NANO LETTERS
Thermooptical Properties of Gold Nanoparticles Embedded in Ice: Characterization of Heat Generation and Melting
2006 Vol. 6, No. 4 783-788
Hugh H. Richardson,*,† Zackary N. Hickman,† Alexander O. Govorov,‡ Alyssa C. Thomas,† Wei Zhang,‡ and Martin E. Kordesch‡ Department of Chemistry and Biochemistry and Department of Physics and Astronomy, Ohio UniVersity, Athens, Ohio 45701 Received January 17, 2006; Revised Manuscript Received February 22, 2006
ABSTRACT We investigate the system of optically excited gold NPs in an ice matrix aiming to understand heat generation and melting processes at the nanoscale level. Along with the traditional fluorescence method, we introduce thermooptical spectroscopy based on phase transformation of a matrix. With this, we can not only measure optical response but also thermal response, that is, heat generation. After several recrystallization cycles, the nanoparticles are embedded into the ice film where the optical and thermal properties of the nanoparticles are probed. Spatial fluorescence mapping shows the locations of Au nanoparticles, whereas the time-resolved Raman signal of ice reveals the melting process. From the time-dependent Raman signals, we determine the critical light intensities at which the laser beam is able to melt ice around the nanoparticles. The melting intensity depends strongly on temperature and position. The position-dependence is especially strong and reflects a mesoscopic character of heat generation. We think that it comes from the fact that nanoparticles form small complexes of different geometry and each complex has a unique thermal response. Theoretical calculations and experimental data are combined to make a quantitative measure of the amount of heat generated by optically excited Au nanoparticles and agglomerates. The information obtained in this study can be used to design nanoscale heaters and actuators.
Introduction. Solid-state nanoparticles (NPs) have great potential in modern bionanotechnology as markers, biosensors, and actuators.1,2 The active use of semiconductor NPs as markers is based on their strong fluorescence emission.1,2 Typically metal NPs do not show strong fluorescence; however, they have useful thermal properties. Under optical illumination, Au NPs generate heat efficiently.3-8 The heating effect becomes especially strong under the plasmon resonance conditions when the energy of incident photons is close to the plasmon frequency of a Au NP. In recent papers, the heating effect in Au NPs has been used for several purposes. Reference 3 reports imaging of proteins labeled with Au NPs in cells using an all-optical method based on photothermal interference contrast. Reference 7 describes remote release of materials (drugs) from a capsule containing Au NPs excited with intense light. In ref 9, the authors have assembled a superstructure Au-NP-polymer-CdTe-NP with novel thermal and optical properties. Because of the excitonplasmon interaction, the optical emission of such a superstructure is strongly temperature-dependent.9 A theoretical * Corresponding author. E-mail:
[email protected]. † Department of Chemistry and Biochemistry. ‡ Department of Physics and Astronomy. 10.1021/nl060105l CCC: $33.50 Published on Web 03/18/2006
© 2006 American Chemical Society
paper (ref 10) describes the mechanisms of melting polymer and ice matrixes with embedded gold NPs. Here we investigate the thermooptical properties of gold NPs imbedded into the ice matrix. By recording the timeresolved Raman signal we observe the melting process and determine the threshold melting power, Pmelting(T), where T is the background temperature. Resonant laser light of relatively weak intensity is able to melt ice with embedded Au NPs, whereas even a very intense laser beam does not melt ice alone. This comes from strong absorption in Au NPs in the regime of plasmon resonance. Along with melting spectroscopy, we perform local fluorescence study, which provides us with information about the averaged spatial density of Au NPs, N h AuNP . Overall we see a clear correlation between the melting process and the local density of Au NPs. At the positions with large Au-NP densities, a beam of relatively weak intensity can melt the ice matrix. The transmission electron microscopy (TEM) micrographs of Au NPs showed that the Au subsystem is not in the form of single NPs but form small complexes of various geometries. Assuming a single NP complex inside a laser spot, we can show theoretically that Pmelting ∝ Tred, where Tred ) T/n2/3 AuNP is the reduced temperature. The appearance of the reduced
Figure 1. Experimental setup with a well that contains ice with embedded Au NPs. Inset: 50-nm Au NP excited with a focused laser beam and heat transfer to the ice matrix.
temperature in our description reflects the fact that a complex composed of several NPs generates more heat and can melt ice at a lower temperature. Indeed, we observe that the experimental dependence Pmelting versus Tred is close to a linear function. Simultaneously, we observe relatively large scattering on the graph Pmelting versus Tred. We can understand this observation using the TEM images of NPs. These images show that the geometry of NP complexes varies a lot. According to the theoretical results described in ref 10, the local temperature inside and around a NP complex strongly depends on its geometry and this leads to large scattering for the measured Pmelting as a function of the reduced temperature for different complexes. The optical spectroscopy of NPs in a solution is very different from that in the case of self-assembled quantum dots imbedded into a solid11 because NPs in a liquid are moving. This makes, for example, single-NP spectroscopy very difficult because the integration time is limited by the time the NP remains in the optical focus. From this point of view, the spectroscopy of NPs embedded in the ice has a lot of potential because, in this ice-based spectroscopy, we can “stop” or “release” the NPs simply by freezing or melting the liquid. Another advantage of the ice matrix is that we can perform thermooptical melting spectroscopy using the well-known thermal properties of ice/water. Experimental Section. An alumel/chromel thermocouple (0.003” diameter purchased from Omega Engineering, Inc) was placed on top of a glass coverslip (300 µm thick). A well was created to hold the gold NPs solution by adhering a Coverwell-Silicone Isolator to the coverslip pinning the thermocouple inside the well (see the sketch in Figure 1). The Coverwell Silicone Isolators were purchased from Electron Microscopy Sciences and were 3.0 mm in diameter and 3.0 mm deep. The 50-nm gold NP solution, purchased from British Biocell International, was diluted with ultrapure 18 MΩ water to a concentration of 4.5 × 109 particles/cm3. 784
Solution (20 µL) was pored into the well and sealed with a thin (100 µm thick) coverslip. The sample apparatus was then placed on a Peltier cooler housed inside a nitrogen controlled climate box and the entire assembly was placed under a WiTec Raman/near-field scanning optical microscope. A 10X lens with a working distance of 7 mm was used to collect Raman and photoluminescence spectra using 532-nm laser excitation. The laser intensity was calibrated against the needle stop valve that restricts the opening aperture of the laser. The laser intensity was controlled from 0.2 to 50 mW using this needle stop. The spatial resolution of the microscope operating in this configuration is 1.3 µm. The temperature of the liquid water and ice film was monitored with the alumel/chromel thermocouple immersed in the water/ice sample. The liquid water was cooled until it froze. Typically, water solutions without gold NPs would freeze around -20 °C and solutions of gold NPs would freeze at higher temperatures. Many freeze/thaw cycles were necessary to degas the sample, but once this had been accomplished, large single crystals of ice (∼200 µm on an edge) could be formed having extended vein structures.12 The sample was then cooled to -20.0 °C and low laser intensity (0.5 mW) Raman images were collected to find the gold NPs. Results and Discussion. Figure 2 a is a Raman intensity map of a 100 × 100 µm2 section of 50-nm gold NPs embedded in ice. The grain boundaries revealed a diminished ice Raman intensity but showed a broad photoluminescence band centered around 600 nm (Figure 2b). This band is assigned to plasmon emission of excited gold NPs. Similar optical emissions of Au NPs were recorded previously in refs 13 and 14. The crystal structure of ice observed in our experiments exhibits veins formed at intersections of grain boundaries (Figures 1 and 2).15 The largest emission from the gold NPs is observed at the grain boundaries with a much smaller clustering of emission within the ice crystals. The Nano Lett., Vol. 6, No. 4, 2006
Figure 2. (a) Raman intensity map of a 100 × 100 µm2 section of ice with embedded 50-nm Au NPs. Inset: the Raman spectrum with the line of ice. (b) Photoluminescence intensity map of the same spatial area. 50-nm Au NPs are mostly trapped at the grain boundaries in the veins and exhibit a broad emission line around 600 nm. Inset: photoluminescence spectrum of 50-nm Au NPs. Lower panel: temperature, power at which the melting process was recorded, and estimated number of Au NPs within the laser spot for several points labeled as 1-11 in figure b.
number of gold particles within the excitation area of 1.3 µm diameter was determined by comparing the photoluminescence from gold NPs in a liquid solution with known concentration to the photoluminescence of gold NPs in ice. An estimate for the number of NPs was obtained simply by dividing the PL signal by the signal related to one NP in a liquid solution. This calibration could not be used to determine the structure or configuration of the gold NPs within the excitation area but only the number of particles within the excitation area. Once the location and number of gold particles within an area of ice were determined, the threshold laser intensity that caused melting at a particular temperature was probed. The melting of ice in the absence of gold NPs was checked by irradiating grains of ice that did not have any gold NPs. These regions did not melt even at the highest laser intensity of 50 mW. The sample was moved to a previously determined spot containing a known number of particles, and time-resolved Raman/emission spectra were collected for a known laser intensity and sample temperature. This isothermal experiment was repeated, increasing the laser intensity until a threshold laser intensity was found that caused a decrease in the ice Raman intensity. Time-resolved Raman spectra were collected and a decreasing Raman signal from ice indicated melting within the excitation area. Figure 3 shows a typical family of time-resolved spectra collected at a particular spot. The melting was confirmed visually as a crater that was optically observable with the microscope. Another spot was selected with a different number of particles, and the entire process was repeated using different temperatures. This Nano Lett., Vol. 6, No. 4, 2006
Figure 3. The normalized Raman intensity of the ice peak as a function of time for three different intensities of the laser beam at position 2 in the map shown in Figure 2b. Solid lines are drawn for guidance. The temporal decrease of the Raman signal for P ) 10 mW indicates the melting process when P > Pmelting.
experiment was repeated for 11 spatial points on the 100 × 100 µm2 section of ice with embedded 50-nm Au NPs (Figure 2b and lower panel). This procedure allowed us to map out the threshold laser intensity that caused melting as a function of temperature and particle number (see also Figure 5). We should stress that the mapping of Au NPs and ice (Figure 2a and b) was performed noninvasively, that is, at a 785
Figure 4. (a) Raman intensity map of section of ice with embedded 50-nm Au NPs performed at low temperature T ) -20 °C and small laser power P ) 0.5 mW. The above conditions were chosen to avoid melting in the system during mapping. Before mapping, the massive melting process was created in the system by scanning with an intense laser beam (P ) 10 mW) at relatively high temperature T ) -2.5 °C. The massive melting led to the formation of a crater, seen as a black dot in the Raman map. The crater appeared around an agglomerate of Au NPs. (b) Photoluminescence intensity map of the same spatial area. We can see a large agglomerate of 50-nm Au NPs as a yellow spot. Inset: photoluminescence spectrum of 50-nm Au NPs.
very small intensity power that is not able to melt ice. Then, for several selected locations, we increased the laser power and observed the melting process as a temporal decrease of the Raman signal of ice. The above experiments were done for the spots with a relatively small number of Au NPs (less than 10; see the lower panel in Figure 2). The follow-up TEM measurements showed that there are also large agglomerates of Au NPs; such agglomerates can cause a very intensive melting process, which was observed. Figure 4 shows the Raman and photoluminescence image of gold NPs located at the grain boundaries and a large Au-NP agglomerate that is embedded in the ice. Excitation of this particle caused a massive melting of the ice that resulted in the formation of a large crater. This type of heating creates large circular liquid fronts of over 200 µm in diameter (Figure 4b) that travel at speeds greater than 2500 µm/s. Theoretical Model. It is interesting to compare the recoded melting laser powers with the results of simple calculations. First we start from a single spherical NP embedded into boundaries media of the dielectric constant ice ) 1.8. As for the dielectric function of gold (Au), we will take it from the tables.16 Because heat conduction is relatively fast inside the ice, we will assume the steady-state regime, that is, the local temperature increase ∆T(r) reaches its steady-state value. The typical times in our time-resolved Raman study are tens of seconds, and the estimated times to establish the thermal steady-state around a single Au NP are in the microsecond range.17 Our theory will describe the system in the solid ice state. In other words, we assume that T(r) < 0 °C at any location r. The maximum Tmax is at the surface of the NP, and the melting process will start there.10 With our calculations, we will estimate at which power (Pmelting) the temperature at the surface of Au NP reaches 0 °C. With further increase of light intensity, a shell of water 786
around the Au NP will form, as described in detail in ref 7. It follows from the solution of heat transfer equations that the function ∆T(r) decays away from the heater as 1/r.18 At the surface of the NP the temperature is maximum and is given by ∆Tmax (I0) )
R2AuNP Re 3kice 1 - Au(ωlaser) 3ice iω 8π 2ice + Au(ω)
[
|
|] x 2
8π‚I0
(1)
c ice
where RAuNP is the NP radius, kice is the thermal conductivity of ice, and I0 and ωlaser are the light intensity flux and frequency, respectively. The light intensity flux is given by the illumination power and the laser-spot area: I0 ) P/A. With a laser power of 1 mW and a spot diameter of 1.3 µm, we obtain light fluxes about 105W/cm2. It becomes obvious from TEM micrographs that NPs in ice form complexes. To determine an average number of Au NPs within the laser spot volume N h AuNP , we divide the local fluorescence intensity by the fluorescence intensity related to one NP. The latter is obtained from the emission experiments with the solution. Figure 5a shows the melting power for different spots (see Figure. 2) at different temperatures. For four spots, we observed that the florescence intensity corresponds to about one Au NP (blue dots in Figure 5a). In this case our theory for a single NP is directly applicable. To compare our theory with the experimental data, we assume that the melting starts at the surface of the NP where the temperature increase is maximum. Then, we can compute the power needed to achieve the temperature T ) 0 °C at the surface of Au NP. The corresponding equation for the melting light power Pmelting has the form ∆Tmax (I0) ) -T, where ∆Tmax (I0) is Nano Lett., Vol. 6, No. 4, 2006
Figure 5. (a) Black and blue dots represent the experimental data for the melting power from positions 1-11 (see Figure 2b) as a function of the thermostat temperature. The blue dots correspond to the four positions with estimate number of Au NPs N h Au NP ≈ 1. We note that the powers shown in the lower panel of Figure 2 correspond to the positions of the upper limit of the error bars in this Figure. The dots are places in the middle of the error bars. Red dots are the calculated data for a single Au NP of 50 nm diameter. Inset: calculated heat power released by a single 50-nm Au NP as a function of the laser power. Arrows show the case for Pmelting ) 4 mW. (b) Dots represent the experimental data for the 2/3 . melting power as a function of the reduced temperature, T0/NAnNP The red curve comes from the simple model described in the text. Insets: TEM micrographs of several 50-nm Au NP complexes found in the solution after melting the ice.
given by eq 1 and T is the background (thermostat) temperature in °C. The red dots in Figure 5a are the calculated melting powers for NAuNP ) 1. We see an overall good agreement with three red dots very close to the corresponding experimental data and one red dot within the experimental error bar. The above calculations were done for the laser wavelength λ ) 530 nm. This wavelength is close to the plasmon resonance of the Au NPs. The inset of Figure 5a shows a calculated magnetite of heat flux intensity from a single Au NP as a function of the light power. Using these theoretical data and experimentally measured melting intensity, we can determine the amount of heat generated by a single optically driven Au NP. For example for the experimental point T ) -21.2 °C, N h AuNP ) 1, and Pmelting ) 4 mW, we estimate that the heat flux intensity of a nanoheater is about 9.6 µW. Nano Lett., Vol. 6, No. 4, 2006
We can also try to estimate the heat generated by complexes made of several/many NPs. In ref 10, the authors theoretically show that the heat and temperature increase in metal-NP assemblies depends strongly on the geometry/ organization of the NPs. The exact theoretical treatment performed in ref 10 includes numerical calculations using the coupled Maxwell and thermal equations and is rather complex. Here we intend to obtain only estimates. We now assume that the NP complex is tightly packed and approximate the complex as a single NP of larger volume: Vtot ) NAuNP‚VAuNP , where NAuNP and VAuNP are the number of NPs in the complex and the volume of a single NP, respectively. The effective radius of the complex then 3 becomes equal to Reff ) x3/4πNAuNP‚VAuNP. The temperature increase at the surface of such a Au NP with an effective radius Reff is given by eq 1 with the substitution RAuNP f Reff. Therefore, the resultant temperature increase is ∆Tmax (I0) ∝ R2eff ∝ N2/3 AuNP. In other words, the temperature increase in this simplified model scales as N2/3 AuNP and Pmelting ∝ T/N2/3 . In Figure 5b, we plot the melting power as a AuNP function of the reduced temperature T/N2/3 and see that AuNP the theoretical line agrees with several experimental points within the error bars, and for other experimental points, the theoretical values are not far from the experimental data. In Figure 5a and b, we observe a relatively large scattering of experimental data points. This can be due to the aforementioned mesoscopic character of the system, where each NP complex has unique geometry and therefore unique thermal properties. In addition, the complexes can be at different distances from the top surface. When a complex is close to the surface, the temperature increase is expected to be larger (approximately by factor 2) because the heat flux becomes “reflected” from the surface. This may be the reason that few experimental data points in Figure 5a and b appear at small melting powers. We also should note that the solution (1) is not noticeably affected by the lower icethermostat interface where the temperature is kept constant (Figure 1). The reason is that the distance from the optically excited NPs to the lower interface is about 2 mm, which is significantly larger than the diameter of a NP (50 nm). At the same time, this ice-thermostat interface plays an important role in the experiment by removing the heat from the sample. We note that the heat generated by Au NPs located closely to the upper ice-air interface reaches the bottom interface within about few seconds,17 while our Raman study is performed during 30 s. Conclusions. To conclude, we performed a study of heat generation from single/few Au NPs using Raman and photoluminescence spectroscopy. Using ice as a matrix and observing the melting effect, we can determine the amount of heat generated by nanoheaters. It turns out that the icebased spectroscopy of NPs is a convenient experimental method. At small light powers, we mapped the sample and estimated an average number of Au NPs within the laser spot. With gradually increasing power, we determined the threshold melting intensity, which can be used to estimate the heat power generated by optically driven NP complexes. Large power beams created irreversible changes in the 787
system, such as craters after an almost explosive melting process. We think that such invasive changes in the system appeared when a laser beam hit large NP agglomerates. Acknowledgment. This work was supported by the BioNanoTechnology Initiative at Ohio University. References (1) For a review see MRS Bull. 2005, 30, 331-408. (2) (a) Milliron, D. J.; Hughes, S. M.; Cui, Y.; Manna, L.; Li, J.; Wang, L.-W.; Alivisatos, A. P. Nature 2004, 430, 190-195. (b) Oh, M.; Mirkin, C. A. Nature 2005, 438, 651-654. (3) Cognet, L.; Tardin, C.; Boyer, D.; Choquet, D.; Tamarat, P.; Lounis, B. PNAS 2003, 100, 11350. (4) Boyer, D.; Tamarat, P.; Maali, A.; Lounis, B.; Orrit, M. Science 2002, 297, 1160. (5) Hu, M.; Hartland, G. V. J. Phys. Chem. B 2002, 106, 7029. (6) Maillard, M.; Pileni, M.-P.; Link, S.; El-Sayed, M. A. J. Phys. Chem B 2004, 108, 5230. (7) Skirtach, A. G.; Dejugnat, C.; Braun, D.; Susha, A. S.; Rogach, A. L.; Parak, W. J.; Mohwald, H.; Sukhorukov, G. B. Nano Lett. 2005, 5, 1372. (8) Pitsillides, C. M.; Joe, E. K.; Xunbin Wei; Anderson, R. R.; Lin, C. P. Biophys. J. 2003, 84, 4023.
788
(9) Lee, J.; Govorov; A. O.; Kotov, N. A. Angew. Chem. 2005, 117, 7605. (10) Govorov, A. O.; Zhang, W.; Skeini, T.; Richardson, H.; Lee, J.; Kotov, N. A. Nanoscale Res. Lett. 2005, 1, 100101. (11) Warburton, R. J.; Urbaszek, B.; McGhee, E. J.; Schulhauser, C.; Ho¨gele, A.; Karrai, K.; Govorov, A. O.; Garcia, J. M.; Gerardot, B. D.; Petroff, P. M. Nature 2004, 427, 135. (12) Richardson, H. J. Mol. Struct., in press, 2006. (13) Dulkeith, E.; Niedereichholz, T.; Klar, T. A.; Feldmann, J.; von Plessen, G.; Gittins, D. I.; Mayya, K. S.; Caruso, F. Phys. ReV. B 2004, 70, 205424. (14) Varnavski, O. P.; Mohamed, M. B.; El-Sayed, M. A.; Goodson, T., III. J. Phys. Chem. B 2003, 107, 3101-3104. (15) Nye, J. F. J. Glaciol. 1989, 35, 17-22. (16) Palik, E. D. Handbook of Optical Constants of Solids; New York, Academic Press: 1985. (17) The laser spot has a diameter of 1.3 µm. The characteristic time to establish the steady-state temperature within the laser spot under constant illumination can be estimated as: ∆ttherm ) d2/Kice, where the thermal diffusivity of ice Kice ≈ 7.8 × 10-7 m2/s. The time for heat to diffuse from the surface regions, excited by light, to the substrate, located 2 mm down, is about 5.2 s. (18) Carslaw, H. S.; Jaeger, J. C. Conduction of Heat in Solids; Oxford University Press: London, 1993.
NL060105L
Nano Lett., Vol. 6, No. 4, 2006
