† Corresponding author. E-mail:
Total ionizing dose induced single transistor latchup effects for 130 nm partially depleted silicon-on-insulator (PDSOI) NMOSFETs with the bodies floating were studied in this work. The latchup phenomenon strongly correlates with the bias configuration during irradiation. It is found that the high body doping concentration can make the devices less sensitive to the single transistor latchup effect, and the onset drain voltage at which latchup occurs can degrade as the total dose level rises. The mechanism of band-to-band tunneling (BBT) has been discussed. Two-dimensional simulations were conducted to evaluate the BBT effect. It is demonstrated that BBT combined with the positive trapped charge in the buried oxide (BOX) contributes a lot to the latchup effect.
In space or other harsh radiation environments, exposure to high energetic particles such as electrons and protons can significantly reduce the lifetime of integrated circuits (ICs) due to radiation effects.[1] When a material is exposed to total ionizing dose (TID) irradiation, the interface traps as well as the electron–hole pairs can be generated. The generated carriers can result in charge buildup in different types of oxides in the metal–oxide–semiconductor field-effect transistors (MOSFETs), resulting in gradual degradation and change in functionality.[2, 3] The basic mechanism for the degradation is the trapping of holes in the oxide and the interface charge at the Si–SiO2 interface. In addition to the TID effect, heavy ions induced single-event effects (SEEs) can also depress long-term reliability and lead to catastrophic device failure.[4] Hence, it is significant for us to research the radiation effects in MOSFETs.
Silicon-on-insulator (SOI) technologies have attracted more popularity for military and space radiation-hardened applications owing to there being some radiation hardness advantages in contrast with the bulk-silicon technology.[5] For instance, SOI devices have manifested their better resistance against dose rate effects and their very low SEEs sensitivity, owing to the limited charge collection volume for cosmic rays and full dielectric isolation between individual transistors.[6] However, SOI technologies are not inherently immune to the total dose irradiation. Actually, the existence of buried oxide (BOX) layer as well as shallow trench isolation (STI) can make SOI devices more vulnerable to TID irradiation.[7–11] With respect to the radiation-hardened applications, partially depleted SOI (PDSOI) devices are extensively used because of their better resistance to the coupling effect and their better controllability of the threshold voltage. To some extent, the floating body effects (FBEs) can weaken the PDSOI advantages, since these effects can make the device characteristic become abnormal.[12] The detrimental FBEs can be suppressed enough but not eliminated via device modifications, such as H-gate or T-gate structures with body ties.[13] The single transistor latchup (STL) effect, as an extreme case of the FBEs, can finally lead to an abnormally abrupt decrease of the subthreshold slope and the sustainability of high drain current state even at typical operating voltage.[14–17]
TID irradiation induced degradation of SOI devices has been reported by many previous published works. In Ref. [18], Ferlet-Cavrois et al. found that TID irradiation can induce the latchup effect for the SOI transistors fabricated with a 180 nm SOI technology, and it was demonstrated that the STL effect is strongly related to parameters such as the supply voltage, gate length, and TID induced positive trapped charge in the BOX. In a later work, Flament et al.[19] evaluated the total dose response of fully depleted SOI transistors, and figured out that the bias configuration during irradiation and the thickness of the buried oxide can have great influence on the coupling effect and STL effect. However, the total dose induced STL effects for PDSOI devices were not involved in that work. Recently, Peng et al.[20] studied the TID induced coupling effect observed in 130 nm PDSOI input/output (I/O) devices with the body floated. However, the authors only analyzed the influence of the coupling effect on the threshold voltage shift, subthreshold slope increase, and transconductance variation in the front channel device; the impact of the coupling effect on STL was not discussed. Fan et al.[21] investigated the bias dependence of TID induced STL effect for 130 nm PDSOI I/O devices, and figured out that TID irradiation can enhance the impact ionization (II) process and thereby make the device more sensitive to the STL effect. However, the basic mechanism of II is not sufficient to explain the STL effect observed at low drain voltage.
In this work, we comprehensively study the TID induced STL effect for 130 nm PDSOI devices. Different parameters like bias conditions, body doping concentration, and onset drain voltage are used to characterize the STL effect. It is found that the STL effect for partially depleted devices can behave like that for the fully depleted ones due to the coupling effect. What is more, the high body doping concentration can make the devices less sensitive to the STL effect, and the onset drain voltage at which latchup occurs can degrade as the total dose level rises. The two competitive mechanisms (impact ionization and band to band tunneling) are also discussed in detail, and band to band tunneling is considered to contribute a lot to the STL effect.
The I/O NMOSFETs devices employed in the following experiments, with H-gate layout pattern and external body contacts, were fabricated with a 130 nm SOI CMOS technology. Processing was performed on a 200 mm diameter UNIBOND
All experiments were conducted at the Xinjiang Technical Institute of Physics and Chemistry, the Chinese Academy of Sciences, by using 60Co γ-rays as the irradiation source. The dose rate is about 120 rad(Si)/s. As listed in Table
![]() | Table 1.
Bias condition definition. . |
Figure
As the total dose level has reached to 300 krad(Si), the window of hysteresis has become so wide that, even though the
Actually, both hysteresis and STL effects can be attributed to the generated excess holes in the floating body region. During the forward gate voltage scan, the generated holes will be accumulated in the floating body region, thus causing the elevation of the body potential. Finally, when the body potential is high enough to activate the conduction of the parasitic source–body–drain bipolar transistor, an abrupt increase of the drain current will be observed. Continuously, followed by a reverse gate voltage scan, the generated excess holes in the body region cannot be released because of the floating body. Hence, the parasitic source–body–drain bipolar transistor can still be in the conduction state to maintain the drain current at high value, even though the gate voltage has been decreased enough during the reverse scan. Eventually, the drain current curves with forward and reverse sweeps do not coincide with each other, leading to very different results.
We further investigated
Radiation-hardened H-gate devices are also investigated as compared with the commercial ones. With respect to the H-gate device, the leakage current is principally due to the positive trapped charge near the back-channel surface. Therefore, an important method to improve the H-gate device’s resistance against TID induced degradation is improving the threshold voltage of the back-gate transistor by adjusting the body doping concentration. As for devices employed in our experiments, the hardened ones have a higher body doping concentration than the commercial ones.
Figure
Figure
In the above discussion, TID induced single transistor latch effects are observed at high drain voltage
In Fig.
Three more experiments were conducted to further investigate the onset drain voltage for TID induced single transistor latchup. The details of experiment scenarios are as follows: 1) same bias conditions (ON bias), but different total dose levels (300 krad(Si) and 500 krad(Si)); 2) same total dose level (500 krad(Si)), but different bias conditions (ON bias and OFF bias); 3) same total dose level (500 krad(Si)) and same bias conditions (OFF bias), but different devices (commercial device with regular body doping concentration and radiation-hardened device with high body doping concentration).
The results are illustrated in Fig.
![]() | Fig. 6. (color online) Comparison of the onset drain voltages measured in three different scenarios. |
Moreover, the results of the second scenario can further demonstrate that the OFF bias condition is worse as to the TID induced STL effect. These results are consistent with the previously published data,[22] which mainly focused on the bias dependence for the shift of the subthreshold voltage or the off-state leakage current. Furthermore, as shown in the third scenario, the radiation-hardened device has a higher onset drain voltage compared with the commercial one, which can also prove that the high body doping concentration can mitigate the degradation induced by TID irradiation.
STL as one of the floating body effects has been depicted by using an impact ionization (II) process in many previous studies. The II is very plausible in light of the fact that the current can be amplified significantly by the parasitic source–body–drain bipolar transistor. At high drain bias, electron–hole pairs are generated near the body–drain junction region owing to the II process, and the generated holes can finally accumulate in the floating body region to elevate the body potential. When the body potential is high enough to forward bias the parasitic body-to-source diode, more electrons can then drift to the body and be collected at the drain region to enhance the II process. Moreover, the increased body potential can also decrease the threshold voltage of the front gate, eventually resulting in an increase of
It should be pointed out that the II process strongly depends on the source current and the electric field being large enough to trigger the avalanche. This high field condition needed for the II process typically requires a high drain bias. However, the onset drain voltage of latchup can degrade to 1.4 V, according to our experiment results in the above discussion, which is not sufficient to form the strong electric field to trigger the II process, and electrons are also limited at the off-state (
Usually, several mechanisms are involved in the excess holes generation: 1) slow electron–hole recombination in the body region; 2) impact ionization process near the drain–body junction; 3) band to band tunneling due to the high electric field under the gate-drain overlap region; 4) thermal/optical generation carriers. In this work, the thermal and optical generation can be neglected for the operation at low temperature and dark conditions. It should be noted that the precise underlying mechanism of the latch effect is still a matter of debate. However, in the above discussion, we have found that the onset drain voltage at which latchup occurs can degrade as the total dose level rises, and the low drain voltage indicates that the probability of impact ionizing is very small. What is more, as the
Hence, it can be inferred that the gate induced drain leakage via band-to-band tunneling could be a critical mechanism that determines the drain current, including the apparent transition to the high current state. The most important difference between II and BBT is apparent that the II is initiated by the carrier supply and the electric field, whereas the BBT requires only the electric field. BBT is characterized by valence band electron tunneling across the silicon bandgap from the inverted drain surface across the Si bandgap into the quasi-neutral drain. The left valence band holes can then transport into the body region and raise its potential. The field dependent carrier generation rate of BBT can be expressed as
![]() | (1) |
![](cpb_26_3_036103/cpb_26_3_036103_eq45.gif)
![](cpb_26_3_036103/cpb_26_3_036103_eq46.gif)
![](cpb_26_3_036103/cpb_26_3_036103_eq47.gif)
In order to investigate the TID induced single transistor latchup and the impact of BBT on it, we have chosen Sentaurus TCAD tools to conduct the two-dimensional (2D) simulations. Process simulation is chosen to produce the device structure and doping profile. Models employed in the device simulations contain high-field saturation mobility and doping dependent mobility degradation models. Band-to-band tunneling and Shockley–Read–Hall (SRH) recombination models are also required in the device simulations.[23] Note that the II model is turned off to exclude its effect on the results. Fixed charge is placed at the interface of the BOX layer and top silicon to perform the post irradiation simulations. The trapped charge density
BBT has been introduced to explain the high level leakage current in the above discussion. In order to examine the influence of BBT on a PDSOI device degraded by total dose, the tunneling model is turned on during the device simulation. Figure
![]() | Fig. 8. (color online) Cross-section of simulated electric field in SOI NMOSFET with the BBT model. Trapped charge density ![]() ![]() ![]() ![]() |
Figure
![]() | Fig. 9. (color online) Simulated profile of band to band tunneling induced generation rate within the high field region. Trapped charge density ![]() ![]() ![]() ![]() |
Further investigation to examine the impact of the increased radiation exposure on a PDSOI device has been conducted as shown in Fig.
![]() | Fig. 11. (color online) Simulated PD SOI front gate ![]() ![]() ![]() |
Though the II model is turned off in the simulation, we cannot eliminate its effects on the final results. None of the BBT or the II can completely explain the total dose latch effect alone. In fact, carriers generated by the BBT can apply as a source for the II to further reinforce this process.
In this work, TID induced STL effects for the body-floated SOI devices have been studied in detail. Two different bias conditions (ON bias and OFF bias) are compared in experiments, and the results show that the OFF bias case is worse as to the total dose latchup effect. In addition, the high body doping concentration can make devices less sensitive to the TID induced single transistor latchup effect. It demonstrates that the onset drain voltage for single transistor latchup has a heightened sensitivity to TID irradiation, and positive trapped charge in the BOX resulting from TID irradiation can reinforce the degradation of the onset drain voltage. Band-to-band tunneling has been proposed as a critical mechanism to interpret the high latch current state at negative gate bias. 2D simulations have been conducted to evaluate the influence of BBT on the single transistor latch effect, and it demonstrates that the combination of BBT and charge buildup in the BOX contributes a lot to the high latch current. It should be noted that we are not claiming that the II cannot account for the latchup effect. Actually, carriers generated by the BBT can apply as a source for the II to further reinforce this process.
[1] | |
[2] | |
[3] | |
[4] | |
[5] | |
[6] | |
[7] | |
[8] | |
[9] | |
[10] | |
[11] | |
[12] | |
[13] | |
[14] | |
[15] | |
[16] | |
[17] | |
[18] | |
[19] | |
[20] | |
[21] | |
[22] | |
[23] | |
[24] |