Article pubs.acs.org/JPCC
Photoluminescence of Mid-Infrared HgTe Colloidal Quantum Dots Sean Keuleyan, John Kohler, and Philippe Guyot-Sionnest* The James Franck Institute at the University of Chicago, 929 East 57th Street, Chicago, Illinois 60637, United States S Supporting Information *
ABSTRACT: The photoluminescence quantum yield of HgTe colloidal quantum dots is measured from 1800 to 6500 cm−1. There is a steep drop to low energy reminiscent of the generic gap law. However, direct evidence of energy transfer to the C−H stretch and overtone vibrations is apparent when temperature tunes the PL wavelength of a given sample through the vibrational resonances. Calculations based on the radiative rate and resonant energy transfer to the ligand vibrations appear to account for much of the quantum yield drop. Power-dependent photoluminescence lifetime measurements on 3.7 nm particles show fast, ∼50 ps, biexciton lifetime similar to other colloidal quantum dot systems of similar sizes.
■
INTRODUCTION Infrared (IR) light is used in many applications including imaging, spectroscopy, gas and atmospheric analysis, and monitoring, tagging, and communications. Recently, colloidal quantum dots (CQDs) were introduced as potential chromophores in the mid-IR1,2 with demonstration of their use as mid-infrared detectors.3 HgTe is a semimetal in the bulk and quantum dots allow in principle tuning through the full range of infrared wavelengths. HgTe CQDs were first synthesized with band gaps tunable in the near-IR,4 followed by extension to 3 μm,5 and to the mid-IR up to 5 μm.1 The photoluminescence (PL) quantum yield (QY) of HgTe CQDs is an important property in evaluating the potential and limitations of the materials for many applications. Previous work showed that HgTe CQDs have high PL QY in the near-IR but that the efficiency drops off significantly to longer wavelengths.5 In this work, we report the PL quantum yield for the reddest emitting CQDs to date, past 5 μm. We also describe the temperature dependence, the coupling to ligand vibrations, and the biexciton lifetimes.
immediate color change. Sample aliquots or the full reaction mixture were then quenched by extraction to a solution of 10% vol dodecanethiol + ∼1% TOP in TCE (2× the extraction volume). This was then mixed several minutes before precipitating with methanol. The precipitate was collected by centrifugation, rinsed 3× with ethanol without redispersing, dried with N2, and redispersed in TCE (typically equal in volume to the original extraction). 2. Absorption, PL, and QYs. In the mid-IR, the determination of PL QY is difficult due to the lack of standards and the weak photoluminescence of the samples. Absorption spectra of the solutions in TCE were measured with a Nicolet Magna IR 660 FTIR and a Cary UV−vis-NIR spectrometer through a 3 mm path cell with CaF2 windows. The absolute PL QY was determined for a sample emitting at 6000 cm−1 (1.6 μm) using a ThorLabs Spectralon coated integrating sphere with a mercury cadmium telluride (MCT) detector on one port. The excitation was provided by an 808 nm laser diode with an average power of 80 mW (peak 160 mW), modulated at 75 kHz. The 808 nm excitation was collimated into the integrating sphere, such that it was incident on an interior wall and not on the sample. The sample absorption (%), A, at 808 nm was determined using a Si photodiode which is insensitive to the PL wavelength on another port of the integrating sphere and comparing the signals with sample (Ssample) and without (Sblank).
■
EXPERIMENTAL METHODS 1. Synthesis. HgTe CQDs were synthesized with a method slightly modified from ref 1. Octadecylamine (ODA) is used to avoid broadening of the absorption and PL, which was found to vary between supplies of oleylamine (70%). ODA (tech. 90%), HgCl2 (99.999% metals-basis), trioctylphosphine (TOP, tech. 70%), Te (99.999%, granules), n-dodecanethiol (98%), tetrachloroethylene (TCE, spectrophotometric, >99%) were obtained from Sigma-Aldrich and used as-received. Briefly, 10 g of octadecylamine and 81 mg of HgCl2 (0.3 mmol) were loaded in a 50 mL round-bottom flask and heated under vacuum to 120 °C for 1 h, forming a clear solution. The mixture was then cooled under Ar to the appropriate reaction temperature, chosen for the desired size. As it cooled below ∼110 °C, the mixture became cloudy-white. A 0.3 mL sample of 1 M Te in TOP was then quickly injected, giving nearly © 2014 American Chemical Society
A=
S blank − Ssample S blank
(1)
The signal on the MCT was determined with lock-in amplification with no sample present, as a measure of the excitation flux (FEx). The PL flux (FPL) was then measured with the sample lowered into the integrating sphere, while a silicon Received: September 10, 2013 Revised: January 13, 2014 Published: January 13, 2014 2749
dx.doi.org/10.1021/jp409061g | J. Phys. Chem. C 2014, 118, 2749−2753
The Journal of Physical Chemistry C
Article
the low quantum yields of the solutions prevented this approach. Instead, we measured the relative quantum yields for all samples between 6000 and 1800 cm−1. We used the same cell as for the absorption measurement and the PL was measured with a step-scan interferometer. The instrument response of the setup was determined using a SiC filament (Newport 80030), placed at the same position as the samples. The temperature of the filament was measured with an optical pyrometer (Leeds & Northrup 8622-C) and the photon flux was taken to be proportional to the blackbody spectrum. The samples were excited in a front face geometry by the modulated 808 nm laser light. The PL spectra were measured with a liquid N2 cooled MCT detector with a lock-in amplifier. The corrected PL spectra were then obtained as
wafer was placed in front of the MCT detector to block the excitation. The relative detection efficiency of the MCT at 1600 and 808 nm was determined using another laser diode of measured power at 1625 nm, indicating a 25 times more efficient detection at 1625 nm. Including a factor of 0.6 to correct for the transmission of the Si wafer, the QY was then determined as QY =
FPL 0.6A 25FEx
(2)
With this procedure, the 1600 nm sample had a QY of 7%. The Spectralon integrating sphere is limited to ∼2.4 μm but this procedure could in principle be extended using a goldcoated sphere. However, the low efficiency of the spheres and Corrected PL Spectrum =
(Measured PL Spectrum) (Calculated Blackbody Spectrum) × (Sample Absorption% at 808 nm) (Measured Lamp Spectrum)
Measured PL spectra intensities were found to be linear with concentration in the range used. The relative QYs of samples were determined by integrating the corrected PL spectra. The absolute QYs in Figure 1 were then obtained by scaling to the absolute determination at 1600 nm described above.
(3)
4. PL Lifetimes. PL lifetime measurements were made by up-conversion. A 1064 nm ∼15 ps 25 Hz Nd:YAG laser pulse was split into two paths. The first path included a delay stage and was used to excite the sample in a 1 cm fused silica cuvette. The second path was brought collinear with the PL emission after the sample with a 45° 1064 nm CaF2 mirror, both passing through a 14 mm long KTP nonlinear crystal that was angle tuned. This results in light being generated at the sum frequency. Narrow filters removed the 1064 and 532 nm and the sum frequency signal was passed into a monochromator and detected with a PMT. This led to typically a few detected photons per pulse. The delay between the PL excitation pulse and the upconversion pulse was limited to 350 ps due to a weak (