Ultra-low light level photoresponse in thermoelectric nanowire arrays

Download full paper

The international telecommunications market is now a $4 trillion-per-year industry. This is not an accident, as the ability to move information quickly and cheaply on a global scale is one of the things that separates profitable companies from those relegated to the trash heap of yesterday’s businesses.

This fact affects international aerospace markets in ways that would not have been anticipated a generation ago. Specific to Boeing, our communications systems increasingly rely on higher bandwidth communications in all segments of the electromagnetic spectrum. As a result, systems involving high speed communications, sensing and imaging require ever more sensitive detection systems coupled with the ability to operate in a trans-spectral architecture. This is increasingly true in the opto-electronic regime, in which rapid conversion of electro-magnetic radiation (including rf, microwave and optical signals) to electrical signals—and the inverse—must occur at high bandwidth to avoid communications and informational bottlenecks.

Detectors in the optical regime historically employ detectors based on bulk absorption in semiconductor media. For general semiconductor devices, photocurrent is generated by the separation of excited electron-hole pairs by a built-in electric field or photovoltaic effect (PV). In other words, the excitation of electrons from valence to conduction band generates an internal field. For semiconductor-based optoelectronic devices, such as silicon solar cells, photon absorption leads to the transfer of charge between different electronic bands, resulting in a photocurrent (PC).

However, there are other physical mechanisms in condensed matter than can be exploited to make a detector. For example, there is a class of materials which exhibit a so-called thermo-electric response. In this case, a temperature gradient is generated by light absorption between two surfaces, a potential difference is created vie the thermoelectric (TE) Seebeck effect. Recent work in two dimensional graphene demonstrated a hot carrier thermoelectric (TE) photoresponse, in which an electrical response was generated by the thermal heating caused by light absorption at the nanoscale.

Boeing has on-going experimentation at strategic universities in the various areas of nanotechnology. One of these nanotechnologies has to do with thermoelectric optical detection. From this, we report on the first demonstration of thermoelectric optical detection in bismuth nanowire arrays (NWA).

NWA devices are composed of thermoelectric bismuth nanowire arrays that are capped with a transparent indium tin oxide electrode. The incident surface features very low optical reflectivity and enhanced light trapping. The unique attributes of the thermoelectric arrays are the combination of a strong temporal and optical wavelength dependence of the photoresponse.

We have shown that, under infrared illumination, the photoresponse can be completely described by thermoelectric effects. At low frequency, quasi-equilibrium is achieved and the signal is f-independent. At high frequencies, the absorbed heat is not dissipated in the back electrode during the illumination cycle, and the signal can rise proportionately with f1/2   to much higher values because the energy is delivered to a thin layer on the front of the sample where the junctions are located.

The TE effect that completely describes the infrared-only illumination appears to be complimented by PV and hot-carrier processes under visible illumination. The TE signal can be fast, with a response time much shorter than the array thermalization time, only limited by optical penetration. The resulting detection arrays may find future optoelectronic application as fast nanoscale thermopiles.

Through this paper, we report the first demonstration of TE response in an array of junctions of nanowires. We discuss the nanoscale optical and TE properties that give rise to the effect. We also find that under illumination there is a photoresponse in both near infrared (NIR) and visible spectral regimes.

Under infrared illumination, the photoresponse is simply governed by heat diffusion and can be uniquely attributed to TE effects. However, under visible illumination, the photo response is a combination of TE and PV effects.

TE effects are caused by the difference in the broadening of the energy distribution in the electronic bands caused by a temperature gradient (See Reference 1). Some traditional TE materials, such as bismuth, have unusually high thermopower values and a large, leading to exceptionally large Z values and high efficiency. However, although it is almost counter-intuitive, bulk TE crystals are not good candidates as TE photoresponse detectors because the same properties that create high Z values cause the material to have high thermal conductivity.

As a result, the energy dissipates quickly in these systems. We chose to address this problem by working in a nanowire geometry. We anticipated that this geometry would show enhanced thermoelectric performance because κ is less in nanowires than in the bulk. In addition, bulk thermoelectric materials have large refractive indices relative to air. This causes large Fresnel reflection at the incident surface of bulk thermoelectrics, which is not conducive for efficient conversion. In fact, in a variety of bulk systems, including bulk crystalline semiconductors and Bi, light induced thermoelectric responses are so weak that they have only been observed under pulsed illumination.

In investigations of PV in solar cells, researchers have discovered that the nanostructuring of bulk materials into wires or sharp points aligned along the optical incident direction results in reduced optical reflection and induced light trapping. There are many mechanisms that may play a role in this effect. The optical reflection of a nanowire array is greatly reduced relative to the well-ordered surface because the electromagnetic field penetrates deep in the material. This property has been attributed to dipole effects because the wires point parallel to the light wave vector, and the photon electric field is perpendicular to the wire length. Nanowire array optical properties have also been discussed in the context of optical meta-materials.

The array electronic properties were determined in separate experiments. X–ray diffraction spectra show that the crystal grains were larger than the wire diameter and were oriented with the c-axis along the wire length (See Figure 3, inset). Electron and hole band parameters were determined via magneto-resistance experiments. As with bulk bismuth, the bismuth in the nanowires was determined to be a semimetal. Thermopower measurements show that the thermopower of the nanowires was diffusive in nature and negative (nanowires, as well as bulk, are n-type because electrons have greater mobility than holes). This value of approximately −90 mV/K did not differ significantly from that attained for single-crystal bulk bismuth along the diagonal orientation.

We also characterized the nanowire arrays under optical light illumination. The optical properties of the array were significantly different from those of bulk single crystal bismuth with its highly reflective front surface. In addition, the 50-μm thick alumina template, prior to processing, was translucent/transparent in the visible and infrared. By contrast, we found that the optical absorption by our nanowire arrays was very high, broadband and anisotropic. The array surface was black and dull without the reflective surface characteristic of single crystal bismuth. This finding is consistent with arrays of many diverse materials (e.g. silicon nanowire arrays) and denotes low reflectivity caused by high levels of light trapping.

We also observed that light reached deep inside the array. Under oblique illumination—that is to say, with the incident light propagation vector non-parallel to the nanowire orientation—the reflected visible light is highly polarized.

Interpretation of this data can draw upon prior TE theoretical work. The physical interpretation of our observations are outlined in the full paper. From these observations, the signal is interpreted as a pure TE response.

The assignment of the signal to TE effects is appropriate in the case of infrared-only illumination. Under infrared illumination, which includes visible light, the inflection point is barely noticeable, and the signal strength is nearly independent of frequency. This behavior strongly suggests that part of the signal is not strictly thermal, indicating the presence of PV or hot electron component in this case.

Semiconductors used in solar cells cannot convert the infrared part of the solar spectrum, which represents approximately half of the total solar energy output, because in PV processes the absorption of a photon with energies below the bandgap energy cannot readily produce electron-hole pairs.

Devices based on TE nanowires can convert the infrared radiation part of the spectrum into usable electrical power, which suggests that their integration in solar cells may increase overall solar energy harvesting efficiency.

By Jeffrey H. Hunt

Figure 3

Nanowire image

Scanning electron microscope side image of NWA. Inset: X-ray diffraction spectra showing presence of crystal grains oriented with the c-axis parallel to wire.