3.1 Dose rate effect on front-gate characteristic

Figure 2 shows the front-gate I_D-V_D curves of 0.8 μm PDSOI NMOS devices before and after irradiation at dose rates of 36 and 1800 Gy/h, under the gate voltage of V_G =1, 3 and 5 V, the total dose of 0.5 and 1 kGy(Si), and the irradiation state of OFF-bias. The drain current at 1 kGy(Si) was greater than that at 0.5 kGy(Si). At V_G = 1 V, the drain current increased significantly with the drain voltage V_D, reaching 0.573 mA at 1 kGy(Si), i.e. the TID effects. Given that the threshold voltage shifts in a negative direction, the conducting channel is formed and the devices are more easily switched at V_G =1 V and TID= 1 kGy(Si).

Fig. 2Full-screen View

(Color online) I_D-V_D curves of front-gate as a function of radiation dose at different dose rates.

From the results of 36 Gy(Si)/h (low dose rate, Figs. 2a–2c) and 1.8 kGy(Si)/h (high dose rate, Figs. 2d–2f), the drain shift at high dose rate is considerably worse than that at low dose rate. The saturation currents are especially sensitive to the dose rate effect at 1 kGy(Si) and V_G=5V.

When V_DS≥V_DSsat and V_GS≥V_th, the saturation current is determined as follows

where I_Dsat and V_Dsat are the saturation current and voltage, respectively; V_GS is the gate-to-source voltage; V_th is the threshold voltage, C_ox is the capacitance of the gate oxide, and μ is the carrier mobility.

The dose rate effects at low dose rate results in more serious degeneration of threshold voltage than that at high dose rate ^[17,18], i.e. V_{th(low-dose-rate)} < V_{th(high-dose-rate)}. Substituting this into Eq.(2), we have V_{DSsat(low-dose-rate)} > V_{DSsat(high-dose-rate)}.

Considering the above factors for Eq. (1), the shift in V_DSsat at low dose rate should be greater than that at high dose rate. However, Fig. 2 shows that the shift in V_DSsat at low dose rate is smaller than that at high dose rate. This inconsistency between the calculated and measured data may be caused by excluding the dose rate effect on carrier mobility.

The drain current, gate-to-source voltage, and drain-to-source voltage can be related by

In this section, we mainly study the TID effect on saturation characteristics of SOI MOS devices. Thus, carrier mobility can be determined from Eq.(3) using the I_D and V_GS data at measured at V_D=5 V as shown in Fig. 3, where the mobility degradation at 36 Gy/h is more serious than that at 1.8 kGy/h.

Fig. 3Full-screen View

(Color online) Normalization mobility as a function of irradiation dose under different dose rates

The radiation-degradation of mobility should be taken into account by^[19]

where μ₀ is the mobility before irradiation, N_ot is the density of trapped oxide charges, N_it is the density of interface-state, and α_ot and α_it are the fitting parameters. Eq. (4) shows that the factor causing the shift in carrier mobility is classified into two parts, namely, trapped oxide charge and interface state. The space-charge model^[20] is a widely accepted model to explain the differences in degeneration of electrical characteristics of devices at different dose rates. The local electrical field formed by trapped hole at high dose rate prevents protons from transporting towards the interface, which enhances the density of interface state at low dose rate than at high dose rate. Thus, only minor differences are found in N_ot for the same irradiation dose at different dose rates and the interface state is the main reason for dose rate effect of mobility.

Other bimolecular mechanisms affecting the buildup of interface-state include the recombination of free electrons and holes; recombination of free electrons and hole trapping defects; and hydrogen dimerization reaction to form hydrogen molecule. The low dose rate irradiation is effective, because bimolecular recombination has little effect. Consequently, effective trapped oxide charges and interface states are increasingly induced. However, the less effective irradiation at high dose rate causes interface traps, because the hole and electron densities increase with the dose rate. Thus, an electron is more likely to recombine with a trapped hole before that hole simulates the release of a proton.

The combination of space-charge model and bimolecular recombination model leads to different densities of interface state at different dose rates. The relationship between dose rate and interface state can be described by a constant at low dose rates, a linear function in the transition region, and another constant at high dose rates ^[21-23]. The first crossover dose rate is 0.36 Gy(Si)/h and the second, 3.6 kGy(Si)/h. In comparison, the dose rates in this work, 36 Gy(Si)/h and 1.8 kGy(Si)/h, belong to the linear region, and the interface state is linearly related to dose rate. In this case, the density of interface state at 36 Gy(Si)/h is greater than that at 1.8 kGy(Si)/h. Therefore, the increase in interface-state density can enhance the scattering effect of the interface state on carriers in the channel, decreasing the carrier mobility at 1.8 kGy(Si)/h. Consequently, the shift in the saturation drain current at 36 Gy(Si)/h is slightly worse than that at 1.8 kGy(Si)/h (Fig.2).

3.2 Dose rate effect of body-current under different irradiation bias states

Figure 4 shows the body current dependence of gate voltage under test conditions at drain bias of 5 V. The curves of empty and solid dots are the I_B-V_G characteristic at 36 Gy(Si)/h and 1 kGy(Si)/h, respectively. Both the pre- and after-irradiation I_B-V_G curves have the upside down hump-shaped variations of the body current with increasing gate voltage.

Fig. 4Full-screen View

(Color online) I_B-V_G curves before and after irradiation for ON-bias (a) and OFF-bias (b).

In the I_B-V_G curves after radiation under ON-bias (Fig.4a), the conductive channel is formed when the gate voltage is greater than the threshold voltage. Electrons can drift from source to drain regions, and then large numbers of ehps(electron- hole-pairs) are created by the electron impact ionization under high electrical field of the depletion region. Most of the electrons generated are swept to the drain side and contribute to the drain current. Meanwhile, holes are collected to the body area because of their heavy equivalent mass. However, the threshold voltage gradually decreases with increasing TIDs. So, the gate voltage of the first detected body current shifts in the negative direction. The holes trapped in the buried oxide can also significantly weaken the lateral electric field^[24], which influences the body current strongly, hence the decrease of the maximum body current with increasing doses. However, the shift value is not affected by the dose rate.

Figure 4(b) shows the test results of I_B-V_GS curves after radiation at OFF-bias. The strong electric fields are mainly located at the space charge region of the drain-body junction, the bottom of body region and source region, and back-interface of buried oxide^[25,26], all of which can promote the trap to capture numerous holes. By contrast, ON-bias just applies the work voltage on the gate, which remarkably reduces the density of trapped hole near the Si/SiO₂ interface on the side of buried oxide than that in OFF-bias. Consequently, OFF-bias facilitates the conduction of the back channel. Body current is collected when the device is OFF state and increases with the density of trapped hole and the TID.

The insert of Fig. 4(b) illustrates a convex-concave peak of body current at 36 Gy(Si)/h and TID= 1 kGy(Si), while it does not appear at 1.8 kGy(Si)/h and TID= 1 kGy(Si). The body current at 36 Gy(Si)/h is greater than that at 1.8 kGy(Si)/h when the gate voltages are negative. This indicates that more holes transport in the body region at 36 Gy(Si)/h. When the gate voltage is negative, the maximum body currents are −5.17 and −4.12 μA at 1 kGy(Si)-36 Gy(Si)/h and 3 kGy(Si)-1.8 kGy(Si)/h, respectively. However, these are not observed under ON-bias conditions because of asymmetrical distribution of the electric potential under OFF-bias conditions, at drain voltage of 5 V with the source and gate being grounded, rather than symmetrical distribution of the electric potential under ON-bias and the same conditions of the gate, drain and source. Thus, ehps induced by total dose irradiation are separated quickly by an applied electric field, which is mainly the gate voltage. Then main effect of a positive gate bias is to remove the electrons, thereby decreasing the fraction that recombines with the trapped holes and resulting in no dose rate sensitivity. Therefore, we can see clearly that OFF-bias can lead to ELDRS of the body current for H-gate PDSOI MOSFET devices.

3.3 Dose rate effect in buried oxide for PDSOI devices

Figure 5 shows the transconductance of back-gate at different dose rates and TIDs. The back-gate voltage of the peak transconductance shifts in a negative direction with TIDs, for both the dose rates. This is because that changes in the position of maximum transconductance are mainly influenced by the shift in the threshold voltage. The threshold voltages at TID< 1 kGy(Si) do not differ obviously.

Fig. 5Full-screen View

(Color online) Transconductance of back-gate at different TIDs at high (a) and low (b) dose rates, and the normalized maximum transconductance (c).

In Fig.5(a), the double peak of back-gate transconductance is seen at 3 kGy(Si). This characteristic is not evident in both the pre-irradiation and doses of ≤ 2 kGy(Si). The interface state at the SiO₂/Si interface between the buried oxide and silicon layer may cause the double-peak phenomenon. The interface state can introduce local permitted state in the band gap of Si, which will further form continuous or discrete electronic energy levels. These energy levels increase with the dose, enhancing the capturing of the channel electron by the donor interface-state of the back gate.

Figure 5(c) shows the normalized maximum transconductance of the back-gate at low and high dose rates as a function of irradiation dose, which are extracted from Fig.5(b). The decay of maximum transconductance at 36 Gy(Si)/h is much worse than at 1.8 kGy(Si)/h, and the decay trend for 1.8 kGy(Si)/h is gradually retarded as the dose increases. MOSFET is a device with a surface channel. On the one hand, the channel electron is scattered by the interface state, resulting in decreasing mobility. On the other hand, the channel electron can fill the interface trap, leading to decreased trap density. These can cause the reduction in transconductance with increasing dose. Meanwhile, the interface state density at 36 Gy(Si)/h is larger than that at 1.8 kGy(Si)/h. Thus, the ELDRS effect appears in the decay of back-gate transconductance.