Advanced
Dependence of Electrons Loss Behavior on the Nitride Thickness and Temperature for Charge Trap Flash Memory Applications
Dependence of Electrons Loss Behavior on the Nitride Thickness and Temperature for Charge Trap Flash Memory Applications
Transactions on Electrical and Electronic Materials. 2014. Oct, 15(5): 245-248
Copyright © 2014, 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 (http://creativecommons.org/licenses/by-nc/3.0) which permits unrestricted noncommercial use, distribution, and reproduction in any medium, provided the original work is properly cited.
  • Received : June 06, 2014
  • Accepted : July 07, 2014
  • Published : October 25, 2014
Download
PDF
e-PUB
PubReader
PPT
Export by style
Share
Article
Author
Metrics
Cited by
TagCloud
About the Authors
Zhenjie Tang
zjtang@hotmail.com
Ma Dongwei
Zhang Jing
Jiang Yunhong
Wang Guixia
Rong Li
Jiang Yin
Abstract
Pt/Al 2 O 3 /Si 3 N 4 /SiO 2 /Si charge trap flash memory structures with various thicknesses of the Si 3 N 4 charge trapping layer were fabricated. According to the calculated and measured results, we depicted electron loss in a schematic diagram that illustrates how the trap to band tunneling and thermal excitation affects electrons loss behavior with the change of Si 3 N 4 thickness, temperature and trap energy levels. As a result, we deduce that Si 3 N 4 thicknesses of more than 6 or less than 4.3 nm give no contribution to improving memory performance.
Keywords
1. INTRODUCTION
Charge Trap Flash (CTF) memory devices, otherwise known as metal-oxide-nitride-oxide-silicon structures, have been the subject of attention in the semiconductor industry due to their advantages over conventional floating gate type memory. These advantages include lower programming voltage, superior programming/erasing speeds, and a simple fabrication process compatible with standard complementary metal-oxidesemiconductor technology [1 - 3] . Recently, CTF memory devices have gained increasing interest in the three dimensional (3D) integration for next generation nonvolatile memory technology [4 , 5] . The tunnel oxide thickness plays a crucial role in regulating the erasing speed, data retention characteristics and charge loss mechanisms for CTF memory devices [6] , while the thickness of the nitride charge trapping layer is less critical. Nevertheless, in 3D architectures, the nitride thickness has a direct effect on charge storage performance and array density [7] . Moreover, temperatures [8] and trap energy levels [9 , 10] also are considerable factors for understanding the electron loss mechanisms. Hence, in this letter, Pt/Al 2 O 3 /Si 3 N 4 /SiO 2 /Si (MANOS) charge trapping memory capacitors with various thicknesses of nitride layer were fabricated. We investigated and analyzed the effect of nitride thickness, trap energy levels and temperatures on electrons loss behavior in the retention state for MANOS capacitors. Also, a reasonable nitride thickness range was obtained through electrical characteristic measurements. Four charge loss mechanisms [11 , 12] are involved in the data retention state for scaled CTF memory devices: trapped electrons tunnel from traps to the silicon conduction band (T-B), trapped electrons tunnel from traps to the Si/SiO 2 interface traps state (T-T), holes tunnel from the silicon valence band to nitride traps (B-T) and thermal excited trapped electrons from the traps to the nitride conduction band followed by tunneling through the tunnel oxide (T-E), as shown in Fig. 1 . However, the T-B and T-E mechanisms are regarded as the two main electron loss mechanisms [11] . Therefore, in our case we only consider the T-B tunneling and T-E mechanisms.
PPT Slide
Lager Image
Energy band diagram of a conventional CTF memory device, showing charge loss mechanisms: T-B, T-T, B-T, and T-E.
2. EXPERIMENTS
The MANOS capacitors were fabricated on p-type (100) Si substrates with a resistivity of 8-12 Ω·cm. Prior to deposition, the p-Si were cleaned by the standard radio corporation of America (RCA) process. Then, these substrates were dipped in HF solution for one minute to remove the native oxide. After a growth of thermal SiO 2 tunnel oxide (TO) with a thickness of 3 nm in dry O 2 ambience, the nitride (Si 3 N 4 ) as charge trapping layers (CTL) ranging from 1 to 6 nm were deposited by low pressure chemical vapor deposition at 700℃. Subsequently, a 8 nm Al 2 O 3 layer was deposited as the blocking oxide (BO) by atomic layer deposition (ALD) using trimethylaluminium (Al(CH 3 ) 3 ) precursor at a substrate temperature of 300℃. The thickness of the samples was measured by a spectroscopic ellipsometer and transmission electron microscopy (TEM). Finally, platinum (Pt) top electrodes with an area of 7.85×10 -5 cm 2 were deposited using magnetron sputtering at room temperature. Figures 2(a) and (b) show a schematic diagram and cross-sectional TEM image, respectively, of the charge trap flash memory device with a 5 nm Si 3 N 4 CTL. The electrical characteristics of these memory capacitors were analyzed by a Keithley 4200 semiconductor characterization system.
PPT Slide
Lager Image
(a) Schematic diagram and (b) cross-sectional TEM image of memory structure with a 5 nm thick Si3N4 layer.
3. RESULTS AND DISCUSSION
In order to have a quantitative understanding of the electrons loss mechanisms in the data retention state, we have calculated the time constant associated with the T-B and T-E processes to characterize the electrons loss behavior. The time constant of T-B, τ T-E , and the time constant of τ T-E are written as [11] :
PPT Slide
Lager Image
PPT Slide
Lager Image
where τ T-E is a time constant [13] ,
PPT Slide
Lager Image
= 0.42 m 0 [14] and
PPT Slide
Lager Image
= 0.25 m 0 [11] are the electron effective mass in the SiO 2 and Si 3 N 4 , respectively, here m 0 is the free electron mass. E T is the trap energy level referenced to the conduction band edge in the Si 3 N 4 (eV), q is the absolute electron charge, E B =1.05 eV [15] is the energy barrier height of electron tunneling (eV), h is Planck’s constant, d TO is the thickness of the SiO 2 (nm), T is the absolute temperature (K), A is the temperature independent constant, k B is Boltzmann’s constant, t is the retention time (s) and is the tunneling distance in the Si 3 N 4 measured from the SiO 2 /Si 3 N 4 interface as follows (nm):
PPT Slide
Lager Image
We assumed that the traps are spatially uniform inside the Si 3 N 4 with an arbitrary energy level distribution, and all traps are initially filled with electrons in the retention state. The BO absolutely restricts electrons tunneling to and from the gate electrode. Fig. 3(a) shows the results of the calculated time constant based on Eqs. (1) and (2). For the electrons captured by shallow trap energy levels (E T ≤ 0.5 eV), τ T-E (<10 -3 s) is at least 2 orders of magnitude smaller than τ T-B , suggesting that T-E is a much quicker electrons loss path than T-B tunneling. These trap energy levels cannot effectively capture electrons due to their smalland have no contribution to electron retention. If > 0.5 eV, T-B tunneling starts to influence electrons loss and the the two electron loss mechanisms compete with each other. At 273 K, a relatively small τ T-B for electrons trapped near the TO (x<1 nm) suggests that T-B tunneling plays a leading role in the electrons loss process of the region. However, increases to 6 nm, τ T-B increases at least 10 orders of magnitude, and the deeper the trap depth, the larger the increase, indicating that T-B tunneling becomes more and more difficult, especially for those deep E T . Moreover, the electron loss path starts to change with increasing, and T-E gradually becomes the important electron loss mechanism from shallow E T to deep E T . It is worth noting that there is little effect of T-B tunneling on electron retention when x>4 nm. It was also found that τ T-E decrease by several orders of magnitude when increasing the temperature, and more and more trapped electrons tend to be de-trapping through T-E, which is the dominant electron loss mechanism when the temperature exceeds 470 K. This means that the dominant region of T-B will gradually reduce to disappear with an increase of temperature from 273 to 500 K. Fig. 3(b) exhibits the tunneling distance of different E T in the retention state based on Eq. (3). It was observed that x decreases with E T increasing from 0.5 to 1.3 eV at the same retention time, indicating that the electrons trapped at deep E T are less affected by T-B tunneling.
PPT Slide
Lager Image
(a) Calculated time constants of andat different trap energy level (0-1.3 eV). τT-B and τT-E are functions of tunneling distance in Si3N4 and temperature, respectively and (b) retention time versus tunneling distance in the Si3N4 (t-x) of MANOS capacitors. In the calculation, dTO= 3 nm was used.
Figure 4 illustrates how the two electron loss processes influences the retention state for MANOS devices. Ⅰ and Ⅱ represent the T-B and T-E dominant region at 273 K, respectively, and the boundary (marked by) is drawn on the basis of the above results. At a retention time t (e.g., 1,000 s), region Ⅰ and Ⅱ should contain empty traps and filled traps. As the electron loss process continues, the empty-filled traps boundary marked by B T-B and B T-E moves toward the BO side and the bottom of the Si 3 N 4 band gap, correspondingly. The boundary of B Ⅰ / Ⅱ , which is nonlinear with tunneling distance, moves down at elevated temperatures, suggesting that T-E gradually dominates the electron loss process.
PPT Slide
Lager Image
Contribution of T-B and T-E tunneling to electron loss behavior in the retention state for MANOS capacitors at 273 K.
The memory window △V FB (flat-band voltage shift) extracted from the 1 MHz capacitance-voltage (C-V) curve under different program voltages is shown in Fig. 5(a) . In the thickness range of our study, much smaller V FB of MANOS devices with an Si 3 N 4 layer less than 3.2 nm are observed, indicating that injected electrons are not trapped effectively but are lost quickly. Though the program speed and memory window increase with a thick Si 3 N 4 layer, the △V FB appears to be fairly similar when the thickness exceeds 4.3 nm. This result is ascribed to the reduced electric field through Si 3 N 4 due to the increasing Si 3 N 4 thickness, which gives rise to a decrease of electron trapping efficiency. The retention characteristics of the samples after 10 hours at different temperatures are shown in Fig. 5(b) . It shows that high charge loss is observable for thin Si 3 N 4 less than 3.2 nm even at 273 K. As illustrated in Fig. 3(a) , the T-B tunneling dominates the electron loss process for thinner Si 3 N 4 due to the small τ T-B , and those trapped electrons near the TO/Si 3 N 4 interface are lost quickly through the TO to the substrate. By increasing the thickness of Si 3 N 4 , the retention characteristics can be improved. However, at and above 4.3 nm, the charge loss is almost independent of Si 3 N 4 thickness, especially for temperatures exceeding 470 K. These results can be attributed to the electron loss mechanisms transition between T-B and T-E, as illustrated in Fig.4 . When T-E dominates the electron loss process, the retention characteristics are not affected by the gradual increase of Si 3 N 4 thickness. Hence, in the thickness range of our study, we deduce that a Si 3 N 4 thickness of more than 6 or less than 4.3 nm has no contribution to improving the memory performance.
PPT Slide
Lager Image
(a) The VFB shift under different program voltages at 0.1 ms for MANOS capacitors with varying Si3N4 thickness and (b) dependence of electron loss on temperature for MANOS capacitors.
4. CONCLUSIONS
In summary, we investigated the effect of nitride thickness, temperature and trap energy levels on the electrons loss behavior of a Pt/Al 2 O 3 /Si 3 N 4 /SiO 2 /Si CTF memory structure. In a data retention state, T-B and T-E tunneling compete with each other to influence the retention characteristics, even at 273 K. Those trapped electrons near TO (<1 nm) and at a shallow E T (<0.5 eV) inside the Si 3 N 4 are lost quickly via T-B and T-E tunneling, respectively. With an increase of Si 3 N 4 thickness and temperature, T-B tunneling is gradually reduced, while T-E dominates the electrons loss process. At a certain E T , electrons may adopt different loss paths, which depend on the Si 3 N 4 thickness and temperature. The electrical measurement results demonstrate that a thin Si 3 N 4 layer of less than 3.2 nm has no effective trapping ability. Though the storage performance can be improved by increasing Si 3 N 4 thickness, the data retention characteristics are almost independent of nitride thickness at elevated temperatures, especially temperatures exceeding 470 K. The above results can pave a way for device design in the future.
Acknowledgements
This work was financially supported by the Science Fundation of Educational Department of Henan Province of China (Grant No. 13A140021 and 14A140008), the National Natural Science Foundation of China (Grant nos. 11347186, 61176124, and 11174122).
References
Tang Z. J. , Zhao D. Q. , Li R. , Zhu X. H. 2014 Trans. Electr. Electron. Mater. http://dx.doi.org/10.4313/TEEM.2014.15.1.16 15 16 -
Tang Z. J. , Liu Z. G. , Zhu X. H. 2010 Trans. Electr. Electron. Mater. http://dx.doi.org/10.4313/TEEM.2010.11.4.155 4 155 -
Liu L. , Xu J. P. , Ji F. , Chen J. X. , Lai P. T. 2012 Appl. Phys. Lett. http://dx.doi.org/10.1063/1.4737158 101 033501 -
Park J. H. , Hur S. H. , Lee J. H. , Park J. T. , Sel J. S. , Kim J. W. , Song S. B. , Lee J. Y. , Lee J. H. , Son S. J. , Kim Y. S. , Park M. C. , Chai S. J. , Choi J. D. , Chung U. I. , Moon J. T. , Kim K. T. , Kim K. , Ryu B. I. 2004 Tech. Dig. Int. Electron Devices Meet. http://dx.doi.org/10.1109/IEDM.2004.1419319 873 -
Hsiao Y. H. , Lue H. T. , Hsu T. H. , Hsieh K. Y. , Lu C. Y. 2010 Proc. IEEE Int. Memory Workshop http://dx.doi.org/10.1109/IMW.2010.5488390 1 -
Gu S. H. , Hsu D. W. , Wang T. , Lu W. P. , Ku Y. H. J. , Lu C. Y. 2007 IEEE Trans. Electron Devices http://dx.doi.org/10.1109/TED.2006.887219 54 90 -
Arreghini A. , Kar G. S. , Van den Bosch G. , Van Houdt J. 2013 IEEE Electron Device Lett. http://dx.doi.org/10.1109/LED.2013.2253442 35 632 -
Tzeng S. D. , Gwo S. 2006 J. Appl. Phys. http://dx.doi.org/10.1063/1.2218025 100 023711 -
Arreghini A. , Akil N. , Driussi F. , Esseni D. , Selmi L. , van Duuren M. J. 2008 Solid State Electron. http://dx.doi.org/10.1016/j.sse.2008.04.016 52 1460 -
Chang M. , Jo M. , Kim S. , Ju Y. , Jung S. , Lee J. , Yoon J. , Lee C. , Hwang H. 2008 Appl. Phys. Lett. http://dx.doi.org/10.1063/1.3041642 93 232105 -
Wang Y. , White M. H. 2005 SolidState Electronics http://dx.doi.org/10.1016/j.sse.2004.06.009 49 97 -
Kim T. H. , Sim J. S. , Lee J. D. , Shim H. C. , Park B. G. 2004 Appl. Phys. Lett. http://dx.doi.org/10.1063/1.1773615 85 660 -
Bernert K. , Oestreich C. , Bollmann J. , MikolajickI T. 2010 Appl. Phys. A http://dx.doi.org/10.1007/s00339-010-5694-0 100 249 -
Hinkle C. L. , Fulton C. , Nemanich R. J. , Lucovsky G. 2004 Microelectronic Engineering http://dx.doi.org/10.1016/j.mee.2003.12.047 72 257 -
Tan Y. N. , Chim W. K. , Cho B. J. , Choi W. K. 2004 IEEE Trans. Electron Devices 51 1143 -