Letter pubs.acs.org/NanoLett
Ultrasensitive Photoreversible Molecular Sensors of AzobenzeneFunctionalized Plasmonic Nanoantennas Gayatri K. Joshi,†,‡ Karl N. Blodgett,† Barry B. Muhoberac,† Merrell A. Johnson,§ Kimberly A. Smith,† and Rajesh Sardar*,†,‡ †
Department of Chemistry and Chemical Biology and ‡Integrated Nanosystems Development Institute, Indiana University-Purdue University Indianapolis, 402 N. Blackford Street, Indianapolis, Indiana 46202, United States § Department of Physics, Indiana University-Purdue University Indianapolis, 402 N. Blackford Street, Indianapolis, Indiana 46202, United States S Supporting Information *
ABSTRACT: This Letter describes an unprecedentedly large and photoreversible localized surface plasmon resonance (LSPR) wavelength shift caused by photoisomerization of azobenzenes attached to gold nanoprisms that act as nanoantennas. The blue light-induced cis to trans azobenzene conformational change occurs in the solid state and controls the optical properties of the nanoprisms shifting their LSPR peak up to 21 nm toward longer wavelengths. This shift is consistent with the increase in thickness of the local dielectric environment (0.6 nm) surrounding the nanoprism and perhaps a contribution from plasmonic energy transfer between the nanoprism and azobenzenes. The effects of the azobenzene conformational change and its photoreversibility were also probed through surface-enhanced Raman spectroscopy (SERS) showing that the electronic interaction between the nanoprisms and bound azobenzenes in their cis conformation significantly enhances the intensity of the Raman bands of the azobenzenes. The SERS data suggests that the isomerization is controlled by first-order kinetics with a rate constant of 1.0 × 10−4 s−1. Our demonstration of light-induced photoreversibility of this type of molecular machine is the firststep toward removing present limitations on detection of molecular motion in solid-state devices using LSPR spectroscopy with nanoprisms. Modulating the LSPR peak position and controlling energy transfer across the nanostructure−organic molecule interface are very important for the fabrication of plasmonic-based nanoscale devices. KEYWORDS: Nanoprisms, molecular sensors, molecular machines, azobenzene, localized surface plasmon resonance, surface-enhanced Raman spectroscopy, photoreversibility
T
from plasmonic energy transfer between the nanoprism and azobenzenes. The reversible photoinduced LSPR change was further investigated with surface-enhanced Raman spectroscopy (SERS). Interestingly, the Raman peak intensity of azobenzene was higher in the cis than the trans conformation, which is opposite to that reported recently in the literature.2 The design and fabrication of molecular sensors functionalized with molecular machines (azobenzene and rotaxanes) that are capable of undergoing reversible switching upon exposure to external stimuli has shown promising applications in the fields of chemistry, biology, and medicine.3−18 In this context, the light-induced response of molecular machines has
his paper describes the investigation of solid-state photoisomerization of azobenzene-containing, self-assembled monolayers (SAMs) of alkylthiols on gold nanoprisms that serve as nanoantennas. For the first time, light-induced reversible switching of azobenzenes between the cis and the trans conformations was detected in the solid state on gold nanoprisms attached onto a glass substrate by monitoring their localized surface plasmon resonance (LSPR) dipole peak. An unprecedented 21 nm reversible shift of the dipole peak (ΔλLSPR) was observed with azobenzene conformational change allowing our fabrication of a highly effective molecular sensor. Such a large ΔλLSPR for a minor ∼0.6 nm increase in the thickness1 of the local dielectric shell surrounding the nanoprism caused by the cis to trans conformational change is unique and caused by the high LSPR sensitivity of this size and geometry gold nanoprisms with perhaps a contribution © 2014 American Chemical Society
Received: September 24, 2013 Revised: December 7, 2013 Published: January 3, 2014 532
dx.doi.org/10.1021/nl403576c | Nano Lett. 2014, 14, 532−540
Nano Letters
Letter
Scheme 1. Fabrication of Molecular Sensors by Functionalization of Gold Nanoprisms Previously Attached to a Substrate of Silanized Glass with a Self-Assembled Monolayer Composed of a Mixture of 9-(Phenyldiazenyl)phenoxy Nonane-1-thiol (ANT) and Nonanethiol (NT) and Photoinduced cis-trans Conformational Change of Azobenzenea
a
The image is not to scale.
Figure 1. (A) UV−visible extinction spectra of gold nanoprisms bound to silanized glass substrate without functionalization (green, λLSPR = 736 nm), with 100% nonanethiol functionalization (black, λLSPR = 754 nm), and with 100% 9-(phenyldiazenyl)phenoxy nonane-1-thiol (red, λLSPR = 788 nm) funtionalization. All spectra were acquired in air. (B) LSPR dipole peak shifts (ΔλLSPR) of molecular sensors prepared from gold nanoprisms attached onto silanized glass substrates after functionalization with different mole ratios of ANT to NT.
several advantages: (1) fast response, (2) selectivity, (3) less production of chemical waste, and (4) nondestructive cycling. The photoresponsive molecule azobenzene has shown promise because it is capable of undergoing a reversible cis to trans photoisomerization, which can be used in optical information storage and other applications.5,10,19−22 Exposing azobenzene to blue light (>400 nm) forces a cis to trans conformational change, whereas exposure to UV light (300−400 nm) causes the reverse transition. In recent years, attaching azobenzenecontaining molecular machines2,5,9,23−25 to plasmonic nanostructures has gained significant attention. Metallic nanostructures display size-, shape-, and local dielectric environmentdependent LSPR properties,26−39 which can be used to design chemical40−42 and biological sensors,43−47 and potentially in the fabrication of other molecular devices using nanostructures as one of their components. Previous studies have shown only irreversible optical shifts from nanostructures upon coupling between the molecular resonance of photochromic molecules and LSPR properties of nanostructures.32,48−50 Zheng et al. reported a reversible LSPR peak shift with lithographically fabricated gold nanodisks during the oxidation and reduction of surface-bound bistable [2] rotaxanes.51 However, photoswitchable solid-state molecular devices have not yet been designed in which the light-induced conformational change of azobenzene is detected by monitoring the LSPR properties of nanostructures in a well-defined physicochemical environment.
Changes in spatial and electronic interactions caused by the conformational change in such molecular machines coupled with the unique optical properties of nanostructures could open up new avenues for design of highly efficient optical switches in molecular electronics and many other nanotechnology applications. Here we use UV−vis spectroscopy and SERS to investigate the spectral response and the reversibility of azobenzenefunctionalized nanoprisms upon cycling exposure to blue and UV light in the solid state. Gold nanoprisms were selected as nanoantennas for our molecular sensor fabrication because of (1) strong electromagnetic (EM) field enhancement at their sharp tips,37,52 (2) LSPR properties that are sensitive to small variation in their surrounding environment,37,53,54 and (3) their atomically flat surface, which allows SAM formation with both a tightly packed lower layer of alkylthiols and a more loosely packed upper layer that provides the required space for the cis− trans conformational change of azobenzenes. Furthermore, we demonstrate a new SERS phenomenon of azobenzene in the solid state and rationalize observation of higher Raman peak intensity for the cis conformation as caused by the combined effects of (1) a stronger interaction between the EM field of the nanoprism and azobenzenes and (2) the possibility of ground state interaction between the lowest unoccupied molecular orbital (LUMO) of azobenzene and the Fermi level of the nanoprism. 533
dx.doi.org/10.1021/nl403576c | Nano Lett. 2014, 14, 532−540
Nano Letters
Letter
Our solid-state molecular sensors were assembled according to Scheme 1. Gold nanoprisms with an average 35 nm edge length and LSPR dipole peak at 750 nm in acetonitrile (Figure S1) were synthesized according to our published procedure55 and then attached onto a silanized glass substrate. The Supporting Information file provides details describing our synthesis and characterization of the nanoprisms, and attachment of the SAM that contains the azobenzenes to their surface. In general, there are a number of molecular level constraints on the geometric parameters for construction of molecular sensors based on conformational change. Both appropriate azobenzene surface density and its mode of attachment to the solid surface are critical for photoisomerization to occur.56 At least an ∼40 nm2 area of conformational freedom per azobenzene is required for efficient isomerization on a planner surface.57 Recently, Weiss and co-workers2 have shown that attaching azobenzenes onto lithographically fabricated gold nanoholes using an alkylthiol SAM increases the efficiency of the cis to trans isomerization while increasing the Raman peak intensity of azobenzenes by reducing surface quenching and steric hindrance. Additionally, the length of the alkylthiol linker was also reported as critical for SAM formation.58,59 As illustrated in Figure 1A, an ∼52 nm LSPR dipole peak red shift was observed when the nanoprisms attached to silanized glass substrate were functionalized with 100% 9(phenyldiazenyl)phenoxy nonane-1-thiol (ANT). The formation of this SAM resulted in a ΔλLSPR consistent with the increased thickness of the local dielectric hydrocarbon shell around the gold nanoprisms, as previously demonstrated by our group and others.37,54,60,61 When the SAM was prepared with 100% nonanethiol (NT), an ∼18 nm ΔλLSPR was detected. This shift is in close agreement with our previous results showing gold nanoprisms displayed a 2.5 nm red shift for each methylene unit (−CH2−), which in the case of NT predicts a 20 nm red shift.60 The heights of fully extended NT and ANT SAMs above the gold surface are expected to be 1.5 nm and 2.4 nm (ChemDraw 3D), respectively. Therefore, the finding that a 0.9 nm difference in SAM thickness between NT and ANT resulted in an additional ∼34 nm ΔλLSPR is unexpected and unique. Furthermore, according to the literature37,54,60,61 an alkylthiol SAM of 2.4 nm thickness (1-hexadecanethiol, HDT) on either silver or gold nanoprisms results in an ∼30 nm LSPR shift. Nevertheless, the 52 nm ΔλLSPR observed for our SAMmodified azobenzene-containing nanoprisms is the largest value reported in the literature. We also prepared nanoprisms coated with several mixed SAMs of ANT and NT of varying mole ratios attached to silanized glass substrate. As shown in Figure 1B, the ΔλLSPR decreased monotonically with decreasing percentage of ANT reflecting a diminished contribution from the longer and more structurally complex ANT. One of the mixed SAM-modified nanoprisms was further characterized by atomic force microscopy (AFM) as illustrated in Figure 2 and Figure S2. The heights of the nanoprisms in the absence and presence of the mixed SAMs of ANT and NT were ∼9.5 and ∼11.8 nm, respectively. The increase of ∼2.3 nm in height is consistent with a single layer SAM formation thickness of 2.4 nm, as mentioned above. The photomechanical behavior of azobenzene was previously employed in molecular sensor design,5,10,15,51,62 but these were either solution state devices or lacked coupling with the LSPR properties of the nanostructures prepared on solid substrates. AFM analysis by others of azobenzene-containing SAMs had
Figure 2. Atomic force microscopy images of gold nanoprisms bound to silanized glass substrate (A) without a SAM and (B) with a 3:7 mole ratio of ANT to NT SAM.
shown that there was ∼0.5 nm difference in height between the cis and trans conformations,1 which in turn would be expected to alter the local dielectric environment of the nanostructures. This dielectric change should be detectable by monitoring the LSPR peak of the nanostructures, but surprisingly it has remained undetected in optical spectroscopic measurements until now (see below). This absence is likely due to a combination of two factors: (1) geometrical constraints of the nanostructures themselves causing weak EM field enhancement and thus low intrinsic LSPR sensitivity2,9,63 and (2) steric hindrance of the cis−trans isomerization in the SAM. Because we have shown (Figure 1A) that the LSPR properties of our particular construction of SAM-coated nanoprisms attached to glass substrate are very sensitive to change of local dielectric environment compared to other anisotropic nanostructures,55 our molecular sensors were next characterized with respect to the azobenzene photoisomerization process. To prevent steric hindrance, we believe that an appropriately formed SAM of alkylthiols containing both ANT and NT is very important. The SAM should form a dense lower layer due to their equal nonane hydrocarbon chain length and equal foot-print area and with variation of their mole ratio should allow reduction of steric hindrance of the azobenzene isomerization by appropriately spacing of the azobenzene groups above the dense lower layer. Our molecular sensors were prepared under normal laboratory conditions that included constant exposure to 534
dx.doi.org/10.1021/nl403576c | Nano Lett. 2014, 14, 532−540
Nano Letters
Letter
Figure 3. (A) Time-dependent changes in UV−visible extinction spectra of molecular sensor prepared from gold nanoprisms with a 3:7 mole ratio of ANT to NT caused by constant exposure to blue light (450 nm). (B) LSPR dipole peak shifts (ΔλLSPR) after one hour of blue light exposure of molecular sensors with different mole ratios of ANT and NT.
To verify that the change in SAM height is caused by the azobenzene conformational change, we performed an AFM analysis on the 3:7 mole ratio of ANT to NT molecular sensor under exposure to UV and then blue light. After analyzing 10 different nanoprisms (see Figure S4), an average 0.6 nm increase in SAM thickness was found between the cis and the trans conformations, in agreement with measurements by others.1 Therefore, the unprecedentedly large 21 nm red shift of the LSPR dipole peak from the cis to trans isomerization is very significant considering there is at most a 0.6 nm increase in the thickness of the local dielectric environment around the nanoprisms. It is known that the LSPR peak of nanostructures undergoes a red shift with increasing alkyl chain length of their SAM coating, that is, the thickness of their hydrocarbon shell.61 However, a 21 nm ΔλLSPR cannot be solely due to an increase of 0.6 nm thickness of the SAMs. Also as mentioned previously, the SAMs composed of pure ANT and HDT are of similar length (2.4 nm), but their attachment to nanoprisms resulted in an ∼22 nm greater LSPR dipole peak shift for ANT. The ΔλLSPR will depend on the sensing volume of the nanoprism where larger sensing volume (larger edge length) will provide greater LSPR response. Therefore, a significant portion of the observed ΔλLSPR is caused by the high LSPR sensitivity of this particular choice of size and geometry of nanoprisms. Other factors, for example, plasmonic energy transfer as a form of resonance energy transfer64−66 between the nanoprism and the azobenzenes likely has a contribution to the LSPR peak shift. Resonance energy trasfer is a nonradiative dipole−dipole interaction caused by the surface plasmon resonance. Gold nanoprisms display a strong localized surface plasmon dipole peak that could induce a strong resonance energy transfer in the near-field, which can be considered to be Forster resonance energy transfer, where localized surface plasmon resonance dipoles replace the fluorescent system. Therefore, we believe a portion of the red shift of the LSPR peak of our nanoprisms could be caused by resonance energy transfer from the nanoprism to attached azobenzenes allowed after the conformational change from cis to trans. The cis and trans azobenzene conformation has two different dipole moments; however which dipole moment strongly interacts with overall dipole of nanoprisms is not known. Nevertheless, in our system a better resonance energy transfer takes place when the azobenzenes are in the trans conformation. To demonstrate such energy transfer, the electronic spectra of a 3:7 mole ratio of ANT to NT SAM on gold-coated glass substrate was compared with our
white light, and as a result a combination of cis and trans conformations of azobenzene were highly probable. To maximize the cis conformation of the azobenzenes attached to the nanoprisms, the sensors were exposed to UV light (365 nm) for an hour. Afterward white light exposure was avoided, and the extinction spectra were collected. The sensors were then illuminated with blue light (450 nm) for an hour to convert the azobenzenes to the trans conformation, and extinction spectra were again collected. Figure 3A follows changes in the extinction spectra recorded from 400 to 1000 nm for the molecular sensor constructed with a 3:7 mole ratio of ANT to NT at 10 min intervals during blue light exposure and shows an unprecedentedly large ∼21 nm LSPR red shift accompanying the cis to trans conformational change. We believe that this is the first example in which a solid state LSPR wavelength shift of a nanoantenna induced by the conformational change of a molecular machine was observed, excluding the reversible oxidation−reduction of rotaxane. Supporting Information (Figure S3) plots the time dependence of the change of wavelength (ΔλLSPR) and maximal intensity of the molecular sensor with 30 mole percent ANT initially in the cis conformation during exposure to blue light. The red shift of the LSPR dipole peak (22 nm) and increase in extinction (∼35%) was completed by an hour. The increase of extinction value is due to an increase in Rayleigh scattering because of the increase in the local refractive index. Figure 3B provides the ΔλLSPR values for sensors fabricated from different mole ratios of ANT and NT after one-hour exposures. Interestingly, the sensor fabricated with 100 mole percent ANT only exhibited a ΔλLSPR of ∼9.5 nm, whereas the highest observed shift of ∼21 nm was with the 3:7 mole ratio of ANT to NT. The reason for the lack of monotonic behavior for the nanoprism LSPR peak shift upon light exposure in the presence of different percentages of ANT is unclear. A complete conformational change from the cis to trans may not take place in the case of the 100 mole percent ANT-containing SAM because of steric hindrance, whereas at the other extreme a relatively small net change in the local dielectric thickness would be expected for 10 mole percent ANT-containing SAM upon light exposure. We believe an optimal concentration of ANT and NT on the surface is required to obtain the highest LSPR shift, where their relative number is critical to maximal isomerization of azobenzenes. Clearly the extent of isomerization modulates the thickness of the dielectric shell and thus strongly affects the LSPR response. 535
dx.doi.org/10.1021/nl403576c | Nano Lett. 2014, 14, 532−540
Nano Letters
Letter
Figure 4. (A) LSPR dipole peak position (λLSPR) of 30 mol percent ANT molecular sensor upon cycling between exposure to UV (365 nm) and blue (450 nm) light. Exposure to UV light caused azobenzene to transition to the cis conformation, while blue light exposure caused a transition to trans. (B) The UV−visible extinction spectra of the same molecular sensor after exposure to 365 nm UV light (green line) and 450 nm blue light (red line) after the first (solid line) and the sixth (dotted line) cycles.
confirmed that the LSPR response of our nanoprisms is due to the photoinduced conformational change of azobenzenes. It is known that irradiation of metallic nanoparticles by light can influence their LSPR properties. More specifically a blue shift of the LSPR peak of spherical silver nanoparticles was detected due to photoexcitation and charge accumulation.67 In another control experiment, our gold nanoprisms attached onto silanized glass substrate without ANT functionalization were exposed to either UV or blue light for 1 h, and no detectable difference in the LSPR peak from unexposed nanoprisms was observed (see Figure S7). Our experimental data are consistent with the literature that photoexcitation has almost no effect on the LSPR properties of gold nanoparticles in comparison to silver nanoparticles.68 This control experiment suggests that photoexcitation of gold nanoprisms did not contribute to the observed LSPR peak shift during the cis−trans isomerization of azobenzenes. The efficiency of photoisomerization depends on the packing of the SAMs, and in most cases a reversible cis−trans conformational change was not observed by others apparently because of steric hindrance.56 Recently, it was shown that appropriate packing of a SAM does allow reversible photoisomerization, but this was monitored by either SERS2 or contact angle measurements.57,69 However, detection of photoinduced reversible isomerization of azobenzenes through monitoring the LSPR peak peak shift of nanostructures has not yet been reported. Figure 4A represents the ΔλLSPR upon repeatedly cycling our molecular sensor between exposure to UV and blue light. Even after six cycles, the sensor still displayed an ∼21 nm shift between the cis and trans conformations, which indicates that our molecular sensor is stable and has the ability to undergo reversible switching without compromising its efficiency. Figure 4B illustrates the complete extinction spectra of the molecular sensor after the first exposure cycle and then after the sixth cycle, where no dramatic change in the extinction value was observed. This suggests that during UV light exposure the gold−sulfur bond remained stable and apparently no azobenzene loss occurred. This is extremely important for design of plasmonic-based molecular sensors with light-controlled LSPR properties. Recently and for the first time, reversible photoisomerization of azobenzene was monitored by SERS in which an azobenzene-containing SAM was prepared on lithographically fabricated gold nanoholes.2 All other earlier SERS studies of
30 mole percent ANT molecular sensors (Figure S5) by monitoring the π−π* transition of azobenzenes in their trans state. Clearly, the electronic transition of azobenzenes is different in the molecular sensor compared to the one constructed with gold-coated glass substrates and no nanoprisms. We also believe resonance energy transfer between metallic nanostructures and functional groups present in close vicinity (
