What ruins an op amp?

3456

What ruins an op amp? [Copy link]

By James Vinson, PhD, Intersil Corporation

Identifying the source of electrical overstress (EOS) events that damage semiconductor devices is difficult, and even more difficult when EOS events occur with no regularity. To illustrate how EOS events can be traced back to their source, here is an example of how we helped a customer identify the cause of failure in two operational amplifiers (op amps).

Initial inspection
The first op amp, OP1, suffered a large EOS event that damaged many of its circuit components. The second op amp, OP2, had an EOS event that damaged only a thin-film resistor. Both op amps had one or more pins that failed the continuity test—the first sign of device failure. In addition, several other pins on OP1 showed functional degradation. Neither OP1 nor OP2 passed functional testing. After running electrical tests, we opened each op amp and inspected the failed device. Visual comparison of the failed device with the working device helped locate the failure.
The OP1 op amp showed damage in multiple locations (Figure 1), with most of the damage associated with the device's output, negative input, and positive supply pins. The blown metal traces at the output of the op amp proved that the op amp received a large amount of energy when the EOS event occurred.
In contrast, the OP2 device showed no signs of the usual EOS—abnormal metal traces and burn marks. Only one resistor was suspicious, showing a slightly different color. Based on this, it was concluded that the EOS event did not cause the failure, and that it was probably caused by oxidation or corrosion of the NiCr resistor. However, other resistors in the same area did not show similar signs of color change, and it is unlikely that a wafer manufacturing problem affected only one resistor. In addition, no other corrosion was observed, and no passivation oxide defects that could have introduced corrosive chemicals to the resistor were found. Analysis revealed that it was an open resistor connected to the negative input of OP2 (Figure 3) that caused the device to malfunction. When this resistor was disconnected, it cut off the feedback path and caused the op amp output to swing high and stay high regardless of the applied input signal. Probing the damaged resistor showed normal traces at the input, indicating that the EOS event did not damage other circuits in the op amp input path.

Figure 1 shows damage in many places in op amp OP1, and it is concluded that a high energy event caused the failure of the op amp

. Figure 2 shows that the output line in the op amp is blown, indicating that the damage is caused by the EOS event. At this time, the fuse is disconnected .
Tracing the cause to the source
After knowing the damage that caused the two failures, we still have to trace the cause to the source. The first step is to identify the cause of the EOS event, which involves obtaining information from the person who reported the failure. Because we need to know which circuits and circuit board configurations were in use when the failure occurred, what the test conditions were when the component was last known to be working properly, and what events occurred after the last test or use when the component was functioning normally. The circuit schematics for each op amp show the connections between the op amp and all other components and "outside" signals.
Looking at the damage patterns observed on each op amp, based on these patterns and the understanding of the circuit components surrounding each op amp, information about the source and intensity of the EOS event is obtained, such as a small external signal passing through a large impedance may be the energy source of the EOS event. Impedance reduces the amount of current and has a certain protection function. Connecting the power and other device pins directly to the op amp creates low impedance, but it makes it easier to conduct EOS energy to the semiconductor device.
The circuit containing OP1 uses the device as a unity gain non-inverting amplifier, with its output connected to a cable conductor on the circuit board. In this configuration, the output of the op amp is connected directly to its negative input. The input signal to the amplifier is connected directly to the positive input of OP1 from the circuit board power supply. Based
on our observations and the application of the op amp, it is believed that the damage occurred because a positive voltage was applied to the output pin of the op amp. The partial schematic of OP1 (Figure 4) shows the path of current from the output pin of the op amp through Q70 and Q75 to the V+ line. Q70 is a large output transistor that can handle the power of the EOS event, but Q75 cannot, as reflected by the aluminum "short" we found at the base-emitter junction of Q75. This small transistor cannot dissipate the large amount of energy of the EOS event without shorting. After the current reaches a critical level, the metal output trace leading to one end of the bond pad burns out, and a large section of the trace is burned out as shown in Figure 2. Burning such a large section of metal requires a large current pulse (1-2A) in a short time.
OP1 also suffered other damage. When the metal output line is disconnected, the current drops to zero very quickly and the voltage increases very quickly. Since the output of the op amp is directly connected to its negative input, damage caused by the EOS voltage pulse was observed around the op amp input connection (Figure 1). In my opinion, it is the parasitic inductance of the EOS signal source that caused the rapid increase in output voltage.
OP2 appeared to be less damaged than OP1 - there was only one open NiCr resistor, which made it difficult to determine the cause of the device failure. Electrical testing showed that the other components connected to the NiCr resistor were working properly. The resistor was connected between the pad and the input stage. There is a lowest breakdown path for the positive voltage from the pad to the negative supply. If the charge took a different path, other circuit damage should have occurred. Therefore, the EOS energy pulse must have entered the negative input pin.
The lowest resistance breakdown path exists between the negative input and the negative supply line, so the EOS current flows through this line. Since we saw no damage to metal lines or other components other than the resistor, we concluded that the EOS event produced only a small amount of energy. Also, a slow pulse should have damaged the center of the NiCr resistor rather than the entire resistor area. Therefore, we determined that the EOS event occurred very quickly, with a fast rise time.

Figure 3 A different type of op amp (OP2) was slightly damaged, destroying only one resistor

Figure 4 A partial schematic of the OP1 op amp shows the path of current during the EOS event
Our
next step in our experimental investigation was to experimentally reproduce the failure. We made certain assumptions about the type of EOS event that caused the damage. For example, we assumed that the test leads provided enough inductance (~2mH) to cause the voltage spike, so that no additional inductance was needed in the test circuit. We also made some guesses about the voltage and current levels, the energy provided to the circuit, and the duration of the EOS event.
For the OP1 device, we used a Tektronix curve tracer to provide 25V pulses with durations ranging from 10 to 50ms. 3 feet of test leads connected the curve tracer to the DUT. Under these conditions, the test part did not fail as we had observed in the OP1 device. A second attempt with the voltage set to 350V and the peak current limited to 2.5A using a series resistor produced damage similar to that seen in OP1. Not only did the pulse damage the same circuit area as OP1, but we also observed more severe damage to the test part. Lowering the voltage level or increasing the series resistor may have reduced the damage, but we felt we had found the cause of the damage, so we did not conduct further experiments.
Our test results led the user to the possible cause of the failure - a failure in the ungrounded test cable on the test bench. An ungrounded cable can charge to extremely high voltages, and when connected to the board, it will discharge into the board circuitry, damaging the op amp and other components. Adding
More Energy
The source of the failure in OP2 was more difficult to pinpoint. First, we applied a voltage to the negative input of the test device and increased this voltage until the op amp's input resistor opened. The +17V signal on the op amp's negative input caused the resistor to burn out, but this did not seem to be the same resistor that failed in OP2.
Rather than showing that the entire resistor had completely failed, the resistor in the test device showed a line across the resistor. We decided to apply more energy to completely burn the resistor out, and to apply the energy quickly to prevent thermal damage to the resistor.
The pulse provided by the curve tracer was too slow to heat the entire resistor quickly, so we tried using a transmission line pulse (TLP) tester. This type of tester charges a length of coaxial cable to a preset voltage and then discharges the cable into the DUT. A TLP tester produces a rectangular current pulse with a rise time of less than 2 ns and a variable pulse width. When we charged the cable to 250V, it produced a peak current of 0.5A, which burned the op amp resistor in 55ns. The results of this pulse test matched the damage seen in OP2.
This result does not mean that the power from the cable assembly caused the damage to the part, but it does indicate that a fast pulse with a fast rise time and a current of about 0.5A can cause similar damage. Further work by the user identified a possible cause as an oscilloscope next to the test board.
Users discovered that the oscilloscope radiation produced a high-energy electric field that induced charges on nearby components. When technicians touched the circuit board with the test instrument, a discharge occurred. Proper shielding was used to remove the charge, eliminating the failure of the operational amplifier on the test board.
Conclusion
In some cases, the failure analyst does not need to fully understand the failed component to understand how it was caused. The supplier knows the component best, while the user knows the actual application of the component best. Therefore, both parties need to share information without reservation to solve the headache-inducing EOS-related problems.
References
1 James Vinson. "Finding the EOS Weird Problem". Proceedings of the 27th International Test and Failure Analysis Conference. Electronic Device Failure Analysis Association, ASM International Organization, 2001.
About the author: James Vinson is a senior principal engineer who works in the field of reliability analysis at Intersil. He has been engaged in reliability certification and failure analysis for more than 18 years, and his expertise is concentrated on electrical overstress failure mechanisms.

zcgzanne

Thank you for collecting such information for sharing. Many of the technical points are very helpful for product development.