Frequently Asked Questions
What is the Device Characterization Lab?
Who manages the laboratory?
How do I get access to the Device Characterization Lab?
How do I get qualified for a probe station?
What are the user's responsibilities?
How do I sign up for a probe station?
What programs are available to control the machines?
How do I check out a wafer?
What can I do to annoy fellow lab users or get kicked out of the 407 lab?
What is the HP4145, HP4155 or HP4156?
What does the HP428X LCR Meter do?
How do I ensure low probe contact resistance?
How do I measure the leakage current on a probe?
How and why should I change the teflon screws on the manipulator?
What kind of probe tips should I order?
How do I tighten a manipulator?
How do I change probe tips?
What's special about Cascade?
How do I make a quasi-static MOS CV measurement?
How do I make a HF MOS CV measurement?
What is the difference between a serial or parallel connection model when doing an impedance measurement?
How can I do measurements on stress-sensitive conducting films?
How do I make a hot carrier measurement?
1. What is the Device Characterization Lab?
The device characterization lab contains several probe stations and various measurement equipment used for making electrical measurements on solid state devices.
2. Who manages the laboratory?
The Device Characterization Lab is run by the Device Group under Professors Chenming Hu, Tsu-Jae King and Jeffrey Bokor. Specific day-to-day running of the laboratory is handled by graduate student managers and probe station superusers. The manager makes decisions concerning the lab as a whole. Each probe station super user has responsibility for a single probestation. They qualify people to use their probe stations and discipline users who break rules or mishandle the station.
3. How do I get access to the Device Characterization Lab?
Each new user must be trained by an existing qualified user. After several training sessions, the new user must ask for approval from one of the Device Group professors and obtain cardkey access into that room.
4. How do I get qualified for a probe station?
The user must make an appointment with the superuser of the probestation to become qualified on their probestation. The superuser will ask the user to demonstrate proficiency in making measurements. The user must be able to answer questions such as the ones answered in this FAQ. After the user has passed, they can use the probe station they are qualified on.
5. What are the user's responsibilities?
By using the device characterization lab, you are agreeing to accept the following responsibilities:
- Follow all lab rules.
Remind other users to follow these rules.
Read current Frequently Asked Questions.
Leave the station/lab neater and better than you find it. Do a small thing for the station/lab every month; make a major improvement every year.
- All users must contribute to the management of the lab by serving as superusers and Lab managers. A superuser is responsible for finding his/her successor. When that is impossible, the Lab Manager appoints the successor. The Lab Manager is responsible for finding his/her successor. When that is not possible, the Device Group professors will appoint the successor.
6. How do I sign up for a probe station?
The user must write his/her name on the board so as to designate the time, the probe station, and the measuring machine to be used. You can only sign up 2 days in advance. Furthermore, you can only sign up for a maximum of 4 hours unless special permission is given for a long measurement. We are currently contemplating using an online reservation system. More info will be posted.
7. What programs are available to control the machines?
The \htb386\bin directory contains the following programs:
Program Author Description 4284 Kelvin Hui HP4284 CV meter control program CVT2 Elyse R. & Joe King Converts LCV data to ASCII and saves Phi_s & Dit data Doping Kelvin Hui Extract Nd(x) Extract Kelvin Hui Convert HP4145 data to PC format HCV Jian Chen High frequency CV Hot Koichi Seki & Khander Quader Hot carrier stress I-V Kelvin Hui Extract intrinsic I-V with specific resistance LCV Kelvin Hui Low frequency C-V Leff Kelvin Hui Leff and Rds measurement
8. How do I check out a wafer?
Wafers, wafer tweezers and storage trays are checked out from the Microlab office and charged to your research advisor. Special wafers such as SOI wafers need to be ordered separately. Check with your grant administrator on how to purchase specialized wafers.
All wafers left in the Device Characterization Lab should be labelled with your name and wafer information.
9. What can I do to annoy fellow lab users or get kicked out of the 407 lab?
- Break things without fixing them or reporting to the superusers and lab manager. Unauthorized use of probe stations, especially P6, P7, and especially the Cascade probe station. Move manipulators and cable wires from one probe station to another.
- Fail to cancel a reservation.
10. What is the HP4145, HP4155 or HP4156?
The HP41XX is a semiconductor parameter analyzer, used primarily to obtain device characteristic curves, such as the traditional Id vs. Vgs or Id vs. Vds curves. Each semiconductor parameter analyzer can save data on a 3.25" floppy disk.
At the heart of the HP41XX are the 4 SMU (Source Monitor Unit) modules. A SMU is a multifunction specialized power supply. Some possible configurations include:
- Provide a voltage subject to a current limit (compliance). Provide a voltage (swept or constant) and measure current, or just provide a constant voltage subject to a current limit.
- Provide a current while measuring voltage subject to a voltage limit.
Each of the 4 SMUs can be configured separately, allowing the user to make different types of measurements.
In order to provide quasi-3D measurements, the 41XX performs a stepped- swept measurement. That is, one independent variable (SMU output) is swept for each step of the other independent variable (SMU output), while the dependent variable (SMU measurement) is logged.
For example, to do a standard Id vs. Vgs curve, you would ground the source, and drive the body, gate and drain. You could elect to either ground the body, or connect it to a SMU operating in a voltage mode with current limit (just in case you forward bias the S/D region). The gate would be connected to an SMU operating in a stepped voltage mode with a very low current limit to protect against gate oxide breakdown. The drain would be connected to an SMU operating in the swept voltage mode with a current limit to protect the device. The current through the drain SMU is the dependent variable Id.
A typical measurement (assuming 1.2um NMOS) may consist of sweeping the drain SMU from 0 to 5 volts in 0.05 volt increments, and stepping the gate SMU from 0 to 5 volts in eleven 500 mV steps. Be aware that the 41XX is a discrete measurement device. It approximates continuous (swept) measurements by taking lots of closely spaced measurements. There is an internal memory limit so going overboard will result in an out-of-memory error. Note that the second independent variable is specified in step size and number. The gate could be connected to ground or to a third SMU supplying a constant voltage (negative in this case).
Setting up the 41XX can be a time-consuming task, so the machine allows one to save a setup (and data) through the floppy drive. Chances are that someone has already written the setup for the type of measurement you want to make, so ask around before re-inventing the wheel.
As specified earlier, the outputs of the SMUs on the 41XX use TRIaxial connectors. DO NOT ATTEMPT to put standard BNCs on it. It is mechanically impossible and attempts to do so will destroy the connectors! Triaxial connectors have three pins on the outside sleeve -- look before you attach. The reason is for this is to bootstrap out the cable capacitance when doing low current measurements. Otherwise the displacement current can cause erroneous current measurements. Internal to the 41XX is a voltage buffer which forces the shield to the same potential as the center signal conductor, thus reducing the cable capacitance by a factor of 1/(1-A) where A is the buffer gain (very close to 1). Needless to say, shorting the shield to the outer braid will cause the 41XX to be unhappy, as will shorting the shield to the center pin. See the question on the box with switch above. For further information, consult the manual for the 41XX located in 407 Cory.
11. What does the HP428X LCR Meter do?
The HP428X LCR Meter is most commonly used to perform impedance parameter measurements on a circuit or device, as a function of frequency, DC bias voltage, or oscillation level. For instance, the impedance can be measured as a function of swept frequency (range = 5 Hz to 13 MHz) at a fixed voltage, or as a function of swept DC bias (range = -35V to +35V) at a fixed frequency. The HP428 simultaneously measures two independent, complementary impedance parameters in each measurement cycle. The following parameter combinations can be measured:
- absolute impedance (or admittance) / phase angle (deg or rad); resistance / reactance; conductance / susceptance; inductance or capacitance / resistance or conductance;
- inductance or capacitance / Q-factor or dissipation-factor.
One very common use of this piece of equipment is that of making capacitance-voltage measurements on MOS-based device structures.
The HP428X can also be used to measure the amplitude, phase, and group delay of various types of network circuits, at frequencies between 5 Hz and 13 MHz.
The HP428X LCR Meter can be programmed either manually using front panel controls or remotely by means of an HPIB and a PC running HT-Basic. Details may be found in the 428X operation manual. In addition, there are several existing HT-Basic programs available for making C-V mesurements.
12. How do I ensure low probe contact resistance?
You will need reasonably sharp probes, > 60 degree probe incident angle, good contact between screw and probe (with care not to touch the probe with bare hands during installation) and debris-free probe tip. A good contact has less than 2 ohms/contact. A lousy contact will have a resistance that fluctuates anywhere from 10-80 ohms/contact. Not only will the measured current be smaller, but it will also be inconsistant. Apply moderate pressure to the pad (with the probe tip sliding about 20-30 um). From the ensuing probe mark you can tell if the contact is good. To ensure even pressure on all probes, adjust the manipulators until the tips barely touch the metal pads. Lower the whole platform using the vernier at the back of the chuck. To move to another die raise the chuck using the vernier instead of adjusting individual manipulators. Sometimes poor metal morphology or organic residue on the surface will prevent a good contact. In such cases, have the probe tips pierce in from the side of the metal.
13. How do I measure the leakage current on a probe?
Lift the probe, and scan voltage using the SMU of the Semiconductor Parameter Analyzer (SPA). Ideally, we would like to make a constant voltage measurement to reduce displacement current and use an high integration rate to get an accurate measurement. However, a constant voltage measurement cannot be done with a high integration rate. Thus, we need to use a low voltage ramp (.1V/step, to reduce displacement current) which is almost like a constant voltage. The SPA should have a leakage current around .05pA.
14. How and why should I change the teflon screws on the manipulator?
To reduce leakage a teflon plate is inserted between the manipulator arm and the base. The arm is also screwed onto the base with teflon screws. However the teflon screws don't have large tensile strength and can break easily. To avoid breaking the screws, care must be exercised in changing probe tips. At no time should any torque be applied to the arm without holding the L-shape arm.
If the screw is broken, unscrew it using a pointed tweezer (pierce and turn). If it fails then use a metal screw of the same thread size to force out the broken teflon screw. Clip off 1-2 mm from the end of the new screw. Put the new screw, teflon plate and manipulator arm into a beaker of acetone to remove metal flakes and grease. Put on polyethylene gloves and screw the arm back onto the manipulator. Tighten the screw just enough to keep the arm in place.
15. What kind of probes tip should I order?
For probe station 6, everyone should supply their own probe tips. You can check out probe tips out from the Microlab. You can also order probe tips from the supplier is Alessi. Order the PTT24 instead of PTT12 if possible. PTT24 means the tip diameter is 2.4 um. A larger diameter will improve contact and more immunity from sliding (though of course nothing will save you if you have bad probing technique). For probe station 7, contact the superuser of that station to order more probe tips for the group.
16. How do I tighten a manipulator?
Loosen the set screws on the side of the loose red knob with an allen wrench. Tighten the knob by tightening the bolt on the opposite side of the loose knob. Retighten the set screws on the side of the loose knob. Tightening the knob without loosening the set screws will strip the bolt. Move manipulator around to expose screw if necessary.
17. How do I change probe tips?
When installing a probe tip, DON'T BEND IT. The incident angle is determined by the slot angle inside the arm (either 75 or 60 degrees). A large incident angle will greatly increase contact resistance and cause sliding. During installation avoid touching the part that is in contact with the metal screw. The length of the probe tip below the screw should be about 0.8 inch. Hold the L-shape arm while screwing in. Do not apply too much force. Check for tautness by gently pulling the probe tip with tweezers.
18.What's special about Cascade (Temperature control capabilities)?
The Cascade Probe Station is designed for temperature control measurements. It is cooled with a special cooling liquid that goes through the chuck. The chamber must be dehydrated for at least five to six hours prior to cooling, so it should be sealed from the surrounding air. The manipulators are moved by the round black knobs on the plastic shield that attach to the manipulators by springs. When cooling or heating the chuck, probes must be lifted up, otherwise they will scratch everything due to thermal expansion. The wafer chuck can move only in the horizontal direction when the chamber is sealed; the mechanism is located on the side of the chamber as is the knob that focuses the microscope.
19. How do I make a quasi-static MOS CV measurement?
For a quasi-static MOS CV, you apply a ramp voltage to one of the terminals from inversion to accumulation and measure the displacement current. Usually you use the 41XX because it can measure low current levels and has a built-in voltage source. The displacement current is proportional to the voltage ramp rate and the capacitance. So if you want to get an accurate measurement you want to use a high ramp rate so your displacement current will be larger compared to noise. But if you ramp it too fast, the 41XX will not be able to follow. Usually this means a measurement around 0.1 to 1 V per second. On the other hand, this measurement is very sensitive to leakage which is not a function of ramp rate since the current levels are low.
20. How do I make a HF MOS CV measurement?
The HP428X LCR meters at P6 and P7 can be used to measure the impedance of a circuit at different biases. Both probe stations have computer-controlled software that allows the user to control the frequency and voltage sweep.
Basically, the LCR meter measures the impedance between the two nodes -- the gate and the substrate at different biases, and read out the capacitance as a function of the bias voltage. Make sure however that you are using the right circuit mode -- serial or parallel (see next question).
21. What is the difference between a serial or parallel connection model when doing a impedance measuremnt?
---+-------- | \ C === / Gp | \ ------------- \ / Rs \ -------------
This is the basic model that the instrument assumes for what it is measuring. When Rs is large compared to 1/Gp, we basically have a C in series with Rs. This is called a serial connection. An example of this would be a thick oxide capacitor which has a large contact resistance. The values given by the instrument are then called Cs and Rs.
In a parallel connection you are assuming the circuit is composed of a parallel connection of a capacitance and resistance. Thus, you are determing Cp and Gp. An example of this is a thin oxide where there is large tunneling current, and thus Gp is large.
22. How can I do measurements on stress-sensitive conducting films?
Probes are usually put down on thick Al films over an insulated region. If stress-sensitive thin films or contacts are being used, it is best to put down a layer of another conducting material on top so that the pressure of the probe is not over the active device. Many films that are already strained (like silicides formed at high temperatures) will appear leaky in IV measurements if the probe is pressed down too hard.
If a buffer layer cannot be made, care must be taken to minimize leakage due to the probe-applied stress. Using a dynamic IV measurement (like using the 4140B, and hooking it up as for an IV measurement, but just turning it on - not running a program) watch the current level displayed on the 4140B as probe is lowered. Once the probe is down, current level goes up (keep light on to make the jump in current more evident). Then begin raising probe slowly until contact is just broken, and then lower just to regain contact. Going slowly and carefully is the only way to get reproducible measurements.
23. How do I make a hot carrier measurement?
For hot carrier measurement, since we want to do accelerated test, we need to determine the stress conditions so devices can reach significant damage in a reasonable stress time.
For NMOS, we need to find a test device, and measure Isub vs. Vgs under various values of Vds. Normally, the plot of Isub vs. Vgs is a bell type plot and we choose the stress Vgs as the value which causes peak Isub. (The reason is that the device hot carrier lifetime Tau is roughly inversely proportional to the cube of Isub, so larger Isub means smaller Tau and we can get significant damage in a reasonable time or use it as a worst case study.) The other concern is the stress condition for Vds. A simple rule of thumb is that the maximum Vds is the Vds which causes the peak Isub, about 10 - 15 uA/um of W. If the stress Vds is too large, Isub would be too large and breakdown might happen (when Isub * Rsub > 0.7).
For PMOS, we need to stress at the Vgs value which causes peak Ig. As for the stress condition for Vds, for the CMOS process, we can use the same Vds as for NMOS; for a PMOS-only run, you need to do a couple of experiments to see what Vd can cause significant damage in a reasonable time. (The peak Ig value might range from a few pA to 20pA per um of W).
In summary, the stress conditions are the following:
Type Vgs Vds NMOS Peak Isub peak Isub ay 10 - 15 uA/um of W PMOS Peak Ig The same as you use for NMOS in a CMOS run; but depends on actual degradation rate in a PMOS-only run (might be 1 - 20 pA/um).