logo

Effects of total dose irradiation on the threshold voltage of H-gate SOI NMOS devices

NUCLEAR ELECTRONICS AND INSTRUMENTATION

Effects of total dose irradiation on the threshold voltage of H-gate SOI NMOS devices

WANG Qian-Qiong
LIU Hong-Xia
CHEN Shu-Peng
WANG Shu-Long
FEI Chen-Xi
ZHAO Dong-Dong
Nuclear Science and TechniquesVol.27, No.5Article number 117Published in print 20 Oct 2016Available online 30 Aug 2016
54200

This work researched the impact of total dose irradiation on the threshold voltage of N-type metal oxide semiconductor field effect transistors (NMOSFETs) in Silicon on Insulator (SOI) technology. Using the sub-threshold separation technology, the factor causing the threshold voltage shift was divided into two parts: trapped oxide charges and interface states, the effects of which are presented under irradiation. Furthermore, by analyzing the data, the threshold voltage shows a negative shift at first, then turns to positive shift when irradiation dose is lower. Additionally, the influence of the dose rate effects on threshold voltage is discussed. The research results show that the threshold voltage shift is more significant in low dose rate conditions, even for a low dose of 100 krad(Si). The degeneration value of threshold voltage is 23.4% and 58.0% for the front-gate and the back-gate at the low dose rate.

Silicon on Insulator (SOI)threshold voltagedose rate effectsinterface states

1 Introduction

Silicon on Insulator (SOI) technology has been regarded as having highly promising military applications for many years. With the structure of completed dielectric isolation, this technology can eliminate latch-up and has demonstrated advantages in radiation tolerance over bulk-silicon technologies [1]. However, the characterization of SOI MOSFETs in irradiation environments is more complex than that of bulk silicon devices because of its buried oxide, which leads to more Si/SiO2 interface in SOI device[2-5]. Therefore, it is challenging to develop radiation-hardened SOI devices[6-9]. The total dose has an effect on electric characteristic parameters under different irradiation dose conditions, especially on the threshold voltage of an SOI MOS device. Nevertheless, there is little concrete analysis regarding the threshold voltage under varying technology and total dose radiation. In order to estimate this influence, experiments and mechanism analysis are needed.

The paper focuses on the total dose effects. This work also researches the degradation of the threshold voltage and mobility for SOI devices, caused by not only the trapped oxide charges but also the interface states. First, the threshold voltage shift shows a “rebound” phenomenon, which usually appears under ultra-high total dose[10,11] or during annealing at high temperatures[12]. However, in this work, the phenomenon happens when the irradiation dose is 200 krad(Si) at room temperature. Two possible reasons account for this phenomenon: the special structure of SOI MOS devices and the effects of buried oxide on carries in channel. Through the use of sub-threshold separation technology[13], the paper quantitatively analyses the effects of positive charges and interface states on threshold voltage degradation in SOI devices, further explaining this “rebound” phenomenon. Moreover, the sub-threshold separation technology can be used to estimate the effects of positive charges and interface states on the electricity characteristic parameters for different irradiation conditions. As far as we know, the dose rate effects in the electrical characteristics of these devices are usually considered for the bulk silicon MOSFETs[14] or bipolar transistors[15-18]. Nevertheless, little work has been conducted to investigate the dose rate effects on SOI MOS devices. At present, dose rate effects on SOI devices have remained a challenge, largely because of insufficient experimental data. In this paper, the research data shows that the threshold voltage shift in low dose rate conditions is more significant when the SOI devices are exposed to γ-rays. At the low dose rate, the threshold voltage shift of the back-gate is 43.0%. By contrast, this threshold voltage shift is just 29.1% at the high dose rate.

2 Experiment and device structure

The experimental samples consisted of Partially Depleted Silicon on Insulator (PDSOI) devices with a H-gate bilateral body-contacted structure, which was fabricated used LDD (Light Doped Drain Source) technology. Fig.1 shows a cross-section of the sample. VG is the gate voltage, VD is drain voltage, and VS is source voltage. tox indicates the thickness of gate oxide, and tbox refers to the thickness of buried oxide, the values of which are 375 nm and 17.5 nm, respectively. The gate width/length ratio is 8 μm/0.8 μm and the work voltage is 5 V.

Fig.1
Cross-section of samples
pic

The experiment uses the γ-rays of 60Co sources. The dose rates in this experiment were 1 rad(Si)/s and 50 rad(Si)/s. The condition of the irradiation was ON-state (i.e. the gate was contacted to Vdd [i.e. 5 V] and other ports were grounded). Two types of body states (i.e. floating-body and body-grounded) were chosen for these experiments. The devices’ electric characteristics were collected by HP4156, which is a precision semiconductor parameter analyzer that the experimental computer controlled. Each test was completed in half an hour.

3 Results and discussion

3.1 The effect of total dose irradiation on threshold voltage

Fig.2 shows the measured I-V characteristics of the front-gate and back-gate for PDSOI NMOSFETs before and after irradiation (VD=0.1 V). The X-axis indicates the gate applied voltage and the Y-axis indicates the logarithm of the drain current.

Fig.2
Measured transfer characteristics of body-grounded device for (a) front-gate and (b) back-gate
pic

Both Fig.2(a) and (b) show that the off-state leakage current increases significantly when the irradiation dose is 300 krad(Si). The radiation bias is ON-state (i.e. VG=5 V), which forms a high electric field in the gate oxide. And then the radiation inducing electrons and holes quickly separate and migrate to the positive and negative electrodes in the oxide, respectively. Electron mobility is greater than hole mobility, so the recombination mechanism of free electrons and holes is weakened. As the density of positive charges increases, more and more effective oxide charges migrate to the SiO2/Si interface and attract electrons near the interface. When the concentration of electrons is large enough, the parasitic conducting channel is formed, which causes the leakage current to increase.

Regarding the I-V characteristics of the front-gate and back-gate, shown in Fig.2, the leakage current reaches up to 10-10~10-9 A when the test gate voltage is 0 V. However, in the front-gate I-V characteristics, a greater number of positive charges migrate to the interface as the gate voltage increases, which leads to a large leakage current as a function of the total dose. When the total dose is 300 krad(Si), there is an exponential growth in the leakage current. However, the test voltage for back-gate I-V characteristics is applied on the substrate; substrate voltage has a slight effect on the electrons in the channel thanks to the isolation of buried oxide in SOI devices.

Using linear extrapolation, this work presents the degradation of threshold voltage as a function of total dose. Fig.3 (a) and (b) indicate the different phenomena that occur between the body-grounded devices and floating-body devices. For the body-grounded devices, shown in Fig.3(a), the negative shift quantity of the threshold voltage (ΔVth) increases after irradiation. When the total dose is 200 krad(Si), the shift is 30.4% for the front-gate. When the total dose is 300 krad(Si), the shift increases to 40.9%. For the back-gate, the degradation of threshold voltage is much more significant.

Fig.3
Normalization ΔVth of front-gate and back-gate as a function of irradiation dose for (a) body-grounded devices, (b) floating-body devices, and (c) body-grounded devices in high dose conditions
pic

When dealing with the floating-body devices, as shown in Fig.3(b), the ΔVth of per unit dose shifts toward the negative direction at the outset. For the front-gate, the negative shift increases from 5.2% to 38.3% as the dose increases from 50 krad(Si) to 200 krad(Si). However, the threshold voltage then shifts toward positive values as the dose increases further. The negative shift decreases from 38.3% to 33.4% as the dose increases from 200 krad(Si) to 300 krad(Si).

The above phenomenon of negative drift turning to positive drift is referred to as a “rebound”. This rebound phenomenon has been mentioned in some studies, appearing for higher doses (i.e., up to 1 Mrad(Si)[10,11]). However, in this paper, the phenomenon arises when the dose increases from 200 krad(Si) to 300 krad(Si). Fig.3(c) shows that the threshold voltage of body-grounded devices shifts in a negative direction as the dose increases, even as it reaches 1 Mrad(Si). For the front-gate, the negative shift increases from 38.2% to 62.2% as the dose increases from 200 krad(Si) to 500 krad(Si). The negative shift can increase even to 91.3% when the dose is 1 Mrad(Si).

The threshold voltage of PDSOI MOS devices is similar to that of bulk silicon devices. The threshold voltage of enhancement-mode nMOSFET is given by

Vth0=VFB0+2ϕF+qNAxdmaxCox (1)

where, VFB0 is the flat band voltage, NA is the doping concentration in silicon film, ϕF is femi potential, ϕF=kTqln(NAni), xdmax(=2εSi(2ϕF+VB)/qNA) is the maximal width of the depletion layer, VB is the voltage of the body region, which is equivalent to the substrate bias of bulk silicon devices, Cox is the oxide capacitance of front-gate.

Additionally, the threshold voltage for the irradiation condition is given by

Vth=Vth0+q(Not+Nit)Cox (2)

where Not is the density of the oxide charge and Nit is the density of the interface state.

The factor that causes the threshold voltage shift is divided into two parts: the trapped oxide charges and interface states, given by

ΔVth=ΔVit+ΔVot (3)

where ΔVit(=qNit/Cox) is the contribution of the threshold voltage shift due to the interface states, and ΔVot(=qNot/Cox) is the contribution of the threshold voltage shift due to the trapped oxide charges.

From the I-V characteristic of MOSFET, the drain current is given by

ID=μ(WL)αCox2β2(niNA)2(1eβVDS)eβϕ(βϕ)1/2 (4)

where ID and VDS are the drain current and voltage, respectively, W/L is the gate width/length, ni is the intrinsic carrier concentration, β=q/kT, and φ is the surface potential.

The constant α is given by

α=2εSitoxεoxLD (5)

where tox is the oxide thickness, LD is the Debye length given by LD=[εSi/(βqNA)]1/2, εSi and εox are the dielectric constants of Si and SiO2 respectively, and μ is the carrier mobility, which is a radiation sensitive parameter. The relationship between threshold and carrier mobility can be shown as

ID=μCoxWL[(VGSVth)VDSVDS2/2] (6)

If the VDS is very small (i.e. VDS2(VGSVth)), (6) can be simplified into (7):

IDμCoxWL(VGSVth)VDS (7)

The carrier mobility μ can be obtained by equation (7) according to the experiment data of ID and VGS. The midgap current is defined to be the current, which occurs when the surface potential equals φF (i.e., ϕ=ϕF(=(kT/q)ln(NA/ni))). Using the carrier mobility μ and femi potential ϕF for equation (4), we can obtain the midgap current. Because the midgap current belongs in the range of 10-14~10-13A, it is essential that the sub-threshold of the I-V curve is linearly extrapolated in the negative direction.

The midgap voltage shift in sub-threshold curves due to trapped oxide charges, which is given by

ΔVot=(Vmg)2(Vmg)1 (8)

where the labels 2 and 1 refer respectively to subthreshold-current curves at different radiation levels.

The difference between the threshold voltage and midgap voltage is defined as stretchout voltage Vso, given by

Vso=VthVmg (9)

The threshold voltage shift due to the interface state is as follows:

ΔVit=(Vso)2(Vso)1=|(Vth)2(Vmg)2||(Vth)1(Vmg)1|=ΔVthΔVot (10)

Using the sub-threshold separation technology, the factors ΔVot and ΔVit for the floating-body and body-grounded devices are shown in Fig.4 (a) and (b) respectively, which is a function of the irradiation dose. For the body-grounded devices, the ΔVit shifts toward a negative value in the first instance. When the dose increases from 150 krad(Si) to 200 krad(Si), this change occurs slowly. The “rebound” phenomenon arises when the dose reaches more than 200 krad(Si). Meanwhile, the ΔVot shifts toward a negative value as a function of the dose, the quantity of which is larger than it is for ΔVit. The above factors cause ΔVth to shift toward a negative value as a function of the dose. For the floating-body device, when the dose increases to over 200 krad(Si), the “rebound” phenomenon arises for both ΔVit and ΔVot, causing the threshold voltage to shift toward a positive value.

Fig.4
ΔVth, ΔVit, and ΔVot as a function of irradiation dose of the front-gate for (a) body-grounded and (b) floating-body
pic

Because of the special structure of SOI devices, the total dose radiation induces a large number of positive charges in the buried oxide layer. For floating-body devices, the potential of buried oxide increases as the charges increase. This attracts electrons to move near the interface, which weakens the channel’s formation. However, for body-grounded devices, the incremental potential of buried oxide has little effect on the body because of its constant voltage. Thus, the threshold voltage shifts in a negative direction.

3.2 The effect of the dose rate on threshold voltage

Fig.5 shows front-gate and back-gate I-V curves for PDSOI nMOSFETs. There are discrepancies under different dose rates (i.e. the high dose rate is 50 rad(Si)/s and the low dose rate is 1 rad(Si)/s). A lager shift of the sub-threshold current at the low dose rate is observed than that at the high dose rate. The total dose radiation induces positive charges that are proportional to the thickness of the oxide. For a measured characteristic of the back-gate, more serious degradation can be found at lower dose rates.

Fig.5
Measured transfer characteristics as a function of irradiation dose for (a) front-gate and (b) back-gate under different dose rate conditions
pic

Fig.6 shows the threshold voltage shift for the front-gate and back-gate at different dose rates. At the beginning of irradiation, we observed a similar shift of threshold voltage at different dose rates. When the dose value increases further, the degeneration of threshold voltage worsens under low dose rate conditions. In addition, as the irradiation dose reaches 50 krad(Si), the threshold voltage shift under low dose rate conditions is larger than the shift under high dose rate conditions. For obtaining the same threshold voltage shift, more dose are needed for high dose rate conditions than that of low dose rate conditions. These results indicate enhanced low dose rate sensitivity (ELDRS) of H-gate partially depleted SOI MOSFET devices for 60Co γ-irradiation.

Fig.6
Normalization ΔVth of (a) front-gate and (b) back-gate under different dose rate conditions
pic

Fig.7 shows the shift of ΔVth, ΔVit and ΔVot for the front-gate using sub-threshold separation technology. It is noted that the trapped oxide charges, as well as the interface states, are responsible for the ELDRS effects in SOI devices. Additionally, the interface states play a significant role in the threshold shift under low dose rate conditions. This threshold voltage shift as a result of the interface states is much larger than those occurring as a result trapped oxide charges (i.e., when the irradiation dose is 100 krad(Si), the ΔVot is -0.07 V and the ΔVit is -0.19 V in the lower dose rate experiments).

Fig.7
ΔVth, ΔVit and ΔVot of the front-gate as a function of irradiation dose under low dose rate conditions
pic

The interface is a transition region that constitutes oxide-nonoxide and ordered-disordered structures. This includes the bending bond of oxide cavities Si3Si and the bond of Si3OH or Si3H, which is formed during the thermal growth process. Both of these are easily broken to form the dangling bond of tri-valent silicon (i.e. the interface state) under ionizing radiation. Edward H Poindexter[19] gave the name “dangling bond” to the Pb center. In most cases, the density of weak bonds is higher when the distance to the interface is shorter.

Irradiation induced excitons react with oxide to form the neutral hydrogen atoms H0. Consequently, the trapped hole becomes H+, which shifts toward the interface because of the additional electric field (VG0). This reaction is given by[20]

H++e+Si3HH2+Si3 (11)

The reaction produces the center of the interface state Pb. The above process is called the H+ model.

Moreover, the relationship between the generation of electron-hole pairs and dose rate is given by

Gr=g0DY(E) (12) Y(E)=(|E|+E0|E|+E1)m (13)

where g0 is the generation rate of electron-hole pairs, D is the dose rate, E is the electric field, and E0, E1 and m are constants.

Because hole mobility is greater than the mobility of H+, trapped oxide holes form electrostatic barriers, which can prevent H+ from reaching the Si/SiO2 interface and generating interface states. At the lower dose rate, the buildup process of oxide charges is weakened. Both the holes and H+ can transport to nearby the Si/SiO2 interface, and participate in the formation of trapped oxide charges and interface states.

By using the experimental data from ID and VGS, the carrier mobility under varying irradiation doses can be obtained from equation (7), as shown in Fig.8, where obvious mobility degradation can be observed. As the trapped oxide charges and interface states increase, the degradation of mobility should be taken into account by [21]

Fig.8
Normalization μ as a function of irradiation dose under different dose rates
pic
μeff=μ01+αoxqNoxCox1+αitqNitCox1 (14)

where αox and αit are fitting parameters.

According to equations (11)-(13), when the density of the interface state increases along with dose, this enhances the scattering effect of the interface state on carriers in the channel, leading to a decrease in the carriers’ mobility, as shown in (14).

Fig.8 also shows that the degradation of carrier mobility worsens at a low dose rate, which is similar to the conclusion of [22]. Thus, this work’s mobility data further verifies that the interface state is the one of the primary factors to cause differences in characteristic degradation under total dose radiation at different dose rates, and also plays an important role in the effect of ELDRS on SOI devices.

4 Conclusion

This paper investigated the impact of total dose effects on the threshold voltage of H-gate SOI NMOS devices. The results show that the threshold voltage shifts in a negative direction as a function of the irradiation dose. However, regarding floating-body devices, when the irradiation dose is not very high, the “rebound” phenomenon occurs, indicating that the threshold voltage shifts from a negative to a positive direction. The positive charges in the buried oxide layer induced by irradiation raise its potential. Electrons are attracted to the Si/SiO2 interface. This weakens the formation of the channel, which causes the threshold to shift to a positive direction. The degradation of the threshold voltage at different dose rates shows the enhanced low dose rate sensitivity of partially depleted SOI MOSFET devices for 60Co γ-irradiation. This work also determined that the interface state is the one of the primary factors to cause a difference in characteristic degradation at different dose rates, which plays an important role in the effect of ELDRS on SOI devices.

References
[1] Schwank J R, Shaneyfelt M R, Fleetwood D M, et al.

Rdiation Effects in MOS Oxides

. IEEE Trans on Nucl Sci, 2008, 55: 1833-1853. DOI: 10.1109/TNS.2008.2001040.
Baidu ScholarGoogle Scholar
[2] Schwank J R, Ferlet-Cavrois V, Shaneyfelt M R, et al.

Radiation Effects in SOI Technologies

. IEEE Trans. on Nucl Sci, 2003, 55: 522-537. DOI: 10.1109/TNS.2003.812930.
Baidu ScholarGoogle Scholar
[3] Flament O, Torres A, Ferlet-Cavrois V.

Bias Dependence of FD Transistor Response to Total Dose Irradiation

. IEEE Trans on Nucl Sci, 2003, 50: 2316-2321. DOI: 10.1109/TNS.2003.822594.
Baidu ScholarGoogle Scholar
[4] Wu X, Lu W, Wang X, et al.

Influence of channel length and layout on TID for 0.18μm NMOS transistors

. Nuclear Science and Techniques, 2013, 24:060202-6. DOI: 10.13538/j.1001-8042/nst.2013.06.019.
Baidu ScholarGoogle Scholar
[5] Potter K, Morgan K, Shaw C, et al.

Total ionizing dose response of fluorine implanted Silicon-On-Insulator buried oxide

. Microelectronics Reliability, 2014, 54:2339-2343. DOI: 10.1016/j.microrel.2014.07.018.
Baidu ScholarGoogle Scholar
[6] Paillet P, Schwank J R, Shaneyfelt M R et al.

Total Dose Hardness Assurance Testing Using Laboratory Radiation Sources

. IEEE Trans on Nucl Sci, 2003, 50: 2310-2315. DOI: 10.1109/TNS.2003.821392.
Baidu ScholarGoogle Scholar
[7] Liu J, Zhou J C, Luo H W, et al.

Total-dose-induced edge effect in SOI NMOS transistors with different layouts

. Microelectronics Reliability, 2010, 50:45-47. DOI: 10.1016/j.microrel.2009.09.003.
Baidu ScholarGoogle Scholar
[8] Mattiazzo S, Battaglia M, Bisello D, et al.

Total Dose Effects on a FD-SOI Technology for Monolithic Pixel Sensors

. IEEE Trans on Nucl Sci, 2010, 57: 2135-2141. DOI: 10.1109/TNS.2009.2038378.
Baidu ScholarGoogle Scholar
[9] Huang H X, Bi D W, Chen M, et al.

Improving Total Dose Tolerance of Buried Oxides in SOI Wafers by Multiple-Step Si+ Implantation

. IEEE Trans on Nucl Sci, 2014, 61: 1400-1406. DOI: 10.1109/TNS.2014.2316017.
Baidu ScholarGoogle Scholar
[10] Faccio F, Heijine E H M, Jarron P, et al.

Study of device parameters for analog IC design in a 1.2μm CMOS-SOI technology after 10 Mrad

. IEEE Trans on Nucl Sci, 1992, 39: 1739-1746. DOI: 10.1109/23.211361.
Baidu ScholarGoogle Scholar
[11] Fleetwood D M, Eisen H A.

Total-Dose Radiation Hardness Assurance

. IEEE Trans on Nucl Sci, 2003, 50: 552-564. DOI: 10.1109/TNS.2003.813130.
Baidu ScholarGoogle Scholar
[12] Gao B, Liu G, Wang L X, et al.

Research on Enhanced Low Dose Rate Sensitivity Effect for PMOSFET Used in Space Dosimeterm

. Atomtic Energy Science and Technology, 2013, 47: 848-853.DOI: 10.7538/yzk.2013.47.05.0848.
Baidu ScholarGoogle Scholar
[13] McWhorter P J, Winokur P S.

Simple technique for separating the effects of interface traps and trapped oxide charge in metal-oxide-semiconductor transistors

. Appl Phys Lett, 1986, 48: 133-135. DOI: 10.1063/1.96974.
Baidu ScholarGoogle Scholar
[14] He B P, Yao Z B, Sheng J K, et al.

Study of the dose rate effect of 180nm nMOSFETs

. Chin Phys C, 2015, 39: 016004(1-5). DOI: 10.1088/1674-1137/39/1/016004.
Baidu ScholarGoogle Scholar
[15] Harold P H, Ronald L P, Steven C W, et al.

Mechanisms for Radiation Dose-Rate Sensitivity of Bipolar Transistors

. IEEE Trans on Nucl Sci, 2003, 50: 1901-1909. DOI: 10.1109/TNS.2003.821803.
Baidu ScholarGoogle Scholar
[16] Hjalmarson H P, Pease R L, Hembree C E, et al.

Dose-rate dependence of radiation-induced interface trap density in silicon bipolar transistors

. Nucl Instrum Methods Phys Res B, 2006, 250: 269-273. DOI: 10.1016/j.nimb.2006.04.122.
Baidu ScholarGoogle Scholar
[17] Chen X J, Barnaby H J, Schrimpf R D, et al.

Nature of Interface Defect Buildup in Gate Bipolar Devices Under Low Dose Rate Irradiation

. IEEE Trans on Nucl Sci, 2006, 53: 3649-3654. DOI: 10.1109/TNS.2006.885375.
Baidu ScholarGoogle Scholar
[18] Boch J, F. Saigné, Schrimpf R D, et al.

Physical Model for the Low-Dose-Rate Effect in Bipolar Devices

. IEEE Trans on Nucl Sci, 2006, 53: 3655-3660. DOI: 10.1109/TNS.2006.886008.
Baidu ScholarGoogle Scholar
[19] Poindexter E H.

MOS interface states: overview and physicochemical perspective

. Semicond Sci Technol, 1984, 4: 961-969. DOI: 10.1088/0268-1242/4/12/001.
Baidu ScholarGoogle Scholar
[20] Saks N S, Klein R B, Griscom D L.

Formation of interface traps in MOSFETs during annealing following low temperature irradiation

. IEEE Trans on Nucl Sci, 1988, 35: 1234-1240. DOI: 10.1109/23.25445.
Baidu ScholarGoogle Scholar
[21] Alvarado J, Boufouss E, Kilchytska V, et al.

Compact model for single event transients and total dose effects at high temperatures for partially depleted SOI MOSFETs

. Microelectronics Reliability, 2010, 50: 1852-1856. DOI: 10.1016/j.microrel.2010.07.040.
Baidu ScholarGoogle Scholar
[22] He B P, Yao Y J, Peng H L, et al.

Effects of Temperature and Dose Rates on Mobility of NMOS Devices

. Microelectronics, 2000, 30: 179-181. DOI: 10.3969/j.issn.1004-3365.2000.03.012.
Baidu ScholarGoogle Scholar