UBC Faculty Research and Publications

Integration of photonic nanojets and semiconductor nanoparticles for enhanced all-optical switching Born, Brandon; Krupa, Jeffrey D. A.; Geoffroy-Gagnon, Simon; Holzman, Jonathan F. Aug 28, 2015

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


52383-Born_B_et_al_Nanojets_enhanced_switching.pdf [ 1.59MB ]
JSON: 52383-1.0319008.json
JSON-LD: 52383-1.0319008-ld.json
RDF/XML (Pretty): 52383-1.0319008-rdf.xml
RDF/JSON: 52383-1.0319008-rdf.json
Turtle: 52383-1.0319008-turtle.txt
N-Triples: 52383-1.0319008-rdf-ntriples.txt
Original Record: 52383-1.0319008-source.json
Full Text

Full Text

ARTICLEReceived 14 Jan 2015 | Accepted 13 Jul 2015 | Published 28 Aug 2015Integration of photonic nanojets andsemiconductor nanoparticles for enhancedall-optical switchingBrandon Born1, Jeffrey D.A. Krupa1, Simon Geoffroy-Gagnon1 & Jonathan F. Holzman1All-optical switching is the foundation of emerging all-optical (terabit-per-second) networksand processors. All-optical switching has attracted considerable attention, but it mustultimately support operation with femtojoule switching energies and femtosecond switchingtimes to be effective. Here we introduce an all-optical switch architecture in the form of adielectric sphere that focuses a high-intensity photonic nanojet into a peripheral coating ofsemiconductor nanoparticles. Milli-scale spheres coated with Si and SiC nanoparticles yieldswitching energies of 200 and 100 fJ with switching times of 10 ps and 350 fs, respectively.Micro-scale spheres coated with Si and SiC nanoparticles yield switching energies of 1 pJ and20 fJ with switching times of 2 ps and 270 fs, respectively. We show that femtojouleswitching energies are enabled by localized photoinjection from the photonic nanojets andthat femtosecond switching times are enabled by localized recombination within the semi-conductor nanoparticles.DOI: 10.1038/ncomms9097 OPEN1 School of Engineering, The University of British Columbia, Okanagan campus, Kelowna, British Coloumbia, Canada, V1V 1V7. Correspondence and requestsfor materials should be addressed to B.B. (email: brandon.born@ubc.ca) or to J.F.H. (email: jonathan.holzman@ubc.ca).NATURE COMMUNICATIONS | 6:8097 |DOI: 10.1038/ncomms9097 | www.nature.com/naturecommunications 1& 2015 Macmillan Publishers Limited. All rights reserved.Logic operations are at the core of data processing, andswitching is the key mechanism for logic operations.Switching is typically realized via electronics, withtransistors relaying the flow of data, but there is pressure tointroduce switches that can operate at optical fibre transmissionrates1,2. The terabit-per-second rates of optical fibres exceed thegigabit-per-second rates of electronics—and the all-optical switch(AOS) has been proposed to alleviate the ensuing optoelectronicbottleneck in front-end networks3,4.The AOS is a photonic analogue to the electronic transistor, asit enables nonlinear mixing of optical beams, but all-opticalswitching must be implemented with consideration to demandsfor low switching energies and ultrafast switching times5–7. ManyAOS technologies have been proposed in contemporary literatureto address these demands7–11.Early AOS studies focused on the development of semi-conductor materials for ultrafast switching times—ideally on afemtosecond level. Charge-carrier photoinjection and recombina-tion were applied in semiconductors having simple planartopologies, that is, bulk wafers, so the all-optical switching timeswere limited by bulk recombination lifetimes. It was later foundthat semiconductors with ultrashort lifetimes could establishultrafast switching between coincident control (pump) andsignal (probe) beams. For example, Ganikhanov et al.12 andGupta et al.13 demonstrated ultrafast all-optical switching usingion-implanted and low-temperature-grown GaAs, respectively.Overall, ultrafast switching times could be achieved withthese material systems, but the use of simple focusing insemiconductors was found to limit the beam intensities,necessitating high switching energies14.More recent AOS studies have focused the development ofdevice geometries for low switching energies—ideally on afemtojoule level. Increased beam intensities, and enhancedmixing of beams, have been introduced through many (typicallyresonant) devices. Almeida et al.9 demonstrated micrometer-scaleAOS devices as ring resonators. Nozaki et al.10 demonstratednanometer-scale AOS devices as photonic crystal resonators.Volz et al.11 demonstrated atomic-scale AOS devices as quantumdots. Such resonant devices support low switching energies, ingeneral, but their implementation must consider the prolongedswitching times that are inherent to resonant cavity lifetimes15.In this study, the demands of all-optical switching, being AOSactivation with femtojoule switching energies and AOS recoverywith femtosecond switching times, are addressed through thedevelopment of a non-resonant AOS architecture. A nanojetfocal geometry is introduced, in the form of a dielectric sphere, tofocus high-intensity photonic nanojets into a semiconductornanoparticle material system that coats the sphere. It is found thatthe localized photoinjection of charge carriers by the nanojetsupports AOS activation with femtojoule switching energies,and the localized recombination of charge carriers in thesemiconductor nanoparticles supports AOS recovery withfemtosecond switching times. A milli-scale AOS architecture isdemonstrated—in support of emerging optical fibre front-endsystems, such as optical time-division multiplexing16, orthogonalfrequency-division multiplexing17, and other all-opticalmultiplexing systems. A micro-scale AOS architecture isdemonstrated—in support of future on-chip optical processors,such as network-on-a-chip systems3.ResultsSemiconductor charge-carrier dynamics. Semiconductor chargecarriers that are photoinjected by a pump beam can be usedto modulate a coincident probe beam. The pump-inducedmodulation to the probe beam’s total transmission, T, is a resultof perturbations to the surface transmission, Ts, and bulktransmission, Tb. The result is seen as a transient on the totaldifferential transmission, DT(t)/TEDTs(t)/TsþDTb(t)/Tb, withopposing polarities for the surface and bulk contributions.According to Drude theory, the pump-induced generation of thecharge-carrier density, N(z,t), increases the differential surfacetransmission, DTs(t)/Ts, and decreases the differential bulktransmission, DTb(t)/Tb. This occurs because N(z,t) decreasesthe refractive index near the surface, Dns(t), and increases theabsorption coefficient in the bulk, Dab(t), which modulates thetotal differential transmission according to18DT tð ÞT¼  ns 1ns nsþ 1ð ÞDns tð Þ dzDab tð Þ; ð1Þwhere ns is the refractive index near the surface, and dz is thephotoinjection depth of charge carriers into the bulk.Both Dns(t) and Dab(t) evolve in proportion to the charge-carrier density, N(z,t), which varies through time, t, along thenormal dimension to the semiconductor surface, z, according to@N z; tð Þ@t¼ N0d tð Þe z=dz  N z; tð Þtb þDr2N z; tð Þ; ð2Þwhere the three terms on the right-hand side characterize therespective processes of photoinjection, recombination, anddiffusion. Photoinjection from an ultrashort laser pulse is appliedby way of the time-varying delta function, d(t). The initialdistribution for N(z,t) is an exponential decay into thesemiconductor, with an initial charge-carrier density of N0 atthe surface and a 1/e photoinjection depth of dz into the bulk. Thephotoinjection depth, dz, is defined here as a general parameterthat can be less than or equal to the semiconductor’s penetrationdepth. Given a suitable focal geometry, with tight focusing and ashort depth of focus penetrating into the semiconductor, dz canbe made to be less than the semiconductor’s penetrationdepth. The boundary condition for N(z,t) at the semiconductorsurface adheres to DqN(z¼ 0,t)/qz¼ SvN(z¼ 0,t), for a diffusioncoefficient, D, and surface recombination velocity, Sv. Given asuitable material system, with a high density of surface states, thisboundary condition can reduce the overall charge-carrier lifetime,t, to a value below the bulk charge-carrier lifetime, tb.Localized photoinjection can enable AOS activation withfemtojoule switching energies, by minimizing the pump pulseenergy, Ep, that is needed to establish the initial charge-carrierdensity, N0¼EpZ/(:opV), in the first term of equation (2). HereZ is the internal quantum efficiency, : is the reduced Planck’sconstant, op is the angular frequency of the pump beam, andVEA|dz is the photoinjection volume set by the pump beam’scross-sectional photoinjection area, A|, and photoinjection depth,dz. In this study, localized photoinjection is applied by way of ananojet focal geometry, to yield a reduced focal spot size, that is,photoinjection area, and reduced depth of focus, that is,photoinjection depth. This leads to reduced photoinjectionvolumes and high initial charge-carrier densities.Localized recombination can enable AOS recovery withfemtosecond switching times, by promoting surface recombina-tion and thus minimizing the overall charge-carrier lifetime, t, asa result of equation (2) and its boundary condition. The charge-carrier lifetime simplifies to the recombination rate relation,1/t¼ 1/tbþ SvR. An increase in the surface recombinationvelocity, Sv, or the specific surface area, that is, surface-to-volumeratio, R, has t decrease below the bulk charge-carrier lifetime, tb.Our prior work has shown that nanoscale cylindrical19 andspherical20 semiconductor forms exhibit this trend, as anincreased surface-to-volume ratio, R, increases the surface statedensity and decreases the charge-carrier lifetime. In this study,localized recombination is applied by way of a nanoparticleARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms90972 NATURE COMMUNICATIONS | 6:8097 | DOI: 10.1038/ncomms9097 | www.nature.com/naturecommunications& 2015 Macmillan Publishers Limited. All rights reserved.material system. It is typically advantageous to apply both amaterial system with an increased surface state density and a focalgeometry that localizes photoinjection in the region of highsurface state density—and this two-fold approach is taken here.The all-optical switching results of this study are defined withrespect to nominal tests using a standard focal geometry with amicroscope objective and targets comprised of the well-established semiconductors, semi-insulating GaAs, float-zone Siand 6H-SiC, with respective wafer thicknesses of 350, 280 and330mm. These semiconductors are particularly advantageous forthis study, given their wide range of bulk lifetimes and surfacerecombination velocities.Figure 1a shows the pump–probe experimental set-up, with thecorresponding details given in the Methods section. The GaAsand Si targets use pump and probe pulses with respectivewavelengths of 780 and 1,550 nm. The SiC target uses pump andprobe pulses with respective wavelengths of 390 and 1,550 nm.Figure 1b shows the experimental differential transmission,DT/T, for the three semiconductor targets. The negative signalpolarities indicate that pump-induced modulation to the probebeam is dominated by absorption in the bulk for all threesemiconductor targets, which agrees with the observed long bulklifetimes. GaAs exhibits a picosecond-scale bulk lifetime oftb¼ 280 ps, Si exhibits a nanosecond-scale bulk lifetime21 oftb¼ 20 ns and SiC exhibits a microsecond-scale bulk lifetime22 oftb¼ 20ms. The fast transient at the onset of the SiC response isdue to charge-carrier scattering. Details on the SiC charge-carrierdynamics are given in the Methods section. All three materialsexhibit unacceptably slow recovery times for AOS operation, asminimal contributions are seen from surface recombination inthe GaAs, Si and SiC targets, despite their wide range of surfacerecombination velocities, Sv¼ 1.2 104m s–1, Sv¼ 500m s–1 andSv¼ 4,000m s–1, respectively20,23.The nanojet focal geometry. Localized photoinjection isconsidered in this section, for enhanced AOS activation, by wayof a nanojet focal geometry. A high-intensity photonic nanojet24can be formed by focusing through an appropriately designeddielectric sphere25,26. The tight transverse constriction andshallow protrusion formed by photonic nanojet focusing, justbeyond the sphere, are well suited to the application of localizedphotoinjection into a peripheral semiconductor. The resultingsmall photoinjection area, A|, and depth, dz, of the pump beam inthe semiconductor establish a reduced photoinjection volume,VEA|dz. This leads to a large initial charge-carrier density, N0,and a small charge-carrier lifetime, t, if the photoinjection ispreferentially applied at a semiconductor surface. The nanojetfocal geometry must be implemented with careful considerationto the properties of the sphere, however, as there is an inherentrelationship between the photonic nanojet’s intensity and thesphere’s diameter, d, and refractive index, n. With this in mind,the properties of the sphere that yield the optimal nanojet arestudied by way of theoretical analyses across two regimes of scale.In the milli-scale regime, with sphere diameters that are muchlarger than the wavelength, focusing is well-described by Raytheory. Figure 2 shows the results from Ray theory applied to thespherical geometry. The intensity at the exit interface of thesphere, intersecting with the optical axis, is displayed as afunction of the sphere refractive index, 1.5ono2 for diametersd4100mm. The curve shows that high-intensity focusing at theexit interface of the sphere is brought about the refractive index ofnE2. An intensity colourmap for this milli-scale regime isgenerated from the Ray theory curve and is shown at the top ofthe figure. The peak intensity for nE2 is shown in white. Thesefindings agree with limiting case of a ball lens in the thick-lensformula27, which predicts focusing of light rays at the exitinterface for nE2.In the micro-scale regime, with sphere diameters that arecomparable to the wavelength, it is necessary to implement arigorous three-dimensional electromagnetic analysis. This can bedone efficiently for a spherical geometry using Mie theory.Figure 3 shows the results from Mie Theory analyses applied tothe spherical geometry. The work presented is based on thealgorithm26 of Lecler et al. Mie theory simulations are carried outto identify the maximum intensities at the exit interface of thesphere, intersecting with the optical axis, as a function of thesphere refractive index, n, and diameter, d. The full distribution ofintensities from Mie theory is shown for the micro-scale regimeas an intensity colourmap map in Fig. 3 for do30 mm and1.5ono2. The trendline for maximum intensities is shown as asolid blue curve. The curve rises from a low refractive index atsmall diameters, nE1.75, toward the Ray theory result at largediameters, nE2. The highest intensities are seen as a broad bandDelay stageDelay780-nm pump beamBandpass filter*DichroicbeamsplitterFocusingelement**SemiconductortargetInGaAs detectorwith bandpass filter1550-nm probe beamBulk target40x0–0.2–0.4–0.6–0.8–1.0–1.2–1.4–1.60 50 100 150 200 250 300 350Time (ps)SHG crystal*ΔT/TGaAs bulk b= 280 psSiC bulk b= 22 μsSi bulk b= 20 nsFigure 1 | Pump–probe experimental set-up and nominal AOS results. (a) Schematic of the pump–probe experimental set-up. A 780-nm pump beamis used for the GaAs and Si targets and a 390-nm pump beam is used for the SiC target. A 1,550-nm probe beam is used for all the targets, and it isre-collimated and focused on the InGaAs detector using two 25-mm-focal length lenses. *A second harmonic generation (SHG) crystal and bandpass filterare placed in the beam path to form the 390-nm pump beam (at a conversion efficiency of 10%). **The focusing element takes the form of a microscopeobjective or dielectric spheres. (b) Experimental differential transmission of the probe beam, DT/T, is shown normalized versus time, for pumpphotoinjection of the bulk Si, SiC and GaAs targets, using a microscope objective as the focusing element, as depicted in the inset.NATURE COMMUNICATIONS | DOI: 10.1038/ncomms9097 ARTICLENATURE COMMUNICATIONS | 6:8097 |DOI: 10.1038/ncomms9097 | www.nature.com/naturecommunications 3& 2015 Macmillan Publishers Limited. All rights reserved.of intensities that peak at nE1.75 for small d then rise towardnE2 for sufficiently large d, where the Mie theory and Ray theoryresults merge.Given the optical responses of Fig. 3 and the requirements forall-optical switching, it is worth commenting on the potential forcontributions from resonance and spherical aberration.Resonance manifests itself as two patterns of interferometricfringes in the intensity colourmap map for the micro-scale regimein Fig. 3. The first fringe pattern comes about from longitudinalmodes that resonate along the optical axis and form the steephigh-spatial-frequency fringes in the intensity map. The secondfringe pattern comes about from coupled longitudinal andwhispering-gallery modes28, that exhibit interferometric beatingalong the perimeter and form the sloped low-spatial-frequencyfringes in the intensity map. It is worth noting that resonance canbe avoided, however, if there is a desire for reduced sensitivity tostructural and thermal fluctuations, by applying laser pulses withsufficiently short durations (being much shorter than the cavitylifetime) or by applying spheres with sufficiently large diameters(being at the upper end of the micro-scale regime or anywherewithin the milli-scale regime).Spherical aberration manifests itself as imperfect focusing atthe exit interface of the sphere. This leads to an increased demandon the switching energy, as the incident beam power must beincreased to compensate for the reduced intensity. Note thatspherical aberration can be compensated, to some degree, in thata sphere, being approximated as a plane-parallel plate betweentwo plano-convex lenses, will have spherical aberration beover-corrected by the plane-parallel plate and under-correctedby the lenses29. In general, spherical aberration decreases as therefractive index increases. Using Ray theory, it can be shown thata sufficiently high refractive index sphere, with nE2, has a 42%reduction in the (transverse) spherical aberration, compared withthat of a sphere with a refractive index of nE1.5.The measured differential transmission, DT/T, is shown inFig. 4, as a function of time, for the spheres having refractiveindices of (a) n¼ 1.51, (b) n¼ 1.76, (c) n¼ 1.83 and (d) n¼ 1.98.The experimental curves are shown with correspondingtheoretical curves, having been generated by a Drude/charge-carrier dynamical model. The model couples the Drude theorycharacteristics for probe beam transmission from equation (1)with the charge-carrier dynamics of equation (2). Mie theorysimulations are seen in the figure insets.1.5 1.6 1.7 1.8 1.9 2.0 2.1 2.2 2.3 2.4 2.510–310–210–1100Intensity colourmapRefractive index (n)Intensity (arb. units)Figure 2 | Photonic nanojet intensity in the milli-scale regime. Thetheoretical intensity of a photonic nanojet, calculated with Ray theory at theexit interface of a dielectric sphere, is shown as a function of the sphere’srefractive index, n. An intensity colourmap of Ray theory simulations in themilli-scale regime (for diameters of d4100mm) is shown at the top of thefigure.n = 1.76 spheren = 1.83 sphere0 2015105 25 302. theorybaRefractive index, nDiameter, d (μm)Figure 3 | Photonic nanojet intensity in the micro-scale regime. The theoretical intensity of a photonic nanojet, calculated with Mie theory at the exitinterface of a dielectric sphere, is shown as a function of the sphere’s refractive index, n, and diameter, d, for pump photoinjection at a wavelength of780 nm. A trendline for the maximum intensity is shown as a solid grey curve. An intensity colourmap of Mie theory simulations in the micro-scale regimeis shown on the left side of the figure—being the result of over a hundred thousand individual Mie theory simulations. The maximum (white) intensity isnormalized for all n with constant d. Select 3D Mie theory simulations, with logarithmic intensities, are shown in the figure insets (a) a sphere withd¼ 30mm and n¼ 1.83, and (b) a sphere with d¼ 3mm and n¼ 1.76.ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms90974 NATURE COMMUNICATIONS | 6:8097 | DOI: 10.1038/ncomms9097 | www.nature.com/naturecommunications& 2015 Macmillan Publishers Limited. All rights reserved.It is apparent from the experimental and theoretical results ofFig. 4 that increasing sphere refractive indices yield increasingcharge-carrier densities at the GaAs surface. This manifests itselfby the observed polarities of the differential transmission curves.The negative curve of the n¼ 1.51 sphere exhibits decreased probetransmission from pump-induced changes primarily to theabsorption coefficient in the bulk—like that seen for the nominalsemiconductor tests with a microscope objective. In contrast, theincreasingly positive curves of the n¼ 1.76, 1.83 and 1.98 spheresexhibit increased probe transmission from pump-induced changesprimarily to the refractive index near the surface. For increasingsphere refractive indices, the preferential deposition of chargecarriers near the semiconductor surface is also seen by way ofreducing charge-carrier lifetimes. Charge-carrier lifetimes oft¼ 210, 120, 60 and 10 ps are measured for the sphere refractiveindices of n¼ 1.51, 1.76, 1.83 and 1.98, respectively. Thedecreasing charge-carrier lifetimes come about from a diminishingdepth of focus and the corresponding preferential deposition ofcharge carriers at the semiconductor surface—where there exists ahigh surface state density and appreciable surface recombinationvelocity, Sv¼ 1.2 104m s–1. Curve-fitting of the Drude/charge-carrier dynamical model results to the experimental resultsconfirms that the increasing sphere refractive indices reduce boththe pump beam’s photoinjection area, A|, and photoinjectiondepth, dz. For sphere refractive indices of n¼ 1.51, 1.76, 1.83 and1.98, the respective photoinjection depths into the semiconductordecrease, according to dz¼ 480, 300, 180 and 70 nm, and this leadsto an increase in the initial charge-carrier densities.Given these experimental findings, it is apparent that a nanojetfocal geometry, with a sphere diameter of d¼ 2.0mm andrefractive index of n¼ 1.98, can support all-optical switching withfemtojoule switching energies and picosecond switching times.The implementation tested here, with a GaAs target beyond thesphere, yields an B10-fJ switching energy (defined for a unitysignal-to-noise ratio) and a 10-ps switching time. It is worthnoting, however, that further reductions in the switching timecan be met by enhancing surface recombination, and suchenhancements are explored in the following section.The nanoparticle material system. Localized recombination isconsidered in this section, for enhanced AOS recovery, byintroducing a semiconductor nanoparticle material system. Theintroduction of semiconductor nanoparticles increases thespecific surface area to its highest theoretical limit, being R¼ 12/dfor a sphere. This maximization of surface-to-volume hastens thecharge-carrier recombination rate, according to 1/t¼ 1/tbþ SvR.The charge-carrier lifetime, t, can ideally be reduced to a valuewell below the bulk semiconductor lifetime, tb.The nanoparticle material system is integrated with the nanojetfocal geometry, by coating nanoparticles over the entire surface ofthe dielectric sphere, to form the complete AOS architecture. AnAOS architecture such as this enables ease of alignment, in thatthe geometry inherently overlaps the coincident pump and probefoci within the nanoparticles at the exit interface of the sphere,and it also enables omni-directionality, in that the coincident210 ps× –0.86 0 50 100 150 200 250 300 350 0 50 100 150 200 250 300 3500Time (ps)120 psn = 1.51 sphere n = 1.76 sphere× 0.22 0Time (ps)0 50 100 150 200 250 300 350 0 50 100 150 200 250 300 350Time (ps) Time (ps)ΔT/TΔT/TΔT/TΔT/Tn = 1.83 sphere60 ps× 0.33 0n = 1.98 sphere10 ps× 1.0 0Figure 4 | All-optical switching with varying sphere refractive indices. Theoretical and experimental differential transmission curves of the probe beam,DT/T, are shown normalized versus time, for photoinjection into a GaAs target, using spheres with a diameter of d¼ 2mm and refractive indices of(a) n¼ 1.51, (b) n¼ 1.76, (c) n¼ 1.83 and (d) n¼ 1.98±0.02. The differential transmission results are shown normalized, with respect to the results in (d)and the relative scaling factors are labelled in the figures. The figures include theoretical curves from the Drude/charge-carrier dynamical model based onequations (1) and (2). Mie theory simulations are shown in the insets to illustrate the varying focal conditions of the coincident pump and probe beams.NATURE COMMUNICATIONS | DOI: 10.1038/ncomms9097 ARTICLENATURE COMMUNICATIONS | 6:8097 |DOI: 10.1038/ncomms9097 | www.nature.com/naturecommunications 5& 2015 Macmillan Publishers Limited. All rights reserved.pump and probe beams can be incident over the full 4p steradianssolid angle of the sphere.The Si and SiC nanoparticles that are used have approximatediameters of 20 and 50 nm, respectively. (GaAs nanoparticlesare not used, due to their high toxicity and limited knowledge forsafe laser excitation and handling30,31.) The nanoparticles arecoated according to the sample preparation steps outlined in theMethods section. The nanoparticle sizes are chosen to be an orderof magnitude smaller than the pump and probe wavelengths, tominimize scattering. Moreover, the thickness of the nanoparticlecoating is kept at or below the length of the photonic nanojet, tomitigate scattering of beams beyond the photonic nanojet.Scattering, if present, would be of greatest concern for theprobe beam, as this beam must enter, exit and propagate wellbeyond the sphere, but it is found that the long wavelength of the1,550-nm probe beam leads to minimal scattering32. Scatteringfor the 390 and 780-nm pump beams is also found to benegligible, given that the pump beam only propagates over thediameter of the nanoparticle-coated sphere, before it initiatesswitching of the probe beam.The spheres used in the AOS architecture must be selected withcareful consideration to their refractive index, n, and diameter, d.A high-intensity focus is required at the exit interface of thesphere, and this intensity depends on both n and d, according toFig. 3. Spheres in the milli-scale regime, with diameters rangingfrom 0.1 to 2.0mm, will simply require that the sphere have arefractive index of nE2. (Such a range is defined for applicabilityand compatibility on the scale of optical fibres.) Spheres in themicro-scale regime, with diameters ranging from 1 to 30 mm, willrequire the sphere to have a refractive index along the blue curveof Fig. 3, between nE1.75 and nE1.83. (Such a range is definedfor applicability and compatibility on the scale of network-on-chip devices.)Given the considerations above for nanoparticles and dielectricspheres, the AOS architecture is tested with spheres in both themilli- and micro-scale regimes using nanoparticle coatings ofboth Si and SiC.Experimental results for the milli-scale AOS architecture areshown in Fig. 5. Figure 5a,b show results for Si and SiC nanoparticle-coated spheres, respectively. The spheres have a diameter of d¼ 2.0mm, being at the upper limit of the milli-scale regime, and arecomprised of S-LAH79 glass, with a refractive index ofn¼ 1.98±0.02 The figure insets show a Mie theory simulation andscanning electron microscope (SEM) image of the nanoparticles.Figure 5a shows the (positive) differential transmission, DT/T,for all-optical switching between pump and probe beams, with Sinanoparticles. The applied Si nanoparticles have a relatively lowsurface recombination velocity, Sv¼ 500m s–1, but a noteworthydecrease in AOS recovery time is still observed—being approxi-mately two thousand times faster than that of bulk Si. This rapidrecovery is due to the promotion of surface recombinationfrom the increased specific surface area of the nanoparticles.Ultimately, the milli-scale AOS architecture with the d¼ 2.0-mmsphere and Si nanoparticles yields a switching time of 10 ps andswitching energy estimated to be 200 fJ.Figure 5b shows the (positive) differential transmission, DT/T,for all-optical switching between pump and probe beams, withSiC nanoparticles. The applied SiC nanoparticles have a relativelyhigh surface recombination velocity, Sv¼ 4,000m s–1, so adramatic decrease in recovery time is observed—being approxi-mately sixty million times faster than that of bulk SiC. (Detailedanalyses and interpretations of the SiC charge-carrier dynamicsare given in the Methods Section.) Ultimately, the milli-scale AOSarchitecture with the d¼ 2.0-mm sphere and SiC nanoparticlesmeets the demands for all-optical switching, as it yields aswitching time of 350 fs and switching energy estimated tobe 100 fJ.Experimental results for the micro-scale AOS architecture areshown in Fig. 6. Figure 6a,b show results for Si and SiCnanoparticle-coated spheres, respectively. The sphereshave a diameter of d¼ 30–40 mm, being at the upper limit ofthe micro-scale regime, and are comprised of N-LASF9glass, with a refractive index of n¼ 1.83±0.02 The figure insetsshow a Mie theory simulation and an SEM image of thenanoparticles.Figure 6a shows the (positive) differential transmission, DT/T,for all-optical switching between pump and probe beams, with Sinanoparticles. The applied Si nanoparticles again produce anoteworthy decrease in the AOS recovery time—being even fasterthan that of the d¼ 2.0-mm sphere in Fig. 5a. The speedenhancement for this micro-scale AOS architecture is attributedto its reduced focal spot area, compared with that of themilli-scale AOS architecture. The reduced focal spot area allowsthe micro-scale AOS architecture to avoid aggregates of larger,that is, 100þ nm, particles that are seen to be interspersed amongthe 20-nm nanoparticles in SEM images. The milli-scale AOSarchitecture, with its larger focal spot area is unable to avoid theaggregates of larger particles, which have a smaller specific arean = 1.98 sphere10 psSinanoparticles0 5 10 15 = 1.98 sphere350 fs0 1 2 3SiCnanoparticlesTime (ps)0ΔT/T0.ΔT/TTime (ps)Figure 5 | Milli-scale AOS architecture with Si and SiC nanoparticles. Experimental (positive) differential transmission for the probe beam, DT/T, isshown normalized versus time, for pump photoinjection of (a) Si nanoparticles and (b) SiC nanoparticles. The nanoparticles coat spheres with a diameterof d¼ 2.0mm and refractive index of n¼ 1.98±0.02. The figures include insets with Mie theory simulations and SEM images of nanoparticles on a200-nm scale.ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms90976 NATURE COMMUNICATIONS | 6:8097 | DOI: 10.1038/ncomms9097 | www.nature.com/naturecommunications& 2015 Macmillan Publishers Limited. All rights reserved.and longer charge-carrier lifetime, and this leads to its longerrecovery time. Ultimately, the micro-scale AOS architecturewith the d¼ 30–40-mm sphere and Si nanoparticles yields aswitching time of 2 ps and switching energy estimated tobe 1 pJ.Figure 6b shows the (positive) differential transmission, DT/T,for all-optical switching between pump and probe beams, withSiC nanoparticles. The applied SiC nanoparticles produce adramatic decrease in AOS recovery time—being even faster thanthat of the d¼ 2.0-mm sphere in Fig. 5b. The speed enhancementfor this micro-scale AOS architecture is brought about from itsreduced focal spot area, as stated above. Ultimately, the micro-scale AOS architecture with the d¼ 30–40-mm sphere and SiCnanoparticles meets the demands for all-optical switching, as ityields a switching time of 270 fs and switching energy estimatedto be 20 fJ.DiscussionAn AOS architecture was introduced in this work. A guidingprinciple of localization was used in the design and analysis tofacilitate AOS activation with femtojoule switching energies andAOS recovery with femtosecond switching times. The AOSarchitecture applied a nanojet focal geometry, in the form of adielectric sphere, to focus high-intensity beams into a nanopar-ticle material system, in the form of a coating of semiconductornanoparticles. The AOS architecture was implemented on milli-scale dimensions, with Si and SiC semiconductor nanoparticles,to yield respective switching energies of 200 and 100 fJ, withrespective switching times of 10 ps and 350 fs. The AOSarchitecture was then implemented on micro-scale dimensions,with Si and SiC semiconductor nanoparticles, to yield respectiveswitching energies of 1 pJ and 20 fJ, with respective switchingtimes of 2 ps and 270 fs.It was found that the AOS architecture could establish localizedphotoinjection of charge carriers, to enable femtojoule switchingenergies, and it could establish localized recombination of chargecarriers, to enable femtosecond switching times. The AOSarchitecture met practical considerations for coupling, includingbeam alignment, directionality and capture cross-section, as wellas practical considerations for stability, including physical andtemperature sensitivity. Future applications of the AOS archi-tecture may be implemented with consideration to coupling ofthe spheres to waveguides and/or fibres. Optical micro-electro-mechanical systems packaging can facilitate this integration,by way of on-chip spherical mounts based on V-grooves33,34,micropits35,36 or suspended microstructures37–39. Futureapplications of the AOS architecture may also include cascadedimplementations, realized as daisy-chained spheres40, whichare particularly challenging for AOS devices41, or parallelimplementations42, which offer omni-directionality withmultiple inputs/outputs. The proposed AOS architecture can bea building block for these future applications.MethodsExperimental set-up. The pump–probe experimental configuration, seen inFig. 1a, utilizes an erbium-doped fibre laser (Toptica Photonics, Inc.), with a 100-fslaser pulse duration, a 90-MHz repetition rate, and wavelengths of 780 and1,550 nm for the respective pump and probe beams. The differential transmission,DT/T, of the probe beam is measured by time-resolved sampling as a function ofthe time delay between pump and probe pulses. The data are acquired with a lock-in amplifier (SR830) using a 100-ms time constant. The GaAs and Si targets usepump and probe pulses with respective wavelengths of 780 and 1,550 nm (that is,photon energies of 1.55 and 0.78 eV). The SiC target uses pump and probe pulseswith respective wavelengths of 390 and 1,550 nm (that is, photon energies of 3.18and 1.55 eV). The 780–1,550-nm pump–probe experiments with d¼ 2.0-mmspheres use a pump–probe power ratio of 1:2. The 390–1,550-nm pump–probeexperiments with d¼ 2.0-mm spheres use a pump–probe power ratio of 1:20.The 780–1,550-nm pump–probe experiments with d¼ 30–40-mm spheres use apump–probe power ratio of 3:1. The 390–1,550-nm pump–probe experimentswith d¼ 30–40-mm spheres use a pump–probe power ratio of 1:1. For thesepump–probe conditions, the pump photon energy is above-bandgap and the probephoton energy is below-bandgap. This leads to probe-power-independence for allreported results, that is, multi-photon contributions from the probe beam are notobserved.A tabletop pump–probe experimental set-up such as this characterizes theimpulse response of the AOS architecture. For future applications, seeking compactdevice profiles and ultrashort laser pulses, the AOS architecture can be integratedwith a sub-millimetre ultrashort pulsed laser, such as a mode-locked laser diode43,with the output response of such a device being a convolution of the impulseresponse measured in this work and the applied ultrashort laser pulse.Sample preparation. Experimental results are collected for focusing throughspheres into a GaAs target and into nanoparticle coatings.For the experiments that focus into a GaAs target, the spheres have a diameterof d¼ 2.0mm. They are mounted in the pump–probe experimental set-up bybonding the spheres to a 22-gauge needle tip with ultraviolet-curable polymer.The GaAs target is positioned in the focal plane of the spheres by collimating theback-reflected beams, as collimated back-reflected beams indicate that the reflectiveGaAs target is positioned in the focal plane of the sphere. Ultimately, the sphereswith refractive indices of n¼ 1.51, 1.76, 1.83 and 1.98 will be separated from theGaAs target by 480, 160, 100 and o1 mm, respectively. The respective sensitivitiesfor these positions, being 1.1, 0.7, 0.6 and o0.5 mm, are dictated by each sphere’sdepth of focus.n = 1.83 sphere2 psSinanoparticles0 5 10 15 2000. (ps)n = 1.83 sphere270 fs0 1 2 3SiCnanoparticlesΔT/T00.ΔT/TTime (ps)Figure 6 | Micro-scale AOS architecture with Si and SiC nanoparticles. Experimental (positive) differential transmission for the probe beam, DT/T,shown normalized versus time, for pump photoinjection of (a) Si nanoparticles and (b) SiC nanoparticles. The nanoparticles coat spheres with a diameterof d¼ 30–40mm and refractive index of n¼ 1.83±0.02. The figures include insets with Mie theory simulations and SEM images of nanoparticles on a200-nm scale.NATURE COMMUNICATIONS | DOI: 10.1038/ncomms9097 ARTICLENATURE COMMUNICATIONS | 6:8097 |DOI: 10.1038/ncomms9097 | www.nature.com/naturecommunications 7& 2015 Macmillan Publishers Limited. All rights reserved.For the experiments that focus into (Si or SiC) nanoparticle coatings, thespheres have diameters of d¼ 2.0mm and d¼ 30–40 mm. The d¼ 2.0-mm spheresare mounted as stated above. The d¼ 30–40-mm spheres are mounted on a planarglass substrate. The nanoparticles are applied to the surface of the sphere with adry-coating process44. It is found that the van der Waals forces in this processprovide sufficient adhesion—although alternative nanoparticle dispersal andadhesion processes may be used, if needed, by way of polymer-basednanocomposite deposition45, chemical vapour deposition46 and/or laser ablation32.Unlike the tests of spheres with the GaAs target, the tests of nanoparticle-coatedspheres are self-aligning, in that coincident pump and probe beams will inherentlyfocus into the nanoparticle coatings.SiC charge-carrier dynamics. The bulk SiC results of Fig. 1b and the nanoparticleSiC results of Fig. 5b,b all exhibit AOS recovery with two distinct time constants.Long recovery time constants are seen atB22 ms for the bulk and at 5 and 1 ps forthe SiC nanoparticles on the micro- and milli-scale architectures, respectively. Thelong recovery time constants are due solely to recombination on the charge-carrierdensity, N(z,t), and it is these time constants that witness pronounced effects fromthe increased specific surface area of the nanoparticles. Short-recovery timeconstants are observed at B3 ps for the bulk and at 350 and 270 fs for the SiCnanoparticles on the micro- and mill-scale architectures, respectively. Theshort-recovery time constants are attributed to ultrafast charge-carrier dynamics inthe SiC bandstructure47, as photoinjected charge carriers populate excited states inthe M-sidevalley and then undergo intervalley and intravalley scattering. Thescattering yields femtosecond and picosecond transients for the effective mass andmobility. The effective mass and mobility transients form ultrafast transients on therefractive index near the surface, Dns(t), and absorption coefficient in the bulk,Dab(t), respectively, and this is ultimately seen as an ultrafast modulation onthe differential transmission of the probe beam, according to equation (1). Suchscattering processes have been observed before for SiC and commented on in theliterature39,48,49.References1. Nazarathy, M. et al. All-optical linear reconfigurable logic with nonlinear phaseerasure. J. Opt. Soc. Am. 26, 21–39 (2009).2. Brzozowski, L. & Sargent, E. H. All-optical analog-to-digital converters,hardlimiters, and logic gates. J. Lightwave Technol. 19, 114–119 (2001).3. Shacham, A., Bergman, K. & Carloni, L. P. in Proceedings of the FirstInternational Symposium on Networks-on-Chip (NOCS 2007) 53–64 (Princeton,NJ, USA, 2007).4. Li, F. et al. All-optical XOR logic gate for 40Gb/s DPSK signals via FWM in asilicon nanowire. Opt. Express 19, 20364–20371 (2011).5. Miller, D. A. Device requirements for optical interconnects to silicon chips.Proc. IEEE 97, 1166–1185 (2009).6. Husko, C. et al. Ultrafast all-optical modulation in GaAs photonic crystalcavities. Appl. Phys. Lett. 94, 021111 (2009).7. Hu, X., Jiang, P., Ding, C., Yang, H. & Gong, Q. Picosecond and low-powerall-optical switching based on an organic photonic-bandgap microcavity. Nat.Photonics 2, 185–189 (2008).8. Piccione, B. et al. All-optical active switching in individual semiconductornanowires. Nat. Nanotechol 7, 640–645 (2012).9. Almeida, V. R., Barrios, C. A., Panepucci, R. R. & Lipson, M. All-optical controlof light on a silicon chip. Nature 431, 1081–1084 (2004).10. Nozaki, T. et al. Sub-femtojoule all-optical switching using a photonic-crystalnanocavity. Nat. Photonics 4, 477–483 (2010).11. Volz, T. et al. Ultrafast all-optical switching by single photons. Nat. Photonics 6,605–609 (2012).12. Ganikhanov, F., Lin, G.-R., Chen, W.-C., Chang, C.-Y. & Pan, C.-L.Femtosecond carrier lifetimes in arsenic-ion-implanted GaAs. Appl. Phys. Lett.67, 3465–3467 (1995).13. Gupta, S. et al. Femtosecond carrier lifetime in GaAs grown by molecular beamepitaxy at low temperatures. Appl. Phys. Lett. 59, 3276–3278 (1991).14. Elezzabi, A. Y., Meyer, J., Hughes, M. K. Y. & Johnson, S. R. Generation of 1-psinfrared pulses at 10.6 mm by use of low-temperature-grown GaAs as an opticalsemiconductor switch. Opt. Lett. 19, 898–900 (1994).15. Heuck, M., Kristensen, P. & Mørk, J. Energy-bandwidth trade-off inall-optical photonic crystal microcavity switches. Opt. Express 19, 18410–18422(2011).16. Vo, T. D. et al. Photonic chip based transmitter optimization and receiverdemultiplexing of a 1.28 Tbit/s OTDM Signal. Opt. Express 18, 17252–17261(2010).17. Hillerkuss, D. et al. 26 Tbit s 1 line-rate super-channel transmissionutilizing all-optical fast Fourier transform processing. Nat. Photonics 5,364–371 (2011).18. Collier, C. M., Born, B., Bethune-Waddell, M., Jin, X. & Holzman, J. F. Ultrafastphotoexcitation and transient mobility of GaP for photoconductive terahertzemission. IEEE J. Quant. Electron. 49, 691–696 (2013).19. Holzman, J. F. et al. Ultrafast carrier dynamics in InP photonic crystals.Nanotechnol. 16, 949–952 (2005).20. Collier, C. M., Born, B. & Holzman, J. F. Ultrafast response of SiC and Sinanocomposite material systems. Electron. Lett. 48, 1618–1619 (2012).21. Hvam, J. M. & Brodsky, M. H. Dispersive transport and recombination lifetimein phosphorus-doped hydrogenated amorphous silicon. Phys. Rev. Lett. 46,371–374 (1981).22. Miyazawa, T., Ito, M. & Tsuchida, H. Evaluation of long carrier lifetimesin thick 4H silicon carbide epitaxial layers. Appl. Phys. Lett. 97, 202106 (2010).23. Lloyd-Hughes, J. Influence of surface passivation on ultrafast carrierdynamics and terahertz radiation generation in GaAs. Appl. Phys. Lett. 89,232102 (2006).24. Chen, Z. G., Taflove, A. & Backman, V. Photonic nanojet enhancement ofbackscattering of light by nanoparticles: a potential novel visible-lightultramicroscopy technique. Opt. Express 12, 1214–1220 (2004).25. Li, X., Chen, Z., Taflove, A. & Backman, V. Optical analysis of nanoparticles viaenhanced backscattering facilitated by 3-D photonic nanojets. Opt. Express 13,526–533 (2005).26. Lecler, S., Takakura, Y. & Meyrueis, P. Properties of a three-dimensionalphotonic jet. Opt. Lett. 30, 2641–2643 (2005).27. Welford, W. T. Aberration of optical systems (Adam Hilger Ltd, 1986).28. Geints, Y., Zemlyanov, A. & Panina, E. Photonic jets from resonantlyexcited transparent dielectric microspheres. J. Opt. Soc. Am. B 29, 758–762(2012).29. Riedl, M. Optical Design Fundamentals for Infrared Systems 2nd edn (SPIEPress, 2001).30. Flora, S. & Dwivedi, N. A toxicochemical review of gallium arsenide. DefenceSci. J. 62, 95–104 (2012).31. Kuempel, E. D., Castranova, V., Geraci, C. L. & Schulte, P. A. Development ofrisk-based nanomaterial groups for occupational exposure control. J. Nanopart.Res. 14, 1029 (2012).32. Fu, Y. H., Kuznetsov, A. I., Miroshnichenko, A. E., Yu, Y. F. & Lukyanchuk, B.Directional visible light scattering by silicon nanoparticles. Nat. Commun. 4,1527 (2013).33. Zhang, Q. X. et al. A silicon platform with MEMS active alignment functionand its potential application in Si-photonics packaging. IEEE J. Selected TopicsQuantum Electron 16, 267–275 (2010).34. Zhang, Q. X. et al. A lens holder in conjunction with a MEMS platform foron-chip aligning and fixing of a ball lens in silicon photonics packaging. IEEEPhotonics Technol. Lett. 23, 588–590 (2011).35. Huang, L.-S, Lee, S.-S., Motamedi, E., Wu, M. C. & Kim, C.-J. in 48th ElectronicComponents & Technology Conference 592–597 (Seattle, WA, USA, 1988).36. Lee, S.-S., Huang, L.-S., Kim, C.-J. & Wu, M. C. Free-space fiber-optic switchesbased on MEMS vertical torsion mirrors. IEEE J. Lightwave Technol. 17, 7–13(1999).37. Lee, C.-C., Hsaio, S.-Y. & Fang, W. Implementation of a micro ball lens on asilicon optical bench using insoluble two-phase liquid immersion technology.J. Micromech. Microeng. 20, 085015 (2010).38. Pan, C. T., Chien, C. H. & Hsieh, C. C. Technique of microball lens formationfor efficient optical coupling. Appl. Opt. 43, 5939–5946 (2004).39. Born, B., Landry, E. L. & Holzman, J. F. Electro-dispensing of micro-spheroidsfor lateral refractive and reflective photonic elements. IEEE Photon. J 2,873–883 (2010).40. Kapitonov, A. M. & Astratov, V. N. Observation of nanojet-induced modeswith small propagation losses in chains of coupled spherical cavities. Opt. Lett.32, 409–411 (2007).41. Ballarini, D. et al. All-optical polariton transistor. Nat. Commun. 4, 1778 (2013).42. Mir, N. F. An efficient switching fabric for next-generation large-scalecomputer networking. Comput. Netw. 40, 305–315 (2002).43. Scollo, R. et al. Mode-locked laser diode with an ultrafast integrateduni-traveling carrier saturable absorber. Opt. Lett. 30, 2808–2810 (2005).44. Borderieux, S., Wu, C. Y., Bonzongo, J.-C. & Powers, K. Control of elementalmercury vapor in combustion systems using Fe2O3 nanoparticles. Aerosol AirQual. Res. 4, 74–90 (2004).45. Collier, C. M., Jin, X. & Holzman, J. F. Ultrafast refractometry forcharacterization of nanocomposite material systems. IEEE Photon. Technol.Lett. 24, 590–592 (2012).46. Song, M. Y., Park, Y. K. & Jurng, J. Direct coating of V2O5/TiO2 nanoparticlesonto glass beads by chemical vapor deposition. Powder Technol. 231, 135–140(2012).47. Gavrilenko, V. I., Postnikov, A. V., Klyui, N. I. & Litovchenko, V. G. EnergyBand structure and optical properties of wurtzite-structure silicon carbidecrystals. Phys. Status Solidi B 162, 477–487 (1990).48. Ding, J.-L. et al. Nonlinear optical properties and ultrafast dynamics ofundoped and doped bulk SiC. Chin. Phys. Lett. 27, 124202 (2010).49. Tomita, T., Saito, S., Suemoto, T., Harima, H. & Nakashima, S.Inter-conduction band electron relaxation dynamics in 6H-SiC. Appl. Phys.Lett. 79, 1279 (2001).ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms90978 NATURE COMMUNICATIONS | 6:8097 | DOI: 10.1038/ncomms9097 | www.nature.com/naturecommunications& 2015 Macmillan Publishers Limited. All rights reserved.AcknowledgementsThe work was supported by the Natural Sciences and Engineering Research Council ofCanada (RGPIN 341487-12) and the Canadian Foundation for Innovation (16659 LOF).Author contributionsB.B. and J.F.H. conceived the idea. B.B. and S.G.-G. developed the simulations. B.B. andJ.D.A.K. fabricated and experimentally tested the device. B.B. and J.F.H. wrote the maintext and all authors contributed to the entire manuscript.Additional informationCompeting financial interests: The authors declare no competing financial interests.Reprints and permission information is available online at http://npg.nature.com/reprintsandpermissions/How to cite this article: Born, B. et al. Integration of photonic nanojets andsemiconductor nanoparticles for enhanced all-optical switching. Nat. Commun. 6:8097doi: 10.1038/ncomms9097 (2015).This work is licensed under a Creative Commons Attribution 4.0International License. The images or other third party material in thisarticle are included in the article’s Creative Commons license, unless indicated otherwisein the credit line; if the material is not included under the Creative Commons license,users will need to obtain permission from the license holder to reproduce the material.To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/NATURE COMMUNICATIONS | DOI: 10.1038/ncomms9097 ARTICLENATURE COMMUNICATIONS | 6:8097 |DOI: 10.1038/ncomms9097 | www.nature.com/naturecommunications 9& 2015 Macmillan Publishers Limited. All rights reserved.


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items