NIST improves technique for detecting transistor defects

3 mins read

Researchers at the National Institute of Standards and Technology (NIST) have revived and improved a once-reliable technique to identify and count defects in transistors.

Over the past decade, transistor components have become so small in high-performance computer chips that the popular method, known as charge pumping, could no longer count defects accurately. NIST’s new method is sensitive enough for the most modern technology, and can provide an accurate assessment of defects that could otherwise impair the performance of transistors and limit the reliability of the chips in which they reside.

The new, modified charge pumping technique can detect single defects as small as the diameter of a hydrogen atom (one-tenth of a billionth of a meter) and can indicate where they’re located in the transistor. Researchers could also use the new capability to detect and manipulate a property in each electron known as quantum spin.

The ability to manipulate individual spins has applications in both basic research and quantum engineering and computing.

Defects in a transistor can interfere with the reliable flow of current and significantly degrade the performance of transistors. These defects could be broken chemical bonds in the transistor material or atomic impurities that trap electrons in the material.

Scientists have devised several ways to categorize defects and limit their impact, tailored to the structure of the transistor under study.

In the traditional design known as the metal oxide semiconductor field effect transistor (MOSFET), a metal electrode called the gate sits atop a thin insulating layer of silicon dioxide. Below the insulating layer lies the interface region that separates the insulating layer and the main body of the semiconductor. In a typical transistor, current travels through a narrow channel, only one billionth of a meter thick, that extends from the source, which lies on one side of the gate, to a “drain” on the other side. The gate controls the amount of current in the channel.

Charge pumping is a two-step process in which the examiner alternately pulses the gate with a positive test voltage, then a negative one - the alternating voltage pulses are applied at a single, set frequency.

In the first step of the test, the positive voltage attracts or pumps electrons, which are negatively charged, to the boundary or interface between the gate’s insulating layer and the body of the transistor. Some of the pumped electrons become trapped in defects at the interface, but there are many electrons left over. In the second step, a negative voltage is applied, to rid the interface of the excess electrons, leaving only the trapped ones behind. The negative voltage also attracts positive charge carriers, known as “holes,” to the region, where they combine with electrons trapped in the defects. This activity generates a current proportional to the number of defects. The greater the output current, the larger the number of defects.

Today, the insulating oxide layer in modern transistors is now so thin — just 10 to 20 hydrogen atoms wide — that an effect from the realm of quantum mechanics comes into play, confounding measurements using the traditional charge-pumping method.

According to quantum theory, electrons and other subatomic particles can never be truly trapped; there’s always some probability they will escape or “tunnel” out of an enclosure or boundary layer. The thinner the material, the higher the probability. As transistor dimensions shrank, the tunnelling current leaking through the insulating oxide layer made it almost impossible to detect defects with ordinary charge pumping. Scientists all but abandoned the technique.

NIST researchers have found a way to salvage the technique so that it not only works for ultrathin transistor components but is also more sensitive, enabling scientists to record signals from a single defect. The solution is derived from the fact that the current that results from quantum tunnelling remains virtually the same, regardless of the frequency at which charge pumping pulses the positive and negative voltages.

As a consequence, the team revised the charge pumping technique by alternately applying the method’s positive and negative voltages at two different frequencies rather than the single frequency used in the traditional method.

Applying the voltages at two different frequencies gave the researchers two different output currents. By subtracting one output current from the other, the constant signal from the quantum tunnelling current dropped out. With the confounding tunnelling current eliminated, the researchers were able to detect defects in transistors with ultrasmall features.

The researchers reported their development of the frequency-modulated charge pumping technique online in the Feb. 2 Applied Physics Letters.

Since only one electron is involved, the output current is equal to multiples of the electron’s charge, a fundamental physics constant calculated by NIST and other institutions.

Because the method can detect single electrons, it may serve as a sensitive probe of an electron’s quantum spin. Modulated-frequency charge pumping may provide a valuable guide to scientists who are now exploring how electron spin might store and transfer information in a computer of the future. It may also prove useful in quantum metrology, as a potential new way of determining a quantum standard of electrical current.