COMMUNICATIONS

A Continuous-Flow Microcapillary Reactor for the Preparation of a Size Series of CdSe Nanocrystals**
By Brian K. H. Yen, Nathan E. Stott, Klavs F. Jensen, and Moungi G. Bawendi* Colloidal semiconductor nanocrystals (NCs) have been extensively studied in recent years for use in a variety of applications including biological fluorescent labels,[1] electroluminescent devices,[2] and lasers.[3] For any of these applications, it is essential to begin with high-quality NCs, and advances in the synthesis of IIVI and IIIV NCs have made it possible to prepare relatively monodisperse, highly crystalline samples. In preparations with organometallic precursors, NCs are often prepared in a batch process in which the precursors are rapidly injected into a heated flask containing a mixture of solvents and coordinating ligands.[46] However, the quality and average size of NCs synthesized in the batch process can depend strongly on factors which are difficult to control such as the injection process, local temperature and concentration fluctuations, rate of stirring, and rate of cooling. In a continuous-flow system, reactions are performed at steady state, making it possible to achieve better control and reproducibility. Further benefits can be realized by scaling down the reactor dimensions to micrometers, thereby reducing the consumption of reagents during the optimization process and improving the uniformity of temperature and residence times within the reaction volume. A microfluidic flow reactor is attractive for NC synthesis because it is possible to rapidly and continuously screen through important reaction parameters, while using minimal amounts of reagents, until NCs of the desired size and monodispersity are produced. In contrast, each set of parameters would represent a separate reaction if the optimization procedure were conducted using a batch process. The inherent advantages of a microfluidic flow system also make it a natural choice for extracting kinetic data on NC nucleation and growth processes, information which has been difficult to obtain using conventional, macroscale batch methods. In spite of the perceived advantages of flow systems, it has been difficult to simply adopt the chemistry used in batch pre-

[*] Prof. M. G. Bawendi, B. K. H. Yen, N. E. Stott Department of Chemistry Massachusetts Institute of Technology 77 Massachusetts Avenue, Room 6-221 Cambridge, MA 02139 (USA) E-mail: mgb@mit.edu Prof. K. F. Jensen Departments of Chemical Engineering and Materials Science and Engineering Massachusetts Institute of Technology 77 Massachusetts Avenue, Room 66-566 Cambridge, MA 02139 (USA)

[**] This work was supported in part by the MRSEC Program of the National Science Foundation under award number DMR 02-13282 and the NSF-CRC program under award number CHE-0209898. B.K.H. Yen acknowledges support from the Department of Defense NDSEG Fellowship Program.

parations of semiconductor NCs to a microfluidic flow reactor. There have been a few reports on the preparation of II VI NCs in flow systems,[79] but these reports have not demonstrated the wide optical tunability, low polydispersities, and high quantum yields attainable in the batch process, nor have they extracted new kinetic data on particle formation. CdSe is probably the most well characterized colloidal semiconductor NC system because its effective bandgap can be tuned over the majority of the visible region. However, existing preparations are generally not amenable to a continuous flow system. In the most widely used preparation of high quality CdSe quantum dots, dimethylcadmium and tri-n-octylphosphine selenide (TOPSe) are rapidly injected into a hot solvent consisting of a mixture of tri-n-octylphospine (TOP) and tri-n-octylphosphine oxide (TOPO).[4] The solvent also serves as the source of surface ligands for the growing NCs. This method ensures that nucleation occurs very rapidly, followed by slower particle growth on existing nuclei, and the particles produced can be reasonably monodisperse and crystalline. Several difficulties arise when this chemistry is implemented in a continuous flow system, the most obvious one being that the solvent can be a solid at room temperature. Also, when (CH3)2Cd is used as a precursor, gas is rapidly evolved during the reaction, making it difficult to achieve reproducible reactor residence times. Finally, at the high reaction temperatures, TOPO can decompose and lead to clogging of the reactor channel, a problem that is only exacerbated as the channel dimensions are made smaller. Therefore, it was necessary for us to develop a new chemistry for CdSe NC synthesis that is more compatible with a microfluidic flow system. Guided by several recent reports on the preparation of semiconductor NCs,[5] cadmium oleate and TOPSe were chosen as the Cd and Se sources, respectively. These precursors are dissolved in a high-boiling solvent system consisting of squalane, oleyl amine, and TOP. This choice of precursors and solvent avoids problems of outgassing and clogging within the reactor channel, making it possible to use a simple capillary reactor to prepare CdSe NCs with excellent size distributions and high photoluminescence (PL) quantum yields. We also demonstrate that we can use the flow system to tune the band-edge absorbance (by varying the average NC size) over a substantial range. The continuous-flow reactor (Fig. 1a) consisted of a miniature convective mixer followed by a heated glass reaction channel (250 lm inside diameter (ID)) maintained at a constant temperature (180320 C). The Cd and Se precursor solutions were delivered in two separate flows and combined in the mixing chamber before they reached the heated reaction section. The presence of the chamber was necessary because once the Cd and Se precursor solutions are combined at room temperature, small CdSe clusters form over several hours. This cluster formation resulted in irreproducibility in the sizes of the final NCs produced by the reactor so the Cd and Se precursors could not be mixed until just prior to reaching the heated section. The volume of the mixer (~ 30 lL) was chosen such that the chamber residence time was long enough

1858

2003 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

DOI: 10.1002/adma.200305162

Adv. Mater. 2003, 15, No. 21, November 4

COMMUNICATIONS

sults.[4,10] The reaction yield and number of nuclei/volume were determined from the optical density and absorbance cross-section at 350 nm as previously reported.[11] As expected, we observed higher reaction yields and the production of larger NCs as the residence time or temperature is increased. Representative spectra for samples prepared at a constant temperature (Fig. 2a) or constant flow rate (Fig. 2c) illustrate some important general trends. Though the average NC size can be tuned by changing the reactor conditions, we

Fig. 2. a) Absorbance and PL spectra for samples produced at various flow rates while keeping temperature (320 C) and precursor concentration (1:1 Se/Cd) constant. b) Corresponding average NC size and size distribution (rr) of the samples represented in (a). c) Spectra for samples prepared at various temperatures while keeping the flow rate (s = 144 s) and precursor concentration constant. d) Corresponding average NC size and width of the size distribution of the samples represented in (c). The size distribution of each sample was determined from the PL peak width assuming a linewidth of ~ 50 meV for the emission of a single NC at room temperature (measured in our group). Calculated size distributions are approximate because they depend on the choice of the single NC emission linewidth, but the trends shown in the figure are correct. Fig. 1. a) Schematic of the capillary reactor. bd) Results for NCs prepared using a 1:1 Se/Cd precursor composition. s is the residence time in the heated section. The spectra, average radii (ravg), and size distributions for samples represented by the diamonds in this figure are shown in Fig. 2a,b. A similar data set for samples along the dashed line is shown in Fig. 2c,d.

(~115 min depending on flow rate) to ensure complete mixing but short enough to avoid formation of small clusters. The excellent stability of our reactor was demonstrated by continuously flowing precursors through the reactor for 8 h. Absorbance spectra taken of the NCs sampled during these long runs were indistinguishable from each other. By systematically varying the temperature, flow rate, and concentration it was possible to finely tune the size of NCs produced in the reactor. Figures 1bd summarize results for a fixed precursor composition. Average NC radii were determined from the position of the band-edge absorbance peak and calibration curves based on TEM and X-ray scattering reAdv. Mater. 2003, 15, No. 21, November 4 http://www.advmat.de

observed a corresponding variation in the size distribution of samples produced (Fig. 2b,d). In particular, the ratio of the size distribution to average radius (rr/ravg) becomes unacceptably large ( > 10 %) at extremes of low temperatures or high flow rates. At the other extremes, as explained below, the kinetics of nucleation and growth place an upper limit on the temperature and a lower limit on the flow rate. In other words, changing reactor conditions makes it possible to vary the average NC size over a wide range, but only at the expense of increasing polydispersity. Given the relatively narrow range of monodisperse sizes available by simply changing the reactor conditions, we sought to access a larger range by systematically varying the precursor concentration. Figure 3a illustrates that increasing the TOPSe concentration, while keeping the cadmium oleate concentration constant, leads to a dramatic increase in the number of nuclei that are formed. With more (less) nuclei on which to grow, smaller (larger)
2003 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

1859

COMMUNICATIONS
Fig. 3. a) Dependence of the number of nuclei on the concentration of TOPSe for a reactor temperature of 280 C. The cadmium oleate concentration was fixed. b) Corresponding reaction yields for the reaction conditions in (a). c) Absorbance and PL spectra for samples along the dashed line (s = 144 s). From bottom to top, the positions of the band-edge absorbance peaks in nm (and average NC radii in nm) are as follows: 552 (1.99), 568 (2.19), 579 (2.32), 586 (2.40).

dots are formed, and the position of the first absorbance peak shifts accordingly (Fig. 3c). By controlling the nucleation event, it is possible to rationally tune the effective bandgap of the NCs continuously over a substantial range while maintaining excellent size distributions. Figure 4 shows the absorbance and PL spectra of a size series of NCs with absorbance maxima ranging from 510 to 606 nm (average radii from ~ 1.5 to ~ 2.7 nm). Four TOPSe concentrations (1:1, 2:1, 5:1, and 18:1 Se:Cd) were used to obtain the size series shown. By controlling the number of nuclei systematically with concentration, we were able to access four size ranges, and within each range, the temperature and flow rates were varied to more finely tune the average NC size. The high PL quantum yields (between 2851 %), primarily due to the presence of amines, and the narrow emission peak widths (full width at half maximum between 2734 nm) indicate the excellent quality of the samples shown in Figure 4. An important result described above is the variation of the size distribution with temperature or flow rate. Understanding this trend provides insight into reactor design issues and underlying nucleation and growth processes. For instance, Figure 2a,b show that the size distribution becomes unacceptably broad as the flow rate is increased (reaction time decreased). Non-uniformity of reaction conditions across the channel can contribute to broadening of the size distribution. A straightforward calculation shows that upon entering the heated section, the fluid at the center of the channel heats to within 5 C of the target temperature in ~0.5 s.[12] Given such a rapid rate of heating, we can conclude that the dispersion in reaction conditions is dominated by the residence time distribution (RTD). For laminar flow in a cylindrical channel, the flow 1860
2003 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Fig. 4. Size series for samples prepared using four TOPSe concentrations: 1:1 Se/Cd, 2:1 Se/Cd, 5:1 Se/Cd, and 18:1 Se/Cd. For each TOPSe concentration, the average NC size was controlled by varying the temperature at a fixed flow rate. From bottom to top, the positions of the band-edge absorbance peaks in nm (and average NC radii in nm) are as follows: 510 (1.52), 535 (1.78), 545 (1.90), 551 (1.98), 555 (2.03), 561 (2.10), 569 (2.20), 575 (2.27), 579 (2.32), 585 (2.39), 592 (2.48), 586 (2.40), 597 (2.55), 606 (2.70).

takes on a parabolic velocity profile so that NCs in the center move faster than those near the channel wall. Superimposed on this flow profile is the radial diffusion of NCs across the channel, and the combination gives rise to a distribution of residence times experienced by NCs exiting the reaction section. The residence time s shown in Figures 13, for example, is technically an average residence time. The width of the RTD function (rt) can be estimated using the Taylor dispersion model.[13] From rt and the size versus time curve (Fig. 2b, triangles), the RTD contribution to the overall size distribution was calculated. At shorter times, the RTD effect is significant, accounting for ~ 60 % of the overall size distribution for s = 14 s. At longer times, the RTD effect becomes negligible, accounting for < 10 % of the overall size distribution for the
http://www.advmat.de Adv. Mater. 2003, 15, No. 21, November 4

COMMUNICATIONS

longest s shown in the figure (144 s). This analysis partially explains the narrowing of the overall size distribution with increased reaction time.[14] The underlying kinetics of NC formation also play an important role in the observed size distributions in Figure 2b. Under certain conditions, there is a natural distribution in growth rates with size that results in narrowing of the size distribution.[15] For steady-state, diffusion-controlled growth of a spherical particle, it can be shown that the growth rate (dr/dt) as a function of particle radius (r) has a maximum at a critical radius rcr. For r > rcr, the growth rate decreases with increasing radius (smaller particles grow faster than larger ones), and the size distribution narrows with time. For r < rcr, smaller particles grow slower or even dissolve compared to larger particles (Ostwald ripening), and this leads to broadening of the size distribution. If the growth rate is also dictated by reaction at the surface (mixed diffusion and reaction control), the growth rate can still have a maximum at some critical radius, and the curve still exhibits focusing and defocusing regions.[16] These arguments have been used to explain the observed narrowing and subsequent broadening of the size distribution during the batch synthesis of semiconductor NCs.[6] In such a preparation, the supersaturation of monomers is high early in the growth so that essentially all of the particle sizes are larger than rcr, and the size distribution sharpens. As the growth continues and concentration of monomers is depleted, rcr increases so that eventually all of the sizes are less than rcr, and the size distribution broadens. For NCs prepared in a flow reactor, the combination of the RTD and growth focusing/defocusing define an optimal range of flow rates that can produce monodisperse samples. For times shorter than this range, the dispersion in residence times is large, and the growth focusing does not proceed long enough to compensate, resulting in poor quality samples. For very short times, nucleation is incomplete, further broadening the size distribution. For times longer than the optimal range, the concentration of monomer decreases enough so that growth proceeds by Ostwald ripening and the size distribution begins to broaden.[17] The dependence of size distribution on temperature (Fig. 2d) can also be explained from kinetic arguments, in particular from the nucleation data shown in Figure 1d. With the exception of the 240 C curve, the number of nuclei increases with time and then saturates to a constant value. This is consistent with existing nucleation and growth models in which NC formation proceeds by a simultaneous nucleation and growth period followed by purely growth on existing nuclei. Since the kinetics of nucleation and growth are strongly temperature-dependent, the formation of nuclei (the rising portion of the curves in Fig. 1d) takes substantially longer as the temperature is lowered. In fact, at 240 C, the nucleation is slowed so much that even at the longest residence time shown (216 s), nucleation is incomplete, and the curve does not saturate. Therefore, at lower temperatures, the kinetics are slowed to such an extent that the majority or all of the residence time is characterized by mixed nucleation and growth (the curves in Fig. 1d have not saturated), resulting in poor size distribu-

tions. At high temperatures, nucleation ends quickly so that the majority of the time spent in the heated section is purely characterized by growth focusing. Another important observation is that the curves in Figure 1d approach essentially the same value. In other words, changing the temperature affects the rate of nucleation, but the final number of nuclei formed remains essentially constant. Thus, with this chemistry there is a fundamental limit on the range of sizes that may be accessed by changing the temperature. In order to access a larger size range while maintaining acceptable size distributions, we systematically varied the concentration of the TOPSe precursor (Fig. 3). Increasing the TOPSe concentration slows the growth rate relative to the nucleation rate, resulting in formation of a larger number of nuclei as shown in Figure 3a, which summarizes the effect of varying the TOPSe concentration while keeping the cadmium oleate concentration and reactor temperature (280 C) constant. Also, a higher TOPSe concentration resulted in higher reaction yields (Fig. 3b). This is due to two effects: firstly, a higher concentration of precursors results in faster overall kinetics, and secondly, a higher concentration of nuclei on which to grow results in faster depletion of monomers in solution. In summary, a combination of the RTD and intrinsic nucleation and growth processes place restrictions on the range of sizes that can be accessed for a given precursor concentration. By changing the TOPSe concentration, it is possible to control the nucleation rate, thereby enabling us to use a simple capillary reactor to produce high quality CdSe NCs with sizes corresponding to a wide spectral range. We have demonstrated that the continuous flow system serves as a powerful kinetic tool for elucidating NC nucleation and growth kinetics. The arguments used to describe the nucleation and growth process and the corresponding variations in size distribution can be applied generally. An obvious extension of the present work is the implementation of our capillary reactor to other nanoparticle systems.

Experimental
Typically, a Cd precursor solution was prepared by heating a suspension of 74.8 mg (0.5 mmol) Cd(OH)2, 350 lL (1.1 mmol) oleic acid (cis-9-octadecenoic acid), and 9 mL squalane (2,6,10,15,19,23-hexamethyltetracosane) at 150 C under vacuum. After approximately 10 min, the solution became optically clear, at which point the temperature was lowered to 100 C. The solution was further degassed for at least 90 min to remove any excess water. Upon cooling to room temperature, 4 mL oleyl amine (cis-1-amino-9-octadecene) were added. The Se precursor solution was prepared by diluting a portion of a 1.5 M TOPSe/TOP stock solution in squalane. The 1.5 M TOPSe/TOP stock solution was prepared as previously reported [4]. All of the above manipulations were performed under a dry N2 atmosphere. The reactor consisted of a length of glass tubing with a 250 lm channel diameter placed inside an aluminum heating chuck. Heating was provided by cartridge heaters inserted within the aluminum block and the temperature was monitored using a small thermocouple inserted next to the glass tubing. The total length of the heating section was 14.6 cm. In order to prevent formation of N2 bubbles in the heated section, the Cd and Se precursor solutions were thoroughly degassed at 90 C before being carefully drawn into syringes. The precursor solutions were then delivered in two separate flows with syringe pumps. After combining the two flows with a tee, the fluid reached a miniature convective mixing chamber containing a magnetic stir bar. After the mixing chamber,

Adv. Mater. 2003, 15, No. 21, November 4

http://www.advmat.de

2003 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

1861

COMMUNICATIONS

the precursors flowed into the heated section, where they quickly reacted to form NCs. The NC solution was then collected for absorption and photoluminescence measurements. Optical absorption spectra were acquired using a Hewlett-Packard 8453 diode array spectrometer. Photoluminescence spectra were acquired using a SPEX Fluorolog 1680 spectrometer, using right-angle collection. Samples were prepared by diluting the raw NC solutions in hexane. Quantum yields were determined by comparing the integrated emission of a given NC sample solution with that of an appropriate reference dye [18]. Received: March 26, 2003 Final version: July 18, 2003 Published online: September 11, 2003

[1] a) M. Bruchez Jr., M. Moronne, P. Gin, S. Weiss, A. P. Alivisatos, Science 1998, 281, 2013. b) W. C. W. Chan, S. Nie, Science 1998, 281, 2016. c) H. Mattoussi, J. M. Mauro, E. R. Goldman, G. P. Anderson, V. C. Sundar, F. V. Mikulec, M. G. Bawendi, J. Am. Chem. Soc. 2000, 122, 12 142. [2] a) V. L. Colvin, M. C. Schlamp, A. P. Alivisatos, Nature 1994, 370, 354. b) B. O. Dabbousi, M. G. Bawendi, O. Onitsuka, M. F. Rubner, Appl. Phys. Lett. 1995, 66, 1316. c) M. C. Schlamp, X. Peng, A. P. Alivisatos, J. Appl. Phys. 1997, 82, 5837. d) N. Tessler, V. Medvedev, M. Kazes, S. Kan, U. Banin, Science 2002, 295, 1506. e) S. Coe, W.-K. Woo, M. G. Bawendi, V. Bulovic, Nature 2002, 420, 800. [3] a) V. I. Klimov, A. A. Mikhailovsky, S. Xu, A. Malko, J. A. Hollingsworth, C. A. Leatherdale, H.-J. Eisler, M. G. Bawendi, Science 2000, 290, 314. b) M. Kazes, D. Y. Lewis, Y. Ebenstein, T. Mokari, U. Banin, Adv. Mater. 2002, 14, 317. c) V. C. Sundar, H.-J. Eisler, M. G. Bawendi, Adv. Mater. 2002, 14, 739. d) H.-J. Eisler, V. C. Sundar, M. G. Bawendi, M. Walsh, H. I. Smith, V. I. Klimov, Appl. Phys. Lett. 2002, 80, 4614. e) A. V. Malko, A. A. Mikhailovsky, M. A. Petruska, J. A. Hollingsworth, H. Htoon, M. G. Bawendi, V. I. Klimov, Appl. Phys. Lett. 2002, 81, 1303. [4] C. B. Murray, D. J. Norris, M. G. Bawendi, J. Am. Chem. Soc. 1993, 115, 8706. [5] a) C. B. Murray, S. Sun, W. Gaschler, H. Doyle, T. A. Betley, C. R. Kagan, IBM J. Res. & Dev. 2001, 45, 47. b) L. Qu, Z. A. Peng, X. Peng, Nano Lett. 2001, 1, 333. c) Z. A. Peng, X. Peng, J. Am. Chem. Soc. 2001, 123, 183. d) M. G. Bawendi, N. E. Stott, US Patent 20 020 071 952, 2002. [6] X. Peng, J. Wickham, A. P. Alivisatos, J. Am. Chem. Soc. 1998, 120, 5343. [7] E. M. Chan, R. A. Mathies, A. P. Alivisatos, Nano Lett. 2003, 3, 199. [8] H. Nakamura, Y. Yamaguchi, M. Miyazaki, H. Maeda, M. Uehara, P. Mulvaney, Chem. Commun. 2002, 2844. [9] J. B. Edel, R. Fortt, J. C. deMello, A. J. deMello, Chem. Commun. 2002, 1136. [10] a) H. Mattoussi, A. W. Cumming, C. B. Murray, M. G. Bawendi, R. Ober, Phys. Rev. B 1998, 58, 7850. b) M. K. Kuno, Ph.D. Thesis, Massachusetts Institute of Technology 1998. [11] C. A. Leatherdale, W.-K. Woo, F. V. Mikulec, M. G. Bawendi, J. Phys. Chem. B 2002, 106, 7619. We have made the assumption that the number of dots is equal to the number of nuclei, which is true as long as growth does not proceed extensively by an Ostwald ripening mechanism. In Ostwald ripening, the concentration of monomers is low enough that smaller dots dissolve at the expense of larger dots, and we would observe a decrease in the number of dots with sufficiently long growth time. In all of the data presented here, the residence time was kept short enough so that this process did not occur. [12] See for example: a) M. A. Edadian, Z. F. Dong, in Handbook of Heat Transfer (Eds: W. M. Rohsenow, J. P. Harnett, Y. I. Cho), 3rd ed. McGraw Hill, New York 1998, Ch. 5. b) J. R. Sellars, M. Tribus, J. S. Klein, Trans. ASME 1956, 78, 441. For our system, we estimated 0.1 W m1 K1, 0.9 g cm3, and 3 J g1 K1 for the fluid thermal conductivity, density, and heat capacity, respectively. We took as the boundary condition that the fluid at the channel wall is at a constant 25 C before the heated region and a constant 320 C within the heated region. [13] See, for example: a) H. S. Fogler, in Elements of Chemical Reaction Engineering, 3rd ed., Prentice Hall, Upper Saddle River, NJ 1999. b) O. Levenspiel, in Chemical Reaction Engineering, 3rd ed., Wiley, New York 1999. In this model, the extent of dispersion in residence times is characterized by a dispersion coefficient, D* = (u2R2)/(48D), where R is the radius of the channel, u is the average flow velocity, and D is the diffusion coefficient. D was estimated using the StokesEinstein equation. The fluid viscosity was extrapolated from experimental values for squalane (U. G. Krahn, G. Luft, J. Chem. Eng. Data 1994, 39, 670) using a modified Arrhenius expression. [14] Introducing a segmented, rather than a homogeneous flow, into the reactor can reduce the dispersion in the RTD. In such an approach, the flow is

[15] [16] [17] [18]

comprised of small, alternating segments of two different phases moving through the heated section. Each segment behaves essentially like a miniature stirred tank reactor, so that molecules within each segment experience a very uniform RTD. In a previous report on the preparation of CdSe NCs with a flow system [8], researchers used a segmented flow by introducing N2 bubbles into the precursor stream, and they observed somewhat improved size distributions in comparison to the homogeneous flow case. We have observed similar results when a lower boiling solvent, di-n-octyl ether rather than squalane, is used. At temperatures greater than ~280 C, the solvent boils within the heated section, forming alternating segments of gas and liquid within the channel. However, it was difficult to estimate the flow velocity of the fluid in the heated region due to the nature of the bubble formation. Experiments are underway to more controllably introduce a gas phase into the liquid precursor stream. T. Sugimoto, Adv. Colloid Interface Sci. 1987, 28, 65. D. V. Talapin, A. L. Rogach, M. Haase, H. Weller, J. Phys. Chem. B 2001, 105, 12 278. Under some conditions, such as a very broad initial size distribution, it is possible for Ostwald ripening to lead to narrowing of the size distribution [16]. Such conditions were not present in the data shown here. The following reference dyes (and quantum yields) were used: Rhodamine 560 chloride in basic ethanol (92 %), Rhodamine 590 chloride in methanol (89 %), Rhodamine 610 chloride in methanol (57 %), and Rhodamine 640 perchlorate in methanol (100 %).

1.3 lm to 1.55 lm Tunable Electroluminescence from PbSe Quantum Dots Embedded within an Organic Device**
, By Jonathan S. Steckel, Seth Coe-Sullivan, Vladimir Bulovic and Moungi G. Bawendi* We demonstrate large area (mm2 in size) infrared electroluminescent devices using colloidally grown PbSe quantum dots (QDs) in organic host materials. By changing the QD size the electroluminescence is tuned from k = 1.33 to 1.56 lm with a full width at half maximum (FWHM) of < 160 nm (< 0.11 eV). This represents only a portion of the accessible QD tuning range, as the lowest energy optical absorption peak of our PbSe solutions can be tuned from 1.1 eV (corresponding to wavelength k = 1.1 lm and 2.6 nm diameter QDs) to 0.56 eV (k = 2.2 lm, 9.5 nm diameter). Our light emitting device fabrication combines the thin film processing techniques available to organic materials with the tunable optical properties of PbSe QDs. Formation of double heterojunction devices is enabled by material phase segregation during the spin-coating step. Such large area emitters in the near infrared have been

[*]

Prof. M. G. Bawendi, J. S. Steckel Department of Chemistry, Massachusetts Institute of Technology 77 Massachusetts Avenue Cambridge, MA 02139 (USA) E-mail: mgb@mit.edu S. Coe-Sullivan, Prof. V. Bulovic Laboratory of Organic Optoelectronics Department of Electrical Engineering and Computer Science Massachusetts Institute of Technology Cambridge, MA 02139 (USA)

[**] J. S. Steckel and S. Coe-Sullivan contributed equally to this work. We thank M. Frangillo for assistance with the high-resolution electron microscopy. This work was funded in part by the NSF-MRSEC program (DMR 0213282) and by the U.S. Army through the Institute for Soldier Nanotechnologies, under Contract DAAD-19-02-0002 with the U.S. Army Research Office.

1862

2003 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

DOI: 10.1002/adma.200305449

Adv. Mater. 2003, 15, No. 21, November 4