Total ionizing dose effects on 12-bit CBCMOS digital-to-analog converters

Xin Wang; Wu Lu; Qi Guo; Xue Wu; Shanbin Xi; Wei Deng; Jiangwei Cui; Jinxin Zhang

doi:10.1088/1674-4926/34/12/124006

J. Semicond. > 2013, Volume 34 > Issue 12 > 124006

SEMICONDUCTOR DEVICES

Total ionizing dose effects on 12-bit CBCMOS digital-to-analog converters

Xin Wang^{1, 2, 3}, Wu Lu^{1, 2,}, Qi Guo^{1, 2}, Xue Wu^{1, 2, 3}, Shanbin Xi^{1, 2, 3}, Wei Deng^{1, 2, 3}, Jiangwei Cui^{1, 2} and Jinxin Zhang^{1, 2, 3}

+ Author Affiliations

Corresponding author: Lu Wu, luwu@ms.xjb.ac.cn

DOI: 10.1088/1674-4926/34/12/124006

Abstract: A digital-to-analog converter (DAC) in CBCMOS technology was irradiated by ⁶⁰Co γ-rays at various dose rates and biases for investigating the ionizing radiation response of the DAC. The radiation responses show that the function curve and the key electrical parameters of the DAC in CBCMOS technology are sensitive to total dose and dose rates. Under different bias conditions, the radiation failure levels were different, and the radiation damage under operation bias conditions was more severe. Finally, test results were preliminarily analyzed by relating the failure mode to DAC architecture and process technology.

Key words: digital-to-analog, CBCMOS, dose rate effect, ionizing radiation

1. Introduction

With the rapid developments in the field of satellite-based telecommunications and the move from analog to digital control of space flights to the international space station, significant importance must be placed on radiation damage to digital-to-analog converters, which are of particular interest due to their complex design, performance, and importance in digital signal processing.

The modern data converter process technologies include the bipolar process, the thin film resistor process, complementary bipolar (CB) processes, CMOS processes, CMOS and LWT thin film resistors, LC$^{2}$MOS and BiCMOS with LWT thin film resistors, etc. In the last decade, CMOS has become a dominant process for data converters--replacing more expensive bipolar laser wafer trimmed devices. The radiation effect and radiation failure mode of CMOS DAC has been studied^[1-7]. However, fewer reports are available on the radiation failure mode and failure mechanism of DACs built by using other processes. Analog devices' CB processes, which are designed for even higher speeds with lower breakdowns, allow high input impedance op amps as well as sample-and-hold amplifiers for data converters. However, there are few results about the radiation and dose effects of CBCMOS DACs, which combine CMOS devices and well-matched high speed PNP and NPN transistors.

In this study, a CBCMOS process DAC under different biases was irradiated at different dose rates and different biases in a $^{60}$Co $\gamma $-ray source. The dose rate effects and radiation failure mode of this CBCMOS DAC were studied, and both time-dependent effects (TDEs) and the dose rate effect in the DAC are observed. In addition, the preliminary study on the relationship of the manufacturing process and device structure of the DACs and the low of radiation induced degradation was carried out in this paper.

2. Device and electrical parameter measurements

2.1 Device description

The tested device is AD8300, which is a complete 12-bit, voltage-output digital-to-analog converter built using a CBCMOS process. It is based on a complementary bipolar process combined with a CMOS process. Compared with the conventional BiCMOS process, complementary devices are added to the CMOS process in the CBCMOS process. The CB process is mainly used to provide high-speed, well-matched PNP and NPN transistors. The fabrication process of modules such as the rail-to-rail output op amp, which contain highly well-matched pair transistors, mostly adopt the bipolar or CB process. AD8300 contains a 12-bit laser-trimmed digital-to-analog converter, a curvature-corrected bandgap reference, a rail-to-rail output op amp, a serial-input register, and a DAC register. The internal DAC is a 12-bit device with an output that swings from GND potential to 0.4 V generated from the internal band-gap voltage. It uses a laser-trimmed segmented R-2R ladder, which is switched by N-channel MOSFETs. The DAC output is internally connected to the rail-to-rail output op amp.

2.2 Electrical parameters measured

There are basically two sets of specifications used for data converters which can be divided simplistically into DC and AC specifications. AC specifications are more common with higher frequency applications and less sensitive than DC specifications when the DACs are used in radioactive environments. So, before and after irradiation and throughout the annealing procedure, static transfer functions and the DC specifications were tested and analyzed. The DC parameters tested during the experiment were dc errors, which were offset error, gain error, two types of linearity error (differential and integral), and power parameters.

All DC parameters were offline tested on an automated Amida-3001XP testing system from AMIDA Technology, Inc. The DAC was full-code tested to obtain the function curve, electrical characteristics, and error parameters. All parameters before and after irradiation, and throughout the annealing procedure, were analyzed.

3. Experimental facilities and setup

The total radiation experiments were carried out at the $^{60}$Co $\gamma $-ray source (Xinjiang Technical Institute of Physics and Chemistry). The radiation dose rates for the test specimens were respectively 0.5 Gy(Si)/s as the high dose rate (HDR) and 0.001 Gy(Si)/s as the low dose rate (LDR), and the total accumulated dose was 800 Gy(Si). According to the MIL-STD-883G, test specimens were enclosed in a Pb/Al container to minimize dose enhancement effects caused by low-energy, scattered radiation. Post-irradiation annealing the HDR-irradiated chip at room temperature was conducted, and the annealing time was equal or greater than the time of LDR-irradiation.

All the samples were irradiated in either non-operation biases or operation biases. The non-operation bias was achieved by grounding all the related pins. The samples were set to the VDD pin $=$ 3 V and the digital input pins were all high level for the operation bias. The bias conditions were kept during the HDR and LDR irradiation and annealing.

Initial electrical measurements were made on 20 samples. There are four experiment statuses during the irradiation test, which are respectively non-operation biases at HDR, non-operation biases at LDR, operation biases at HDR, and operation biases at LDR, and 5 samples for each status. Experimental results show that the failure modes of all 5 samples for each status are the same, but the degrees of irradiation damage to them show a little difference. The results of the samples are detailed below; the radiation damage is most serious.

4. Results and discussion

4.1 The radiation response of the static transfer function

Figures 1 and 2 show function curves, which are the input-output relationships for different total ionizing dose (TID) under HDR and LDR at different biases. The function curve of the DAC at non-operational bias has no significant distortion, whereas the function curve at operating bias is significantly distorted. So the DAC irradiated at the operating bias was more sensitive than the non-operational bias under both HDR and LDR. Under LDR and operating bias, the initial shape of the function curve was a straight line, but then the slope of the curve became smaller and smaller as the total ionizing dose increased. And with the deeper irradiation, the curve looked like a staircase with fewer and fewer steps. Finally, it became a horizontal line with a value close to the ground. Under HDR and operating bias, the shape of the function curve was not significantly distorted until the TID was 800 Gy(Si), when the shape of function became a horizontal line with a value close to the ground.

Figure 1. Input-output relationship for different TID under (a) HDR of 0.5 Gy(Si)/s and (b) LDR of 1 $\times$ 10$^{-3}$ Gy(Si)/s at operating bias.

DownLoad: Full-Size Img PowerPoint

Figure 2. Input-output relationship for different TID under (a) HDR of 0.5 Gy(Si)/s and (b) LDR of 1 $\times$ 10$^{-3}$ Gy(Si)/s at non-operational bias.

DownLoad: Full-Size Img PowerPoint

4.2 The radiation response of electric parameters

Figure 3 shows the evolution of the power consumption of the DAC. The power consumption of a DAC can be described in terms of three kinds of power parameters: Pw0 (the power consumption of DAC when the digital input is 0), Pw1 (the power consumption of DAC when the digital input is 4095) and Pw $=$ max (Pw0, Pw1). By comparison, Pw0 is more sensitive which began increasing after 200Gy(Si) where the other two parameters almost unchanged during HDR radiation under operational bias. There was no significant increase in Pw0 during HDR radiation and annealing under non-operational bias, as well as during LDR radiation. Pw0 began increasing after 200Gy(Si) and became greater than 21 mW which was more than five times as much as the initial value when the total dose is 800 Gy(Si) during HDR radiation under operating bias. Then Pw0 gradually decreased during subsequent annealing and eventually settled to a value close to the initial.

Figure 3. Pw0 versus (a) total dose and (b) room-temperature annealing time.

DownLoad: Full-Size Img PowerPoint

Through the above analysis of the variation of Pw0 under different TID and during annealing time, the degradation of device performances were different under LDR and HDR, and the difference can be eliminated by room-temperature annealing. This dose-rate effect is known as the time-dependent effect (TDE).

Figure 4 shows the evolution of full-scale voltage, $V_{\rm out1}$. $V_{\rm out1}$ is determined by applying the all "1" digital codes to the DAC and measuring its output. The defined limit of $V_{\rm out1}$ is clearly described in the data sheet and is no more than 2.056V and no less than 2.039 V. There was no great decrease in the $V_{\rm out1}$ during HDR and LDR irradiation under non-operating bias. Under operating bias, $V_{\rm out1}$ obviously decreased after 200 Gy(Si) during LDR irradiation, whereas it was stable in the defined limit until a critical TID which was greater than 600 Gy(Si) and reached during the HDR irradiation.

Figure 4. $V_{\rm out1}$ versus (a) total dose and (b) room-temperature annealing time.

DownLoad: Full-Size Img PowerPoint

In addition, the zero-scale error, $V_{\rm out0}$, is measured by applying the all "0" digital codes to the DAC digital input and measuring the output. The typical value of $V_{\rm out0}$ for AD8300 is $+0.5$ mV and no more than $+3$ mV according to the data sheet. There was no a significant change in $V_{\rm out0}$ during the HDR and LDR irradiation.

At the beginning of annealing after the HDR irradiation, there was little increase in the $V_{\rm out1}$ until the annealing time was 180 min, when $V_{\rm out1}$ suddenly returned to the original value.

Through the analysis of the variation of $V_{\rm out1}$ under different TID and during annealing time, the device performance degraded more at LDR than at HDR under the same total dose level, and this is a "true" dose-rate effect that cannot be eliminated by the long-term annealing.

4.3 The radiation response of error parameters

In a DAC, we are concerned with two measures of the linearity of its transfer function: integral nonlinearity, INL (or relative accuracy), and differential nonlinearity, DNL. Figure 5 shows the DNL for different TID under HDR of 0.5 Gy(Si)/s (the right images) and LDR of 1 $\times$ 10$^{-3}$ Gy(Si)/s (the left images) at operating bias. Differential nonlinearity is the maximum deviation of an actual analog output step, between adjacent input codes, from the ideal step value of $+$1LSB (least significant bit), calibrated based on the gain of the particular DAC. The value of the DNL is calculated for each of the 4096 possible output voltages in Fig. 5. The typical value of DNL for AD8300 is $\pm $1/2 LSB and the absolute value of DNL is no more than 1LSB according to the data sheet.

Figure 5. DNL for different TID under an LDR of 1 $\times$ 10$^{-3}$ Gy(Si)/s and an HDR of 0.5 Gy(Si)/s at operating bias.

DownLoad: Full-Size Img PowerPoint

Figures 5(a) and 5(a') show the value of DNL before irradiation, and all the DNL values are within normal range. The values of DNL at some high bits became more negative outside the normal range when TID reached 100 Gy(Si) during both HDR and LDR irradiation. The values of DNL at more high bits showed negative direction increase and became more negative with the accumulation of TID as shown in Figs. 5(c) and 5(c'), in which TID reached 200 Gy(Si). The variation trend of DNL and the failure mode of DAC were basically the same during both HDR and LDR irradiation until TID reached 200Gy(Si).

The values of DNL at lots of high bits positive direction increased significantly, and the negative values did not increase any further when TID reached 400 Gy(Si) during LDR irradiation. During HDR irradiation, the values of DNL at higher bits showed a negative direction increase and became more negative when TID reached 400 Gy(Si), whereas positive values were still within the normal range. The different variation trend of the DNL after 400 Gy(Si) showed that the failure mode of DAC changes for LDR irradiation after a certain TID between 200 Gy(Si) and 400 Gy(Si).

Figure 6 shows the INL for different TID under HDR of 0.5 Gy(Si)/s (the right images) and LDR of 1 $\times$ 10$^{-3}$ Gy(Si)/s (the left images) at operating bias. Integral nonlinearity is the maximum deviation, at any point in the transfer function, of the output voltage level from its ideal value--which is a straight line drawn through the actual zero and full-scale of the DAC. The value of the INL is calculated for each of the 4096 possible output voltages in Fig. 6. The typical value of DNL for AD8300 is $\pm \frac1 2$ LSB and the absolute value of DNL is no more than 2 LSB according to the data sheet.

Figure 6. INL for different TID under HDR of 0.5 Gy(Si)/s and LDR of 1 $\times$ 10$^{-3}$ Gy(Si)/s at operating bias.

DownLoad: Full-Size Img PowerPoint

Figures 6(a) and 6(a') show the value of INL before irradiation, and all the INL values are within a normal range. The values of INL at some middle bits became more positive outside the normal range when TID reached 100 Gy(Si) during both HDR and LDR irradiation. The values of INL at more bits showed a positive direction increase and became more positive with the accumulation of TID, and some values at low bits became more negative outside the normal range as shown in Fig. 4(c'), in which TID reached 200 Gy(Si). Meanwhile, during LDR irradiation the values of INL at low bits showed a negative direction increase and were outside the normal range, whereas positive values were still within the normal range.

The values of INL at lots of middle and high bits continued to increase to become more positive, while the INL at more low bits became more negative when TID reached 400 Gy(Si) during HDR irradiation. During LDR irradiation the values of INL at almost all bits showed negative direction increase and became significantly more negative when TID reached 400Gy(Si).

Figure 7 shows the evolution of the DNL and INL of the DAC. Through the analysis of the variation of the two parameters under different TID and during annealing time, the error parameters of a device degraded differently at LDR and at HDR under the same total dose level, and this is a "true" dose-rate effect that cannot be eliminated by long-term annealing.

Figure 7. DNL and INL versus (a) total dose and (b) room-temperature annealing time.

DownLoad: Full-Size Img PowerPoint

5. Discussion

5.1 Total ionizing effects

5.1.1 The failure mechanism of the AD8300

(1) The malfunction of the switches as one of the causes of DAC degradation (aberrance of the function curve) during LDR irradiationp

Figure 1(b) shows the input-output relationship for different TID under LDR at operating biases. The initial shape is a straight line but, as the total ionizing dose increases, the function looks like a staircase with fewer and fewer steps. This might be caused by both the decrease of the threshold voltage and the increase of the leakage current of the NMOSFET that separates the ground from the segment R/2R ladder network^[1]. See Fig. 8 for an equivalent circuit schematic of the analog section. The internal DAC is a 12-bit device with an output that swings from GND potential to 0.4 V generated from the internal band-gap voltage. It uses a laser-trimmed segmented R-2R ladder which is switched by N-channel MOSFETs. Because of the malfunction of the NMOSFETs, the ratios between the current of each branch inside the converter were not a power of 2. The values of output voltage could not be fitted to a straight line since the values would not be an integer multiple of the LSB unit. Therefore, the ratio between the output and the input code will be non-linear regularly and the size of the non-linearity must be estimated. But the non-linearity of the function curve is a staircase with erratic steps. This irregularity does not occur because of a malfunction of the switches, but is due to the degradation of other sections of the DAC.

Figure 8. Equivalent AD8300 schematic of the analog portion.

DownLoad: Full-Size Img PowerPoint

(2) The degradation of the reference and output amplifier causing the decrease of the full-scale voltage and gain of the DAC

As the total ionizing dose increases, in addition to the distortion of the functional curve mentioned above, the slope of the function curve and the full-scale voltage decreases which leads to an irregularity of the non-linearity of the function curve.

The internal DAC is a 12-bit device with an output that swings from GND potential to 0.4 V generated from the internal band-gap voltage. The output is buffered by a low power consumption precision amplifier. The rail-to-rail amplifier is configured with a gain of approximately five in order to set the 2.0475 V full-scale output. Degradation of either the reference or the rail-to-rail amplifier can result in a decrease of the slope of the function curve and the full-scale voltage.

5.1.2 The failure mechanism of the switches

Figure 9 shows the structure of an internal NMOS switch which consists of two inverters and two NMOSFETs. When the NMOS switch is exposed to ionizing radiation, a number of electron-hole pairs are produced in the SiO$_{2}$ of the NMOS. After initial electron-hole recombination, the electrons migrate out of the regions quickly, but the holes remain in SiO$_{2}$ due to their slower mobility. Some holes react with oxygen vacancies to form positive oxide-trapped charges, while others migrate to the Si-SiO$_{2}$ interface and react with dangling bonds to create the interface state. The radiation-induced oxide trapped charge, which is expressed as $\Delta N_{\rm ot}$, results in negative drift of the threshold voltage and an increase of leakage current. The radiation-induced interface state, which is expressed as $\Delta N_{\rm it}$, results in positive drift of the threshold voltage and an increase of leakage current^[7]. The total dose response of the NMOS switch results from radiation-induced oxide trapped charges and a subsequently induced interface state. During the initial periods of irradiation, radiation-induced oxide trapped charges are more than an interface state, that is $\Delta N_{\rm ot}$ $>$ $\Delta N_{\rm it}$, which results in the threshold voltage drifting negatively and an increase of leakage current. If the threshold voltage of $N_{2}$ becomes lower than 0, it will be always ON even though the gate voltage is 0 V.

Figure 9. Structure of an internal NMOS switch in a DAC.

DownLoad: Full-Size Img PowerPoint

Although both the oxide trapped charge and interface state contribute to the degradation of the devices, their performances are different during the post-irradiation annealing. The oxide-trapped charge can be eliminated by annealing at room temperature, but the interface state cannot be significantly removed unless annealing above 100 ℃ takes place^[9-11]. The dose rate effect of the NMOS switch results from the different annealing performances of the two kinds of radiation-induced charge.

5.2 Dose rate effects

The dose rate effects and radiation failure mode of this CBCMOS DAC were studied, and it is found that the degree of degradation and radiation failure were both different in the DAC under different dose rates.

The dose rate effect shown by the three parameters, $V_{\rm out1}$, DNL and INL, is an obvious "true" dose-rate effect which is different from the dose rate effect shown by the parameters of DAC devices fabricated by using traditional CMOS^{[1, 5, 7]}or bipolar processes^[6]. Through preliminary analysis, the author believes that the difference is a result of the mixed process which is based on the CB process combining the CMOS process.

The difference of radiation failure modes during HDR irradiation and LDR irradiation is shown as:

(1) The different variation tendency of the input-output relationship with the accumulation of TID.

(2) The different variation tendency of the value of linearity errors with the accumulation of TID.

Furthermore, the radiation failure modes of a DAC are likely to be different under initial irradiation and deep irradiation according to the variation tendency of the value of linearity errors. More research should be carried out to find out the causal factors for this difference.

6. Conclusion

According to above analysis, we can see that the radiation mode of the CBCMOS DAC is very complex because of the advanced manufacturing process and the complex device structure. The complexity manifests as the complicated variation tendency of both the functional curve and the static parameters. By comparative analysis of the variation of the functional curve and the static parameters under different dose rates and different biases, the results obtained are as follows:

(1) The malfunction of the switches is one of the causes of DAC degradation during LDR irradiation which is shown by the function curve becoming like a staircase with fewer and fewer steps.

(2) The degradation of the reference and output amplifier causes the decrease of the full-scale voltage and the gain of the DAC.

(3) Both TDE and the "true" dose rate effect in the AD8300 are observed, and both the degree of degradation and the degradation mode of the functional curve and static parameters are very different under different dose rate. Both kinds of dose rate effect shown by the testing parameters result from the process characteristics of the CBCMOS process, and the kinds of dose rate effect shown by different parameters are related to the fabrication processes of the relevant modules which are sensitive to TID.

(4) The degree of degradation of the functional curve and static parameters is very different under different biases, and worsen under an operating bias.

References

[1]	Franco F J, Lozano J, Agapito J A. Radiation effects on CMOS R/2R ladder digital-to-analog converters. Noordwijk, Netherlands:RADECS, 2003:571
[2]	Aghara S, Fink R J, Charlton W S, et al. Degradation of commercially available DAC ICs in a mixed-radiation environment. IEEE Rad Effects Data Workshop Rec, Piscataway, NJ, 2002:34 http://ieeexplore.ieee.org/document/1281318/?reload=true&arnumber=1281318
[3]	Ampe J, Thai V, Buchner S, et al. COTS ADC & DAC selection and qualification for the GLAST mission. IEEE Rad Effects Data Workshop, Piscataway, NJ, 2005:79
[4]	Franco F J, Zong Y, De Agapito J A, et al. Radiation tolerant D/A converters for the LHC cryogenic system. Nuclear Instruments and Methods in Physics Research Section A:Accelerators, Spectrometers, Detectors and Associated Equipment, 2005, 553(3):604 http://cat.inist.fr/?aModele=afficheN&cpsidt=17255658
[5]	Prater J S, Brown B, Trinh T. Analysis of low dose rate effects on parasitic bipolar structures in CMOS processes for mixed-signal integrated circuits. IEEE Trans Nucl Sci, 2011, 58(3):1023 doi: 10.1109/TNS.2011.2116041
[6]	Wang Y, Lu W, Ren D, et al. Total dose effect of 10-bit bipolar D/A converter under different ⁶⁰Co γ dose rates. Atomic Energy Science and Technology, 2009, 43(10):951
[7]	Liu Y, Yang S, Lin D, et al. Synergistic effect of neutron and gamma irradiation on 10-bit CMOS digital-to-analog converter. High Power Laser and Particle Beams, 2010, 22(9):2186 doi: 10.3788/HPLPB
[8]	Winokur P S, Schwank J R, Mcwhrter P J, et al. Cr relating the radiation response of MOS capacitors and transistors. IEEE Trans Nucl Sci, 1984, 31(6):1453 doi: 10.1109/TNS.1984.4333529
[9]	Witczak S C, Schrimpf R D, Galloway K F, et al. Gain degradation of lateral and substrate pnp bipolar junction transistors. IEEE Trans Nucl Sci, 1996, 43(6):3151 doi: 10.1109/23.556919
[10]	Nowlin R N, Fleetwood D M, Schrimpf R D, et al. Hardness-assurance and testing issues for bipolar/BiCMOS devices. IEEE Trans Nucl Sci, 1993, 40(6):1686 doi: 10.1109/23.273492
[11]	Schrimpf R D, Graves R J, Schmidt D M, et al. Hardness-assurance issues for lateral PNP bipolar junction transistors. IEEE Trans Nucl Sci, 1995, 42(6):1641 doi: 10.1109/23.488761

Fig. 1. Input-output relationship for different TID under (a) HDR of 0.5 Gy(Si)/s and (b) LDR of 1 $\times$ 10$^{-3}$ Gy(Si)/s at operating bias.

DownLoad: Full-Size Img PowerPoint

Fig. 2. Input-output relationship for different TID under (a) HDR of 0.5 Gy(Si)/s and (b) LDR of 1 $\times$ 10$^{-3}$ Gy(Si)/s at non-operational bias.

DownLoad: Full-Size Img PowerPoint

Fig. 3. Pw0 versus (a) total dose and (b) room-temperature annealing time.

DownLoad: Full-Size Img PowerPoint

Fig. 4. $V_{\rm out1}$ versus (a) total dose and (b) room-temperature annealing time.

DownLoad: Full-Size Img PowerPoint

Fig. 5. DNL for different TID under an LDR of 1 $\times$ 10$^{-3}$ Gy(Si)/s and an HDR of 0.5 Gy(Si)/s at operating bias.

DownLoad: Full-Size Img PowerPoint

Fig. 6. INL for different TID under HDR of 0.5 Gy(Si)/s and LDR of 1 $\times$ 10$^{-3}$ Gy(Si)/s at operating bias.

DownLoad: Full-Size Img PowerPoint

Fig. 7. DNL and INL versus (a) total dose and (b) room-temperature annealing time.

DownLoad: Full-Size Img PowerPoint

Fig. 8. Equivalent AD8300 schematic of the analog portion.

DownLoad: Full-Size Img PowerPoint

Fig. 9. Structure of an internal NMOS switch in a DAC.

DownLoad: Full-Size Img PowerPoint

[1]	Franco F J, Lozano J, Agapito J A. Radiation effects on CMOS R/2R ladder digital-to-analog converters. Noordwijk, Netherlands:RADECS, 2003:571
[2]	Aghara S, Fink R J, Charlton W S, et al. Degradation of commercially available DAC ICs in a mixed-radiation environment. IEEE Rad Effects Data Workshop Rec, Piscataway, NJ, 2002:34 http://ieeexplore.ieee.org/document/1281318/?reload=true&arnumber=1281318
[3]	Ampe J, Thai V, Buchner S, et al. COTS ADC & DAC selection and qualification for the GLAST mission. IEEE Rad Effects Data Workshop, Piscataway, NJ, 2005:79
[4]	Franco F J, Zong Y, De Agapito J A, et al. Radiation tolerant D/A converters for the LHC cryogenic system. Nuclear Instruments and Methods in Physics Research Section A:Accelerators, Spectrometers, Detectors and Associated Equipment, 2005, 553(3):604 http://cat.inist.fr/?aModele=afficheN&cpsidt=17255658
[5]	Prater J S, Brown B, Trinh T. Analysis of low dose rate effects on parasitic bipolar structures in CMOS processes for mixed-signal integrated circuits. IEEE Trans Nucl Sci, 2011, 58(3):1023 doi: 10.1109/TNS.2011.2116041
[6]	Wang Y, Lu W, Ren D, et al. Total dose effect of 10-bit bipolar D/A converter under different ⁶⁰Co γ dose rates. Atomic Energy Science and Technology, 2009, 43(10):951
[7]	Liu Y, Yang S, Lin D, et al. Synergistic effect of neutron and gamma irradiation on 10-bit CMOS digital-to-analog converter. High Power Laser and Particle Beams, 2010, 22(9):2186 doi: 10.3788/HPLPB
[8]	Winokur P S, Schwank J R, Mcwhrter P J, et al. Cr relating the radiation response of MOS capacitors and transistors. IEEE Trans Nucl Sci, 1984, 31(6):1453 doi: 10.1109/TNS.1984.4333529
[9]	Witczak S C, Schrimpf R D, Galloway K F, et al. Gain degradation of lateral and substrate pnp bipolar junction transistors. IEEE Trans Nucl Sci, 1996, 43(6):3151 doi: 10.1109/23.556919
[10]	Nowlin R N, Fleetwood D M, Schrimpf R D, et al. Hardness-assurance and testing issues for bipolar/BiCMOS devices. IEEE Trans Nucl Sci, 1993, 40(6):1686 doi: 10.1109/23.273492
[11]	Schrimpf R D, Graves R J, Schmidt D M, et al. Hardness-assurance issues for lateral PNP bipolar junction transistors. IEEE Trans Nucl Sci, 1995, 42(6):1641 doi: 10.1109/23.488761

1	Subthreshold analog techniques Nanjian Wu Journal of Semiconductors, 2019, 40(2): 020204. doi: 10.1088/1674-4926/40/2/020204
2	Total ionizing dose radiation effects on NMOS parasitic transistors in advanced bulk CMOS technology devices Baoping He, Zujun Wang, Jiangkun Sheng, Shaoyan Huang Journal of Semiconductors, 2016, 37(12): 124003. doi: 10.1088/1674-4926/37/12/124003
3	The total ionizing dose effect in 12-bit, 125 MSPS analog-to-digital converters Xue Wu, Wu Lu, Yudong Li, Qi Guo, Xin Wang, et al. Journal of Semiconductors, 2014, 35(4): 044008. doi: 10.1088/1674-4926/35/4/044008
4	The effects of electron irradiation on the optical properties of the organic semiconductor polypyrrole J. V. Thombare, M. C. Rath, S. H. Han, V. J. Fulari Journal of Semiconductors, 2013, 34(9): 093001. doi: 10.1088/1674-4926/34/9/093001
5	Total ionizing dose effects on a radiation-induced BiMOS analog-to-digital converter Xue Wu, Wu Lu, Yiyuan Wang, Jialing Xu, Leqing Zhang, et al. Journal of Semiconductors, 2013, 34(1): 015006. doi: 10.1088/1674-4926/34/1/015006
6	Total dose ionizing irradiation effects on a static random access memory field programmable gate array Gao Bo, Yu Xuefeng, Ren Diyuan, Li Yudong, Sun Jing, et al. Journal of Semiconductors, 2012, 33(3): 034007. doi: 10.1088/1674-4926/33/3/034007
7	A low power 3.125 Gbps CMOS analog equalizer for serial links Ju Hao, Zhou Yumei, Jiao Yishu Journal of Semiconductors, 2010, 31(11): 115003. doi: 10.1088/1674-4926/31/11/115003
8	The Bipolar Field-Effect Transistor: VIII. Longitudinal Gradient of Longitudinal Electric Field (Two-MOS-Gates on Pure-Base) Jie Binbin, Sah Chihtang Journal of Semiconductors, 2010, 31(7): 071001. doi: 10.1088/1674-4926/31/7/071001
9	Effects of the reciprocating parameters of the carrier on material removal rate and non-uniformity in CMP Wang Cailing, Kang Renke, Jin Zhuji, Guo Dongming Journal of Semiconductors, 2010, 31(12): 126001. doi: 10.1088/1674-4926/31/12/126001
10	Effect of total ionizing dose radiation on the 0.25 μm RF PDSOI nMOSFETs with thin gate oxide Liu Mengxin, Han Zhengsheng, Bi Jinshun, Fan Xuemei, Liu Guang, et al. Journal of Semiconductors, 2009, 30(1): 014004. doi: 10.1088/1674-4926/30/1/014004
11	The Bipolar Field-Effect Transistor: VII. The Unipolar Current Mode for Analog-RF Operation (Two-MOS-Gates on Pure-Base) Jie Binbin, Sah Chih-Tang Journal of Semiconductors, 2009, 30(3): 031001. doi: 10.1088/1674-4926/30/3/031001
12	Method of simulation of low dose rate for total dose effect in 0.18 μm CMOS technology He Baoping, Yao Zhibin, Guo Hongxia, Luo Yinhong, Zhang Fengqi, et al. Journal of Semiconductors, 2009, 30(7): 074009. doi: 10.1088/1674-4926/30/7/074009
13	Total Ionizing Dose Radiation Effects of RF PDSOI LDMOS Transistors Liu Mengxin, Han Zhengsheng, Bi Jinshun, Fan Xuemei, Liu Gang, et al. Journal of Semiconductors, 2008, 29(11): 2158-2163.
14	Total Dose Radiation Hardened PDSOI CMOS 64k SRAMs Guo Tianlei, Zhao Fazhan, Liu Gang, Li Duoli, Li Jing, et al. Chinese Journal of Semiconductors , 2007, 28(8): 1184-1186.
15	Total Ionizing Dose Radiation Effects on MOS Transistors with Different Layouts Li Dongmei, Huangfu Liying, Gou Qiujing, Wang Zhihua Chinese Journal of Semiconductors , 2007, 28(2): 171-175.
16	A New CMOS Image Sensor with a High Fill Factor and the Dynamic Digital Double Sampling Technique Liu Yu, Wang Guoyu Chinese Journal of Semiconductors , 2006, 27(2): 313-317.
17	Total Dose Gamma Irradiation Effects and AnnealingCharacteristics of a SiGe HBT Niu Zhenhong, Guo Qi, Ren Diyuan, Liu Gang, Gao Song, et al. Chinese Journal of Semiconductors , 2006, 27(9): 1608-1611.
18	Total Dose Radiation-Hard 0.8μm SOI CMOS Transistors and ASIC Xiao Zhiqiang, Hong Genshen, Zhang Bo, Liu Zhongli Chinese Journal of Semiconductors , 2006, 27(10): 1750-1754.
19	Influence of Floating Body Effect on Radiation Hardness of PD SOI nMOSFETs Zhao Hongchen, Hai Chaohe, Han Zhengsheng，and Qian He Chinese Journal of Semiconductors , 2005, 26(2): 234-237.
20	Sensitivity of Total-Dose Radiation Hardness of SIMOX Buried Oxides to Doses of Nitrogen Implantation into Buried Oxides Zheng Zhongshan, Liu Zhongli, Zhang Guoqiang, Li Ning, Li Guohua, et al. Chinese Journal of Semiconductors , 2005, 26(5): 862-866.

Search

GET CITATION

Xin Wang, Wu Lu, Qi Guo, Xue Wu, Shanbin Xi, Wei Deng, Jiangwei Cui, Jinxin Zhang. Total ionizing dose effects on 12-bit CBCMOS digital-to-analog converters[J]. Journal of Semiconductors, 2013, 34(12): 124006. doi: 10.1088/1674-4926/34/12/124006

X Wang, W Lu, Q Guo, X Wu, S B Xi, W Deng, J W Cui, J X Zhang. Total ionizing dose effects on 12-bit CBCMOS digital-to-analog converters[J]. J. Semicond., 2013, 34(12): 124006. doi: 10.1088/1674-4926/34/12/124006.

Export: BibTex EndNote

Article Metrics

Article views: 2445 Times PDF downloads: 17 Times Cited by: 0 Times

History

Received: 28 March 2013 Revised: 07 August 2013 Online: Published: 01 December 2013

Article Navigation > Journal of Semiconductors > 2013 > 34(12): 124006

Xin Wang, Wu Lu, Qi Guo, Xue Wu, Shanbin Xi, Wei Deng, Jiangwei Cui, Jinxin Zhang. Total ionizing dose effects on 12-bit CBCMOS digital-to-analog converters[J]. Journal of Semiconductors, 2013, 34(12): 124006. doi: 10.1088/1674-4926/34/12/124006 ****X Wang, W Lu, Q Guo, X Wu, S B Xi, W Deng, J W Cui, J X Zhang. Total ionizing dose effects on 12-bit CBCMOS digital-to-analog converters[J]. J. Semicond., 2013, 34(12): 124006. doi: 10.1088/1674-4926/34/12/124006.

Citation:

PDF( 1131 KB)

Total ionizing dose effects on 12-bit CBCMOS digital-to-analog converters

DOI: 10.1088/1674-4926/34/12/124006

Xin Wang^1,2,3,
Wu Lu^1,2,,
Qi Guo^1,2,
Xue Wu^1,2,3,
Shanbin Xi^1,2,3,
Wei Deng^1,2,3,
Jiangwei Cui^1,2,
Jinxin Zhang^1,2,3

1.
Xinjiang Technical Institute of Physics & Chemistry, Chinese Academy of Sciences, Urumqi 830011, China
2.
Xinjiang Key Laboratory of Electronic Information Materials and Devices, Urumqi 830011, China
3.
University of Chinese Academy of Sciences, Beijing 100049, China

More Information

Corresponding author: Lu Wu, luwu@ms.xjb.ac.cn
Received Date: 2013-03-28
Revised Date: 2013-08-07
Published Date: 2013-03-01

Abstract

Abstract

A digital-to-analog converter (DAC) in CBCMOS technology was irradiated by ⁶⁰Co γ-rays at various dose rates and biases for investigating the ionizing radiation response of the DAC. The radiation responses show that the function curve and the key electrical parameters of the DAC in CBCMOS technology are sensitive to total dose and dose rates. Under different bias conditions, the radiation failure levels were different, and the radiation damage under operation bias conditions was more severe. Finally, test results were preliminarily analyzed by relating the failure mode to DAC architecture and process technology.
- digital-to-analog,
- CBCMOS,
- dose rate effect,
- ionizing radiation

FullText(HTML)

1. Introduction

With the rapid developments in the field of satellite-based telecommunications and the move from analog to digital control of space flights to the international space station, significant importance must be placed on radiation damage to digital-to-analog converters, which are of particular interest due to their complex design, performance, and importance in digital signal processing.

The modern data converter process technologies include the bipolar process, the thin film resistor process, complementary bipolar (CB) processes, CMOS processes, CMOS and LWT thin film resistors, LC$^{2}$MOS and BiCMOS with LWT thin film resistors, etc. In the last decade, CMOS has become a dominant process for data converters--replacing more expensive bipolar laser wafer trimmed devices. The radiation effect and radiation failure mode of CMOS DAC has been studied^[1-7]. However, fewer reports are available on the radiation failure mode and failure mechanism of DACs built by using other processes. Analog devices' CB processes, which are designed for even higher speeds with lower breakdowns, allow high input impedance op amps as well as sample-and-hold amplifiers for data converters. However, there are few results about the radiation and dose effects of CBCMOS DACs, which combine CMOS devices and well-matched high speed PNP and NPN transistors.

In this study, a CBCMOS process DAC under different biases was irradiated at different dose rates and different biases in a $^{60}$Co $\gamma $-ray source. The dose rate effects and radiation failure mode of this CBCMOS DAC were studied, and both time-dependent effects (TDEs) and the dose rate effect in the DAC are observed. In addition, the preliminary study on the relationship of the manufacturing process and device structure of the DACs and the low of radiation induced degradation was carried out in this paper.

2. Device and electrical parameter measurements

2.1 Device description

The tested device is AD8300, which is a complete 12-bit, voltage-output digital-to-analog converter built using a CBCMOS process. It is based on a complementary bipolar process combined with a CMOS process. Compared with the conventional BiCMOS process, complementary devices are added to the CMOS process in the CBCMOS process. The CB process is mainly used to provide high-speed, well-matched PNP and NPN transistors. The fabrication process of modules such as the rail-to-rail output op amp, which contain highly well-matched pair transistors, mostly adopt the bipolar or CB process. AD8300 contains a 12-bit laser-trimmed digital-to-analog converter, a curvature-corrected bandgap reference, a rail-to-rail output op amp, a serial-input register, and a DAC register. The internal DAC is a 12-bit device with an output that swings from GND potential to 0.4 V generated from the internal band-gap voltage. It uses a laser-trimmed segmented R-2R ladder, which is switched by N-channel MOSFETs. The DAC output is internally connected to the rail-to-rail output op amp.

2.2 Electrical parameters measured

There are basically two sets of specifications used for data converters which can be divided simplistically into DC and AC specifications. AC specifications are more common with higher frequency applications and less sensitive than DC specifications when the DACs are used in radioactive environments. So, before and after irradiation and throughout the annealing procedure, static transfer functions and the DC specifications were tested and analyzed. The DC parameters tested during the experiment were dc errors, which were offset error, gain error, two types of linearity error (differential and integral), and power parameters.

All DC parameters were offline tested on an automated Amida-3001XP testing system from AMIDA Technology, Inc. The DAC was full-code tested to obtain the function curve, electrical characteristics, and error parameters. All parameters before and after irradiation, and throughout the annealing procedure, were analyzed.

3. Experimental facilities and setup

The total radiation experiments were carried out at the $^{60}$Co $\gamma $-ray source (Xinjiang Technical Institute of Physics and Chemistry). The radiation dose rates for the test specimens were respectively 0.5 Gy(Si)/s as the high dose rate (HDR) and 0.001 Gy(Si)/s as the low dose rate (LDR), and the total accumulated dose was 800 Gy(Si). According to the MIL-STD-883G, test specimens were enclosed in a Pb/Al container to minimize dose enhancement effects caused by low-energy, scattered radiation. Post-irradiation annealing the HDR-irradiated chip at room temperature was conducted, and the annealing time was equal or greater than the time of LDR-irradiation.

All the samples were irradiated in either non-operation biases or operation biases. The non-operation bias was achieved by grounding all the related pins. The samples were set to the VDD pin $=$ 3 V and the digital input pins were all high level for the operation bias. The bias conditions were kept during the HDR and LDR irradiation and annealing.

Initial electrical measurements were made on 20 samples. There are four experiment statuses during the irradiation test, which are respectively non-operation biases at HDR, non-operation biases at LDR, operation biases at HDR, and operation biases at LDR, and 5 samples for each status. Experimental results show that the failure modes of all 5 samples for each status are the same, but the degrees of irradiation damage to them show a little difference. The results of the samples are detailed below; the radiation damage is most serious.

4. Results and discussion

4.1 The radiation response of the static transfer function

Figures 1 and 2 show function curves, which are the input-output relationships for different total ionizing dose (TID) under HDR and LDR at different biases. The function curve of the DAC at non-operational bias has no significant distortion, whereas the function curve at operating bias is significantly distorted. So the DAC irradiated at the operating bias was more sensitive than the non-operational bias under both HDR and LDR. Under LDR and operating bias, the initial shape of the function curve was a straight line, but then the slope of the curve became smaller and smaller as the total ionizing dose increased. And with the deeper irradiation, the curve looked like a staircase with fewer and fewer steps. Finally, it became a horizontal line with a value close to the ground. Under HDR and operating bias, the shape of the function curve was not significantly distorted until the TID was 800 Gy(Si), when the shape of function became a horizontal line with a value close to the ground.

Figure 1. Input-output relationship for different TID under (a) HDR of 0.5 Gy(Si)/s and (b) LDR of 1 $\times$ 10$^{-3}$ Gy(Si)/s at operating bias.

DownLoad: Full-Size Img PowerPoint

Figure 2. Input-output relationship for different TID under (a) HDR of 0.5 Gy(Si)/s and (b) LDR of 1 $\times$ 10$^{-3}$ Gy(Si)/s at non-operational bias.

DownLoad: Full-Size Img PowerPoint

4.2 The radiation response of electric parameters

Figure 3 shows the evolution of the power consumption of the DAC. The power consumption of a DAC can be described in terms of three kinds of power parameters: Pw0 (the power consumption of DAC when the digital input is 0), Pw1 (the power consumption of DAC when the digital input is 4095) and Pw $=$ max (Pw0, Pw1). By comparison, Pw0 is more sensitive which began increasing after 200Gy(Si) where the other two parameters almost unchanged during HDR radiation under operational bias. There was no significant increase in Pw0 during HDR radiation and annealing under non-operational bias, as well as during LDR radiation. Pw0 began increasing after 200Gy(Si) and became greater than 21 mW which was more than five times as much as the initial value when the total dose is 800 Gy(Si) during HDR radiation under operating bias. Then Pw0 gradually decreased during subsequent annealing and eventually settled to a value close to the initial.

Figure 3. Pw0 versus (a) total dose and (b) room-temperature annealing time.

DownLoad: Full-Size Img PowerPoint

Through the above analysis of the variation of Pw0 under different TID and during annealing time, the degradation of device performances were different under LDR and HDR, and the difference can be eliminated by room-temperature annealing. This dose-rate effect is known as the time-dependent effect (TDE).

Figure 4 shows the evolution of full-scale voltage, $V_{\rm out1}$. $V_{\rm out1}$ is determined by applying the all "1" digital codes to the DAC and measuring its output. The defined limit of $V_{\rm out1}$ is clearly described in the data sheet and is no more than 2.056V and no less than 2.039 V. There was no great decrease in the $V_{\rm out1}$ during HDR and LDR irradiation under non-operating bias. Under operating bias, $V_{\rm out1}$ obviously decreased after 200 Gy(Si) during LDR irradiation, whereas it was stable in the defined limit until a critical TID which was greater than 600 Gy(Si) and reached during the HDR irradiation.

Figure 4. $V_{\rm out1}$ versus (a) total dose and (b) room-temperature annealing time.

DownLoad: Full-Size Img PowerPoint

In addition, the zero-scale error, $V_{\rm out0}$, is measured by applying the all "0" digital codes to the DAC digital input and measuring the output. The typical value of $V_{\rm out0}$ for AD8300 is $+0.5$ mV and no more than $+3$ mV according to the data sheet. There was no a significant change in $V_{\rm out0}$ during the HDR and LDR irradiation.

At the beginning of annealing after the HDR irradiation, there was little increase in the $V_{\rm out1}$ until the annealing time was 180 min, when $V_{\rm out1}$ suddenly returned to the original value.

Through the analysis of the variation of $V_{\rm out1}$ under different TID and during annealing time, the device performance degraded more at LDR than at HDR under the same total dose level, and this is a "true" dose-rate effect that cannot be eliminated by the long-term annealing.

4.3 The radiation response of error parameters

In a DAC, we are concerned with two measures of the linearity of its transfer function: integral nonlinearity, INL (or relative accuracy), and differential nonlinearity, DNL. Figure 5 shows the DNL for different TID under HDR of 0.5 Gy(Si)/s (the right images) and LDR of 1 $\times$ 10$^{-3}$ Gy(Si)/s (the left images) at operating bias. Differential nonlinearity is the maximum deviation of an actual analog output step, between adjacent input codes, from the ideal step value of $+$1LSB (least significant bit), calibrated based on the gain of the particular DAC. The value of the DNL is calculated for each of the 4096 possible output voltages in Fig. 5. The typical value of DNL for AD8300 is $\pm $1/2 LSB and the absolute value of DNL is no more than 1LSB according to the data sheet.

Figure 5. DNL for different TID under an LDR of 1 $\times$ 10$^{-3}$ Gy(Si)/s and an HDR of 0.5 Gy(Si)/s at operating bias.

DownLoad: Full-Size Img PowerPoint

Figures 5(a) and 5(a') show the value of DNL before irradiation, and all the DNL values are within normal range. The values of DNL at some high bits became more negative outside the normal range when TID reached 100 Gy(Si) during both HDR and LDR irradiation. The values of DNL at more high bits showed negative direction increase and became more negative with the accumulation of TID as shown in Figs. 5(c) and 5(c'), in which TID reached 200 Gy(Si). The variation trend of DNL and the failure mode of DAC were basically the same during both HDR and LDR irradiation until TID reached 200Gy(Si).

The values of DNL at lots of high bits positive direction increased significantly, and the negative values did not increase any further when TID reached 400 Gy(Si) during LDR irradiation. During HDR irradiation, the values of DNL at higher bits showed a negative direction increase and became more negative when TID reached 400 Gy(Si), whereas positive values were still within the normal range. The different variation trend of the DNL after 400 Gy(Si) showed that the failure mode of DAC changes for LDR irradiation after a certain TID between 200 Gy(Si) and 400 Gy(Si).

Figure 6 shows the INL for different TID under HDR of 0.5 Gy(Si)/s (the right images) and LDR of 1 $\times$ 10$^{-3}$ Gy(Si)/s (the left images) at operating bias. Integral nonlinearity is the maximum deviation, at any point in the transfer function, of the output voltage level from its ideal value--which is a straight line drawn through the actual zero and full-scale of the DAC. The value of the INL is calculated for each of the 4096 possible output voltages in Fig. 6. The typical value of DNL for AD8300 is $\pm \frac1 2$ LSB and the absolute value of DNL is no more than 2 LSB according to the data sheet.

Figure 6. INL for different TID under HDR of 0.5 Gy(Si)/s and LDR of 1 $\times$ 10$^{-3}$ Gy(Si)/s at operating bias.

DownLoad: Full-Size Img PowerPoint

Figures 6(a) and 6(a') show the value of INL before irradiation, and all the INL values are within a normal range. The values of INL at some middle bits became more positive outside the normal range when TID reached 100 Gy(Si) during both HDR and LDR irradiation. The values of INL at more bits showed a positive direction increase and became more positive with the accumulation of TID, and some values at low bits became more negative outside the normal range as shown in Fig. 4(c'), in which TID reached 200 Gy(Si). Meanwhile, during LDR irradiation the values of INL at low bits showed a negative direction increase and were outside the normal range, whereas positive values were still within the normal range.

The values of INL at lots of middle and high bits continued to increase to become more positive, while the INL at more low bits became more negative when TID reached 400 Gy(Si) during HDR irradiation. During LDR irradiation the values of INL at almost all bits showed negative direction increase and became significantly more negative when TID reached 400Gy(Si).

Figure 7 shows the evolution of the DNL and INL of the DAC. Through the analysis of the variation of the two parameters under different TID and during annealing time, the error parameters of a device degraded differently at LDR and at HDR under the same total dose level, and this is a "true" dose-rate effect that cannot be eliminated by long-term annealing.

Figure 7. DNL and INL versus (a) total dose and (b) room-temperature annealing time.

DownLoad: Full-Size Img PowerPoint

5. Discussion

5.1 Total ionizing effects

5.1.1 The failure mechanism of the AD8300

(1) The malfunction of the switches as one of the causes of DAC degradation (aberrance of the function curve) during LDR irradiationp

Figure 1(b) shows the input-output relationship for different TID under LDR at operating biases. The initial shape is a straight line but, as the total ionizing dose increases, the function looks like a staircase with fewer and fewer steps. This might be caused by both the decrease of the threshold voltage and the increase of the leakage current of the NMOSFET that separates the ground from the segment R/2R ladder network^[1]. See Fig. 8 for an equivalent circuit schematic of the analog section. The internal DAC is a 12-bit device with an output that swings from GND potential to 0.4 V generated from the internal band-gap voltage. It uses a laser-trimmed segmented R-2R ladder which is switched by N-channel MOSFETs. Because of the malfunction of the NMOSFETs, the ratios between the current of each branch inside the converter were not a power of 2. The values of output voltage could not be fitted to a straight line since the values would not be an integer multiple of the LSB unit. Therefore, the ratio between the output and the input code will be non-linear regularly and the size of the non-linearity must be estimated. But the non-linearity of the function curve is a staircase with erratic steps. This irregularity does not occur because of a malfunction of the switches, but is due to the degradation of other sections of the DAC.

Figure 8. Equivalent AD8300 schematic of the analog portion.

DownLoad: Full-Size Img PowerPoint

(2) The degradation of the reference and output amplifier causing the decrease of the full-scale voltage and gain of the DAC

As the total ionizing dose increases, in addition to the distortion of the functional curve mentioned above, the slope of the function curve and the full-scale voltage decreases which leads to an irregularity of the non-linearity of the function curve.

The internal DAC is a 12-bit device with an output that swings from GND potential to 0.4 V generated from the internal band-gap voltage. The output is buffered by a low power consumption precision amplifier. The rail-to-rail amplifier is configured with a gain of approximately five in order to set the 2.0475 V full-scale output. Degradation of either the reference or the rail-to-rail amplifier can result in a decrease of the slope of the function curve and the full-scale voltage.

5.1.2 The failure mechanism of the switches

Figure 9 shows the structure of an internal NMOS switch which consists of two inverters and two NMOSFETs. When the NMOS switch is exposed to ionizing radiation, a number of electron-hole pairs are produced in the SiO$_{2}$ of the NMOS. After initial electron-hole recombination, the electrons migrate out of the regions quickly, but the holes remain in SiO$_{2}$ due to their slower mobility. Some holes react with oxygen vacancies to form positive oxide-trapped charges, while others migrate to the Si-SiO$_{2}$ interface and react with dangling bonds to create the interface state. The radiation-induced oxide trapped charge, which is expressed as $\Delta N_{\rm ot}$, results in negative drift of the threshold voltage and an increase of leakage current. The radiation-induced interface state, which is expressed as $\Delta N_{\rm it}$, results in positive drift of the threshold voltage and an increase of leakage current^[7]. The total dose response of the NMOS switch results from radiation-induced oxide trapped charges and a subsequently induced interface state. During the initial periods of irradiation, radiation-induced oxide trapped charges are more than an interface state, that is $\Delta N_{\rm ot}$ $>$ $\Delta N_{\rm it}$, which results in the threshold voltage drifting negatively and an increase of leakage current. If the threshold voltage of $N_{2}$ becomes lower than 0, it will be always ON even though the gate voltage is 0 V.

Figure 9. Structure of an internal NMOS switch in a DAC.

DownLoad: Full-Size Img PowerPoint

Although both the oxide trapped charge and interface state contribute to the degradation of the devices, their performances are different during the post-irradiation annealing. The oxide-trapped charge can be eliminated by annealing at room temperature, but the interface state cannot be significantly removed unless annealing above 100 ℃ takes place^[9-11]. The dose rate effect of the NMOS switch results from the different annealing performances of the two kinds of radiation-induced charge.

5.2 Dose rate effects

The dose rate effects and radiation failure mode of this CBCMOS DAC were studied, and it is found that the degree of degradation and radiation failure were both different in the DAC under different dose rates.

The dose rate effect shown by the three parameters, $V_{\rm out1}$, DNL and INL, is an obvious "true" dose-rate effect which is different from the dose rate effect shown by the parameters of DAC devices fabricated by using traditional CMOS^{[1, 5, 7]}or bipolar processes^[6]. Through preliminary analysis, the author believes that the difference is a result of the mixed process which is based on the CB process combining the CMOS process.

The difference of radiation failure modes during HDR irradiation and LDR irradiation is shown as:

(1) The different variation tendency of the input-output relationship with the accumulation of TID.

(2) The different variation tendency of the value of linearity errors with the accumulation of TID.

Furthermore, the radiation failure modes of a DAC are likely to be different under initial irradiation and deep irradiation according to the variation tendency of the value of linearity errors. More research should be carried out to find out the causal factors for this difference.

6. Conclusion

According to above analysis, we can see that the radiation mode of the CBCMOS DAC is very complex because of the advanced manufacturing process and the complex device structure. The complexity manifests as the complicated variation tendency of both the functional curve and the static parameters. By comparative analysis of the variation of the functional curve and the static parameters under different dose rates and different biases, the results obtained are as follows:

(1) The malfunction of the switches is one of the causes of DAC degradation during LDR irradiation which is shown by the function curve becoming like a staircase with fewer and fewer steps.

(2) The degradation of the reference and output amplifier causes the decrease of the full-scale voltage and the gain of the DAC.

(3) Both TDE and the "true" dose rate effect in the AD8300 are observed, and both the degree of degradation and the degradation mode of the functional curve and static parameters are very different under different dose rate. Both kinds of dose rate effect shown by the testing parameters result from the process characteristics of the CBCMOS process, and the kinds of dose rate effect shown by different parameters are related to the fabrication processes of the relevant modules which are sensitive to TID.

(4) The degree of degradation of the functional curve and static parameters is very different under different biases, and worsen under an operating bias.

References(11)

References

[1]	Franco F J, Lozano J, Agapito J A. Radiation effects on CMOS R/2R ladder digital-to-analog converters. Noordwijk, Netherlands:RADECS, 2003:571
[2]	Aghara S, Fink R J, Charlton W S, et al. Degradation of commercially available DAC ICs in a mixed-radiation environment. IEEE Rad Effects Data Workshop Rec, Piscataway, NJ, 2002:34 http://ieeexplore.ieee.org/document/1281318/?reload=true&arnumber=1281318
[3]	Ampe J, Thai V, Buchner S, et al. COTS ADC & DAC selection and qualification for the GLAST mission. IEEE Rad Effects Data Workshop, Piscataway, NJ, 2005:79
[4]	Franco F J, Zong Y, De Agapito J A, et al. Radiation tolerant D/A converters for the LHC cryogenic system. Nuclear Instruments and Methods in Physics Research Section A:Accelerators, Spectrometers, Detectors and Associated Equipment, 2005, 553(3):604 http://cat.inist.fr/?aModele=afficheN&cpsidt=17255658
[5]	Prater J S, Brown B, Trinh T. Analysis of low dose rate effects on parasitic bipolar structures in CMOS processes for mixed-signal integrated circuits. IEEE Trans Nucl Sci, 2011, 58(3):1023 doi: 10.1109/TNS.2011.2116041
[6]	Wang Y, Lu W, Ren D, et al. Total dose effect of 10-bit bipolar D/A converter under different ⁶⁰Co γ dose rates. Atomic Energy Science and Technology, 2009, 43(10):951
[7]	Liu Y, Yang S, Lin D, et al. Synergistic effect of neutron and gamma irradiation on 10-bit CMOS digital-to-analog converter. High Power Laser and Particle Beams, 2010, 22(9):2186 doi: 10.3788/HPLPB
[8]	Winokur P S, Schwank J R, Mcwhrter P J, et al. Cr relating the radiation response of MOS capacitors and transistors. IEEE Trans Nucl Sci, 1984, 31(6):1453 doi: 10.1109/TNS.1984.4333529
[9]	Witczak S C, Schrimpf R D, Galloway K F, et al. Gain degradation of lateral and substrate pnp bipolar junction transistors. IEEE Trans Nucl Sci, 1996, 43(6):3151 doi: 10.1109/23.556919
[10]	Nowlin R N, Fleetwood D M, Schrimpf R D, et al. Hardness-assurance and testing issues for bipolar/BiCMOS devices. IEEE Trans Nucl Sci, 1993, 40(6):1686 doi: 10.1109/23.273492
[11]	Schrimpf R D, Graves R J, Schmidt D M, et al. Hardness-assurance issues for lateral PNP bipolar junction transistors. IEEE Trans Nucl Sci, 1995, 42(6):1641 doi: 10.1109/23.488761