† Corresponding author. E-mail:
Project supported by Funds of Key Laboratory, China (Grant No. y7ys011001) and Youth Innovation Promotion Association, Chinese Academy of Sciences (Grant No. y5yq01r002)
In this study, we investigate the single-event transient (SET) characteristics of a partially depleted silicon-on-insulator (PDSOI) metal-oxide-semiconductor (MOS) device induced by a pulsed laser.We measure and analyze the drain transient current at the wafer level. The results indicate that the body-drain junction and its vicinity are more SET sensitive than the other regions in PD-SOI devices. We use ISE 3D simulation tools to analyze the SET response when different regions of the device are hit. Then, we discuss in detail the characteristics of transient currents and the electrostatic potential distribution change in devices after irradiation. Finally, we analyze the parasitic bipolar junction transistor (p-BJT) effect by performing both a laser test and simulations.
The development of silicon-on-insulator (SOI) devices has been motivated by the full dielectric isolation of individual transistors, which prevents the occurrence of latch-up. Moreover, because the concentration of sensitive charges that accumulate in SOI ICs is considerably less than that in bulk-silicon integrated circuits (ICs) SOI ICs are less sensitive to single-event effects (SEE).[1–3] However, the parasitic bipolar junction transistor (p-BJT) effect inherent in the main metal-oxide semiconductor (MOS) transistors reduces the hardness of the single-event upset (SEU) of SOI ICs, especially those ICs that utilize partially depleted silicon-on-insulator (PD-SOI) technology.[4,5]
SET is a significant error mechanism in high-speed digital complementary MOS (CMOS) IC devices.[6,7] As the technology node and concomitant voltage scaling, SETs have become a contributor to the soft error rate for submicron devices. Particle accelerator testing is known as the standard method to be employed when characterizing the sensitivity of modern device technology to SEEs.[8] However, accelerator testing is usually expensive and is not easily accessible. For the past few years, pulsed laser systems have been widely used to characterize SEE behavior in semiconductor devices and logic ICs.[9–11] Studies have shown that pulsed laser excitation may be a reasonable approximation to ion excitation if the working wavelength could be chosen properly.[12] Recently, many local studies have been done on the SET characteristics of partially depleted (PD)-SOI MOSFETs.[13,14] Most of them analyze the SET characteristics using simulation tools, and few results have been obtained using pulsed laser systems.
This paper is arranged as follows. In Section
The pulsed laser system used in this study is utilized at the Institute of Microelectronics, Chinese Academy of Sciences (IMECAS). Figure
The transistors used in our experiment arentype devices because of the obvious SET response compared to P-type devices under the same radiation conditions.[15,16] These devices were fabricated at the IMECAS using 0.35-μm PD-SOI CMOS technology. As shown in Fig.
We applied two body contacts in the device structure at each edge along the gate width direction. The devices used in this experiment are irradiated in the off-state, with the drain biased at 3.3 V and other electrodes grounded. We tested two types of transistors, and the layouts are summarized in Fig.
The devices are scanned in the source-drain axis with a small displacement step by the pulsed laser (20 pJ). The laser-hit locations are illustrated in the layout schematic in Fig.
The transient currents of device_1 induced by the pulsed laser are described in Fig.
![]() | Fig. 4. (color online) Variations of the transient currents when the laser hit different locations of device 1. |
In Fig.
The collected charge quantity measured on the drain when the laser strikes the nine locations in device 1 is shown in Fig.
![]() | Fig. 5. (color online) (a) Charge quantity calculated by integration of the transient current (b). The transient peak amplitude as a function of laser hit locations. |
The transient current of device_2 is illustrated in Fig.
![]() | Fig. 6. (color online) (a) Variations of the transient currents when the laser strikes different locations of device_2. (b) Charge quantity calculated by the integration of the transient current. |
Figure
Reference [18] showed the solution to the diffusion equation of drain current I(t), and I(t) can be modeled as follows:
![]() |
![]() |
We can conclude that the body-drain junction and its surroundings are more SET sensitive than the other regions in PD-SOI devices. When long-channel devices are hit by the pulsed laser, the body region near the source is less sensitive than the other short-gate devices. In order to further study the p-BJT effect, in the next section, we use the ISE TCAD simulation tools.
In this section, we use ISE simulation tools to illustrate the phenomenon in the laser test experiment. We used a limited set of physical models to achieve a trade-off between a large realistic structure and an affordable simulation time. As the parasitic elements of the actual device are not considered, the simulated results may not be comparable with the experimental data. Instead, our goal is to qualitatively analyze the mechanism involved in the carrier transport process. The device used in the simulation has the same structure as the device_1 in the laser test experiment.
We used the Heavy Ion module in the ISE TCAD tool to simulate the charge-deposition process in the PD-SOI device. The linear energy transfer (LET) was characterized with the parameter LET_f with a constant amplitude of 0.1 pc/μm. W used the Gaussian shape spatial distribution for the charge-deposition process.
Figure
![]() | Fig. 7. (color online) Variations of drain transient current when heavy ions strike different locations obtained during the simulation. |
Figure
![]() | Fig. 8. (color online) Variations of source transient current when heavy ions strike different locations obtained during the simulation. |
Figure
![]() | Fig. 9. (color online) Distributions of electrostatic potential at (a) 0 ps, (b) 1 ps, (c) 3 ps, and (d) 10 ps when the body-drain junction (location #3) was hit. |
From Fig.
![]() | Fig. 10. (color online) Distributions of electrostatic potential at 3 ps when (a) the source region (#1) and (b) drain region (#5) was hit (legend is the same as Fig. |
After the simulation analysis, we can conclude that when the body-drain region and its vicinity are hit, the p-BJT effect is triggered, and the charges that are generated are amplified. The heavy-ion simulation and the pulsed laser test experiment have a large degree of similarity in terms of charge collection and transport.
In this paper, we studied the SET characteristics of PD-SOI MOSFETs using the pulsed laser system and ISE 3D simulation tools. The experimental results obtained for device_1 indicate that the drain transient current reached the highest amplitude of 2.2 mA when the laser hit the body-drain region. For short-channel devices, both the body-under-gate and the drain region are sensitive to the SET effect. We found that the p-BJT effect amplifies the charges when sensitive regions are hit by the laser. 3D simulation tools illustrate the transient currents collected by the source and drain. The transient current result obtained by performing the simulation has the same trend as that obtained with the laser experiment. Moreover, the body electrostatic potential distribution after irradiation confirmed that the increase in the body potential is caused by hole accumulation. The body-drain junction is the most sensitive region in PD-SOI MOSFETs, and the p-BJT effect should be minimized to decrease the SET current. These results have good practical significance for the study of radiation-hardening technology of other semiconductor devices and integrated circuits.
[1] | |
[2] | |
[3] | |
[4] | |
[5] | |
[6] | |
[7] | |
[8] | |
[9] | |
[10] | |
[11] | |
[12] | |
[13] | |
[14] | |
[15] | |
[16] | |
[17] | |
[18] |