Test results of single-event effects conducted by the Jet Propulsion Laboratory

This paper reports recent single-event effects results for a variety of microelectronic devices that include ADC, DAC, DDS, MOSFET driver, analog switch, oscillator, and zero delay buffer. The data was collected to evaluate these devices for possible use in NASA spacecraft. Six of ten devices under study were sensitive to SEL, and the latchup was destructive.

I. INTRODUCTION lectronic devices used in satellites and other spacecraft are exposed to cosmic radiation. To insure reliability of these devices, the effects of radiation should be carefully studied. The primary method used to evaluate the radiation reliability of an electronic device is to measure its single event effect (SEE) cross section as a function of ionizing power of the ion beam. Many SEE results for space applications have been published previously [1][2][3][4].
The studies discussed in this paper were undertaken to establish the sensitivity, of the electronic devices, to single event latchup (SEL) and single event upset (SEU). SEE measurements were performed on ten different types of CMOS devices including ADC's, DACs, analog switches, MOSFET drivers, digital synthesizers, delay buffers and crystal oscillators.

A. Test Facilities
Heavy ion SEU measurements were performed at two facilities. The SEU Test Facility located at the Brookhaven National Laboratory (BNL) and the Radiation Effects Facility located at the Cyclotron Institute Texas A&M University (TAM). The BNL facility uses a twin Tandem Van De Graaff accelerator while the TAM facility uses an 88" cyclotron. Both facilities provide a variety of ion beams over a range of Manuscript Table I for BNL and Table II  At both facilities, DUTs were tested at room temperature as well as at an elevated temperature. The elevated temperature depended on the DUT specification [5]. To determine each cross section point, either a minimum of fifty latchup events were accumulated or a beam fluence of 10 7 ions/cm 2 was used.
The SEU measurements included the saturation cross sections and the linear energy transfer (LET th ) threshold. The LET th is the minimum LET value necessary to cause a SEU at a fluence of 1x10 7 ions/cm 2 .

III. TEST RESULTS AND DISCUSSION A) LTC1604
The Linear Technology LTC1604 is a 333ksps, 16-bit Analog to Digital Converter, (ADC), that draws 220mW from ±5V supplies.
Two of the available supplies on the HP6629 were used to power the DUT. One supply was used for V digital , and one supply for V analog . Both supplies were set to +5.25V. A third supply was used to power a support circuit, Demonstration Circuit 718 (DC718), provided by Linear Technology. The DC718 is a computer-hosted digital data acquisition system that when used with the evaluation software, QuickEval-II, provided the functionality feedback used during the test.
The measurements were performed at the TAM facility. The LTC1604 was tested at room temperature and at 85ºC. SELs were observed at a LET of 70.0 MeV-cm 2 /mg but no latchups were observed at a LET of 55.0 MeV-cm 2 /mg. The latchup LET threshold is between 55 and 70 MeV-cm 2 /mg. In figure 1, we compare the result of room temperature measurements with that of the heated measurements. As one might expect, the latchup cross section is higher for the heated device [5].
Previously, SEL measurements were performed on the LTC1604 at BNL facility [6]. In these measurements one device (different date code) was tested. The I 127 ion beam at 3 different incident angles was used to perform the measurements. Incident angles of 0, 37 and 60 degrees resulted in an effective LET of 59.5, 74.5 and 119.1 MeVcm 2 /mg which corresponds to the range of 29.5, 23.6 and 14.75 µm. No SEL event was detected for LTC1604 up to a LET 119.1 MeV-cm 2 /mg. The discrepancy between the BNL and TAM data is caused by the fact that LTC1604 is a commercial part and there might be variation in the radiation characterization from different date codes. Furthermore, the range of the ion used in the BNL measurements is not enough to cause latchup in a CMOS device. In general, SEL involves deeper regions in the CMOS device (greater than 40 µm) as contrasted with Single event Upset (SEU). The ion ranges used in the BNL measurements were below 29.5 µm and at highest LET (119.1 MeV-cm 2 /mg) it was 14.75 µm.
Our latchup measurements indicate that LTC1604 is not highly sensitive to latchup. The cross section is relatively small, and rises to about 10 -6 at high LETs. The LTC1604 was tested for destructive latchup by turning off latchup protection. When the device went into a latchup state, the supply current increased to 600mA. The lack of an output signal from the device indicated that the device was not functioning in this high current state. To determine if this condition was recoverable, the beam was turned off and the device was power cycled. The device did not recover, after power cycling, indicating that device had been destructively failed.

B) LTC1609
The Linear Technology LTC1609 is a 200ksps, serial 16bit ADC that draws 65mW from ±5V supplies.
The LTC1609 serial ADC was tested for SEL at both facilities. For the test at the BNL facility, we used an inhouse fabricated PC board. This board provides necessary voltages, clock, input signal to the DUT and monitors output signal. Two devices were tested. Both devices were tested at room temperature and one of the two devices was also tested at an elevated temperature of 85ºC. At room temperature, the latchup LET threshold is between 8.0 and 11.7 MeV-cm 2 /mg. For the heated device, the latchup LET threshold is between 5.3 and 8.0 MeV-cm 2 /mg. Figure 2 shows the results obtained at the BNL facility for the two devices that were tested at room temperature. This data indicates that LTC1609 is sensitive to latchup, and has an LET threshold of about 10 MeV-cm 2 /mg. The cross section is relatively small, at low LETs, and gradually increases to about 5x10 -4 at high LET (saturation cross section).
Also, the LTC1609 was tested for latchup at the TAM facility using the manufacturer's evaluation board. For the test at the TAM facility, only one of the four supplies available on the HP6629 was needed to provide the 5.25 V required by the evaluation board. Another of the four supplies was used to power a support circuit, a DC718, provided by Linear Technology.
In figure 3, we compare room temperature measurements and heated measurement taken at both facilities. The square symbols represent the room temperature measurements taken at the BNL and the TAM facilities. The triangle symbols represent the heated measurements taken at the BNL and the TAM facilities. Although, the TAM facility ions have longer range, there is very little difference between the two sets of data. As one might expect, the latchup cross section is higher for the heated measurements. There was a factor of 1.5 -2 increase in latchup cross section for the heated device which is much less than the increase between heated and room temperature for LTC1604. We tested LTC1609 for destructive latchup. During the irradiation, with latchup current protection turned off, the supply current increased to 1.2A and the lack output signal from the device indicated that the device was not functioning. To determine if this condition was recoverable, the beam was turned off and the device was power cycled. The device did not recover and therefore had been destructively damaged.

C) LTC1595
The Linear Technology LTC1595 is a 16-bit serial Digital to Analog Converter (DAC). A printed circuit board was fabricated to test the device. This board provides necessary voltages and a serial input to the DUT. Two devices were tested at room temperature. One of the two devices was also tested while heated to 85ºC. At room temperature, the latchup LET threshold is between 10.0 and 11.7 MeV-cm 2 /mg. For the heated device the latchup LET threshold is between 8.0 and 10 MeV-cm 2 /mg.
In figure 4, we show our results for two devices that were tested at room temperature. This data indicates that LTC1595 is sensitive to latchup, and has an LET threshold of about 10 MeV-cm 2 /mg. The cross section is relatively small at low LETs, gradually increasing to about 10 -4 at high LET (saturation cross section).
In figure 5, we compare the average result of two room temperature measurements with the heated measurements. As one might expect, the latchup cross section is higher for the heated measurements. There was a factor of 1.5 to 2 increase in latchup cross section for the heated device.
We tested LTC1595 for destructive latchup. When the device was irradiated without latchup protection the device destructively failed.
One of the supplies available on the HP6629 was needed to provide the 15V required by the evaluation board. The second of the four supplies was used to power DC718 support circuit, provided by Linear Technology. Latchups were observed at a LET of 8.5 MeV-cm 2 /mg with the device at room temperature. The latchup saturated cross section was 5x10 -5 cm 2 . We tested the LTC1864 for destructive latchup. The device destructively failed when tested without latchup protection.

E) AD9858
The Analog Devices AD9858 is a direct digital synthesizer (DDS).
Evaluation boards and related software supplied by the manufacturer, Analog Devices, was used to interrogate the registers and monitor device functionality. The AD9858 was tested at the TAM facility for SEU and SEL.

1-SEU Test Results
The device exhibited SEU in the registers that change the operating conditions of the device. The effect of the SEU in the working registers was a change in the output waveform of the device. The LET threshold of the device was 7.5 MeVcm 2 /mg while the saturation cross section was 8.8x10 -5 cm 2 . Two samples were tested at the TAM facility and showed statistically identical results. In figure 6, we show the SEU results for the two samples.

2-SEL Test Results
Two devices were tested for latchup at the TAM facility. These devices experienced SEL at a LET of 10 MeVcm 2 /mg with a saturation cross-section of 4x10 -5 cm 2 . The result of the SEL measurements for two samples is shown in figure 7. The latchup was seen to be non-destructive in all cases and conditions. The device was left in the latchup state for 24 hours with no obvious degradation in operation once removed from latchup state.
We note that recent SEE measurements of the older DDS devices, AD9850 and AD9851 [7], provided similar results in terms of the SEU and SEL cross sections compared to our measurements for AD9858.

F) TC4423
The Microchip TC4423 is a 3A, dual output buffers/MOSFET driver.
Printed circuit boards were fabricated to mount the test devices as well as the resistors, and bypass capacitors used to configure the test circuit.
Two devices were tested at room temperature and one device was tested at an elevated temperature of 125ºC. The TC4423 was tested at both TAM and BNL facilities. No latchups were observed for either the heated or the nonheated devices up to a LET of 86.3 MeV-cm 2 /mg.
The four analog switches were daisy-chained together so that an input signal could be fed into the first switch and the signal sequentially fed into the other three switches. Printed circuit boards were fabricated to test the devices. This board provides necessary voltages and input signal buffer to the DUT. This board also, provides visual feedback that allows the DUT to be monitored during testing. The DG412 and DG413 were tested at both TAM and BNL facilities. Two devices were tested at room temperature and one of the two devices was also tested at an elevated temperature of 125ºC. No latchups were observed for either the heated or the nonheated devices up to a LET of 86.3 MeV-cm 2 /mg.

H) CY23FS08 Failsafe Buffer
The Cypress CY23FS08 is a Zero Delay Failsafe Buffer. A printed circuit board was fabricated to test the device. This board provides necessary voltages, clocks to the DUT and monitors its functionality. Two devices were tested at the TAM facility. Both devices were tested at room temperature and one was also tested at an elevated temperature of 85ºC.
Latchups were observed at a LET of 70 MeV-cm 2 /mg at an elevated temperature of 85ºC. Latchups were observed at a LET of 86.3 MeV-cm 2 /mg at room temperature and an elevated temperature of 85ºC.
In figure 8, we compare the results from room temperature measurements with that of heated measurements. The latchup LET threshold for room temperature is between 75 to 86 MeV-cm 2 /mg. The latchup LET threshold for the heated measurements is between 55 to 70 MeV-cm 2 /mg.
The device was tested for destructive latchup by turning off the latchup protection. During irradiation, the device current went up to 220mA. After stopping the irradiation and power cycling, the current went down to 7mA and the device was functioning. This device did not experience destructive latchup.
This device shows extreme sensitivity to SEU, even at a low LET such as 8.5 MeV-cm 2 /mg.

I) VPC1 Crystal Oscillator
The VPC1 programmable crystal-controlled oscillators operate at 160 MHz while providing a jitter as low as 7 ps.
The crystal, which was placed on top of the IC, was moved to the side to allow the ions to strike the IC unimpeded (without having to pass thought the crystal). The VPC1 was tested at the TAM facility. Two devices were tested at room temperature and one was also tested at an elevated temperature of 85° C. No latchups were observed for either the heated or the non-heated devices. The V+ and V-supply voltage was increased to 3.6 V and latchup measurements were repeated. Again, no latchups were observed for either the heated or the non-heated devices at a LET of 86.3 MeVcm 2 /mg.

IV. SUMMARY
We have presented SEE data for a variety of commercial CMOS devices. The latchup measurements were performed at room temperature and an elevated temperature. The latchup cross section is higher for the heated measurements. Latchups were destructive for the following parts; LTC1604, LTC1609, LTC1595 and LTC1864, except for the AD9858 and CY23FS08. The TC4423, DG412, DG413 and VPC1 were latchup immune up to a LET of 86.3 MeV-cm 2 /mg. We summarized our SEE test results in table III.