Single Molecular Stamping of a Sub-10-nm ... - ACS Publications

We introduce a nanoscale stamping technique of sub-10-nm colloidal quantum dot (QD) arrays to highly localized areas of three-dimensional nanostructur...
0 downloads 0 Views 1MB Size
13804

Langmuir 2008, 24, 13804-13808

Single Molecular Stamping of a Sub-10-nm Colloidal Quantum Dot Array Kazunori Hoshino,*,† Travis C. Turner,† Sunmin Kim,‡ Ashwini Gopal,† and Xiaojing Zhang*,† Department of Biomedical Engineering and Microelectronics Research Center, The UniVersity of Texas at Austin, 10100 Burnet Road, J. J. Pickle Research Campus, Austin, Texas 78758 ReceiVed September 6, 2008. ReVised Manuscript ReceiVed October 6, 2008 We introduce a nanoscale stamping technique of sub-10-nm colloidal quantum dot (QD) arrays to highly localized areas of three-dimensional nanostructures using a quartz tuning fork employed as the stamp pad (the “Nano Stamp”). CdSe/ZnS core-shell nanoparticles with diameters of 9.8 nm were deposited on microfabricated silicon probe tips. The number of transferred QDs, which ranged from several thousands down to single molecular order (less than 10), was precisely controlled by adjusting the stamping depths and angles. The stamping areas were varied from 1.2 µm × 1.2 µm down to 30 nm × 30 nm. Using the Nano Stamp, QDs can be transferred to a variety of protruding nanostructures. The amount of particles transferred to the tip was assessed by fluorescence intensity measurements, and the number of particles was estimated by direct transmission electron microscopy (TEM) observation. Correlation between the fluorescence intensity and the observed stamping depth and the approaching angle of the tip was found, demonstrating the efficacy of our Nano Stamp technique.

Introduction Nanoparticle attachment to a scanning probe tip has been studied to add supplementary functions to atomic force microscopy (AFM) probes. Colloidal quantum dots (QDs) are promising materials for tip modification, especially for near-field scanning optical microscopy (NSOM). The extremely small dimensions and the highly tunable luminescence properties of QDs have found applications in high resolution labeling of biosamples.1,2 The high locality of colloidal QDs can potentially enhance NSOM resolution to the order of single quantum dot molecule (with typical diameter of 3-10 nm) as previously suggested with a single fluorescent molecule in ref 3. Several approaches have been considered to utilize colloidal QDs in NSOM. CdSe/ZnS core-shell QDs were deposited on the entire surface of the probe tip in refs 4 and 5. A diluted solution of QDs in poly(methyl methacrylate) (PMMA) was used to coat a fiber-based standard NSOM probe with a small number of quantum dots in ref 6. In both cases, QDs were attached without any position control, and the location of the QDs effective in actual imaging is not clear. Position control of QD attachment has yet to be thoroughly investigated to exploit the fascinating nature of colloidal quantum dots. Recent studies have been reported on attaching nanowires or particles at the probe tip,7-9 mostly through labor-intensive * To whom correspondence should be addressed. [email protected]. edu (K.H.); [email protected] (X.J.Z.). † The University of Texas at Austin. ‡ Present address: Department of Biological Engineering, Cornell University. (1) Kloepfer, J. A.; Cohen, N.; Nadeau, J. L. J. Phys. Chem. B 2004, 108, 17042. (2) Voura, E. B.; Jaiswal, J. K.; Mattoussi, H.; Simon, S. M. Nat. Med. 2004, 10, 993. (3) Michaelis, J.; Hettich, C.; Mlynek, J.; Sandoghdar, V. Nature 2000, 405, 325. (4) Vickery, S. A.; Dunn, R. C. J. Microsc. 2001, 202, 408. (5) Shubeita, G. T.; Sekatskii, S. K.; Dietler, G.; Potapova, I.; Mews, A.; Basche´, T. J. Microsc. 2003, 210, 274. (6) Chevalier, N.; Nasse, M. J.; Woehl, J. C.; Reiss, P.; Bleuse, J.; Chandezon, F.; Huant, S. Nanotechnology 2005, 16, 613. (7) Dai, H.; Hafner, J. H.; Rinzler, A. G.; Colbert, D. T.; Smalley, R. E. Nature 1996, 384, 147.

manual assembly and manipulation under microscopic observation.7 Some methods utilize force sensing capability of the probe itself to control tip-nanoparticle interactions,8 while others require the probe tip to be surface modified to trap nanoparticles.9 Many of these techniques are no longer applicable to particles as small as 3-10 nm such as colloidal quantum dots. Our nanoscale stamping transfer (the “Nano Stamp”) is a highly localized and controlled version of microcontact printing.10 The number of transferred particles can be reduced down to single molecular order. A force sensing tuning fork is introduced to precisely perform the stamping transfer. The process is highly automated, and practically any kind of protruding nanostructure can be used as the destination substrate. Our method holds the same benefits as microcontact printing. First, particles to be transferred can be prepared by well-studied methods such as Langmuir-Blodgett (LB)11 or Langmuir-Schaefer (LS)12 technique. Well-packed layers of functional nanoparticles can be reliably deposited without any damage to the substrate. Second, practically any kind of solid material can be used for the target substrate.

Experimental Section Experimental Setup. Figure 1 shows the experimental setup of our nanoscale stamping transfer. The unique feature of the Nano Stamp is that the force sensing capability is integrated on the source side, rather than on the target as shown in many previous studies of nanowire/nanoparticle manipulations with scanning probes. This significantly reduces requirements on modifying the scanning probe’s physical characteristics, such as size, shape, stiffness, and resonant frequency. In addition, high-throughput stamping experiments can be carried out without the need to install the probe onto an already complicated scanning microscopy setup. (8) Junno, T.; Deppert, K.; Montelius, L.; Samuelson, L. Appl. Phys. Lett. 1995, 66, 3627. (9) Vakarelski, I. U.; Higashitani, K. Langmuir 2006, 22, 2931. (10) Santhanam, V.; Andres, R. P. Nano Lett. 2004, 4, 41. (11) Dabbousi, B. O.; Murray, C. B.; Rubner, M. F.; Bawendi, M. G. Chem. Mater. 1994, 6, 216. (12) Santhanam, V.; Liu, J.; Agarwal, R.; Andres, R. P. Langmuir 2003, 19, 7881.

10.1021/la802936h CCC: $40.75  2008 American Chemical Society Published on Web 11/08/2008

Molecular Stamping of Sub-10-nm Colloidal QD Arrays

Figure 1. Schematics of the Nano Stamp. (a) Experimental setup. A self-assembled array of colloidal QDs was prepared on a PMMA precoated tuning fork. The amplitude of a mechanically oscillated tuning fork is measured by a lock-in amplifier to control the probe tip position. (b) Control parameters used in the experiment. Approaching angle was adjusted to 70° or 60° to ensure the QD deposition on the probe’s top surface.

A quartz tuning fork, oscillated at its resonant frequency, serves as the force sensor.13 An external piezo vibrator is used to induce the oscillation amplitude roughly estimated to be 1 nm. The typical resonant frequency and the Q factor are 32.6 kHz and 3000, respectively. The oscillation dampening, through the impression of the approaching probe tip on the QD layers, is monitored with a lock-in amplifier built in a LabVIEW virtual instrument, which enables the position control of the probe tip with a resolution greater than 0.5 nm. Silicon probe tips were prepared by standard wet etching of a single crystal silicon wafer. They were fabricated based on the design reported in ref 14. Scanning electron microscopy (SEM) photographs of the tip are shown in Figure 2. The probe retains a flat top surface, which is used as the destination surface for the transfer. A piezo stage precisely displaces the probe tip to impress the surface of a tuning fork prepared with self-assembled QDs, which results in highly localized transfer of QDs at the probe tip. Unlike the conventional noncontact feedback control introduced for scanning probe microscopy, the piezo stage allows the tip to penetrate the QD layer until the depth reaches the preset value. The PMMA coating, whose Young’s modulus (typically 2-3 GPa) is smaller than that of silicon (130-190 GPa) by approximately 2 orders of magnitude, serves as a “stamp pad” which reduces the likelihood of damage to the probe tip while stamping. QD Film Preparation. We used the CdSe/ZnS core-shell particles acquired from Evident Technologies, NY. QDs with different average diameters, namely, 9.8, 9.0, 7.8, and 7.4 nm (emission wavelength peaks 620, 600, 560, and 540 nm, respectively), were tested and successfully stamped using the same stamping procedures. The smaller band gap of the larger QDs allows easier excitation as reported in QD-based light emitting diode studies15,16 and is suitable for our future testing. Here, we use the QDs with an average diameter of 9.8 nm for further measurement and evaluation as a near-field light emitting source. Close-packed colloidal QDs were prepared on a tuning fork based on a method reported in ref 12. First, the QDs were precipitated and (13) Karrai, K.; Grober, R. D. Appl. Phys. Lett. 1995, 66, 1842. (14) Hoshino, K.; Rozanski, L. J.; VandenBout, D. A.; Zhang, X. J. J. Microelectromech. Syst. 2008, 17, 4. (15) Dorn, A.; Huang, H.; Bawendi, M. G. Nano Lett. 2008, 8, 1347–1351. (16) Caruge, J. M.; Halpert, J. E.; Wood, V.; Bulovic, V.; Bawendi, M. G. Nat. Photonics 2008, 2, 248.

Langmuir, Vol. 24, No. 23, 2008 13805

Figure 2. SEM photographs of the silicon probe. (a) Oblique-view and (b) side-view. The probe was made by anisotropic wet etching of a silicon wafer. It has a flat top surface which originates from the surface of the wafer.

Figure 3. TEM micrograph of self-assembled QDs. They were prepared on a carbon film coated TEM grid following the same recipe as that for the tuning fork. The image suggests that particles are close-packed with a thickness variation of 1-4 monolayers.

resuspended in hexane to remove excess ligands. Then 0.032 nmol of the quantum dots was mixed into 400 µL of a 50:50 (v/v) solvent of hexane and 1,2-dichloroethane. This hydrophobic colloidal suspension was dispensed onto a 2 cm diameter convex water surface, pinned at the edge by a Teflon disk. As the solvent evaporates, the QDs form a uniform array due to capillary immersion and convective forces. This floating film was then gently touched and picked up with the PMMA precoated tuning fork. Figure 3 shows a transmission electron microscopy (TEM) image of QDs on a carbon film coated TEM grid prepared by the technique described above. TEM and AFM measurements suggest the thickness of the prepared film to

13806 Langmuir, Vol. 24, No. 23, 2008

Figure 4. Diagram showing typical force curves for (a) shallow stamping and (b) deep stamping. As the tip started impressing the tuning fork, oscillation dampening was measured in the force curve at point A. Hystereses were found for both cases as the longer backward distances (distance from B to C) than the stamping depth (distance from A to B).

be 1-4 monolayers. The particles on the tuning fork can be transferred to destination surfaces within a few hours after sample preparation.10 Here, we conducted all the stampings within 30-60 min to maintain temporal consistency for performance comparison.

Experimental Results Figure 4 shows typical response curves of the tuning fork oscillation, with Figure 4a showing a shallower stamping and Figure 4b the deeper case. After the tip contacts the surface, as is clearly shown at point A, the oscillation decreases as the tip penetrates the QD layers. The stamping depth was precisely controlled by monitoring the tuning fork amplitude. The amplitude was dampened to almost zero with a deeper penetration, as shown by point B in Figure 4b. Another control parameter is the angle with which the probe approaches the tuning fork, as shown in Figure 1b. While the piezo stage draws the tip back after the stamping, hysteresis was found in the oscillation amplitude. This is due to the sticking of the tip to the surface, which causes the backward distance to be larger than the penetration depth. After the stamping procedures, each probe tip was directly observed by TEM to estimate the area and the amount of transferred particles. Figure 5 shows TEM micrographs of four typical samples. Successful transfer of close-packed particles was observed for each sample. No physical damage of the tip was observed. For samples (a) and (b), multiple layers of particles are also visible as darker areas. Sample (d) shows single molecular order (less than 10) transfer of colloidal quantum dots. The estimated number of transferred particles was 8000, 600, 30, and

Hoshino et al.

5 for samples (a), (b), (c), and (d), respectively. The particles tended to be deposited onto the most proximal area of the probe tip, which touched the stamp pad first. In cases where small bumps were formed around the tip, denser deposition of particles was found on the bumps. In the experimental setup, the approaching angle of the probe toward the stamp pad can be adjusted to control both the amount and the location of QDs to be deposited. In cases with larger approaching angles, we observed deposition of particles at the tip apex. Here, in order to better control and quantify the amount of particles ranging from a single molecular order to thousands, we adjusted the approaching angles so that the particles were deposited on the top surface of the probe, while at the same time particles were deposited slightly away from the apex. Occasionally, small pieces of PMMA are found to attach to the tip, especially when the stamping depth was larger than 500 nm and the approaching angle was close to 90°. These PMMA pieces were supposedly taken away from the stamp pad. We decided to use stamping depths smaller than 400 nm with the approaching angles of 60° and 70° in the following measurements. The amount of transferred particles was evaluated for different stamping depths and approaching angles by fluorescence measurement. A standard mercury lamp with a filter set of an excitation filter (peak wavelength, 535 nm; bandwidth, 50 nm), a dichroic mirror (565 nm long-pass), and an emission filter (peak wavelength, 610 nm; bandwidth, 75 nm) was used. Figure 6 shows fluorescence images of probe tips with transferred particles. The stamping depths were 290, 150, 85, and 35 nm, with an approaching angle of 60° for all the cases. The brightness of the image in Figure 6d is enhanced from the original only for printing purposes. We did not observe the blinking emission as reported in ref 6 with the cases of smaller numbers of deposited QDs. This is mainly due to the fact that we used an excitation setup with a standard color CCD, which covers a wide range of fluorescent intensities but is not sensitive enough to directly record single molecular-order fluorescence in real time. The required relatively long integration time (typically more than 4 s) of the CCD may have made the on-and-off states of the QDs less evident. We believe that direct TEM observation has proved the single molecular deposition at the probe tip. For further characterization of the emission from fewer numbers of QDs, time-resolved fluorescent measurement with photon counting sensitivity will be needed. Fluorescence intensities were plotted as functions of the stamping depth (Figure 7) and backward distance (Figure 8) with two different approaching angles of 70° (Figures 7a and 8a) and 60° (Figures 7b and 8b). If we assume the number of deposited QDs to be proportional to the contacting area of the tip, the amount of QDs (y) should be proportional to the second power of the stamping depth d. If we also consider the stamping depth, the amount of QDs should be proportional to the third power of the stamping depth. Since we have only 1-4 monolayers of QDs on the tuning fork, transferred QDs cannot always be proportional to the stamping depth. The actual curve should be approximated as somewhere between y ) kd2 and y ) kd3, where k is a constant. In either approximation, a linear fit around the shallower stamping depths still seems appropriate to show the positive correlation in the measurements with relatively large variations as in Figures 7 and 8. A positive correlation with R2 ) 0.622 and 0.362 was found for a linear fit of the intensity-stamping depth relation as shown in Figure 7. A possible main cause of the variation may be the nonuniformity of the QD thickness (approximately 1-4 monolayers) in the QD film preparation as found in Figure 3. An

Molecular Stamping of Sub-10-nm Colloidal QD Arrays

Langmuir, Vol. 24, No. 23, 2008 13807

Figure 5. TEM images of typical probes. Estimated numbers of transferred particles are (a) 8000, (b) 600, (c) 30, and (d) 5 on the probes.

Figure 6. Fluorescence images of probe tips with transferred particles. The stamping depths were 290, 150, 85, and 35 nm, and the approaching angle was set at 60° for all the cases.

exceptionally bright case indicated by a black arrow in Figure 8b was probably made of a large chunk of QDs, which was sometimes made in the QD film preparation. Without the brightest point, the linear correlation for the 60° case is R2 ) 0.663, which is almost the same as that in the case of 70°. The 60° case tends to transfer more QDs than the 70° case. Better correlations (R2 ) 0.7906 and 0.493) were found for the intensity-backward distance plot in Figure 8 than for the intensity-stamping depth plot in Figure 7 for both approaching angles. The backward distance, which was found as the hysteresis in the force curves in Figure 4, is related to the force required to detach the probe tip from the tuning fork. This also means that the backward distance can indirectly indicate the number of transferred particles that are detached from the tuning fork. Since the precise values of backward distances are measurable but unpredictable, the stamping depth, which is controllable by the proposed setup, is practically more important. Nevertheless, the good linearity of the backward distance still holds important

Figure 7. Fluorescence intensities plotted for stamping depths with approaching angles of (a) 70° and (b) 60°. The bracketed equations in (b) are calculated without an exceptionally bright case indicated by the black arrow.

information in characterizing the particle preparation and the transfer.

13808 Langmuir, Vol. 24, No. 23, 2008

Hoshino et al.

Manipulation of colloidal QDs in a very small number is becoming an important issue in fabricating colloidal QD-based devices such as light emitting diodes (LEDs),15,17 transistors,18 and biochemical sensors.19,20 Creation of a nanometer-scale LED at the scanning probe tip21 is one of our promising applications of Nano Stamp. The Nano Stamp can be further applicable to several kinds of particles and surfaces as have been tested in previous studies of microcontact printing.10,12,22,23 Our test stamping on the tip of drawn optical fibers, which are commonly used in standard NSOM, showed promising results. A vast variety of particles, including polymer, metal, semiconductor, or diamond nanoparticles, can be deposited using a similar procedure. Attachment of fluorescent nanodiamonds at the probe tip is a topic of current interest in magnetometry at the nanoscale.24-26 Successful single molecular order transfer of nanoparticles on three-dimensional nanostructures leads to the creation of future scanning probes, sensors, and quantum logic devices.

Figure 8. Fluorescence intensities plotted for several backward distances with the approaching angles of (a) 70° and (b) 60°. These show better correlation than the stamping distances, which indicates that the backward distance is highly related with the amount of particles detached from the stamp pad. The bracketed equations in (b) are calculated without the exceptionally bright case indicated by the black arrow.

Conclusions We developed a new technique, the Nano Stamp, for highly localized nanoparticle stamping on three-dimensional nanostructures. A self-assembled layer of sub-10-nm QDs was transferred to the destination probe tip as the tip impressed the particles prepared on the PMMA precoated force-sensing tuning fork. Well-controlled amounts of particles were deposited ranging from several thousands of particles down to single molecular order (less than 10) with deposition areas of approximately 1.2 µm × 1.2 µm and 30 nm × 30 nm. The amount of deposited QDs can be fine-tuned through stamping control, as shown by the measured correlation curve between the stamping depth and the fluorescence intensity. Considering the sub-nanometer control capability of the tuning fork based the positioning system,13 we did not see major technical difficulties which may impede the realization of single molecule deposition at the probe apex in the near future.

Acknowledgment. This research was performed in part at the Microelectronics Research Center (MRC) at the UT Austin and the National Nanofabrication Infrastructure Network (NNIN) supported by National Science Foundation (NSF NNIN-0335765). We also thank NSF EPDT Program (ECS-26112892), NSF IMR Program (DMR-0817541), NSF CMMI Program (CMMI0826366), UT Research Grant, the Welch Foundation, and The Strategic Partnership for Research in Nanotechnology (SPRING) for partial financial support of this work. LA802936H (17) Hoshino, K.; Yamada, K.; Matsumoto, K.; Shimoyama, I. J. Micromech. Microeng. 2006, 16, 1285. (18) Klein, D. L.; Roth, R.; Lim, A. K. L.; Alivisatos, A. P.; McEuen, P. L. Nature 1997, 389, 699. (19) Fritz, J.; Baller, M. K.; Lang, H. P.; Rothuizen, H.; Vettiger, P.; Meyer, E.; Gu¨ntherodt, H. J.; Gerber, Ch.; Gimzewski, J. K. Science 2000, 288, 316. (20) Medintz, I. L.; Clapp, A. R.; Mattoussi, H.; Goldman, E. R.; Fisher, B.; Mauro, J. M. Nat. Mater. 2003, 2, 630. (21) Hoshino, K.; Rozanski, L. J.; VandenBout, D. A.; Zhang, X. J. Appl. Phys. Lett. 2008, 92, 131106. (22) Yan, X.; Yao, J.; Lu, G.; Chen, X.; Zhang, K.; Yang, B. J. Am. Chem. Soc. 2004, 126, 10510. (23) Cayre, O.; Paunov, V. N.; Velevb, O. D. J. Mater. Chem. 2003, 13, 2445. (24) Degen, C. L. Appl. Phys. Lett. 2008, 92, 243111. (25) Sonnefraud, Y.; Cuche, A.; Faklaris, O.; Boudou, J. P.; Sauvage, T.; Roch, J. F.; Treussart, F.; Huant, S. Opt. Lett. 2008, 33, 611. (26) Taylor, J. M.; Cappellaro, P.; Childress, L.; Jiang, L.; Budker, D.; Hemmer, P. R.; Yaocoby, R.; Walsworth, R.; Likin, M. D. Nat. Phys. 2008, 4, 810.