Factors To Consider When Making Sub-Femtoamp Measurements.
For currents of about 100 [micro]A, the limitation on how small of a change in current can be detected is often the digital resolution of the instrument (1 nA in a 5.5-digit instrument). At lower signal levels, noise often limits the measurement. For example, if noise in the system is 100 femtoamps (1 fA = [10.sup.-15] A), and a 100-picoamp (1 pA = [10.sup.-12] A) signal is being measured, the limitation is the 100-fA noise, not the digital resolution of 1 fA.
At picoamp currents and lower, a number of effects in the cabling, test fixtures, ammeter, and the device or material under test can generate significant currents that limit the measurement. Motion in a cable, for example, causes frictional forces at the boundaries of the cable and insulation material. This removes charges from the insulation, generating a flow of up to a nanoamp of current. Low-noise cable has graphite at the interface that reduces friction and allows local currents to resolve the charge imbalance, thus reducing the tribo-electrically generated noise.
Even if a cable that has been moved recently is now at rest, charge can continue to flow for many minutes or hours after motion ceased, generating a slowly declining noisy current, often tens of femtoamps in magnitude. This is due to a variety of dielectric absorption effects in which residual charges come to equilibrium with a long time constant. Any mechanical vibration in the system can cause noisy currents to continue to be generated. To minimize this effect, it is important to keep cables as short as possible and minimize any mechanical stresses on them by attaching them to rigid structures. Similarly, insulating feedthroughs and support structures generate or absorb charge when force is applied. In some materials, this is due to mechanical action, such as in cables. In others, it is due to the peizoelectric nature of the material.
Printed circuit boards, which are commonly used in electronic circuitry, can also generate nanoamps of current due to electrochemical action of ionic substances on the boards reacting with ambient moisture. Cleaning a board with de-ionized water will reduce this, as will keeping the board in a sealed enclosure so that dust does not collect on it.
In designing circuits involving low currents, guarding is often used. If a metal trace with voltage V on it is near a low-current trace, a current equal to V/R will be added to the signal current [I.sub.S]. By adding a guard conductor between the conductor having voltage V on it and the sensitive input to the ammeter, the current V/R flows to ground through the guard conductor and the ammeter reads only the signal current [I.sub.S].
The easiest way to quantify generated currents in an experiment is to directly measure the currents using the ammeter used for the experiment. Start by verifying that the ammeter has acceptable noise and DC current error with just a cap (shielded open circuit) on the input. Then add the cabling, any test fixturing or sample holder, and then the sample with no test voltage applied, verifying that the DC current and current noise at each step of the process are still acceptable. If at some point they aren't, measure the elements of the added portion of the circuit and take steps to reduce the generated current until it is acceptable. Note the magnitude of the offset and noise. This will limit the uncertainty in the final measurement.
Special considerations are needed for measuring currents below 10 fA. Clearly all of the effects mentioned above can create significant errors at femtoamp and lower levels. For example, it is difficult to find a 1-m length of low-noise cable that generates a current less than 1 or 2 fA DC current even after resting for hours. Noise from such a cable is a few hundreds of an aA (attoamperes; 1 aA = [10.sup.-18] A). Cables are often the limiting factor in femtoamp measurements. Use the shortest cable possible to connect the input, and use an instrument with a remote head to get the measuring terminals close to the current source. Using a Remote SourceMeter with 400-aA peak-to-peak (p-p) noise, we tested low-noise cables of various lengths for noise. The result was that cables longer than about 6 in materially added to the system noise of 400 aA or less.
All insulators are suspect and must be individually evaluated for acceptably low generated currents. All substrates must be thoroughly cleaned to remove current-generating ionic material. Often different cleaning methods must be tried. Any motion or noise on voltage leads with capacitive coupling to the input create current noise due both to I=C [multiplied by] (dV/dt) and I=V [multiplied by] (dC/dt). An ideal resistance also creates a noise [i.sub.N] = [[4kTF/R].sup.1/2]. [I.sub.N] is the noise generated by an ideal metallic resistance, k is Boltzman's constant, T is absolute temperature, R is the resistance in ohms, and F is the bandwidth of the measuring circuitry in Hz. For example, at room temperature (300 K), [10.sup.12] ohms generates 0.65-fA p-p noise in a 1-Hz bandwidth. All insulation paths must be considered as noise-generating resistances and evaluated in this framework.
Variations below the femtoamp level occur because of effects that are innocuous at higher levels. These include noise from low-noise semiconductor devices, noise from slight variations in temperature, or variations in a "light leak" in a dark environment, creating a small variation in photoelectrically generated currents. Temperature stability over a few minutes can usually be obtained by putting the sample in an enclosed box. Light leaks may need to be eliminated by tighter construction, opaque tape, etc. Again, all of these effects can be evaluated, quantified, and reduced by using the sensitive ammeter as a troubleshooting tool.
In addition, some effects that are not observed at higher levels are important at femtoamp and attoamp levels. For example, any space with an electric field and a conductor in it may act as an ion chamber. The electric field will accelerate any charged particles, and they will create a current when striking the sensitive input conductor. The particles may be charged due to background ionizing radiation or they may be outgassing from some part of the experiment in the space. Minimizing enclosed volumes around the sensitive terminal reduces this effect. In some experiments, removing materials in the space surrounding the input one at a time may be necessary to determine the source of the current due to outgassing or radiation.
As to the ultimate measuring possibilities for low current, experiments have recently been done testing the limits of a new SourceMeter. When using this instrument to average many readings of a repetitive waveform over a long time, changes of 10 aA ([10.sup.-17] A) are measured. Most often, the limit to measuring low currents is well above this level, due to effects discussed above. The troubleshooting techniques described here can be used to obtain valid readings of femtoamp and lower currents with known uncertainties.
More information on sources of generated currents and techniques for reducing them can be found in the 5th edition of "Low Level Measurements," published by Keithley Instruments Inc. To order a handbook, call 800-552-1115, or visit Keithley's Web site at www.keithly.com and click on Free Literature.
Erdman is a senior market development manager and Daire is a core technology scientist at Keithley Instruments Inc, Solon, Ohio.
|Printer friendly Cite/link Email Feedback|
|Author:||Erdman, Bob; Daire, Adam|
|Publication:||R & D|
|Article Type:||Brief Article|
|Date:||Jan 1, 2000|
|Previous Article:||New Products.|