Near field infrared absorption of plasmonic semiconductor
microparticles studied using atomic force microscope
infrared spectroscopy
Jonathan R. Felts1, Stephanie Law2, Christopher M. Roberts3, Viktor Podolskiy3, Daniel M.
Wasserman2, William P. King1
1
2
Department of Mechanical Science and Engineering, University of Illinois Urbana-Champaign,
Urbana, Illinois 61801, USA
Department of Electrical and Computer Engineering, University of Illinois Urbana-Champaign,
Urbana, Illinois 61801, USA
3
Department of Physics, University of Massachusetts Lowell, Lowell, Massachusetts, 01854,
USA
S1. Fabrication and Characterization of heavily Si-doped InAs Micropillar array
The sample investigated consisted of a 1.1 m thick layer of highly n-doped indium
arsenide (InAs) grown by molecular beam epitaxy (MBE) above a 0.75 µm thick undoped InAs
buffer layer on a single-side polished undoped gallium arsenide (GaAs) wafer. The buffer layer
was required in order to separate the plasmonic material from the defects formed at the
GaAs/InAs interface resulting from the significant lattice mismatch of the two material systems.
The doping density of the highly-doped InAs layer was designed to result in a plasma frequency
in the mid-IR wavelength range [1]. Transmission and reflection measurements were made on
the as-grown film using a Bruker v70 FTIR spectrometer in order to determine the doped layer
optical properties. Reflection spectra were normalized to a Au-coated Si wafer, which gives
close to 100% reflection across the wavelength range of interest, while transmission spectra were
normalized to unblocked, free-space transmission.
The resulting spectra were modeled in
accordance with the transmission matrix method described in [1] and [2], where the doped InAs
1
layer’s permittivity is given by the Drude formalism, with only the plasma wavelength and
scattering time as fitting parameters. Using this technique, our doped InAs layer was determined
to have a plasma wavelength of p=5.48 m, with a scattering time of approximately Г = 1x10-13
s. The resulting wavelength is a function of the silicon doping, and does not depend upon the
dimensions of the pucks [1].
The top doped layer was then processed into an array of pucks with a nominal diameter
of 1.7 m and a periodicity of 3.4 m using standard photolithographic techniques. The pucks
were etched using a mixture of 1:1:10 HBr:HNO3:H2O, leaving the undoped InAs layer intact
underneath. Reflection and transmission spectra for these pucks were taken in a similar fashion
to the as-grown wafers, as described above. Weak dips in transmission and reflection were
observed at  ~ 5.73µm, in spectral agreement with our finite-element model of the fabricated
structure, using the permittivity determined from the measurements of our as-grown material. A
slight spectral shift between transmission and reflection dips is observed in our data, which was
also observed in previous samples and in our COMSOL model. This is believed to result from
the combination of (angle-dependent) scattering and reflection/transmission from the puck
sample. The size and spacing of the pucks in the array preclude the excitation of collective
resonances by the laser used in experiments [3].
We removed the GaAs substrate from the sample to ensure the laser light was absorbed
mainly by the pillars and not the substrate.
Following initial, far-field, transmission and
reflection measurements, the processed film was attached face-down to a glass slide with crystal
bond wax, leaving the GaAs substrate exposed. The sample was then etched overnight in a
solution of 1:3:16 NH4OH:H2O2:H2O, which selectively removed the GaAs from the epitaxiallygrown InAs. The remaining material and glass slide carrier was placed in acetone to remove the
2
wax and release the patterned InAs, leaving the InAs film freely floating. After diluting the
acetone with water, the film was carefully placed onto a submerged zinc selenide (ZnSe) prism
and allowed to air-dry. The final structure, now ready for near-field optical characterization,
consisted of the doped InAs pucks supported by a thin undoped InAs layer, resting on the ZnSe
prism.
S2. Numerical calculations of light propagation through micropillar arrays
Numerical solutions of Maxwell equations in the pillar arrays were performed by a
commercial finite-element method (FEM), partial differential equation (PDE) solver, COMSOL
multiphysics. The system was approximated as periodic planar array of cylindrical pillars
(pucks). The permittivity of an individual puck was approximated by the Drude model, whose
parameters were deduced from measurements of the transmission and reflection of an unpatterned structure.
These 3D calculations allow us to compute both microscopic field
distributions across the unit cell, and macroscopic parameters (transmission and reflection) of the
structure.
In a separate set of calculations, a transfer-matrix technique was used to compute the
spectrum of guided modes in the structure comprising of only a GaAs substrate, an undoped
InAs buffer layer and air. The sharp satellite resonances in the reflection spectrum from FEM
calculations were then matched to the guided modes in the puck-less system, shifted by the
integer multiples of the grating vectors corresponding to periodicity of a unit cell.
References
[1] S. Law, D. C. Adams, A. M. Taylor, and D. Wasserman, Opt. Express 20, 12155 (2012)
3
[2] Y. B. Li, A. Stradlng, T. Knight, J. R. Birch, R. H. Thomas, C. C. Phillips, and I. T.
Ferguson, Semicond. Sci. Technol. 8, 101 (1993)
[3] R. Adato, A. A. Yanik, J. J. Amsden, D. L. Kaplan, F. G. Omenetto, M. K. Hong, S.
Erramilli, and H. Altug, PNAS 106, 46, 19227 (2009)
4