Electronic Supporting Information Upconverting Nanoparticle to Quantum Dot Förster Resonance Energy Transfer: Increasing the Efficiency Through Donor Design R. Marin a,b, L. Labrador-Paéz c, A. Skripka a, P. Haro-González c, A. Benayas a,d, P. Canton b, D. Jaque* c,e, F. Vetrone* a,f a Institut National de la Recherche Scientifique, Centre Énergie Matériaux Télécommunications (INRS - EMT), Université du Québec, 1650 Boul. Lionel-Boulet, Varennes, Québec, J3X 1S2, Canada b Dipartimento di Scienze molecolari e Nanosistemi, Università Ca' Foscari, Venezia, Via Torino 155/B - 30172 Venezia-Mestre, Italy c Fluorescence Imaging Group, Departamento de Física de Materiales, Instituto Nicolás Cabrera, Facultad de Ciencias, Universidad Autónoma de Madrid, 28049 Madrid, Spain d Department of Physics and CICECO Aveiro Institute of Materials, University of Aveiro, 3810-193, Aveiro, Portugal e Instituto Ramón y Cajal de Investigación Sanitaria, IRYCIS, Ctra. Colmenar km. 9.100, Madrid 28034, Spain f Centre for Self-Assembled Chemical Structures, McGill University, Montréal, Québec, H3A 2K6, Canada Corresponding authors: daniel.jaque@uam.es and vetrone@emt.inrs.ca Table of Contents Procedure for squared residuals minimization... 2 Estimation of the average interparticle distance in sol... 3 Interpretation of the different decay rate in L-UCNPs, s-ucnps and CS-UCNPs... 4 Absorption of LiYF 4: 0.5 mol% Tm 3+, 25 mol% Yb 3+ L-UCNPs... 5 Structural and optical characterization of CIS QDs... 6 Structural characterization of L-UCNPs, s-ucnps, and CS-UCNPs... 7 UCNPs emission change in sols titrated with CIS QDs... 9 Interparticle distance distribution in sols... 10 Radiative nature of ET in sols... 11 Upconversion emission change of dried UCNPs in the presence of CIS QDs... 12 Numerical considerations of FRET between UCNPs and CIS QDs... 13 R 0 dependence on the refractive index n... 14
Procedure for squared residuals minimization The recorded signal of the upconversion nanoparticles (UCNPs) + quantum dots (QDs) mixed sols obtained exciting with 980 nm laser (y obs,i) was fit point by point via a linear combination (y i) of the emission spectra of QDs (Q i - directly excited at 440 nm) and UCNPs (U i) alone. The mathematical procedure relies on the minimization of a squared residual function S: S = (y obs,i y i ) 2 i = [y obs,i (aq i + bu i )] 2 i a and b are weighing coefficients that account for the contribution of each luminescent species. The condition to be imposed to minimize S is that the first order derivative of this function calculated with respect to the weighing coefficients is equal to zero: S a = S b = 0 This condition leads to the homogeneous system: And, rearranging: ( i Q i 2 i U i Q i Q i U i i U i 2 i ( a b ) = ( i Q i 2 i U i Q i ) ( a b ) = ( i Qiy obs,i ) i U i y obs,i 1 i Q i U i 2 ) i U i ( i Qiy obs,i ) i U i y obs,i By solving this system, a and b values are obtained, thus the relative contribution to the total luminescence signal from the two moieties. 2
Estimation of the average interparticle distance in sol In the case under study, the mean interparticle distance distribution (P) between donors and acceptors in the sol was estimated utilizing the formula: P(R) = 3 2 a (R a ) e (R a )1 3 Where R is the mean interparticle distance of the nearest neighbors and a is the so-called Weigner- Seitz radius. This parameter is a function of the particle density n, according to the following relation: 3 a = ( 4πn ) This parameter corresponds to the volume available per particle in the system. In our particular case, the particle density corresponds to the concentration of UCNPs and QDs together, with the largest contribution to this value coming from the latter species. The estimated interparticle distance corresponds to the distance between the center of spheres of radius a at which centers the nanoparticles (NPs) are located. The k th moment of the distribution is given by: < R k > = P(R)R k dr 0 1 3 = a k Γ(1 + k 3 ) Where Γ is the gamma function. Specifically, here the first moment is given by the simplified relation <R> = 0.893a. 3
Interpretation of the different decay rate in L-UCNPs, s-ucnps and CS-UCNPs The fit of the photoluminescence decay curves shown in Figure 4 (main text) returns lifetime (LT) values different for the three UCNP samples (Table 1; main text). Both small and core/shell UCNPs (L-UCNPs and CS-UCNPs, respectively) feature a slower decay rate compared to the core only large UCNPs (L-UCNPs). In the case of CS-UCNPs, this behavior stems from the distribution of Yb 3+ ions over a larger volume (both core and shell). Hence, after a sensitizer ion has absorbed the excitation light, energy migration events happen on a longer length scale than in the case of core-only particles. Such spatial extension, in this case, directly corresponds to an increase of the LT value. Regarding s-ucnps, the explanation is not as straightforward since smaller NPs are expected to suffer from a more severe quenching due to more prominent surface effects following from a larger surface-tovolume ratio. We tentatively imputed the longer LT value with respect to that of L-UCNPs to the use of both OA and OLAm during the synthesis process. These molecules remain attached on the UCNP surface and their effect is expected to stem from the electrostatic interaction between the positively and negatively charged heads (of OLAm and OA, respectively). This interaction allows for a better packing of ligand molecules on the UCNP surface, resulting in a better surface coverage, and thus a lesser fraction of surface/defect trap states that act as luminescence quenchers. 4
Absorption of LiYF 4: 0.5 mol% Tm 3+, 25 mol% Yb 3+ L-UCNPs Figure S1. Absorbance of a L-UCNPs sol. Near-infrared absorption band, centered at 960 nm, arises from the Yb 3+ : 2 F 7/2 2 F 5/2 transition. 5
Structural and optical characterization of CIS QDs Figure S2. QDs synthesized for this study have the typical optical features of CuInS 2 (CIS) QDs (A): the broadband absorption extends up to around 650 nm, while a rather broad emission (full width at half maximum of approximately 100 nm) featuring a large Stokes shift peak at approximately 710 nm. The crystalline structure of the QDs is tetragonal chalcopyrite (PDF #00-047-1372), as confirmed from the XRPD pattern (B). The TEM micrograph shows NPs of 3 to 4 nm in size (C). Fourier Fast Transform (FFT) of the imaged area (inset in C) returns the presence of spots indexed as {112} crystalline plane. 6
Structural characterization of L-UCNPs, s-ucnps, and CS-UCNPs Figure S3. XRPD patterns of the three UCNPs batches (A, D, G) feature reflections corresponding to the tetragonal LiYF 4 phase (PDF #01-078-2179). TEM observations show UCNPs with a bipyramid habitus (B, E, H; scale bars correspond to 200 nm). In the case of CS-UCNPs, the size increase and the absence of a second population of UCNPs (H.i C-UCNPs H.ii CS-UCNPs) corroborates the growth of a core/shell architecture. In the HRTEM micrographs of single UCNPs, crystalline fringes are observable (C, F, I), and the electronic diffraction patterns are characterized by the presence of intense diffraction spots that are indexed to tetragonal LiYF 4 (insets in C, F, I). 7
Figure S4. The bipyramid was conveniently described considering the base edge a and the full height l (A). Following this geometrical model, statistical analysis was performed on 100 particles in each of the three batches, returning the respective size distributions (L-UCNPs B, s-ucnps C, CS-UCNPs - D). Values and associated error at which each of the distributions peak are reported in tables beside the corresponding graphs. 8
UCNPs emission change in sols titrated with CIS QDs Figure S5. L-UCNPs sol was titrated with pure toluene solvent as a control experiment for the titration of the same sample with the QD sol (Figure 2A; main text). The addition of increasing amounts of toluene does not lead to any significant variation of the upconversion luminescence (UCL) signal. Figure S6. As in the case of core-only L-UCNPs, the titration of a CS-UCNPs sol in toluene with either pure solvent (A) or a sol of CIS QDs in toluene (B) shows that the UCL is influenced by the presence of the acceptors depending on the concentration of this species. Specifically, the higher photon-order transitions experience a steep intensity decrease upon addition of increasing QD amount (C and D). Tm 3+ : 1 G 4 3 F 4 transition intensity does not increase as much as in the case of L-UCNPs (i.e. the increase is only of the order of 4-5%). This makes it impossible to obtain quantitative information about the QD emission and to investigate the relationship between this quantity and the average interparticle distance according to the model outlined in the main body of the manuscript. 9
Interparticle distance distribution in sols Figure S7. The average interparticle distance distribution P(R) in the case of the higher (3.7 μm) and lower (38 nm) concentration of NPs tested were calculated using the mathematical approach reported above. It is worth noting that even at the higher concentration, the distribution is peaked at approximately 43 nm, a value that is well above distances usually considered suitable for observing Förster resonance energy transfer (FRET) phenomenon. Figure S8. When considering the allowed interparticle distances, L-UCNPs characteristic lengths were taken into account according to the results shown in Figure S3. These considerations were used to select proper values for the parameter C in Equation 5 (main text). The closest an acceptor can get to the UCNP surface is in correspondence of the barycenter of the bipyramid lateral face (A). An average minimum interparticle distance that takes into account the steric hindrance of the particles was obtained considering the radius of an equivalent sphere, i.e. a sphere whose volume is the same as that of a bipyramid (B). In this particular case (L-UCNPs), these distances are respectively 23 (C = 23) and 27 nm (C = 27). 10
Radiative nature of ET in sols Figure S9. A first visual inspection of the UCL decay curves of a L-UCNPs sol titrated with different amounts of QD sol does not show an appreciable shortening of the UCNPs decay rate (A). This is confirmed by the data analysis, which does not return any lifetime shortening of the UCL signal in the presence of increasing amount of acceptors (B). Decay curves were acquired exciting the sample at 960 nm and recording the emission at 453 nm. 11
Upconversion emission change of dried UCNPs in the presence of CIS QDs Figure S10. The evidence of FRET between UCNPs and CIS QDs in dry form, gathered using timeresolved methods, were also corroborated by steady state measurements. Specifically, it is evident from the difference between the behavior of L-UCNPs (A) compared to s-ucnps (B) and CS-UCNPs (C). In the two former UCNP types, the UCL spectra recorded in absence of QDs and in the presence of the highest QD amount investigated showed a marked decrease of the emission intensity in the blue spectral region. This is accompanied by a simultaneous appearance of a broad emission feature in the red region, ascribed to CIS QDs. The emission profile in the spectral region from 670 to 750 nm has been magnified in A (x4) and C (x10) to better appreciate the contribution of indirectly excited QDs to the overall photoluminescence signal. 12
Numerical considerations of FRET between UCNPs and CIS QDs Figure S11. The simulation of the Förster radius (R 0) dependence, respectively, upon the dipole orientation factor (κ 2 ) and the photoluminescence quantum yield (PLQY) for s-ucnps (A and E) and cs-ucnps (B and F) returns a similar behavior to that observed for L-UCNPs (Figure 5B and D; main text). 13
R 0 dependence on the refractive index n Figure S12. The Förster radius value (R 0) does not vary significantly considering values of the refractive index, n, comprised between those of LiYF 4 (1.449) and DDT or OA (both 1.459). The effective refractive index was chosen to fall in between these values, since FRET takes place between Tm 3+ ions embedded in the UCNP crystalline lattice and QDs that are directly in contact with the UCNP surface. Specifically, the difference in the calculated Förster radius for the two extremes is approximately 0.5%. 14