Jinxiang Li1, Kevin F MacDonald1, Nikolay I Zheludev1,2. 1. Optoelectronics Research Centre, University of Southampton, Highfield, Southampton SO17 1BJ, United Kingdom. 2. Centre for Disruptive Photonic Technologies and The Photonics Institute, SPMS, Nanyang Technological University Singapore 637371, Singapore.
Abstract
Nanomechanical photonic metamaterials provide a wealth of active switching, nonlinear, and enhanced light-matter interaction functionalities by coupling optically and mechanically resonant subsystems. Thermal (Brownian) motion of the nanostructural components of such metamaterials leads to fluctuations in optical properties, which may manifest as noise, but which also present opportunity to characterize performance and thereby optimize design at the level of individual nanomechanical elements. We show that nanomechanical motion in an all-dielectric metamaterial ensemble of silicon-on-silicon-nitride nanowires can be controlled by light at sub-μW/μm2 intensities. Induced changes in nanowire temperature of just a few Kelvin and nonthermal optical forces generated within the structure change the few-MHz Eigenfrequencies and/or picometric displacement amplitudes of motion, and thereby metamaterial transmission. The tuning mechanism can provide active control of frequency response in photonic metadevices and may serve as a basis for bolometric, mass, and micro/nanostructural stress sensing.
Nanomechanical photonic metamaterials provide a wealth of active switching, nonlinear, and enhanced light-matter interaction functionalities by coupling optically and mechanically resonant subsystems. Thermal (Brownian) motion of the nanostructural components of such metamaterials leads to fluctuations in optical properties, which may manifest as noise, but which also present opportunity to characterize performance and thereby optimize design at the level of individual nanomechanical elements. We show that nanomechanical motion in an all-dielectric metamaterial ensemble of silicon-on-silicon-nitride nanowires can be controlled by light at sub-μW/μm2 intensities. Induced changes in nanowire temperature of just a few Kelvin and nonthermal optical forces generated within the structure change the few-MHz Eigenfrequencies and/or picometric displacement amplitudes of motion, and thereby metamaterial transmission. The tuning mechanism can provide active control of frequency response in photonic metadevices and may serve as a basis for bolometric, mass, and micro/nanostructural stress sensing.
By virtue of their low mass
and fast (MHz–GHz frequency) response times, nanomechanical
oscillators actuated and/or interrogated by light are of fundamental
and applied interest in numerous applications, ranging from mass and
force sensors to photonic data processing and quantum ground state
measurements.[1−13] As the dimensions of such systems decrease, their thermal (i.e.,
Brownian) motion assumes increasing importance. By adding noise to
induced/controlled movements that underpin the functionality, it can
constrain performance, but it also presents opportunity by directly
linking observable (far-field optical) properties to geometry, composition,
and temperature at the nanoscale. We show here that such motion can
be optically controlled at μW/μm2 intensities
in nanomechanical photonic metamaterials. In an array of mechanically
independent and (by design) alternately dissimilar dielectric nanowires,
which are at the same time of identical bilayer (i.e., asymmetric)
material composition and part of an optically resonant ensemble subject
to the fundamental constraint of linear transmission reciprocity,
dependences of motion Eigenfrequencies and picometric displacement
amplitudes on local light-induced temperature changes can be accurately
determined. We further show that the amplitude of high-frequency oscillatory
motion driven by nonthermal optical (gradient and radiation pressure)
forces is resonantly enhanced, leading to larger changes in transmission,
when the nanowires’ natural frequencies are photothermally
tuned to coincide with the pump modulation frequency.In the
present study, we employ an all-dielectric metamaterial
comprising pairs of dissimilar (by length and width) silicon nanobricks
on a free-standing array of flexible silicon nitride nanowires (Figure a). It is fabricated
on a 200 nm thick Si3N4 membrane coated by plasma-enhanced
chemical vapor deposition with a 115 nm thick layer of amorphous Si.
This bilayer is then structured by focused ion beam milling to define
rows of alternately short, narrow (720 nm × 210 nm) and long,
wide (780 nm × 300 nm) nanobricks in the Si layer on parallel
21 μm long nanowire beams cut through the Si3N4 layer with a gap size between neighboring nanowires of 170
nm.
Figure 1
Detecting thermal (Brownian) motion of nanowires within an all-dielectric
nanomechanical metamaterial. (a) Scanning electron microscope image
of the metamaterial, fabricated on a 21 μm wide free-standing
silicon nitride membrane. The inset enlarged section shows detail
of the supported silicon nanobricks, and the dashed line denotes a
unit cell of the structure. (b) Schematic of experimental apparatus
for recording frequency spectra of metamaterial transmission. Other
than between the two collimators, light is carried in polarization-maintaining
single-mode optical fiber with the MEMS switch providing for inversion
of the light propagation direction through the sample. (c) Exemplar
measurement of optical transmission amplitude spectral density [for
light incident on the silicon nitride side of the sample at a power
level of 15.9 μW], showing a pair of peaks associated with the
mechanical resonances of two individual nanowires within the array:
①/② a narrower/wider wire decorated with shorter/longer
Si bricks. The overlaid magenta curve and calibrated displacement
spectral density scale [to the right-hand side] are obtained by fitting eq to the experimental data.
Derived values of f0, Q, and meff are shown inset.
Detecting thermal (Brownian) motion of nanowires within an all-dielectric
nanomechanical metamaterial. (a) Scanning electron microscope image
of the metamaterial, fabricated on a 21 μm wide free-standing
silicon nitride membrane. The inset enlarged section shows detail
of the supported silicon nanobricks, and the dashed line denotes a
unit cell of the structure. (b) Schematic of experimental apparatus
for recording frequency spectra of metamaterial transmission. Other
than between the two collimators, light is carried in polarization-maintaining
single-mode optical fiber with the MEMS switch providing for inversion
of the light propagation direction through the sample. (c) Exemplar
measurement of optical transmission amplitude spectral density [for
light incident on the silicon nitride side of the sample at a power
level of 15.9 μW], showing a pair of peaks associated with the
mechanical resonances of two individual nanowires within the array:
①/② a narrower/wider wire decorated with shorter/longer
Si bricks. The overlaid magenta curve and calibrated displacement
spectral density scale [to the right-hand side] are obtained by fitting eq to the experimental data.
Derived values of f0, Q, and meff are shown inset.The metamaterial structure supports a near-infrared closed
mode
optical resonance[14] at a wavelength of
1542 nm (see Supplementary Figure S1),
underpinned by the excitation of antiparallel displacement currents
in adjacent dissimilar silicon nanobricks by incident light polarized
parallel to the long axis of the bricks. In the vicinity of this optical
resonance, thermal (Brownian) motion of the nanowires, mutual positional
fluctuations of pico- to nanometric amplitude, translate to fluctuations
of metamaterial transmission (of order 0.1%) at their few MHz natural
mechanical resonance frequencies.[15,16] These thermomechanical
oscillations are detected as peaks in frequency spectra of transmission
amplitude spectral density (Figure b,c): the metamaterial is mounted in a vacuum chamber
at a pressure of 4 × 10–3 mbar to exclude air
damping of mechanical motion. (It should be noted here that while
a classical Brownian particle in a fluid is thermally perturbed by
“external” collisions with ambient atoms, the thermal
motion of objects under vacuum is driven “internally”
by momentum transfer from the annihilation, creation, and interference
of phonons.) The sample chamber is located between a confocal pair
of 20× (NA 0.4) microscope objectives, via which incident light
at a wavelength of 1550 nm is focused onto the sample (to a spot of
diameter ∼5 μm at the center of the metamaterial array)
and transmitted light is collected. The arrangement includes a fiber-optic
MEMS switch to enable transmission measurements in both directions
through the sample without disturbance of its position/alignment relative
to the beam path. (The “forward” direction of light
propagation is designated as that for which light is incident on the
Si side of the sample; “backward” is the Si3N4 side.)Figure c shows
a representative measurement of optical transmission amplitude spectral
density (ASD) in which peaks associated with the fundamental out-of-plane
flexural modes of a pair of individual nanowires, one narrow and one
wide decorated, respectively, with short and long Si nanobricks, are
seen. (Attribution to this oscillatory mode is confirmed through computational
modeling; see Supporting Information.)Nanowire displacement ASD can be expressed as[17,18]where kB is the
Boltzmann constant, T is temperature, and for each
mode meff, f0, and Q are, respectively, the effective mass, natural
frequency, and quality factor of the oscillator. Experimental data
can thus be calibrated: the vertical scale on the left of Figure c converted from
signal measured in μV to nanowire displacement in picometres
(on the right) by fitting eq to the data. Specifically, we fit a linear superposition
of two instances of the expression, one for each of the spectral peaks
with co-optimized values of f0, Q, and meff. Root mean square
(RMS) thermal motion amplitudes can then be evaluated as the square
root of an integral of power spectral density (ASD2) over
frequency. These calculations yield amplitudes of 76 and 67 pm, respectively,
for the lower and higher frequency peaks in Figure c, which compare extremely well with analytical
values of 76 and 68 pm derived from energy equipartition theorem[19]Figure shows how
the Brownian motions characteristics of nanowires,
as manifested in the ASD of optical transmission, depend upon (i.e.,
can be controlled by tuning) incident laser power, and how this dependence
differs for the two directions of incident light propagation. With
increasing laser power, mechanical Eigenfrequencies redshift (Figure a) and RMS displacement
amplitudes increase (Figure b), both in direct proportion and more rapidly for the backward
propagation direction. The behaviors are consistent with a photothermal
tuning mechanism, whereby laser-induced heating decreases tensile
stress in the nanowires. The effect is more pronounced for the backward
direction of light propagation because while forward and backward
transmission levels are identical (as they must be in a linear, reciprocal
medium) reflectivity and absorption are not (Figure S1).
Figure 2
Optical control of nanowire Eigenfrequencies. Transmission amplitude
spectral density, showing peaks ① and ② as assigned
in Figure c, for opposing
directions of light propagation through the sample (a) forward and
(b) backward [light incident respectively on the silicon and the silicon
nitride side], and for a range of laser power levels [as labeled].
Figure 3
Optical control of nanowire Eigenfrequencies and Brownian
motion
amplitudes. Dependences for peaks ① and ② in Figure [i.e., for narrow
and wide nanowires as identified in Figure c, under nominally forward and backward directions
of illumination] of (a) resonance frequency, (b) RMS displacement
amplitude, and (c) light-induced nanowire temperature change, on incident
laser power [total power incident on the metamaterial sample]. Square
symbols are experimental data points with error bars given by the
standard deviation over three repeated measurement cycles. Solid lines
are derived from an analytical description of the photothermal tuning
mechanism via a simultaneous best-fit to the four experimental data
sets in (a).
Optical control of nanowire Eigenfrequencies. Transmission amplitude
spectral density, showing peaks ① and ② as assigned
in Figure c, for opposing
directions of light propagation through the sample (a) forward and
(b) backward [light incident respectively on the silicon and the silicon
nitride side], and for a range of laser power levels [as labeled].Optical control of nanowire Eigenfrequencies and Brownian
motion
amplitudes. Dependences for peaks ① and ② in Figure [i.e., for narrow
and wide nanowires as identified in Figure c, under nominally forward and backward directions
of illumination] of (a) resonance frequency, (b) RMS displacement
amplitude, and (c) light-induced nanowire temperature change, on incident
laser power [total power incident on the metamaterial sample]. Square
symbols are experimental data points with error bars given by the
standard deviation over three repeated measurement cycles. Solid lines
are derived from an analytical description of the photothermal tuning
mechanism via a simultaneous best-fit to the four experimental data
sets in (a).An analytical model for optical
control of thermomechanical (Brownian)
motion resonances, tightly constrained by the requirement to describe
the properties of two independent, similar (related) but not identical
oscillators - narrow and wide Si3N4/Si bilayer
nanowires, under two similar (related) but not identical regimes of
optical excitation - forward and backward directions of illumination,
provides for accurate quantitative evaluation of light-induced temperature
changes in the individual nanowires and of the corresponding relationships
between their resonance frequencies/amplitudes, illumination conditions,
and the optical properties of the metamaterial array. From Euler–Bernoulli
beam theory,[20] the stress-dependent fundamental
frequency of a doubly clamped beam of homogeneous rectangular cross-section
iswhere t and L are the thickness
and length of the beam, E is
Youngs’s modulus, ρ is density, and σ is tensile
stress along the beam length. The temperature-dependence of stress
can be expressed in the form[21]where σ0 is the stress at ambient temperature, ΔT is
the difference between average beam and ambient temperatures, and
α is the beam’s thermal expansion coefficient.Assuming the presence of a heat source uniformly distributed over
the rectangular cross-section at the midpoint of the beam, while the
two ends are held at ambient temperature (T0 = 298 K), the equilibrium difference ΔT between
average beam and ambient temperatures is[21]where κ is thermal conductivity, A is the cross-sectional area, and P is
the power of the heat source.For the purpose of applying these
analytical expressions to the
present case, we approximate the silicon nitride nanowires decorated
with silicon nanobricks as simple rectangular-section beams of the
same length with effective values of E, ρ,
α, κ, and A derived from the material
parameters of Si and Si3N4 and the volume fractions
of the two materials in each nanowire type (see Supporting Information). P is taken as the
optical power absorbed by a nanowire: Pabs = μγPin, where Pin is the total power incident on the metamaterial, γ
is the absorption coefficient of the metamaterial, and μ is
the “absorption cross-section” of an individual nanowire.By substituting eqs and 5 into eq and then fitting that expression simultaneously to
all four sets of experimental data points in Figure a (see Supporting Information) under constraints thatwe obtain the four
solid curves plotted in each panel of Figure . The eigenfrequency
fitting (Figure a)
is extremely good and yields zero-power (i.e., ambient temperature)
resonant frequencies of 3.18 and 3.13 MHz for the wider and narrower
nanowires, respectively. Derived values of μ, 4.9% and 5.6%,
respectively, for the narrower and wider nanowires, are consistent
to a first approximation with their areas of geometric intersection
with the ∼5 μm diameter incident laser spot. However,
they are not in proportion simply to the ratio of nanowire widths
(0.7:1). This is because the near-field distribution of the electromagnetic
field around the metamaterial at resonance is not homogeneous, and
the absorption cross section of constituent nanowires is therefore
not expected to be directly proportional to its geometric cross-section.
Derived values of γ, 10.6% and 26.0%, respectively, for the
forward and backward directions of light propagation, correspond well
to measured (far-field) values of metamaterial absorption at 1550
nm (Figure S1a) of 13.5% and 23.7%.μ must be identical for forward
and backward directions of illumination for a given nanowire (i.e.,
the same nanowire will intercept the same fraction of incident light
in both directions);γ must be identical for the
two nanowires for a given illumination directions (i.e., as a metamaterial
ensemble property, absorption can only have a single value in each
direction);σ0 must take a single fixed value (ensuring
degeneracy of nominally
forward- and backward-illumination zero-power resonant frequencies
for each type of nanowire);Light-induced changes in nanowire temperature (Figure c) depend upon the direction
of illumination and nanowire dimensions, that is, upon the strength
of optical absorption and the rate at which heat is dissipated (the
latter being lower for the nanowire of smaller cross-section). The
narrow nanowire changes temperature at a rate of 27 K/μW of
absorbed power, and the wide nanowire at 19 K/μW. These derived
dependences of induced temperature change ΔT on laser power map to the theoretical dependences of RMS Brownian
motion amplitude presented in Figure b via eq , using values of meff for the two nanowires
established in the above (Figure c) calibration of displacement spectral density. The
picometrically accurate correlation with experimental data points
separately derived from integrals of PSD over frequency is remarkably
good.With the introduction of a second,
modulated pump laser, one can
observe in probe transmission the interplay between thermomechanical
fluctuations and nonthermal optically driven motion, that is, the
action of optical forces (Figure ). Here, we employ a probe laser at a fixed power of
22 μW at a wavelength of 1540 nm, and a pump laser at 1550 nm
electro-optically modulated at a fixed frequency of 2.97 MHz. The
beams are coincident on the sample in the forward direction (coupled
in fiber before the MEMS switch input), and a narrow bandpass filter
(at the switch output) ensures that only transmitted probe light reaches
the detector. Under these conditions, we observe again in the frequency
spectrum of probe transmission peaks relating to Brownian motion at
nanowire Eigenfrequencies dependent upon average pump power (i.e.,
pump-induced temperature change). But also a sharp peak at the pump
modulation frequency relating to structural reconfiguration, and thereby
probe transmission change, driven by nonthermal optical forces. (There
is no coherent temperature oscillation induced by the pump; modulation
period is some 2 orders of magnitude shorter than the nanowires’
thermal relaxation time.) The magnitude of this optically induced
transmission change is resonantly enhanced by more than an order of
magnitude (reaching ∼30%) when the pump modulation frequency
coincides with a nanowire’s mechanical eigenfrequency (as shown
in the inset in Figure ).
Figure 4
Optical control of nanomechanical motion: thermal and nonthermal
mechanisms. Amplitude spectral density of probe transmission (in the
forward direction) for a range of pump peak power levels with a fixed
pump modulation frequency.
Optical control of nanomechanical motion: thermal and nonthermal
mechanisms. Amplitude spectral density of probe transmission (in the
forward direction) for a range of pump peak power levels with a fixed
pump modulation frequency.In summary, we have shown that fluctuations in the resonant optical
properties of a photonic metamaterial, which are associated with the
mechanically resonant Brownian motion of its constituent elements,
can be controlled by light at sub-μW/μm2 intensities.
In an all-dielectric metamaterial ensemble of free-standing silicon
nitride nanowires (mechanical oscillators) supporting an array of
silicon nanobricks (optical resonators), the few MHz Eigenfrequencies
and picometric amplitudes of individual nanowires’ motion are
directly proportional to incident laser power, changing, respectively,
in consequence of light-induced heating by up to 0.7% and 0.9% per
K.An analytical model for the photothermal tuning mechanism,
simply
but effectively constrained by the requirements of optical transmission
reciprocity in a linear medium, links the local, nanoscopic properties
and behaviors of individual nanowires (i.e., at subwavelength scale)
to the far-field optical properties of the (micro/macroscopic) metamaterial
ensemble. It provides for accurate evaluation of light-induced changes
in nanowire temperature and of ambient condition (zero-illumination)
Brownian motion Eigenfrequencies and displacement amplitudes.We further illustrate how nonthermal optical forces generated at
the near-IR optical resonance of the metamaterial structure, again
at sub-μW/μm2 intensities, can be engaged to
drive motion at its (photothermally tuned) mechanical Eigenfrequencies
and thus to deliver strong light-induced changes in transmission.The ability to finely tune the nanomechanical resonance characteristics
of photonic metamaterials may be beneficial in a variety of metadevice
applications where, for example, the frequency of nanostructural oscillation
is required to match (or avoid matching) another frequency, such as
that of a pulsed laser. The fact that tuning characteristics can (as
here) depend strongly upon the direction of light propagation through
a metamaterial by simple virtue of bilayer material composition (leading
to different levels of reflection and absorption for light incident
on opposing sides) may find application in devices to favor/select
a single direction of propagation. The accurately quantifiable sensitivity
of optical response to nanomechanical properties in such structures
also suggests applications to bolometric sensing and detection of
changes in mass (e.g., through adsorption/desorption) or micro/nanostructural
stress.
Authors: A D O'Connell; M Hofheinz; M Ansmann; Radoslaw C Bialczak; M Lenander; Erik Lucero; M Neeley; D Sank; H Wang; M Weides; J Wenner; John M Martinis; A N Cleland Journal: Nature Date: 2010-03-17 Impact factor: 49.962