 10.1007/s10853-010-4562-9
Chen J. X. , Xu J. P. , Liu L. , Lai P. T. 2013 Applied Physics Express 6 084202 -    DOI : 10.7567/APEX.6.084202