The procedure for BTS is:

1. Execute a high frequency CV curve at less than 50º C. (Extract the flatband voltage.)

2. Bias the device to positive voltage, then temperature stress at 250º-275º C (for 2-10 min. Bias should be >1MV/cm of dielectric thickness.)

3. While maintaining the positive bias, cool the device back to the low temperature.

4. Repeat step 1, and extract the new flatband voltage.

5. Bias the device to negative voltage, then temperature stress at 250º-275º C (for 2-10 min.)

6. While maintaining the negative bias, cool the device back to the low temperature.

7. Repeat step 1, and extract the new flatband voltage.

8. The shift in flatband voltage is directly proportional to the mobile ionic concentration.

Some of the primary problems with the BTS method are the time required and errors due to probe needles losing contact during thermal cycling. Another problem is erroneous capacitance measurements due to cabling effects and series resistance. A capacitance meter that supports transmission line and series resistance correction should be used. Inaccuracies in the bias source will also affect the flatband voltage extraction, the CV meter should include an accurate voltage source and meter.

The procedure for TVS is:

1. Heat the device to 250º-275º C.

2. Apply a positive bias of greater than 1MV/cm and soak the device for 2-4 minutes.

3. Perform a quasi-static CV sweep from positive to negative bias.

4. Several methods of calculating the mobile ionic concentration from the quasi-static CV data are available. The most common is to compare the measured CV curve to the oxide capacitance (Cox) and integrate the result over the applied gate bias (2).

The primary problem with the TVS method is that the algorithm compares the measured CV data to oxide capacitance (Cox). Since mobile ions typically move while the device is in the depletion region, comparing the measured data to Cox in depletion can cause large errors in the calculated ionic concentration, particularly at low contamination levels. This problem has been solved by using the Simultaneous Triangular Voltage Sweep technique.

Another source of error in both TVS and STVS measurements is noise from the electrical heaters in the thermal chuck. Chuck heaters that are powered by DC voltages are superior to AC heaters. The chuck may be of coaxial construction if care is taken in maintaining high insulation resistance and low shunt capacitance. Triaxial chuck designs are more costly, but generally give better noise characteristics. A feedback charge quasi-static capacitance meter, such as the Keithley Model 595, will give better noise performance than a linear ramp, differential picoammeter capacitance meter.

If the device under test exhibits leakage current, this can lead to errors in the calculations for ionic concentration. The software controlling the test should include capability for detecting and correcting for leakage currents. If leakage is too high, the TVS technique cannot be used and the BTS technique may be needed.

• No temperature cycling, resulting in faster tests and no lost probe contact.
• More sensitive to ionic contamination - 1e9 ions/cm^2 for STVS vs. 1e10 ions/cm^2 for BTS
• More accurate contamination calculation since not compared to Cox in depletion
• Can differentiate between species of mobile ion
• Since both high frequency and quasi-static C are available, the system may be used for BTS, TVS or STVS.

The procedure for STVS is:

1. Heat the device to 250º-275º C.

2. Apply a positive bias of greater than 1MV/cm and soak the device for 2-4 minutes.

3. Perform a simultaneous CV sweep from positive to negative bias.

4. Integrate between the high frequency and quasi-static curves versus applied gate bias.

As device geometries continue to shrink, lower levels of ionic contamination in dielectrics are required. Some fabrication facilities have begun monitoring for mobile ions at several points in the process, such as at oxidation furnaces, after metal deposition and after interlayer dielectrics, using a combination of BTS, TVS and STVS techniques where appropriate. A Keithley System 83 CV measuring system operating under Metrics-ICS software provides all the tools necessary for CV monitoring of mobile ions in insulating layers.

References:

1. ASTM F1153-92 - Test Method for Characterization of MOS Structures by C-V Measurement.

2. Kuhn, M., and Silversmith, D.J., "Ionic Contamination and Transport of Mobile Ions in MOS Structures", J. Electrochemical Society, Vol 118 (June 1971), p.996

3. Stauffer, L., Wiley, T., "Mobile Ion Monitoring By Simultaneous Triangular Voltage Sweep", Solid State Technology - Metrology Supplement, Vol 38, Num.8, (August 1995)