Jaeyoun
Kim
*
Department of Electrical and Computer Engineering, Iowa State University, Ames, Iowa, 50011, USA. E-mail: plasmon@iastate.edu; Tel: 1-515-294-4214
First published on 3rd August 2012
Along the advances in optofluidics, functionalities based on the surface plasmon-polariton have also been finding an increasing level of involvement within micro/nano-fluidic systems, gradually forming a new field of plasmo-fluidics. This survey of the burgeoning field reveals that judicious selection and combination of plasmonic and micro/nano-fluidic features render the plasmo-fluidic integration useful and mutually beneficial to the point of inevitability. We establish categories for the level of integration and utilize them as a framework for surveying existing work and extracting future perspectives.
Jaeyoun Kim | Jaeyoun Kim received his Ph.D. in Electrical Engineering from the University of Michigan at Ann Arbor in 2003. He then joined Professor Luke P. Lee's BioPOETS Lab at the University of California at Berkeley as a postdoctoral researcher. There he studied optofluidics, plasmonics, and biomimetic optical devices. In 2006, he joined the Electrical and Computer Engineering Department of Iowa State University where he is currently an Associate Professor. His current research topics include bio-inspired microsystems, plasmonic photovoltaic devices, and bio-photonic sensors. He was the recipient of NSF’s CAREER award in 2010 and Air Force Summer Faculty Fellowship in 2009. |
Fig. 1 Organization chart with the survey topics arranged in accordance with the proposed framework of categorization. |
At the interface between metallic and dielectric materials, an optical wave in the dielectric side can initiate a periodic fluctuation in the metal's charge density. An evanescent wave also appears in the dielectric side, linking the positively and negatively charged areas. If a set of resonance conditions were satisfied, this evanescent wave can be phase-matched to the original excitation wave and the fluctuation in the charge density can be sustained. Such a resonantly coupled state of charge density fluctuation and optical waves is commonly referred to as the surface plasmon-polariton (SPP). Depending on the geometry and dimensions of the metallic hosting structure, the SPP can take either a propagating (P-) or localized (L-) form, giving rise to a P- or L-SPP, respectively. For planar, unbound metal-dielectric interfaces, the P-SPP is the dominant form, although its excitation requires strict phase-matching through the use of Otto or Kretschmann setup. Fig. 2a schematically illustrates the optical excitation of P-SPP using the results of numerical simulations. On the other hand, the L-SPP develops around metallic structures that are small with respect to the wavelength of the optical wave, as shown Fig. 3a. Regardless of the type, the optical wave portion of the SPP, shown schematically in Fig. 2b, enjoys benefits such as localization below the diffraction-limit and dramatic intensity enhancement in the direct vicinity of the interface. The former comes from the wave's evanescent nature and the latter from the extremely tight confinement of electrons within thin and/or small metallic regions. Thanks to the involvement of highly interactive electrons, the SPPs can couple with each other much more actively than purely optical waves could, resulting in a wealth of hybrid modes. Best of all, the efficiencies associated with these plasmonic phenomena depend strongly on the nanoscale dielectric environment of the metallic hosting structure, providing facile routes for optical biosensing at molecular levels.
Fig. 2 (a) Numerical simulation results of failed (upper) and successful (lower) excitation of P-SPP. The P-SPP field profile above the Au layer clearly reveals its evanescent nature. (b) A schematic view of the electric fields associated with the charge density fluctuation. Again, the evanescent nature is evident with sub-micron tail extent. (c) A typical setup for plasmo-fluidic SPR and SPRi with microfluidic systems integrated directly over the phase-matching prism. Reproduced from ref. 9. (d) An example of critical-level-integrated SPRi device comprising multi-well array with a matching reconfigurable valve system. Reproduced from ref. 9. |
Fig. 3 (a) A schematic diagram explaining the physical origin and optical characteristics of L-SPP developed around a metallic nanostructure (MNS). (b) A plasmonic nanohole-based biosensor for use in the flow-through configuration. The transmission-based detection allows the adoption of a collinear arrangement in which the light source and the detector can be placed on opposite sides across the nanohole sensor. (c) A plot showing the superiority of the flow-through configuration over the flow-over one in terms of time response. (b) and (c) reprinted with permission from ref. 32. Copyright (2009) American Chemical Society. |
Dielectric materials in the liquid phase also offer features useful for optical applications. Thanks to their fluidic nature, they can create a perfectly conformal dielectric environment around an optical component simply by surrounding it. The same fluidic nature also enables rapid and dynamic modification of the dielectric environment through injection and/or mixing of liquids exhibiting different dielectric characteristics. The level of the resulting refractive index change greatly exceeds those attainable through electro-, acousto-, or nonlinear optical methods.1 The microfluidic architecture further strengthens these advantages with its tight fluidic confinement and precise fluidic control at a length-scale ideally matching those of integrated optic components, such as waveguides and resonators. Microfluidically confined liquid can also exert a physical force on microstructures in its way, enabling their actuation and transport. Interestingly, liquid medium also provides forces countering externally applied ones, such as the fluidic drag force in microscale and Brownian motion in nanoscale. These can be exploited to our advantage, most notably for refining particle motions in liquids.
Early stage plasmo-fluidic integration started out with in-channel implementations of SPP-based biosensing, a logical target with self-evident technical advantages. As shown in Fig. 2b, in the dielectric side, the SPP's electric fields are evanescent with only ∼300 nm extent above the metal-dielectric interface. Since the reagents and analytes in the volume beyond the SPP's extent would simply be wasted, the use of low-profile microfluidic channels for their delivery made a perfect sense. The fact that the extra volume, while not contributing to signal generation, can still obscure the measurement by casting temperature and/or concentration gradient further strengthen the plasmo-fluidic integration's benefit,2 most notably in the SPP-based studies of molecular interaction kinetics in which the diffusion rate is the main factor.3 The use of optical waves which requires no shielding or insulation proved ideal for liquid-based sensing tasks as well. This type of straightforward co-utilization of plasmonic and microfluidics features has led to the first stage of plasmo-fluidic integration that can be named convenience-level integration. The length-scale of microfluidic structures, typically measured in microns, and their easy patterning have been the main targets of exploitation at this level.
With the research growing both in size and diversity, such intuitively obvious integration has gradually evolved into more sophisticated ones. This new stage of integration is marked by the proactive search for useful yet unexploited functionalities in microfluidics or plasmonics and their utilization to bring in significant enhancements to their counterparts. The process usually involves substantial modification of the original configuration and/or the introduction of extrinsic techniques or materials. This phase can be named convergence-level integration. In both the convenient- and convergence-level integrations, reconciling the fundamental difference in the plasmonic and microfluidic length-scales, one measured in nm while the other in microns, remains the biggest challenge.
Recently, the integration has further progressed into a higher level at which totally new plasmo-fluidic functionalities are created with the combined use of microfluidics and plasmonics presumed from their very conception. In this case, the plasmo-fluidic integration becomes inevitable. On the other hand, as the plasmo-fluidic device miniaturized towards the nanoscale, the plasmonic and microfluidic effects become difficult to separate as well, whether intended or not. Critical-level may be an adequate term to label this phase. At this level, the challenge has been extended to the separation of the two inseparable functionalities, for their better design and analysis. Of course, there are numerous emerging research outcomes whose levels of integration are yet to be determined.
In this review, we survey the recent progress in plasmo-fluidics using the framework of convenience-convergence-critical level categorization and try to extract future perspectives from the emerging ones. Areas and sub-areas of plasmo-fluidics to be reviewed are summarized in Fig. 1 along with their typical length-scales. They stem from three major types of SPPs, i.e., P-SPP, L-SPP, and LR (long range)-SPP, and branched into different sub-areas in response to demands and technical limitations. Microfluidic synthesis of plasmonic nanostructures is also to be surveyed using the same framework for completeness.
Soon these evolved into a convergence-level integration in which features specific to microfluidics were proactively exploited. The laminar nature of the microflow was the first target. Providing an on-chip self-reference has been an important goal, especially for the realization of differential SPR sensing schemes, and the microfluidic system's capability to form a set of co-propagating flows in their channels proved ideal for its accomplishment.7 A more complex functionality, the microfluidic chemical signal generation, has also been integrated with SPR. Through microfluidic synthesis of chemical signals, the impacts of noise and background thermal drift could be mitigated, leading to a substantial decrease in the minimum detection limit.8 In these convergence-level examples, the microfluidics functionalities played active roles to improve the performance of the plasmonic counterparts.
It is the advent of SPR imaging (SPRi) which rendered the plasmo-fluidic integration inevitable, eventually leading it to the critical level. SPRi pursues parallel processing of SPR using a broadly collimated excitation beam, an array of sample reservoirs, and a wide field-of-view camera in combination. With high throughput, real-time monitoring, and process automation set as the main technical goals,9 the use of microfluidic architectures and functionalities is a must, not an option. The microfluidic devices in the SPRi setups can be subcategorized into multi-functional or multi-layered ones. The former focuses on maximizing the process automation and achieves it by installing as many microfluidic functionalities as possible. For instance, Lee et al. integrated microscale pumps, valves, flow sensors, heaters, and temperature sensors along with passive channels and reservoirs.10 Krishnamoorthy et al. also integrated on-chip generators of electro-osmotic and electro-kinetic flows.11 On the other hand, the multi-layered microfluidics focuses on realizing dynamically reconfigurable and/or individually addressable sample array patterns. The use of crossed flow channels,9,12 microfluidic printing,13 and 3D flow cell arrays14 all fall in this subcategory and are now indispensable in SPRi. Fig. 2d shows the reconfigurable microfluidic SPRi by Luo et al.9
Currently, the biggest challenge in plasmo-fluidic SPR is the growing complexity in the microfluidic subsystems, especially for those intended for on-demand reconfiguration. New microfluidic techniques are demonstrated to address the issue. Recent advances in droplet-based microfluidics and electro-wetting actuation have been immediately adopted for dynamically reconfigurable SPRi pattern generation and referencing.15 The utilization of SPR for profiling nanofluids in microfluidics channels16 is of special interest since the direction of benefit, which usually flows from microfluidics to plasmonics, was reversed in this case. To which level these emerging applications should be categorized will be determined as their uses expand. This section also validates the proposed survey framework.
There have also been efforts to find the optimum microfluidic length-scale in the plasmo-fluidic SPR and SPRi. The analysis by Gervais and Jensen17 and Squires et al.18 revealed that the efficiency of SPR- or SPRi-based molecule detection in microfluidic settings depends on the interplays between convection, diffusion, and surface-bound reaction. Their relative strengths are signified by two Peclet numbers (PeH and PeS, the channel and sensor Peclet number, respectively) and Damköhler number (Da). The Peclet numbers represent the relative thickness of the depletion layer with respect to the channel and sensor size, respectively. On the other hand, Da indicates the relative strength of diffusion and surface-bound reactions. In typical microfluidic settings, the channel height h becomes the most critical parameter governing these ratios. Shallow channels, with h ∼ 1 μm, can drive the operation into the diffusion-dominant, surface reaction-limited regime. The low channel profile also better matches the short extent of the P-SPP's evanescent tail. For microfluidic channels made with elastomers such as PDMS, however, the sagging of the channel ceiling towards the floor also needs to be considered. The empirical rule is to avoid any channel aspect ratio greater than 20:1,19 which sets the minimum channel height at a typical value of 10 μm.
Recently, the use of the nanoscale-recessed metallic structures, such as the nanowells or nanoholes shown in the inset of Fig. 3b, has prompted the emergence of L-SPP-based biosensors integrated at the convergence-level. In these new biosensing setups, the nanofluidic reservoir nature of the nanoscale recess is combined with its ability to host a variety of L-SPP modes, leading to a significant improvement in the sensitivity. The relative ease in the fabrication of such nanoscale recessed structures, through the use of self-assembled sacrificial nanosphere arrays22 or imprinting,23,24 has also contributed to their popularity.
Among the two nanoscale recessed structures, the nanohole array, a punched-through version of the nanowell array, has been more popular for plasmo-fluidic biosensing. In contrast to the nanowell, which relies on the L-SPP resonance within the inner volumes of individual wells, the nanohole array-based sensing utilizes a collective phenomenon called extraordinary transmission (EOT). EOT arises from the near-field coupling between the SPPs and manifests itself as a high level transmission of the excitation optical wave through otherwise opaque metallic films. As is usual with this type of collective SPP excitation, the EOT conditions are extremely sensitive to the dielectric environment of the nanohole array, allowing their utilization as a transduction mechanism for detecting molecular interactions. Since the EOT-based transducers produce sensing signals in the form of transmission, the experimental setup can be totally collinear with the light source on one side and the detector on the other side of the nanohole array, as shown in Fig. 3b, which is much simpler than the reflection-based monitoring adopted for the nanowell arrays. The potential of EOT in biosensing has been recognized from the early days of plasmo-fluidics25 but the recent leap in its popularity can be attributed to the simplification in their fabrication2,26 and the emergence of sophisticated detection methodologies relying on dual-wavelength27 or differential28 schemes. Recently the EOT-based sensing has been extended to viral detection29 and lipid membrane studies.30 At this level, the analyte fluid flows over the sensing spots, constituting the flow-over mode of operation. In this case, the impact of the microfluidic system's length-scale is similar to those in SPR or SPRi, leading to a similarity in channel dimensions.
In the face of this problem, the use of nanoholes perforated on free-standing metal films was proposed and demonstrated.31–34 In this architecture, it becomes possible to effectively transport analytes to the nanohole's inner volume by forcing the analyte-containing solution to flow through the nanoholes. As shown in the insets of Fig. 3c, this is equivalent to dividing a microfluidic channel-based sensor into a larger number of nanofluidic channels with the inner walls covered with L-SPP-supporting metal layers. Since the molecular binding mainly occurs within the nanohole's inner walls where the L-SPP's intensity reaches its peak, the transduction signals can be very high. We can safely categorize this flow-through scheme into the critical-level integration because the plasmonic and microfluidic functionalities are coupled inseparably, making it impossible to design or analyze one without taking the other into consideration.
In the length-scale point of view, the flow-through scheme offers the advantage of higher surface-to-volume ratio. A detailed analysis was performed by Escobedo et al.33 The results showed that the flow-through scheme's advantage in the time response over that of the flow-over scheme depends on the ratio between the Peclet numbers of the two systems which, in turn, depends on the ratio between the microfluidic sensor length and the thickness of the metal film perforated with nanoholes. Under the assumption of finite reaction kinetics, an improvement in the vicinity of 10 was predicted. The value is close to the experimentally observed 6-fold improvement shown in Fig. 3c. In principle, the improvement can be made greater through the use of metallic nanoholes with smaller diameters which allow denser packing. However, as the diameter approaches the Debye length, which is typically in the range of tens of nm, one needs to take the electrokinetic aspects of the channel into consideration. Most nanoholes reported in the literature avoided this cumbersome task by using nanoholes with a diameter greater than 100 nm.
In fact, the flow-through concept was so successful that it has also been utilized for non-EOT-type nanohole-based biosensing.35 The sieve-like geometry of flow-through nanohole arrays and the metallic film's ability to function as an electrode were utilized in combination to achieve an optofluidic concentration of molecules.36 By integrating the microfluid's fluidic nature, the EOT capability of the nanohole, and the electric gradient force altogether within one platform, the work represented a culmination of the critical-level plasmo-fluidic integration.
Fig. 4 (a) The typical metal stripe waveguide setup for guiding a LR-SPP mode. The width of the stripe is usually in the range of a few microns. The schematic view of its electric field profile is also included. The large size of the LR-SPP mode, often measured in microns, can be advantageous for plasmo-fluidic integration. (b) and (c) SEM images of cytop ultrathin film-based LR-SPP plasmo-fluidic biosensor. Reproduced from ref. 45. |
The existence of such a beneficial form of SPP has been known for decades in 1D settings37 but it was Berini who showed that it can also be supported by a 2D, finite-width metal stripe38 and exhibits advantageous properties for plasmo-fluidic sensing. Its long propagation distance mentioned above was the first to be exploited. From an optics point of view, such a long waveguide length facilitates the realization of complex optical components, such as branches, combiners, and interferometers, enriching the device functionality. In microfluidics, a propagation distance ranging from a few hundred microns to a few millimeters is ideal for crossing a typical microfluidic channel or to trace a section of it. With judicious design, the increase in the propagation distance can be translated into a longer interaction length, eventually contributing to the sensitivity enhancement. Another important feature of the LR-SPP is its relatively weak mode confinement which renders its mode field diameter comparable to that of single-mode optical fibers. This characteristic allows a direct excitation of LR-SPP with optical fibers through end-fire coupling. Moreover, such a larger mode size is also beneficial for matching the length-scales of typical microfluidic channels, usually measured in tens of microns, and leads to an increase in the sensing volume within the microflow.
These size advantages of LR-SPP could have led to convenience-level integrations with microfluidics for possible applications in biosensors, had it not been for a technical challenge arising from the very source of all the advantageous properties: the symmetric cladding. To support the LR-SPP, the difference in the refractive indices of the dielectric materials in contact with the upper and lower sides of the metal stripe must be controlled within 10−3.39 In microfluidic biosensing, for which one side of the metal stripe is frequently exposed to aqueous (n ∼ 1.33) environments, the requirement becomes difficult to satisfy due to the scarcity of solid materials with equally low refractive indices.
The issue, however, has been tackled in a variety of proactive ways, prompting the emergence of several convergence-level integrated systems. In the most straightforward approach, low-index materials, such as cytop (n ∼ 1.336)40–42 or Teflon (n ∼ 1.31)43 have been adopted to form the lower cladding. A number of different schemes for microfluidic integration have been proposed and implemented. A recent paper reported its use for bacterial detection in static conditions.44 The metal stripe has also been made on the top of an ultra-thin film to form a free-standing waveguide that can be immersed within aqueous solutions for symmetric cladding.39 Researchers tried to further increase the degree of symmetrization by making the ultra-thin film with cytop.45 As shown in Fig. 4b and c, this free-standing architecture has the additional advantage of making both faces of the metallic stripe available for microfluidic biosensing, effectively doubling the sensing area.
Some researchers also tried to bypass the requirement of symmetric cladding by indirectly modifying the metal stripe's environment. Konopsky et al. used 1D photonic crystal structure to effectively lower the refractive index experienced by the LR-SPP mode down to unity, enabling the use of LR-SPP for hydrogen detection.46,47 Joo et al. also demonstrated, both numerically and experimentally, that the requirement of symmetric cladding can be relaxed by installing a planar metal layer below the metal stripe.48 The lower dielectric layer, now of finite thickness between the two metallic structures, can actually be the microfluidic channel containing the analyte.
The tightly localized nature of the effect suggests that its impact can be enhanced by the microscale dimensions of microfluidic structures. In fact, localized heating within microfluidic channels constitutes the technical basis for convenience-level integration of photothermal effect with microfluidics. Soon, the integration began to strengthen a number of lab-on-a-chip functionalities including mixing,54 lysing,55,56 and thermorheological flow-control.57,58 The degrees of their functional improvement are high enough to put them into the category of convergence-level integration. At this level, the photothermal effect can also help the fabrication of microfluidic devices. One example is the nanoscale milling of polymer using L-SPP-based photothermal ablation.59
Recent years have seen the emergence of new photothermal applications featuring critical-level plasmo-fluidic integration. One example is the all-optical guiding of liquid body along microfluidic channels with a low-power laser beam60 illustrated schematically in Fig. 5a. In typical polymeric microfluidic channels, their hydrophobicity resist the influx of liquid, eventually stopping it at a point where the pressure applied to the liquid is balanced by the hydrophobic resistance. The same applies to liquid suspended with metallic nanoparticles. When the liquid–air interface gets illuminated with a laser beam, an in-channel photothermal heat generation occurs. Then the microfluidic channel first confines the photothermally generated heat, expediting the boiling and evaporation of the liquid. Then it confines the liquid vapor, forcing it to condense in front of the original interface. Since the newly condensed liquid layer lowers the hydrophobic resistance, the liquid body can advance towards the new balancing point, resulting in photothermal guiding of liquid body. This application utilizes the flow resistance of a microfluidic channel and its ability to confine multiple entities in an inseparable and irreplaceable fashion, categorizing itself into critical-level integration.
Fig. 5 (a) A schematic illustration of the all-optical guiding of microflow based on photothermally induced evaporation-condensation-coalescence cycle. Reproduced with permission from ref. 60. (b) An example of critical-level plasmo-fluidic integration in which the gap between “nanocaps” play triple roles as a SERS hot-spot, an L-SPP resonator, and a nanofluidic channel at which the target molecules are trapped and accumulated. The diameter of the nanosphere is 100 nm. (c) The numerical simulation results show that the cusp between the metallic nanocaps indeed performs plasmonic and nanofluidic functionalities simultaneously. (b) and (c) reproduced with permission from ref. 85. |
Another emerging application of the photothermal effect with critical-level dependence on plasmo-fluidic integration is the plasmonic nanoparticle-mediated heterogeneous photocatalysis.61 It is currently under intense research as an assisting technology for the heterogeneous photocatalysis.61–63 In this new photocatalysis scheme, light energizes the electrons within a nanoparticle through L-SPR to improve the catalytic efficiency. There, the photothermal effect provides highly localized heat that can expedite the catalysis without affecting the overall thermal environment. The microfluidic channel functions as a continuous flow reactor with its length-scale providing an extremely high surface-to-volume ratio, small thermal mass, and fast transport of the substances involved. The eminent targets of the field, solar energy-based in situ water purification,64 renders the integration of plasmonic photothermal effect and microfluidic structures unique and essential.
Thanks to its molecular fingerprinting capability, SERS has been enjoying substantial popularity, as shown by the publication of more than ten review papers during the past three years alone.65–75 It is clear from the reviews that the recent trend is moving towards the design and realization of SERS hot-spots with built-in considerations on the efficient transport of analyte molecules towards them.76 The surveys and references therein need not be repeated here. Even though the latest review on SERS was performed in January 2012, the rapid progress in the field has already accumulated enough innovations worthy of another survey, particularly from a plasmo-fluidic integration point of view. Some of them were at the convenience level. A prominent example is the development of SERS optrode which facilitates sequential measuring of the same channel by keeping the SERS hot-spots on the facet of the probing fiber rather than placing them within the channel.77
At the convergence level, the majority of the emerging work focused on how to give SERS extra amplification capability. Surface-enhanced Raman excitation spectroscopy (SERES), in which the Raman scattering is induced not by an isolated L-SPP but by their collective resonance supported by well-organized, periodically arrayed nanostructures,78 has been extensively exploited. Metallic gratings have been frequently chosen for the realization of SERES.79,80 Other types of resonance, such as the dielectric grating-based guided mode resonance, have also been utilized to strengthen the SERS effect.81,82 For flat metal surfaces with no textures, the plasmonic lensing effect has also been utilized to strengthen SERS.76
Critical-level plasmo-fluidic integration has been mainly concerned with the utilization of metallic nanoholes for SERS. As surveyed in Section 3.1, nanoscale holes can excite and sustain L-SPPs and the corresponding change to the optical excitation is a good reporter of the molecular activities occurring inside them. With the light-amplifying L-SPPs developed within and around its inner volume, the nanohole can also function as an excellent SERS hot-spot. A combined utilization of these two effects, with the former reporting on the occurrence of a binding event and the latter on the binding molecule's fingerprint spectrum, offers an opportunity to achieve molecular detection and identification simultaneously. Examples include the work by Xu et al. which utilized quasi-3D nanohole arrays to combine SERS and plasmonic Fabry-Perot resonance83 and the ones by Oh et al. which utilized the gaps between nanocaps85 for a combined excitation of SERS and L-SPP. The latter is of special interest in its proactive incorporation of nanofluidic functionalities for the plasmonic detection process, as illustrated in Fig. 5b and c. Due to the fact that the simultaneous excitation of those two effects is inevitable at this length-scale, these types of integration all belong to the critical level. The cited reports are examples of turning the inevitable integration of heterogeneous functionalities at the hot-spot into advantages through judicious design. It is also interesting to point out that, in plasmo-fluidic settings, this nanogap- or nanohole-based SERS/L-SPP co-excitation almost always accompanies trapping of objects suspended in fluidic medium. This is another example of inevitable integration in need of proactive exploitation. Previous sections emphasized the importance of transporting analytes towards the SERS hot-spot and trapping can provide a solution to this issue. Reports of simultaneous achievement of trapping and SERS have begun to appear in the literature.86,87 The following section elaborates more on this effect.
Fig. 6 (a–c) The scannable plasmo-fluidic trap based on the interaction between two counter-propagating SPPs. Reprinted with permission from ref. 92. Copyright (2010) American Chemical Society. (d) A schematic illustration of the SIBA-enabled two-particle trapping setup. (e,f) The rectangular aperture supports two trapping spots. (g–i) The numerical simulation results show that the resonator was originally detuned from its peak resonance slightly and the strength of L-SPP resonance becomes stronger with each successive capture of nanoparticle. This leads to a plasmo-fluidic trap pre-programmed to capture two particles. (d) to (i) reprinted with permission from ref. 94. Copyright (2012) American Chemical Society. |
The plasmo-fluidic trapping became a critical-level integrated phenomenon when researchers began to take the impact of the trapped object into consideration, heralding the introduction of the self-induced back-action (SIBA) concept.93 The SIBA effect hinges on the sensitivity of the L-SPP's resonance conditions to the dielectric environment of the trapping structure. At nanoscale, the trapping event itself certainly modifies the resonance conditions, producing a feedback to the L-SPP which may strengthen or weaken the trapping force. Judicious incorporation of the SIBA has reportedly lowered the trapping power by an order of magnitude.88 The concept is still evolving, as exemplified by the recent work by Chen et al. in which the SIBA concept is used for optimizing a plasmonic resonator to trap two particles.94 As shown in Fig. 6d–f, the work started with a specially designed rectangular nanoaperture capable of supporting two trapping spots. Fig. 6g–i show that the resonator was initially detuned from the peak L-SPP resonance. As the aperture kept trapping particles, the strength of the resonance increased. In other words, the resonator was programmed for continuous trapping. We anticipate that such pre-programmable functionalities will gain further research interests.
At this critical-level, the plasmo-fluidic trapping effect reveals another intriguing aspect by linking itself with the topic of Section 5.1: the photothermal conversion. In fact, the effort to promote the interplay between the trapping and photothermal effects constitutes a big part of emerging work at the critical level. Ohlinger et al. experimentally confirmed that photothermal conversion can destabilize 3D optical traps through a sudden temperature increase.95 In the face of this issue, Donner et al. provided theoretical framework96 and Zhang et al. proposed an experimental method to separate the influence of the photothermal effect from that of the plasmonic action.97 The plasmonic nanobowtie-based work by Roxworthy et al. showed that the photothermal effect can actually produce some counter-intuitive results, such as the maximization of trapping strength at wavelength other than the L-SPP's resonance wavelength, which we can utilize to our advantage.98,99
After Wagner et al. reported their pioneering work on microfluidic metallic NP synthesis,101 a large number of reports appeared with topics closely related to the yield and utility of the produced NPs, such as in-channel aggregation,102 size tunability,103–105 hetero-structuring,106 and in situ functionalization with surfactants.102,107 Recent reports show a more intimate integration of NP synthesis and microfluidics. For instance, Jamal et al. synthesized Au NPs in microfluidic channels and then immobilized them on the wall of the channel reactor that follows immediately.108 Such Au NP-immobilized channels are useful for incorporating L-SPP-based plasmonic functionalities into lab-on-a-chip configurations. Throughout this work, well-established microfluidic unit processes, such as mixing and separation, have been proactively utilized in addition to the obvious length-scale advantages, such as reduced reagent consumption and large surface-to-volume ratio, categorizing them into the convergence level.
Unfortunately, the resonance wavelength of the spherical NPs exhibits a very weak dependence on the particle size.50 Two approaches have emerged to red-shift the inherently short resonance wavelength. Schemes belonging to the first approach modify the geometry of the individual NP so that each NP can host more than one L-SPP modes. When designed properly, the intra-particle coupling of the L-SPP modes leads to the formation of a new, red-shifted mode. Highly anisotropic metallic nanostructures, such as nano-rods, also support L-SPP modes with longer resonance wavelengths. This tailoring of the NP's geometric features is also important for its non-optical properties, such as the cell uptake efficiency.110 In the second approach, the red-shift is induced by assembling NPs, spherical or non-spherical, so that their L-SPP modes can achieve an inter-particle coupling. In this case, the assembly pattern is of utmost importance.
In accordance, the focus of interest in recent microfluidic NP synthesis has been extended towards the production of shape-tailored NPs capable of inducing intra- and/or inter-particle coupling. Fabricating such irregularly shaped and/or assembled NPs requires accurate control over temporal, spatial, and environmental parameters, at levels far above those for achieving monodispersity in particle size alone. Microfluidics helped overcoming the difficulty by providing accurate control in both spatial and temporal domains.
In 2005, Kohler et al. fabricated a Si microfluidic chip with three individually controllable micromixers, each followed by an inter-diffusion zone, to systematically study the impact of processing conditions on the Au NP synthesis. By inducing NP clustering with the microfluidic system, they could red-shift the optical extinction peak from 550 to 700 nm.111 Among the follow-up work, of special interest to plasmonics is the synthesis of bi-metallic, Au–Ag NPs with non-spherical and/or clustered morphologies.112–115 In particular, the 4-branched star-like Au–Ag NP pushed the extinction peak towards 720 nm.112 Boleininger et al. also reported microfluidic fabrication of Au and Ag nano-rods whose aspect ratio and extinction spectra were tuned by regulating the duration of reaction between the metal seeds and surfactant CTAB.
The advances in micromixers, among many other microfluidic functionalities, have been essential in diversifying the NP's morphology and enhancing its controllability. Their high mixing speeds, achieved through funneling116 or pulsing117 of multi-laminated flows, greatly narrowed the NP's size distribution width. The accuracy of the micromixer also improved control over the shape118 and size119 of the NPs. Recent advances in 3D architecture further bolstered the advantages of micromixers.120 The parallelization potential inherent in microfluidic systems also came into play. Simultaneous installation and operation of a large number of reactors not only increases the NP production but also enables massively combinatorial fabrication, expediting the identification of optimal process conditions.121,122
Fig. 7 (a) A schematic diagram of droplet-based Au nano-rod synthesis. The inset shows that changes in the process parameters red-shifted the L-SPP resonance wavelength as far away as 750 nm. Reproduced with permission from ref. 123. (b) A schematically illustrated example of integrated SPR based on the combined utilization of microfluidic, plasmonic, and surface-acoustic functionalities. Reproduced from ref. 131. (c) An illustration of on-demand SERS hot-spot formation utilizing electrooptofluidic functional integration. Reproduced from ref. 137. |
Two approaches have been taken to address the issue. Amarie et al. utilized sub-micron-scale spherical metal cavities that are big enough for simple, sacrificial microsphere-based fabrication yet small enough to host plasmonic resonance modes.145,146 The second scheme utilized the Au NP as an optical antenna that catches the impinging excitation light and transfers it to the WGM-hosting microsphere. Simultaneous exploitation of the large extinction cross-section and strong evanescent field of an Au NP under plasmonic resonance for WGM excitation was the main motivation of this scheme. Since Au NPs are significantly smaller than the microsphere WGM resonator, they were raised to the height of the microsphere's equatorial plane by a set of dielectric rods fabricated with MEMS technology.147 It is anticipated that there will be more efforts to develop this type of MEMS-assisted, opto-plasmonically hybridized functionalities for use in microfluidics.
Footnote |
† Published as part of a themed issue on optofluidics. |
This journal is © The Royal Society of Chemistry 2012 |