Access Official Technical PDF

Share Summary:LinkedInEmail

RF plasma treatment offers a low-temperature way to modify important electrical properties in InGaAs junctionless MOSFETs. In a study of devices with an Al2O3 gate dielectric, treatment in forming gas substantially reduced source/drain resistance and increased channel mobility at low plasma power density. It also shifted threshold voltage in a positive direction and improved subthreshold behavior. For engineers working with compound-semiconductor devices, the findings show both the potential of plasma processing and the importance of controlling its conditions.

Why junctionless InGaAs devices need careful contact engineering

A junctionless (JL) MOSFET uses a doped semiconductor channel without the separately formed source/drain junctions used in conventional inversion-mode devices. This architecture can simplify process integration, and InGaAs is of interest because it can retain comparatively high electron mobility at substantial doping levels. Avoiding source/drain implantation or selective epitaxial regrowth may also address some fabrication challenges associated with III-V devices.

However, source/drain contact resistance is a significant concern. Forming effective ohmic contacts conventionally can require annealing at temperatures of 300°C or higher. Such temperatures may create unwanted material movement, including diffusion of indium into the gate dielectric. A process that improves contacts while limiting thermal exposure is therefore attractive for InGaAs transistor fabrication.

How the RF plasma treatment was evaluated

The devices used a 20-nm-thick n-In0.53Ga0.47As channel, formed by thinning an initially 32-nm layer, with an Al2O3 gate dielectric and a junctionless layout. The samples received no post-fabrication thermal anneal before treatment. Instead, they were exposed for 10 minutes to 13.6 MHz RF plasma in forming gas containing 10% hydrogen and 90% nitrogen. The sample holder was heated to 150°C, while plasma power density was varied from 0.5 to 1.75 W/cm2.

To distinguish effects associated with plasma exposure from those caused by heating and other influences, some samples were treated from the device front side and others from the substrate back side. Circular transmission line structures were used to assess semiconductor sheet resistance and specific contact resistance. Transistor measurements then tracked characteristics including drain current, threshold voltage, subthreshold slope and mobility.

RF plasma treatment reduced contact resistance

Before treatment, the metal-to-InGaAs contacts showed nonlinear current-voltage behavior, consistent with rectifying rather than ohmic contacts. After front-side treatment at 0.5 W/cm2, the measured current-voltage response became linear. The sheet resistance of the 20-nm InGaAs layer fell from approximately 330 kΩ per square to about 2 kΩ per square, while specific contact resistance decreased by more than two orders of magnitude.

The transistor measurements supported this result. At 0.5 W/cm2, extracted source/drain series resistance fell from 14.4 kΩ in untreated devices to 140 Ω after treatment. This reduction is important because parasitic source/drain resistance limits the current that can be delivered through the channel, complicating assessment of the transistor itself.

Front-side and back-side results differed. Back-side exposure did not produce a comparable improvement in contact resistance or sheet resistance. The contrast indicates that heating alone does not account for the contact changes. The researchers attribute the front-side effect to non-thermal plasma-related factors, including low-energy particle bombardment and radiation, which can affect the metal-semiconductor interface and reduce its potential barrier.

Mobility and gate characteristics also changed

At low plasma power density, the extracted volume electron mobility increased from about 90 cm2/V·s in the untreated devices to approximately 1,300 cm2/V·s. A separate capacitance-based evaluation after treatment at 0.5 W/cm2 found an effective mobility reaching approximately 1,500 cm2/V·s. The study connects the mobility improvement to plasma-assisted changes in bulk and surface defects in the InGaAs layer.

The treatment also affected the Al2O3 gate dielectric and its interface with the channel. At 0.5 W/cm2, threshold voltage shifted from −0.49 V to 0.05 V, corresponding to a reduction in effective fixed positive charge under the gate. The inverse subthreshold slope improved from 190 to 150 mV per decade, which the study associates with a lower interface trap density.

Together, reduced series resistance and higher mobility produced a substantial increase in on-state current. The on-to-off current ratio after treatment at 0.5 W/cm2 was approximately 1.7 × 105. These electrical changes should be considered together: contact improvement affects measured drive current, while mobility, interface traps and dielectric charge influence channel and gate behavior.

Why plasma power density matters

The results were not a case of more plasma power producing better performance. Although the threshold-voltage shift tended to increase with power density, higher power eventually reduced on-current and the on-to-off current ratio. Mobility declined at higher treatment levels, and source/drain resistance rose as power increased beyond the low-power range. The researchers attribute this deterioration to defect formation associated with more energetic bombardment and increased radiation intensity.

For process development, this makes plasma power density a key control variable rather than a simple means of increasing treatment strength. The most favorable overall results were obtained around 0.5 W/cm2, while specific contact resistivity reached its minimum around 0.75 W/cm2. The best setting therefore depends on which measured device characteristics are being prioritized.

Practical implications for device processing

Low-temperature RF plasma treatment demonstrated a route to improve InGaAs junctionless MOSFET contacts without relying on a conventional high-temperature post-fabrication anneal. It also modified channel mobility and gate-stack electrical behavior, showing that a plasma process can influence several parts of a device at once. That combination makes it essential to evaluate contact resistance, mobility, threshold voltage and subthreshold response together.

The central process lesson is to optimize exposure conditions rather than assume plasma treatment is uniformly beneficial. In these devices, low-power front-side treatment delivered the strongest combined gains, while increasing power introduced degradation. For engineers assessing plasma steps in III-V transistor fabrication, the results identify a promising process direction—but one whose effectiveness depends on the device structure, treatment geometry and careful power control.

Access Official Technical PDF

About The Author