Passing through materials that are opaque to visible light, terrahertz waves offer advantages in imaging applications. However, near-field imaging, beneficial for achieving high spatial resolution and better sensitivity, can be problematic in the terrahertz region with the detector located remotely from the probe.

An integrated device combining aperture, probe and detector on a single semiconductor chip has been developed at the Advanced Device Laboratory, Saitama, Japan, removing the need for optical and mechanical alignments [Kawano et.al., Nature, doi: 10.1038/nphoton.2008.157].

Near-field imaging is well established in the visible and microwave regions but sensitivity is commonly degraded by the separation of detector and probe, allowing the influence of far-field waves. Near-field terrahertz systems use a small aperture, with a consequent weak field and small wave area, and a collecting tip that is prone to picking up the stronger far-field background signal.

The new device efficiently detects only the evanescent wave with a planar probe that is integrated with the detector, eliminating the need for separate waveguide and lens. The 8 μm aperture and planar probe, separated by a layer of silicon dioxide insulation, are deposited onto a GaAs/AlGaAs chip with a rectangular detector of 2D electron gas (2DEG) 60nm below the chip surface. The evanescent wave induces resistance heating of the 2DEG and, because it is a two-dimensional layer, it is unaffected by the far-field wave. The planar probe effectively changes the spatial distribution of the evanescent field so that it is better coupled to the detector in the same plane.

“A resolution of 9 μm has been obtained using the system and further improvements are expected with the use of a detector based on carbon nanotubes, expected to exhibit higher sensitivity and resolution as well as operating temperature improvements,” explains Yukio Kawano. The resulting sub-micron device could be used at room temperature and is likely to open up new applications in chemistry, physics and engineering.