12142
Langmuir 2008, 24, 12142-12149
Articles Colloidal Crystals from Surface-Tension-Assisted Self-Assembly: A Novel Matrix for Single-Molecule Experiments Wen Cong Yeon,† Balakrishnan Kannan,‡ Thorsten Wohland,‡ and Vivian Ng*,† Information Storage Materials Laboratory, Department of Electrical and Computer Engineering, 4 Engineering DriVe 3, National UniVersity of Singapore, Singapore 117576, Singapore, and Department of Chemistry, 3 Science DriVe 3, National UniVersity of Singapore, Singapore 117543, Singapore ReceiVed January 3, 2008. ReVised Manuscript ReceiVed August 19, 2008 In this work, we develop a new method of creating colloidal crystals with cavities for the entrapment and long-term observation of single biomolecules. Colloidal crystals are first fabricated using surface-tension-assisted self-assembly. Surface tension helps to reduce the interparticle distance between dispensed colloids. Subsequently, the colloids are used as a matrix in which single fluorescently tagged molecules can be tracked using fluorescence microscopy. This method has a high efficiency of self-assembly for small volumes (4 µL) of colloidal suspensions (polystyrene colloids with diameters of 1000, 500, 200, and 100 nm) at low concentration (1% w/w). The spatial hindrance effect on the diffusion of molecules and their entrapment is discussed on the basis of fluorescence correlation spectroscopy results from the diffusion of molecules with different hydrodynamic radii in the cavities of colloidal crystals formed from micrometer- to nanometer-sized polystyrene spheres. Single horseradish peroxidase molecules turning over fluorescent products are tracked over a few seconds. This shows that colloidal crystals can be used to test the function of single molecules of enzymes and protein under controlled spatial confinement.
Introduction Three-dimensional self-assembled colloidal crystals have been reported to have potential applications in diverse fields such as photonics,1 optoelectronics,2 data storage,3 chemical and biochemical sensors,4 and all-optical computers.5,6 The attractiveness of colloidal crystals lies in its face-centered cubic geometry with excellent translational periodicity in the micrometer- to nanometer-scale regime. For instance, the resultant periodicity in optical indices between the voids and the constituent spheres (usually of material of high refractive index) traps light for photonic applications. However, proteins that exist in living cellular compartments are often found in a fluid phase with a matrix of membranes and/or structural fibers in close juxtaposition.7 This spatially hindered environment in living cells differs significantly from the dilute solutions typically used in in vitro studies of proteins, and hence the reported behavior of protein from those studies * Corresponding author. E-mail:
[email protected]. † Department of Electrical and Computer Engineering, National University of Singapore. ‡ Department of Chemistry, National University of Singapore.
(1) Chutinan, A.; John, S.; Toader, O. Phys. ReV. Lett. 2003, 90, 123901. (2) Painter, O.; Lee, R. K.; Scherer, A.; Yariv, A.; O’Brien, J. D.; Dapkus, P. D.; Kim, I. Science 1999, 284, 1819–1821. (3) Cumpston, B. H.; Ananthavel, S. P.; Barlow, S.; Dyer, D. L.; Ehrlich, J. E.; Erskine, L. L.; Heikal, A. A.; Kuebler, S. M.; Lee, I.-Y.S.; McCord-Maughon, D.; Qin, J. Q.; Rockel, H.; Rumi, M.; Wu, X.-L.; Marder, S. R.; Perry, J. W. Nature 1999, 398, 51–54. (4) Asher, S. A.; Peteu, S. F.; Reese, C. E. Anal. Bioanal. Chem. 2002, 373, 632–638. (5) Liu, J.-M. Photonic DeVices; Cambridge University Press: Cambridge, U.K., 2005. (6) Dalton, L.; Canva, M.; Stegeman, G. I.; Gubler, U.; Bosshard, C.; Shim, H.-K.; Jin, J.-I. In Polymers for Photonics Applications ; Lee, K.-S., Ed.; Advances in Polymer Science 158; Springer: Berlin, 2002; Vol. 1. (7) Nickell, S.; Kofler, C.; Leis, A.; Baumeister, W. Nat. ReV. Mol. Cell Biol. 2006, 7, 225–230.
may differ from the actual behavior in living cells.8 For instance, the confinement of proteins in cages with sizes comparable to the size of the confined protein affects the folding rates and improves protein stability.9,10 This motivated us to assess the spatial hindrance effect on proteins. Owing to the excellent translational periodicity of their face-centered cubic geometry of constituent spheres, colloidal crystals offer a confinement medium with well-defined cavities and interconnecting pores of uniform size and shape. The choice of self-assembled colloidal crystals is further strengthened by the simplicity of bottom-up methods to fabricate a matrix with controlled micrometer- to nanometer-scale features overcoming the size limitation in top-down methods. Among the bottom-up methods used to fabricate colloidal crystals, the conventional horizontal deposition method11 is attractive for work with biological agents because it is less invasive. Less invasive methods are advantageous when delicate biological samples such as proteins12,13 or DNA14 are used with colloids. Other techniques such as centrifugation15 and confined cell16,17 require the exertion of a physical force. Thermophoresis18 -related methods require the application of heat, an electrical (8) Ellis, R. J. Trends Biochem. Sci. 2001, 26, 597–604. (9) Zimmerman, S. B.; O Trach, S. J. Mol. Biol. 1991, 222, 599–620. (10) Zhu, H.-X.; Dill, K. A. Biochemistry 2001, 40, 11289–11293. (11) Yan, Q. F.; Zhou, Z. C.; Zhao, X. S. Langmuir 2005, 21, 3158–3164. (12) Kreuter, J. Microcapsules and Nanoparticles in Medicine and Pharmacy; Donbrow, M., Ed.; CRC: Boca Raton, FL, 1992. (13) Zhang, Y. Colloids Surf. 2006, 48, 95. (14) Kumar, A.; Pattarkine, M.; Bhadbhade, M.; Mandale, A. B.; Ganesh, K. N.; Datar, S. S.; Dharmadhikari, C. V.; Sastry, M. AdV. Mater. 2001, 13, 341–344. (15) Scroden, R. C.; Al-Daous, M.; Blanford, C. F. Chem. Mater. 2002, 14, 3305–3315. (16) Gates, B.; Qin, D.; Xia, Y. N. AdV. Mater. 1999, 11, 466–469. (17) Park, S. H.; Xia, Y. N. Langmuir 1999, 15, 266–273. (18) Epstein, P. S. Z. Phys. 1929, 54, 537.
10.1021/la800016h CCC: $40.75 2008 American Chemical Society Published on Web 10/08/2008
Colloidal Crystals from Self-Assembly
Langmuir, Vol. 24, No. 21, 2008 12143
Figure 1. Schematic diagram showing the trenchlike cavity fabricated by attaching a mechcanical mask to a substrate. Measurements are in millimeters. The thickness is not drawn to scale.
field, and charging of the colloids.19,20 Despite being less invasive and simple to implement, the horizontal deposition method requires either a large volume (100 µL) or high concentration (10% w/w) of colloidal suspensions. Here we develop a costefficient technique, called surface-tension-assisted self-assembly, to make colloidal crystals with a small volume (4 µL) of colloids at a low concentration (1% w/w). Colloidal crystals were successfully formed using 100-1000nm-diameter polystyrene spheres separately. The interconnected cavities are of a length scale comparable to the size of biological macromolecules, and this is suitable to the study of the spatial hindrance effect on diffusion. The diffusion of fluorescently labeled molecules in the crystals was measured using fluorescence correlation spectroscopy21,22 (FCS), a technique with singlemolecule sensitivity. One observation from our diffusion study is a strong slowing down of molecules in instances where the cavity size is very close to the molecular size of the diffusing species. Coupled with the nanomolar concentration of protein molecules used in our experiments, we were able to detect and observe single protein molecules actively turning over substrates within the cavities of the colloidal crystals.
Experimental Section Materials. Suspensions of polystyrene colloids with diameters of 100, 200, 500, and 1000 nm (concentration 1% w/w) purchased from Agar Scientific Corporation were directly used without further treatment. Si[100] and glass substrates (10 mm × 10 mm × 1 µm) were cleaned in an ultrasonic bath for 20 min each in acetone and 2-isopropanol, rinsed copiously with deionized (DI) water, and then dried with nitrogen. Si substrates were used for scanning electron microscopy whereas glass substrates were used for FCS experiments. All of the fluorescent molecules used in this work were used without further purification. These include 5-(and 6)-carboxytetramethyl rhodamine, succinimidyl ester (TMR) (Pierce, Singapore), fluorescently labeled dextran with molecular weights of 4.4, 40, and 155 kDa (T1037, R9131, and T1287, Sigma-Aldrich, Singapore). (19) Tien, J.; Terfort, A.; Whitesides, G. M. Langmuir 1997, 13, 5349–5355. (20) Masuda, Y.; Itoh, T.; Itoh, M.; Koumoto, K. Langmuir 2004, 20, 5588– 5592. (21) Magde, D.; Webb, W. W.; Elson, E. L. Phys. ReV. Lett. 1972, 29, 705– 708. (22) Krichevsky, O.; Bonnet, G. Rep. Prog. Phys. 2002, 65, 251–297.
Dihydrorhodamine (Invitrogen, Singapore) was used as the substrate for horseradish peroxidase (Invitrogen). The excitation and emission wavelengths of TMR are λex) 544 nm and λem ) 576 nm. The fluorescently labeled dextran molecules have λex) 490 nm and λem ) 520 nm. The excitation wavelength of dihydrorhodamine is λex) 500 nm, and the emission wavelength is λem ) 525 nm in 0.1 M Tris pH 8.9 (after oxidizing with H2O2 and horseradish peroxidase). Fabrication of Colloidal Crystals. For our modified horizontal deposition, the colloidal suspension was deposited into a template with a trenchlike cavity. A mechanical mask with an 8 mm × 1 mm × 1 mm (l × w × t) hole was wetted with photoresist and was attached to a substrate coated with photoresist. The mask-substrate assembly was prebaked at 90 °C for 20 min in an oven and then exposed to UV radiation using a mask aligner system. The exposed photoresist was washed off by soaking the assembly in a developer solution for 40 s, leaving a trenchlike cavity of 8 mm × 1 mm × 1 mm as shown in Figure 1. A colloidal suspension was deposited in the trenchlike cavity placed horizontally in a Petri dish. Low concentrations (1% w/w) and small volumes (4 µL) of polystyrene nanospheres were used. Samples were left to dry overnight. To give the colloidal crystals a resistance to water, thermal annealing was performed at a temperature slightly higher than the glass-transition temperature of polystyrene (95 °C). The nanospheres are joined together at their interface as a result of the slight viscoelastic deformation of their surfaces23 without changing the shape of the nanospheres and the geometry of the crystals. The mechanical mask was detached from the substrate by applying 2-isopropanol on all the four sides of the interface between the mechanical mask and the substrate. Characterization of Colloidal Films. The morphology of the colloidal crystals was imaged with a field-emission scanning electron microscope (SEM) (JSM-6700F, JEOL) operating at 3 kV. The height profile of the colloidal crystal was measured in a KLA-Tencor P-15 profiler. Optical images were taken with a Leica DC 100 microscope. FCS Instrument. FCS measurements were performed on a selfdeveloped FCS instrument.24 It is built around a laser scanning confocal microscope (LSCM) (FV300, Olympus, Singapore). For fluorescence excitation, 543 nm light from a He-Ne Laser (Melles Griot, Singapore) was used. The laser light was reflected by a mirror and a dichroic mirror (488/543/633) and scanned by a pair of galvanometric mirrors (G120DT, GSI Lumonics). A water-immersion (23) (a) Mazur, S.; Beckerbauer, R.; Buckholz, J. Langmuir 1997, 13, 4287– 4296. (b) Gates, B.; Park, S. H.; Xia, Y. N. AdV. Mater. 2000, 12, 653–656. (24) Pan, X.; Yu, H.; Shi, X.; Korzh, V.; Wohland, T. J. Biomed. Opt. 2007, 12, 14034.
12144 Langmuir, Vol. 24, No. 21, 2008
Yeon et al.
objective (60×, 1.2 NA, UPlanApo, Olympus, Singapore) focuses the laser light into the sample creating a small focal volume (∼0.3 × 10-15 L).24 The focal volume will be somewhat larger within the colloidal crystals because of the different refractive index of polystyrene, but because all measurements are made at constant depth, the focal volume will be constant during the measurements in a particular crystal. The fluorescence light is collected by the same objective and is imaged over a 3× magnification stage on a pinhole with 150 µm diameter. A 580DF30 (Omega, Brattleboro, VT) emission filter selected the wavelength from the emitted light. An imaging lens (Achromat f ) 60 mm, Linos, Goetingen, Germany) focuses the collected light onto the active area of an avalanche photodiode (APD) in a single-photon-counting module (SPCMAQR-14, Pacer Components, Berkshire, U.K.). The TTL output signal from the APD was fed into an autocorrelator (Flex02-01D, correlator.com, Zhejiang, China) that calculates and gives autocorrelation curves. Curve fitting was performed in IgorPro (WaveMetrics, Lake Oswego, OR) using self-written programs. Laser power measured after the dichroic mirror was maintained at 100 µW to prevent photobleaching. The background count owing to scattering by the polystyrene spheres in the crystals, measured with DI water dispensed onto the crystals, was higher (3-10 kHz for all spheres except 200 nm spheres that had a background of 10-20 kHz) than the background measured on a cover slide with a DI water droplet (0.6 kHz), but no autocorrelation was found from the scattering data. Diffusion times of up to 1 s were obtained for an acquisition time of 10-30 s. A freshly prepared sample on a glass substrate was used for every measurement. FCS Data Analysis. The autocorrelation function of a freely diffusing species in a 3D Gaussian volume element can be described by the analytical expression25
G(τ) )
(
1 τ 1+ N τD
)( -1
1+
τ K τD 2
) ( -0.5
1+
)
Ftrip -(τ⁄τtrip) e + 1 - Ftrip G∞ (1)
where N is the number of particles within the focal volume weighted by the correction factor for the experimental geometry, τ is the lag time, τD is the diffusion time in the focal volume, and K is the structure factor that takes into account the asymmetry in the focal volume and is given by z0/ω0. z0 and ω0 are the distances in the axial and radial directions, respectively, at which the excitation intensity reduces to 1/e2 of its value at the center of the focal volume. K was determined for our setup to be 5.1 ( 0.8. G∞ is the convergence value of G(τ) for long times and is expected to be 1. In all measurements, G∞ did not differ from 1 by more than 3%. Diffusion time τD is given by
τD )
ω02 4D
(2)
where D is the diffusion coefficient. By measuring τD of a fluorophore with a known diffusion coefficient (e.g., TMR with D ) 2.8 × 10-6 cm2/s), the value of ω0, which is specific to the instrument, can be determined using eq 2. Using the calibration value of ω0, D of any species can be determined. Ftrip is the fraction of fluorophores in the triplet state, and τtrip is the triplet state relaxation time. In the above equations, it is assumed that the molecules undergo Brownian diffusion and D is a characteristic of the fluorophores, independent of the duration of measurement. However, there are situations where the molecular diffusion is hindered by the obstacles within the focal volume.26 The diffusion coefficient in this case depends on the duration of the measurement, and the diffusion is said to be anomalous. Depending on the distribution in the size and ¨ .; Widengren, J.; Kask, P. Eur. Biophys. J. 1993, 22, (25) Rigler, R.; Metz, U 169–175. (26) (a) Weiss, M.; Hashimoto, H.; Nilsson, T. Biophys. J. 2003, 84, 4043– 4052. (b) Banks, D. S.; Fradin, C. Biophys. J. 2005, 89, 2960–2971. (c) FatinROGUE, N.; Starchev, K.; Buffle, J. Biophys. J. 2004, 86, 2710–2719.
the number of obstacles within the focal volume, D will show a distribution of values. In the case of normal diffusion where D is independent of time, the mean-square displacement can be written as27,28
〈r2(t) 〉 ) 2nDt
(3)
where n denotes the dimensionality of the system and t is the time. When the diffusion is anomalous,
〈r2(t) 〉 ) 2nΓtR
(4)
Γ is the transport coefficient and R is the temporal exponent. If R ) 1, then the diffusion is normal and eq 4 reduces to eq 3. If R < 1, then it is called anomalous subdiffusion (ASD). Equating espressions 3 and 4, the time-dependent diffusion coefficient is given by
D(t) ) ΓtR - 1
(5)
The autocorrelation function in the presence of anomalous diffusion can be written as28
G(τ) )
( ( )) (
τ 1 1+ N τD
R -1
1+
( ))
1 τ K2 τD
(
1+
R -0.5
×
)
Ftrip -(τ⁄τtrip) e + G∞ (6) 1 - Ftrip
Single-Molecule Detection. A freshly prepared crystal was used for each experiment. For the detection of entrapped fluorescent molecules inside the colloidal crystals, fluorescence images were obtained using an avalanche photodiode (APD) as the detector owing to their superior quantum efficiency compared to that of photomultiplier tubes. The wavelength of the laser for fluorescence excitation was 543 nm and was maintained at 100 µW power. The focal volume was kept at a height of 3 µm from the top surface of the cover slide on which the crystals are grown. Line scans (the focal volume is traced along a line of length 235 µm) were carried out for 30 s in fast scanning mode, and the photon arrival times were recorded directly from the APD using the software PhotonCount (Flex02-01D, correlator.com, Zhejiang, China). A line scan is a standard mode in confocal microscopes in which the confocal volume is scanned over a single line in contrast to the normal imaging mode. For the detection of single molecules entrapped in the crystal, the catalytic reaction29,30 of horseradish peroxidase (HRP) with the substrate dihydrorhodamine 6G in the presence of H2O2 was used. During the oxidation of dihydrorhodamine by HRP, the fluorescent product, rhodamine 6G, is first bound to the HRP molecule, forming a fluorescent enzyme-product complex before rhodamine is released. The fluorescence of the complex is detected by the APD. A 1 nM HRP (40 µL) solution is dispensed on the colloidal crystals fabricated with our smallest nanospheres (100 nm diameter) and left to incubate for 10 min at room temperature. Repeated flushing with PBS was carried out to remove HRP molecules outside the colloidal crystals. Then, 120 µM (40 µL) H2O2 and 22 nM (40 µL) dihydrorhodamine 6G were dispensed onto the colloidal crystals, and line scanning was started immediately.
Results and Discussion With the conventional horizontal deposition method, owing to the high solvent content, the liquid meniscus spreads out evenly to form a large circular blot. For spontaneous self-assembly, the favored geometrical packing of the nanospheres is the facecentered cubic geometry. When dried, two distinct zones are (27) Masuda, A.; Ushida, K.; Okamoto, T. Phys. ReV. E 2005, 72, 060101(R). (28) Wachsmuth, M.; Waldeck, W.; Langowski, J. J. Mol. Biol. 2000, 298, 677–689. (29) Hewson, W. D. Hager, L. P. The Porphyrins; Academic Press: New York, 1979; Vol. VIII, p 295. (30) Edman, L.; Fo¨ldes-Papp, Z.; Wennmalm, S.; Rigler, R. Chem. Phys. 1999, 247, 11–22.
Colloidal Crystals from Self-Assembly
Langmuir, Vol. 24, No. 21, 2008 12145
Figure 3. (a) Schematic illustration of the cross-sectional profile of the deposited suspension. The ellipse corresponds to the region where the optical image measurement shown in panel b is taken. (b) Light microscopy image (magnification 50×) of the top view of a colloidal film formed on a glass substrate with the surface-tension-assisted selfassembly technique.
Figure 2. (a) Optical photograph of part of the circular blot of the colloidal crystal film self-assembled from polystyrene spheres of 200 nm diameter on a horizontal glass substrate. Two distinct zones are identified. Zone 1 shows the crystallization of colloids into a regular fcc arrangement. Zone 2 shows colloids packed with short-range order. (b) SEM image of colloidal crystals in zone 1 shown in panel a. Line defects and point defects are observed amidst regular fcc packing. (c) SEM images of colloidal film in zone 2 of panel a. Short-range order of colloids in predominantly the bcc arrangement is observed.
identified as seen in Figure 2a. A very thin outermost ring surrounds a much larger circular region. In the outer ring as in Figure 2b, colloidal crystals are found. Beyond the thin ring, the long-range periodicity of the colloidal crystal breaks down. The colloids are packed in multilayers of both face-centered cubic (fcc) and body-centered cubic (bcc) arrangements with shortrange order (Figure 2c). The efficiency of the self-assembly for low colloidal concentration by the horizontal deposition technique is too low for practical purposes. An outward radial transport of spheres and the spreading of the dispensed suspension lead to larger interparticle distances between the colloids. With the
increased interparticle distance, attractive capillary forces31,32 are no longer effective, and periodicity breaks down. In the case of surface-tension-assisted self-assembly, we observe the colloidal crystals climbing in discrete steps resembling a long flight of stairs. The stairs plateau after 100 µm, and we observe a region with grains of colloidal crystals. As shown in Figure 3a, nucleation begins at the thinnest part of the liquid meniscus, which is a line that runs parallel to the cavity wall at the center of the cavity. Immersion capillary forces pin down a line of spheres from the bulk solvent along the interface between the solvent meniscus, air, and the Si substrate. Convective forces transport the colloids toward this nucleation line. Owing to the reduced spreading of the deposited suspension caused by the cavity walls in the template, the interparticle distance between the colloids is reduced such that capillary forces can act effectively to pack all of the colloids in a regular fcc lattice. As seen, the introduction of the template modifies the liquid meniscus, and surface tension helps to reduce the contact area of the deposited blot. Hence we term this method surface-tension-assisted selfassembly. Figure 4 shows SEM images of the colloidal crystals formed from 200 and 100 nm colloids. They are formed with fcc packing and can extend over a lateral length scale of a few hundred micrometers with a height of ∼50 layers of spheres, giving rise to 5-10 µm depending on the size of the spheres. All of the crystals have colloids arranged in an fcc lattice. The results are very reproducible over different specimens. Unlike with the conventional horizontal deposition technique, we have eliminated areas of colloidal aggregations with shortrange order that are not useful for our purpose. With surfacetension-assisted self-assembly, the efficiency of colloidal crystal formation is increased with the modified meniscus (owing to the mechanical mask attached to the substrate) that reduces the contact area of the dispensed colloidal suspension. (31) Ross, C. Annu. ReV. Mater. Res. 2001, 31, 203. (32) Yablonovitch, E. Phys. ReV. Lett. 1987, 58, 2059–2062.
12146 Langmuir, Vol. 24, No. 21, 2008
Yeon et al.
Figure 4. SEM image of the top and side profiles of colloidal crystal films on a Si substrate formed by surface-tension-assisted self-assembly using 200- and 100-nm-diameter colloids.
Figure 6. Normalized autocorrelation curves of TMR molecules showing the slower temporal decay when the crystal diameter is decreased from 1000 to 100 nm. The noisiness of the 200 nm sample can be attributed to the strong autofluorescence of the 200 nm crystals. This was reproducibly found for this size of spheres for all samples. Whereas it decreases the S/N ratio and thus increases the standard deviation of the parameters determined, it does not influence the average value, and the 200 nm sphere measurements fit very well with the other measurements. (Note that the noise is contributing mostly in the flat initial part of the correlation function at short times but is moderate on the slope and the convergence plateau for longer times.)
Figure 5. (a) Two layers of spheres arranged in fcc fashion. Triangles represent tetrahedral cavities with the pink ones formed from three spheres in the bottom layer whereas the yellow ones are formed from three spheres in the top layer. Octahedral cavities are marked by the green rhombuses. (b) Inverse structure showing an octahedral cavity linked to a tetrahedral cavity. The aperture of the passage is highlighted where the largest circle (with radius rp) that can be drawn in the aperture is also shown.
Diffusion in Colloidal Crystals. Figure 5a shows two layers of spheres arranged in fcc packing. Tetrahedral cavities are formed by one sphere from the top layer with three spheres from the bottom layer or vice versa. Octahedral cavities are formed by six spheres, three from the top and three from the bottom layer. The inverse colloidal crystal geometry in Figure 5b contains tetrahedral and octahedral cavities. Each octahedral cavity is connected to
eight tetrahedral cavities, and each tetrahedral cavity is connected to four octahedral cavities. Owing to the translational periodicity of the crystal structure, it forms a network of interlinked tetrahedral and octahedral cavities. The radius of the pores that link the cavities is denoted as rp (Figure 5b). Using Gambit33 software to draw the crystal structure, we find that rp works out to be 0.156r for spheres of radius r. The particle size is given by the hydrodynamic radius of the diffusing species (rs). TMR has an rs of 0.56 nm. The dextrans have rs values of 1.4, 4.5, and 8.5 nm for the 4.4, 40, and 155 kDa molecular weights. To relate the spatial hindrance effect to the diffusion, the molecule-topore size ratio (λ) is defined as λ ) rs/rp. For a particular particle when the polystyrene sphere diameter is decreased from 1000 to 100 nm, λ increases, which indicates that the spatial hindrance increases. For the range of diffusing molecules and pores that we used, λ varies from 0.007 to 1.087. Figure 4a,c shows strong dislocations consisting of line and point defects that would result in a diffusion time close to free diffusion. Therefore, in FCS measurements we checked for unusual deviations in the measurement. However, diffusion times are not close to free diffusion because defects are only rarely seen. We made at least nine independent readings for each case. The spatial hindrance effect on the diffusion is manifested in the autocorrelation function (ACF) curves as shown for TMR in Figure 6. When the polystyrene sphere diameter is decreased, the diffusion is hindered, and the temporal decay of ACF becomes slower. The ACF curves for all of the molecules were fit to eq 6. (33) Gambit is a software product from Fluent Inc. http://www.fluent.com.
Colloidal Crystals from Self-Assembly
Langmuir, Vol. 24, No. 21, 2008 12147
Table 1. Summary of Physical Parameters of Molecules and Colloids and Fit Parameters Obtained Using Equation 6a species name
MW (Da)
rs (nm)
nanosphere diameter (nm)
rp (nm)
λ
TMR TMR TMR TMR DS DS DS DS DM DM DM DM DL DL DL DL
527 527 527 527 4400 4400 4400 4400 40 000 40 000 40 000 40 000 155 000 155 000 155 000 155 000
0.56 0.56 0.56 0.56 1.4 1.4 1.4 1.4 4.5 4.5 4.5 4.5 8.5 8.5 8.5 8.5
1000 500 200 100 1000 500 200 100 1000 500 200 100 1000 500 200 100
78 39 15.6 7.8 78 39 15.6 7.8 78 39 15.6 7.8 78 39 15.6 7.8
0.007 0.014 0.036 0.072 0.018 0.036 0.089 0.179 0.058 0.115 0.288 0.578 0.109 0.218 0.543 1.087
τD (µs)
R
60 ( 27 0.38 ( 0.07 284 ( 32 0.58 ( 0.02 533 ( 100 0.77 ( 0.05 1196 ( 88 0.87 ( 0.01 182 ( 61 0.53 ( 0.02 472 ( 115 0.75 ( 0.14 1170 ( 50 0.73 ( 0.03 1236 ( 27 0.86 ( 0.01 420 ( 53 0.68 ( 0.07 826 ( 26 0.73 ( 0.01 entrapment entrapment 783 ( 24 0.68 ( 0.01 1270 ( 88 0.86 ( 0.02 entrapment entrapment
a TMR, tetramethyl rhodamine; DS, dextran small; DM, dextran medium; DL, dextran large; MW, molecular weight; rs, hydrodynamic radius; rs, pore radius; λ, molecule-to-pore size ratio; τD, diffusion time; R, temporal exponent in eq 6.
The fit parameters are summarized in Table 1. Above a certain value of λ (>0.28), rs of the diffusing molecules (for 40 and 155 kDa dextrans) becomes comparable to the space for escape rp into the neighboring cavities (in 200 and 100 nm colloids). In this size regime, the molecules start getting entrapped in the cavities. This is similar to the diffusion in agarose gels where the diffusion coefficient showed a steep decrease when the molecular size became comparable to the pore size in the gel. For λ > 0.28, no correlation can be measured (i.e., the correlation curve becomes essentially flat) because the movement of the particles is too slow to be measured by FCS and photobleaching is faster than the hopping of the particle to the next cavity. These cases are marked as “entrapment” in Table 1. When the particle size rs approaches the pore size rp, it becomes more difficult for the particles to pass from one cavity to the next, and diffusion is hindered. At a value of λ > 0.28, free movement can no longer be observed, and particles are entrapped or move so slowly that they are bleached before they can hop to the next cavity. In general, one would expect that R decreases with λ because a larger λ means a more difficult obstacle to overcome for the particle when hopping from one cavity to the next. However, the behavior of R for λ < 0.28 is more complex. From Table 1 and Figure 7, it can be seen that on average R increases with λ but periodically decreases in between. The periodic decrease of R with λ is the expected trend, but the overall increase of R with λ needs further explanation. This more complex behavior stems from the fact that in order to change λ we change the polystyrene sphere size (rsphere) and thus the pore size (rp ) 0.156 rsphere) as well as the particle size (rs). However, FCS measurements depend on the size of the observation volume (robs) in relation to the pore size rp (Figure 8). If the observation volume is much larger than the spheres (robs > rsphere), then the diffusion measured is completely dominated by the intercavity movement (i.e., the time that particles need to hop from one cavity to the next). The value of R is expected to be large, and the diffusion time is expected to be larger than for free diffusion. If the observation volume is much smaller than the spheres (robs < rsphere), then the diffusion measured is the intracavity diffusion, and the value of R is expected to be large as well. For intermediate cases (robs ≈ rsphere), both the intercavity as well as the intracavity movement contribute to the observed diffusion coefficient (i.e., a particle is trapped in a cavity and moves around until it can hop to the next cavity, and thus R is expected to be smaller). The overall increase in R with λ is thus a consequence of the particle size rsphere decreasing in relation to the observation volume
Figure 7. (A) Anomalous diffusion factor R versus the pore size (rp) in the colloidal crystals. The value of R decreases with increasing rp. This corresponds to a transition from situation A to C in Figure 8. (B) In contrast, the value of R increases with increasing particle size rs. This corresponds to a transition from situation B/C to E in Figure 8. (C) Overall, R increases with the ratio of particle to pore size λ ) rs/rp. The dashed lines are guides for the eye to show that R decreases intermittently for increasing λ, although the overall trend is an increase in R with λ. (D) Diffusion time τD versus λ. The closer the particle size comes to the pore size, the slower the diffusive movement measured.
robs. This means that in Figure 8 we are moving from right (rs < rp and robs ≈ rsphere) to left (rs < rp and robs > rsphere), until for very small rp and very large λ we come to the region where particle and pore sizes become comparable (rs ≈ rp and robs < rsphere) and where particles are finally entrapped. These findings are in accordance with the literature, where slow diffusion and entrapment are found when the particles become similar in size to the pores in a gel network.26 Single-Molecule Studies in Colloidal Crystals. To ascertain further that the molecules are indeed confined and slowed in their movement within the cavities of the colloidal crystals when λ > 0.28, single molecules entrapped within the crystals were tracked in time. The concentration of HRP molecules dispensed was 1 nM. Furthermore, because the incubation time was short (10 min), a much lower concentration of particles within the cavities is expected. This means that statistically fluorescence from only single molecules is detected. A strong slowing down of HRP is possible because in the case of 100-nm-diameter colloidal crystals
12148 Langmuir, Vol. 24, No. 21, 2008
Yeon et al.
Figure 8. Schematic of the changes in FCS when the size of the polystyrene spheres of the nanocrystals changes in relation to the FCS observation volume. Whereas for small spheres (left) the average diffusive movement seen is dominated by particle hopping from cavity to cavity (i.e., the intercavity movement), for larger spheres (right) the diffusion measured by FCS is dominated by intracavity movements. In addition, the factor λ ) rs/rp, describing the size of the diffusing particles compared to the size of the pores in the network, determines whether particles can move freely, undergo hindered diffusion, or are entrapped.
Figure 9. Single-molecule detection of entrapped HRP in a 100-nmdiameter colloidal crystal fabricated with surface-tension-assisted selfassembly. Control experiments (traces 1-3) show the absence of fluorescence bursts when (1) DI water, (2) H2O2 + dihdrorhodamine 6G, and (3) HRP molecules were dispensed respectively on the 100 nm crystal. Fluorescence bursts at pixel number 23 in traces 4-9 show the presence of HRP molecules within the cavities by the turning over of dihydrorhodamine 6G in the presence of H2O2. The lateral size of one pixel is 2.2 µm.
λ ) 0.578 > 0.28 and hence a strong slowing down of the HRP is expected. (HRP has a molecular weight of 44 kDa. The molecular dimensions of HRP can be estimated to be 6.0 nm × 3.5 nm × 3.0 nm.) In Figure 9, the control experiments are shown in traces 1-3. The photon counts detected when (1) DI water, (2) H2O2 + dihydrorhodamine, and (3) HRP molecules were dispensed in isolation on a 100 nm crystal show no burst in fluorescence, and the maximum count in these traces is 5. Trace 1 shows that the colloidal crystals do not emit autofluorescence. Trace 2 shows that the low concentration of the substrates employed ensured that no autocatalysis of the dihydrorhodamine occurs, which would increase the background count for the whole line scan.
Trace 3 shows that HRP alone did not have detectable autofluorescence. Trace 4 was recorded on a 100 nm crystal with HRP molecules in the presence of H2O2 + dihydrorhodamine. HRP is entrapped in the cavities and detected when it converted the nonfluorescent substrate (dihydrorhodamine 6G) into fluorescent rhodamine 6G in the presence of H2O2 (Experimental Section). During consecutive scans (traces 5-9), which are 1.1 s apart, the fluorescence persisted at the same location (pixel number 23), indicating that the HRP molecules are entrapped over a period of 5.5 s in the cavity. The difference in intensity of the peaks can be attributed to differences in enzymatic activity or availability of the substrate. After trace 9, the HRP either moved to another cavity or became inactive because no significant fluorescence intensity is detected. We attempted the formation of colloidal crystals using our surface tension self-assembly with both protein and nanospheres dispensed simultaneously and left to dry in order to encapsulate the protein during the formation process of the colloidal crystals. Experiments with 500, 200, and 100 nm nanospheres did not form colloidal crystals. This can be attributed to the fact that the protein introduced into the colloid suspension changes the colloidal surface potential and even small perturbations on the membrane surface induce dramatic changes in the macroscopic organization of the colloids,34 disrupting the formation of wellordered colloidal crystals.
Conclusions Colloidal crystals with nanosized cavities were successfully fabricated by a novel method, called surface-tension-assisted self-assembly, with small volumes of colloidal suspensions at (34) Baksh, M. M.; Jaros, M.; Groves, J. T. Nature 2004, 427, 139–141.
Colloidal Crystals from Self-Assembly
low concentrations. With this method, extra surface tension introduced by the cavity reduces the contact area of the dispensed suspension and increases the efficiency of self-assembly by removing regions of short-range order. The colloidal crystals fabricated provided us with a matrix to measure the effects of hindrance on the diffusion of molecules with different hydrodynamic radii in well-defined but varying sizes of interconnecting cavities. Diffusion in the crystals is found to be anomalous for a molecule-to-pore size ratio ranging from 0.007 to 1.087. When the ratio is larger than 0.28, the molecules are slowed to such an extent that no correlation curves can be measured. This was further corroborated by performing line scans inside 100-nmdiameter colloidal crystals incubated with HRP. By employing a nonfluorescent substrate that forms a fluorescent enzyme-
Langmuir, Vol. 24, No. 21, 2008 12149
product complex, we were able to observe single active HRP molecules over a period of a few seconds. In conclusion, we have established a method of forming colloidal crystals that can act as a controlled matrix for future single-molecule experiments. Acknowledgment. We thank Dr. Lesieur Chungkham Claire for her help and advice on the biological work, and W.C.Y. thanks the National University of Singapore for his research scholarship. B.K. thanks the Singapore Ministry of education for support (R-143-000-230-112). T.W. gratefully acknowledges the Singapore Bioimaging Consortium for support (SBIC 003/2005, R-143-000-284-305). LA800016H
