Chapter 7 Flash Memory Device Using Ge Quantum Dot
7.3 Ge Quantum Dot Flash Memory
Ge nanocrystal flash memory devices using SiO2 tunneling oxide were demonstrated a few years ago and good retention performance was shown [51, 52]. The fabrication of Ge quantum dot on silicon dioxide became possible in recent years.
Fig.7.3 The impact of trap energy on the retention time of Ge flash memory using HfO2 dielectric.
For pure Ge quantum dot flash memory, the impact of trap energy on the retention time is illustrated in Fig.7.3, in which the diameter of the quantum dot is 5 nm and the conduction band shift is 0.15 eV. The result demonstrates again that the trap energy has a very important effect on the retention time. A 0.2 eV difference between trap energies results in near 4 orders of magnitudes difference of the retention time. The explanation is that the electron is localized in the traps of the quantum dot, and hence it is difficult to be injected from the quantum dot. As a result, retention time becomes longer and the information can be stored longer.
When the trap energy is fixed, we study the impact of the barrier height on the retention time in Fig. 7.4 with tunnel oxide thickness 4.5 nm. Compared with Fig. 7.3, it is obvious that the barrier height has less effect on the retention time compared to the trap energy. It is believed that, with the same conduction band shift, the contribution of Ge quantum dot may be larger than the contribution of high-k dielectric to the good retention. Therefore, it is possible that Ge quantum dot flash memory with HfAlO will provide a better retention than Si quantum dot flash memory with HfO2 as shown in Fig.7.7.
Fig.7.4 The impact of barrier height on the retention time.
The programming and retention times of Ge quantum dot flash memory with HfO2 are illustrated in Fig.7.5, considering a large square quantum dot with conduction band shift 0 eV. For Ge quantum dot flash memory with HfO2 dielectric, the tunnel oxide thickness of near 4.5 nm can provide 10 years retention time. As we discussed previously in chapter 6, for Si quantum dot flash memory with HfO2 dielectric, the tunnel oxide thickness is required to be 6.1 nm. Therefore, with the use of Ge
quantum dot, the HfO2 tunnel oxide thickness can be scaled down from 6 to 4.5 nm, which is a great improvement for the scaling of tunnel oxide thickness.
Fig.7.5 Programming and retention times of Ge quantum dot flash memory (dot line is ten years retention standard).
The impact of the dot size on the programming and retention times are discussed in Fig. 7.6. The programming and retention performance of HfO2 flash memory with the quantum dot of 2 nm, 3 nm and 5 nm diameter is simulated in Fig.7.6. The result shows that 5nm quantum dot provides faster programming time and better retention time at all tunnel oxide thicknesses. The reason is that the larger the size of the quantum dot, the smaller is the quantum confinement effect, therefore it results in larger tunneling current probability. Hence, for faster programming time, larger size quantum dot is suggested. Therefore, 5 nm quantum dot flash memories with 4.3 nm HfO2 tunnel oxide are selected to provide 10 years retention times and at the same time the programming time 2×10-2 s at 2 V control gate voltage. However, the larger
size of the quantum dot will decrease the reliability of the flash memory. Hence, for the flash memory with larger size of the quantum dot, it will not be good for providing better reliability compared with the flash memory with a smaller quantum dot. And in Fig.7.6, when the size of the quantum dot increases to 7 nm which the conduction band shift is assumed to be 0 eV, the gain in programming/retention is insignificant.
For the quantum dot with larger size (larger than 7 nm), the quantum effect is reduced and the memory performance will be worse than that of the dot with smaller size.
3.4 3.6 3.8 4.0 4.2 4.4 4.6
10-3 10-2 10-1 100 101 102 103 104 105 106 107 108 109
Retention Time (S)
Tunnel Oxide Thickness (nm) 5nm
3nm 2nm
Programming @2V retention 10 years
Fig.7.6 The impact of dot size on programming and retention times.
The comparison of the retention time of flash memories with different dielectrics, considering Si and Ge quantum dots are illustrated in Fig. 7.7. A larger quantum dot (6nm×10nm×10nm) is embedded between control oxide and tunnel oxide and the conduction band shift is taken as 0 eV. It is significant that the use of Ge quantum dot improves the retention time greatly. Using Ge quantum dot, the flash memory with HfAlO dielectric can even have better retention than the device with HfO2 as we have
predicated in Fig. 7.4. That means Ge quantum dot has more contribution to optimizing the retention time than the use of high-k dielectrics under the same condition. With the increase of EOT of tunnel oxide, it seems that the contribution of high-k dielectrics becomes larger. When EOT is less than 1.6 nm, the retention time of Ge quantum dot flash memory with SiO2 is better than Si quantum dot flash memory with HfO2 and HfAlO. However, when EOT is larger than 1.6 nm, the high-k dielectric shows obvious predominance and gives more contribution to optimizing retention time. With the continuous increase of tunnel oxide thickness, the difference between high-k dielectrics is enlarged. Hence, Ge quantum dot will play important role in providing good retention time in the flash memory with smaller dimension.
Fig.7.7 The comparison of retention time of flash memories with various dielectrics and quantum dots.