Speed Limit for Triplet-Exciton Transfer in Solid-State PbS Nanocrystal-Sensitized Photon Upconversion

Transcription

1 SUPPORTING INFORMATION Speed Limit for Triplet-Exciton Transfer in Solid-State PbS Nanocrystal-Sensitized Photon Upconversion Lea Nienhaus, Mengfei Wu, Nadav Geva, # James J. Shepherd, Mark W. B. Wilson,, Vladimir Bulović, Troy Van Voorhis, Marc A. Baldo, Moungi G. Bawendi, * Department of Chemistry, Department of Electrical Engineering and Computer Science, # Department of Materials Science and Engineering, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, United States Present Address Department of Chemistry, University of Toronto, Toronto, Ontario, M5S 3H6, Canada *correspondence should be addressed to: mgb@mit.edu

2 Ligand τ (µs) 18C 2760 OA C C C C C C C 2340 Table S1: Mono-exponential tail fits to the NCs Only kinetics. As reported in Ref. 56 in the main text, an increased dielectric environment yields shorter PbS NC PL lifetimes τ. Dielectric constants Ligand Dielectric constant OA C C C C C C C C 3 Toluene 2.4 Table S2: Dielectric constants for the ligands used in main text to determine the effective medium dielectric constant. 8C Laser power (nw) Counts per second cps/nw Quenching efficiency (%) NC only NC+BCP OA Laser power (nw) Counts per second cps/nw Quenching efficiency (%) NC only NC+BCP Table S3: Observed quenching of the IRPL of PbS NCs using a higher bandgap organic material bathocuproine (BCP) instead of rubrene which cannot function as a triplet acceptor. The additional power needed is likely due to scattering in the organic layer. The PbS kinetics are unchanged in the NC vs. NC/NC+DBP region; see Figure S10.

3 4C Laser power Counts per second cps/nw Quenching efficiency (%) Corrected quenching efficiency (%) NC only NC+rubrene C Laser power (nw) Counts per second cps/nw Quenching efficiency (%) Corrected quenching efficiency (%) NC only NC+rubrene C Laser power (nw) Counts per second cps/nw Quenching efficiency (%) Corrected quenching efficiency (%) NC only NC+rubrene C Laser power (nw) Counts per second cps/nw Quenching efficiency (%) Corrected quenching efficiency (%) NC only NC+rubrene C Laser power (nw) Counts per second cps/nw Quenching efficiency (%) Corrected quenching efficiency (%) NC only NC+rubrene C Laser power (nw) Counts per second cps/nw Quenching efficiency (%) Corrected quenching efficiency (%) NC only NC+rubrene C Laser power (nw) Counts per second cps/nw Quenching efficiency (%) Corrected quenching efficiency (%) NC only NC+rubrene C Laser power (nw) Counts per second Normalized cps Quenching efficiency (%) Corrected quenching efficiency (%) NC only NC+rubrene OA Laser power (nw) Counts per second cps/nw Quenching efficiency (%) Corrected quenching efficiency (%) NC only NC+rubrene Table S4: Laser power, counts per second and observed quenching efficiencies of the IRPL counts of PbS NCs when adding the rubrene layer. The corrected quenching efficiency accounts for the estimated quenching caused by scattering due to the addition of an organic layer, compare Table S3. Figure S1: Absorption (black) and emission (red) spectra of the PbS NCs in toluene (oleic acid ligands).

4 Figure S2: AFM image of a monolayer PbS NC film. The NCs shown here have been exchanged to a C8 ligand. Figure S3: Solution lifetime of PbS NCs (oleic acid) in toluene. The red line is the mono-exponential fit to the long time component of the PL decay, showing a lifetime: τ PbS =3.1µs.

5 Figure S4: Parasitic IRPL signal when exciting a solid-state rubrene:dbp sample at 785 nm. These fast decay dynamics are clearly visible for NC+rubrene samples where a high pump power is necessary due to low IRPL emission because of quenching or a low QY due to damage caused during ligand exchange (4C, 6C, 8C, 18C). If the sample itself is bright, the parasitic IR emission from the sample background is hidden within the noise. The grey line is a mono-exponential fit and yields an early-time component of τ rubrene IR = 2 ns.

6 Extracted TET lifetimes for all aliphatic ligand lengths (6C, 18C in main text) Figure S5: Extracted transfer times for the remaining aliphatic ligand lengths not shown in the main text. The black curve represents the NC-only decay. The red curve is the quenched IRPL in presence of rubrene. By subtracting a scaled copy of the neat NC decay to account for the inactive NCs, we obtain the TT dynamics (dark gray) of the active NCs. The extracted TT transfer time τ, and the resulting characteristic time of transfer τ TET are shown for each ligand length.

7 Extracted lifetime: Oleic acid Figure S6: extracted transfer time for the native oleic acid. The black curve represents the NC only decay, the gray line is a monoexponential fit to the late component (τ DotsOnly =2.55 µs). The red curve is the quenched IRPL in presence of rubrene. By subtracting a scaled copy of the neat NC decay we obtain the TT dynamics (purple curve, τ=599 ns). This yields a characteristic TT time: τ TET,OA =780ns when accounting for the competition with the intrinsic PL lifetime. Direct observation of the visible rubrene dynamics for C8 ligands Figure S7: dynamics of the upconverted visible light (610 nm) at 60 khz repetition rate. The dynamics are fit to a biexponential. The rising exponential (τ rise =313ns) portrays the combined timescale of TET, diffusion-mediated TTA, FRET to DBP and subsequent emission. The decaying exponential (τ decay =13µs) reflects the long lifetime of isolated triplets in rubrene. Obvious laser scatter (green circles) is excluded from the analysis. The discrepancy between the lifetime extracted by the PbS quenching experiment and the visible rise primarily stems from the additional time needed for diffusion-mediated TTA to occur. The inset shows the dynamics at a repetition rate of 200 khz. A higher repetition rate leads to a buildup of excitations in the rubrene, as the long triplet lifetime does not allow all excitations to decay between pulses. As a result, the diffusion time for triplets is reduced, and the visible rise time is reduced.

8 Figure S8: Solution decay dynamics of PbS NCs which have been ligand exchanged in toluene, showing similar decay dynamics in solution before and after ligand exchange. Figure S9: a, shows the PL emission peak shift to lower energies when going from solution (black) to solid state (red). b, highlights the constant FWHM of the PL emission when going from solution to solid state indicating the redshift is not primarily caused by resonance energy transfer.

9 Figure S10: IRPL decay dynamics for a C8 ligand on 790 nm PbS NCs, with bathocuproine (BCP) added as a wide bandgap organic, which cannot function as a triplet acceptor. The decay dynamics are unchanged, indicating that the material evaporation does not harm the PbS NCs, and that there is no energy transfer. Scaled subtraction for samples where exciton backtransfer is strong: Due to the long lifetime of rubrene triplets and the additional time needed for diffusion and TTA, backtransfer via far-field reabsorption or FRET will only become prominent after 1-2 µs. As a result, this will not affect the kinetic analysis in the monoexponential fitting range chosen. As the long tails cannot be matched as in Ref. 4 in the main text, the scaling is chosen to have a tangent point between the NCs+Rubr and the scaled copy of the NCs only curve. Figure S11: NCs only subtraction when there is a large long-time mismatch of the NCs only and NCs+Rubr decay dynamics, shown here for a 4C ligand. Here, the scaling is chosen to have a common tangent point (red and green). This results in an extracted curve (maroon) which falls into the noise before rising again due to the backtransfer of singlet excitons from rubrene:dbp.