You are on page 1of 4

Effect of Vds Variation on the Trapped Charge

Distribution of a SONOS Memory


Dary Mochamad Rifqie1, Muhammad Amin Sulthoni1,2, Akhmadi Surawijaya1,2, Irman Idris1,2
1
School of Electrical Engineering and Informatics, Bandung Institute of Technology
2
Centre of Excellence on Microlectronics, Bandung Institute of Technology
Bandung, Indonesia

Abstract - In this paper, simulation of SONOS memory cell achieved by the channel hot electron injection method. The
has been performed. We investigate the effect of various electrons are injected above the drain junction of the cell
Vds values in order to observe the charge distribution in transistor in case when high voltage values are applied at the
the floating gate. By varying drain voltage from 0 to 5 V gate and drain. Variation of the voltage applied to the drain and
and maintaining gate voltage constant at 8 volt, we gate resulted the placement of the trapped electrons in different
observed a decreasing programming threshold voltage. I-V part of the storage layer due to the kinetic energy variation. The
curve shows threshold voltage shifts according to the charge injection placement is a key performance parameter for
variation of voltage drain. Furthermore, we also observed a program, read and erase. The threshold voltage value and its
spatial distribution of trapped electrons in floating gate stability depends on the position of the trapped charge in the
which is changing as the voltage drain is varied. storage layer, injected charge and the quality (energy depth) of
the traps. Futhermore, a 4-bit operation has been performed in
Keyword – SONOS memory, trapped charge, Threshold silicon-on-insulator (SOI)-based SONOS memory which store
voltage, floating gate electrons on the top and bottom side of oxide-nitride-oxide
I. INTRODUCTION layers [4].
A concept of Silicon-oxide-nitride-oxide-silicon (SONOS) is In this paper, we will investigate the distribution of electrons
interesting for use in silicon-based nonvolatile memory. as trapped in the storage layer with respect to the drain voltage
declared in International Technology Roadmap for variations by using Synopsis simulator.
Semiconductor (ITRS), floating gate silicon-based memory
II. DEVICE STRUCTURE
will reach thickness limit of tunnel oxide around 8 nm[1]. The
By using Structure Device Editor (SDE) simulation, we
key construct of this technology is based on trapped charge
simulate a SONOS memory device with the design rule as
storage inside a nitride layer[2].
follow:
Compared to floating gate memory, resilience toward
Tabel 1.design rule of SONOS cell
physical defects in SONOS memory storage is one of the
Physical part Dimension (nm)
important aspect. Charge trapping storage device like SONOS
Tunnel Oxide thickness 2
memory seems to be technologically promising to replace
Blocking oxide thickness 4
floating gate flash technology. Double-bit operation has been
Nitride layer thickness 8
achieved in SONOS technology electrons placement in the
Channel length 130
storage layer[3]. Localizing electron in SONOS can be

1
obtain the probability of hot electron which can passed over
The SONOS memory device structure shown in tecplot oxide barrier.
view can be seen in Figure.1 below:
IV. RESULTS AND DISCUSSION
Figure 2 to 5 shows the distribution of trapped electrons
inside nitride layer after programming. The figure shows that
the distribution of trapped electrons is changing as the voltage
is varied. Higher drain voltage will result in location of the
trapped electrons in the storage layer to shift to right side of
the nitride layer. This can be explained that as the voltage in
drain increase, the electron gains more energy from source
side.

Figure 1. Details of SONOS memory device


Figure 2. Distribution of trap occupation in 0 volt drain voltage, 8
In this simulation, we use Sdevice to varied voltage value volt gate voltage
for program condition in drain. The voltage variation are
listed below:

Table 2.Voltage variations for program state Figure 3. Distribution of trap occupation in 1 volt drain voltage, 8
Sample Gate voltage Drain voltage volt gate voltage

1 8 volt 0 volt
2 8 volt 1 volt
3 8 volt 3 volt Figure 4. Distribution of trap occupation in 3 volt drain voltage, 8

4 8 volt 5 volt volt gate voltage

III. SIMULATION
Figure 5. Distribution of trap occupation in 5 volt drain voltage, 8
For device simulations, the general idea is to give a voltage
volt gate voltage
bias to source,drain and gate electrodes. In order to simulate
the programming and erasing function in synopsis CAD tools, From Figure 6 to 9, we found that I-V curve threshold
simulation of transient mode is applied. Using different voltage shifts from 1.45, 1.15, 0,8 and 0.6 Volt for voltage
voltages at the drain and gate electrodes for programming drain of 0, 1, 3 and 5 Volt respectively. Inspection of I-V
(shown at table 2), we then observe the distribution of trpped curve of the programmed states indicate that the threshold
charges in the storage layer. This states is saved as voltage have larger value for a lower voltage drain. As voltage
programmed state and it is use for the other sdevice command drain is rising, electrons trapped in the nitride layer will not
in order to calculate its I-V characteristic. This characteristic spread and became more concentrated in a specific location
will be calculated by applying a constant drain voltage (5 volt) within the nitride layer. Therefore, the value of threshold
and varying gate voltage (from -1.25 to 3.25 volt). voltage is shifting to the left side as the programming drain
Futhermore, we examined the resulting I-V characteristic voltage increase.
curve to determine the threshold voltage shifting. Moreover, The shifting of the values of the threshold voltage can be
we also defined channel hot electron injection for the applied to determine the memory performance. An I-V curve
programming mechanism. In order to simulate the channel hot for the erased state is provided for comparison. The change in
electron injection, we use spherical harmonics expansion to shifting of threshold values is wide enough so we can easily
2
select the read voltage. A low valued threshold voltage ensure
that the Vread will be low enough, so the power consumption
also will be low[5].

Figure 9. Threshold voltage shift for erase (red) and programmed


Figure 6. Threshold voltage shift for erase (red) and programmed state (black) at 5 volt drain voltage
state (black) at 0 volt drain voltage

V. CONCLUSION
A SONOS nonvolatile memory device has been simulated. The
results show that spatial distribution of electron in nitride layer
shift as the voltage drain is varied. Futhermore, by varying
drain voltage from 0 to 5 V and maintaining gate voltage
constant at 8 volt, we observed a decreasing programming
threshold voltage. I-V characteristic curve shows threshold
voltage shift to the left side as the drain voltage is increased.

REFERENCE
Figure 7. Threshold voltage shift for erase (red) and programmed [1]. ([Online] Available: http://public.itrs.net/ 2001) ITRS Roadmap.
state (black) at 1 volt drain voltage [2]. Fukuda, M., Nakanishi, T., & Nara, Y. (2003). New nonvolatile
memory with charge-trapping sidewall. IEEE Electron Device
Letters, 24(7), 490–492. doi:10.1109/led.2003.815002.
[3]. Detlev Richter, Flash Memory : economic principle of
performance, cost and reliability optimization., springer, 2013
[4]. Datta, A., Kumar, P. B., & Mahapatra, S. (2007). Dual-Bit/Cell
SONOS Flash EEPROMs: Impact of Channel Engineering on
Programming Speed and Bit Coupling Effect. IEEE Electron
Device Letters, 28(5), 446–448. doi:10.1109/led.2007.895406
[5]. Sentaurus Device (sdevice) user manual..
[6] Kartiwa, G. M., Hendrayana, Y. H., Prihatiningrum, N.,
Sulthoni, M. A., Surawijaya, A., & Idris, I. (2017). Effect of
Lsg/Lfg ratio variation to the IV curve of split-gate 1st

Figure 8. Threshold voltage shift for erase (red) and programmed generation superflash. 2017 International Symposium on
state (black) at 3 volt drain voltage Electronics and Smart Devices (ISESD).
doi:10.1109/isesd.2017.8253345

3
we do also a simulation varying gate voltage(5 volt/6 volt/ 7
volt) and then tune the thickness of the tunneling layer so we
can find almost the same amount of electron traped in the
floating gate. The voltage and thickness variation are listed
below:
Example Gate voltage (v) Tunnelling layer
thickness (nm)
1 5 1.8
2 6 3.66
3 7 4.3

Tabel 3. voltage variation for program state

You might also like