Performance of Precision FH

1
Performance of Precision FH

• Joe T. Evans, Jr.
• Radiant Technologies, Inc.
• June 19, 2005

2
High Speed Testing of Ferroelectric Capacitors

• Ferroelectric capacitors are difficult to test at
  high speed because of their non-linear
  polarization response to the stimulus voltage
• Three difficulties arise
• Reflections
• Overshoot during switching
• Artificial increases in the coercive voltage.

3
Symptoms of Too Fast

• When the measured signal is too fast for the
  measurement circuitry, a reflection occurs after
  the peak voltage at Vmax.

• The plot to the left shows reflections on a 100pF
  linear capacitor.
• The test period was 3.2µs and the reflection
  oscillated for 0.5µs.
• The measurement circuitry of the FH does reflect
  within the capacitor areas specified for the
  system.

4
Symptoms of Too Square

• When the measured capacitor is very square, the
  current generated by the capacitor increases very
  quickly as it begins to switch. The output
  voltage of the measurement circuitry can
  overshoot its proper level trying to keep up
  with the change.

• Overshoot decreases with longer test periods.
• The test frequency at which overshoot begins for
  a particular capacitor area increases as the
  capacitor becomes less square.
• Overshoot determines the test frequency limit on
  the Precision FH.

6µs test on 50µx 50µ PZT capacitor.
5
Overshoot and Pmax

• Overshoot does not appear to affect the final
  value of Pmax measured on the sample.

• To the right is plotted the capacitor from the
  previous page at 6µs compared to the same
  capacitor at 3µs, 10µs, and 50µs.

6
Coercive Voltage and Amplification

• The effect of the amplifier speed on hysteresis
  can be seen on the pages 16 and 17 showing the
  comparison of 50µ x 50µ and 10µ x 10µ PZT
  capacitors measured on the Precision Premier and
  the Precision FH.
• The absolute value of the Vc of the capacitor as
  measured on the Precision FH is slightly less
  than that measured on the Precision Premier but
  both testers arrive at the same Pmax at Vmax.
• The Precision FH integrator has a 50MHz amplifier
  with a 350V/µs slew rate while the integrator of
  the Precision Premier has a 4MHz amplifier with a
  350V/µs slew rate.
• The trade off is that the Precision Premier is
  very stable for long measurements so it has a
  wider frequency range than the FH. But, the FH
  is much faster!

7
Coercive Voltage and Amplification

• All amplifiers slow down the signal they amplify.
  
• On hysteretic capacitors, the amplifier delay
  causes the measured coercive voltage to move out
  from its true value.
• The faster the amplifier, the more accurate the
  apparent coercive voltage.
• But, the faster the amplifier, the less stable it
  is for longer measurements.
• The design of the Precision FH is focused on high
  speed hysteresis loops with the speed and range
  of PUND and Leakage being secondary.

8
Stability and Speed

• Pulse and leakage measurements are more sensitive
  to amplifier stability than are hysteresis
  measurements.
• Hysteresis loops are affected as well but can go
  longer than pulse or leakage measurements while
  achieving the same accuracy.
• This effect is true on all Precision testers.

9
Description of Samples

• Capacitor Structure (from bottom to top)
• 5000Å silicon dioxide on lt100gt silicon
• 400Å titanium dioxide
• 1500Å polycrystalline lt111gt and lt200gt platinum
• 2550Å 20/80 SOG PZT deposited in seven layers
• 1000Å polycrystalline platinum
• 400Å titanium dioxide
• 2500Å silicon dioxide
• Metal interconnect
• 200Å chromium
• 2000Å gold
• A single wafer has capacitors ranging from 16mm2
  to 25µ2.

10
Recovery

• As has been reported in the past, a DC bias
  applied to a ferroelectric capacitor can reverse
  some imprint effects.
• The Radiant capacitors are capable of
  withstanding long periods at 9V across their
  2550Å of thickness.
• Imprint and process degradation effects may be
  100 reversed using the recovery procedure
  recommended below.
• The capacitors tested in this experiment were
  recovered prior to testing, making them
  extremely square.
• Recommended Recovery Procedure
• 9V 1Hz square wave for 100 seconds at room
  temperature.

11
Capacitor Properties as Tested

• Capacitors with very square switching
  characteristics are difficult to test at high
  frequency because the sudden increase in current
  flow from the capacitor when it starts to switch
  may overwhelm the current cancellation capacity
  of the amplifiers on the RETURN input of the
  tester.

• The squareness of a ferroelectric capacitor can
  be determined by executing the Normalized
  Capacitance vs Voltage plot filter on a measured
  hysteresis loop of the sample.
• The capacitors tested for this evaluation
  exceeded 100µF/cm2 during the switch. Typical
  research capacitors usually exhibit about
  30µF/cm2.

12
Capacitor Properties as Tested

• Radiant focused on qualifying the Precision FH
  for very square capacitors.
• Ferroelectric capacitors that generate the same
  total charge (Pmax at Vmax) but are not as square
  will more easily be measured by the Precision FH
  at higher speeds.

13
Description of Evaluation

• The Precision FH is designed to execute high
  speed hysteresis and PUND measurements on
  ferroelectric capacitors.
• To minimize the effect of cable loading and
  reflections, the dimensions of the FH were set to
  allow it to be placed directly on the probe
  station next to the probes.
• To evaluate the Precision FH, Radiant executed
  tests of extremely square PZT capacitors directly
  on wafer on a standard probe station.
• Capacitors with areas within the performance
  range of the FH were tested for hysteresis from
  1µs to 1ms in period.
• The Precision FH and the Precision Premier
  overlap in their performance ranges at 1KHz
  hysteresis. The same capacitors were measured at
  1KHz on the Premier for comparison to the FH.

14
The FH on the Probe Station

• The Precision FH enclosure so it can be attached
  to probe stations with magnets.

15
The FH on the Probe Station

• The coax cable connections from the probes to the
  FH can be very short.

16
Compare Precision FH to Premier

• 50µ x 50µ PZT capacitor measured on both the FH
  and the Premier.

17
Compare Precision FH to Premier

• 10µ x 10µ PZT capacitor measured on both the FH
  and the Premier. Averaging was used to minimize
  noise from the Premier. As well, the Premier has
  about 1.5pF of parasitic capacitance which was
  measured and subtracted from its hysteresis loop.
  The parasitic capacitance on the FH is 33fF and
  did not affect this measurement.

18
2500µ2 Capacitor 10µs to 1ms

• The 2500µ2 capacitor was tested on the x1
  amplification stage of the Precision FH. It has
  a range for square capacitors (gt80µF/cm2 dynamic
  capacitance) from 10µs to 20ms. Capacitors with
  dynamic capacitance around 80µF/cm2 may run
  faster without overshoot.

19
1000µ2 Capacitor 6µs to 1ms

• The 1000µ2 capacitor was tested on the x1
  amplification stage of the Precision FH.

20
400µ2 Capacitor 4µs to 1ms

• The 400µ2 capacitor was tested on the x1
  amplification stage of the Precision FH. Note
  the reduced overshoot at the higher speed. The
  smaller area translates to a lower overall charge
  transfer during the test, reducing the overshoot.

21
100µ2 Capacitor 1µs to 1ms

• The 100µ2 capacitor was tested on the x20
  amplification stage of the Precision FH. This
  area is the optimal area for measurements to 1µ2.
  

22
25µ2 Capacitor 1µs to 1ms

• The 25µ2 capacitor was also tested on the x20
  amplification stage of the Precision FH. The
  25µ2 is small enough that it will go as fast as a
  480ns period without overshoot but with some
  extension of the coercive voltage.

23
Summary of Precision FH Performance
Parameter Minimum Maximum 1. Voltage
Range 10V 2. Maximum Hysteresis
Frequency a) 25µ2 500Hz 2MHz
b) 100µ2 500Hz 1MHz c) 400µ2 50Hz 200KHz
d) 1000µ2 50Hz 200KHz e) 2500µ2 50Hz 100KH
z 3. PUND Pulse Widths a) x1 amplification
level 10µs 2ms b) x20 amplification level
200ns 200µs 4. Leakage Measurement
period a) x1 amplification level 200µs 2ms
b) x20 amplification level 200µs 200µs
5. Leakage Measurement Range 50nA 5µA 6. Area
Range (PZT) 5µx5µ 50µx50µ 7. Acquisition Rate per
Point 10ns 8. Clock Rate 100MHz 9. Output
Current 50mA 9. Rise time to 5V 200ns 10. Small
Signal rise time 20ns
24
Summary of Precision FH Performance

• Power Self-adjusting from 100V to 220V
• Number of ADC bits 14
• Number of DAC bits 14
• Switch over point from the 100MHz clock to the
  200KHz clock for hysteresis 100µs
• Switch over point from the 100MHz clock to the
  200KHz clock for PUND 40µs
• Number of points in hysteresis
• 500ns 50
• 1µs 100
• 100µs 5000
• 1ms 200
• 20ms 500