Concentration Quenching in Upconversion Nanocrystals.

Despite considerable effort to improve upconversion (UC) in lanthanide-doped nanocrystals (NCs), the maximum reported efficiencies remain below 10%. Recently, we reported on low Er3+- and Yb3+-doped NaYF4 NCs giving insight into fundamental processes involved in quenching for isolated ions. In practice, high dopant concentrations are required and there is a trend toward bright UC in highly doped NCs. Here, additional quenching processes due to energy transfer and migration add to a reduction in UC efficiency. However, a fundamental understanding on how concentration quenching affects the quantum efficiency is lacking. Here, we report a systematic investigation on concentration-dependent decay dynamics for Er3+ or Yb3+ doped at various concentrations (1-100%) in core and core-shell NaYF4 NCs. The qualitative and quantitative analyses of luminescence decay curves and emission spectra show strong concentration quenching for the green-emitting Er3+ 4S3/2 and NIR-emitting 4I11/2 levels, whereas concentration quenching for the red-emitting 4F9/2 level and the IR-emitting 4I13/2 level is limited. The NIR emission of Yb3+ remains efficient even at concentration as high as 60% Yb3+, especially in core-shell NCs. Finally, the role of solvent quenching was investigated and reveals a much stronger quenching in aqueous media that can be explained by the high-energy O-H vibrations. The present study uncovers a more complete picture of quenching processes in highly doped UC NCs and serves to identify methods to further optimize the efficiency by careful tuning of lanthanide concentrations and core-shell design.

Introduction

Upconversion (UC) luminescence has become an active field of research since the pioneering work by Auzel, Ovsyankin, and Feofilov in 1960s. Especially in the past two decades, the advent of UC nanocrystals (NCs) has triggered renewed interests by emerging applications of UC NCs in bioimaging, solar cells, sensors, anticounterfeiting, and three-dimensional (3D) displays.[1−5] Upconverting two (or more) low-energy photons to one high-energy photon has been demonstrated for a variety of combinations of lanthanide ions and in different spectral regions, ranging from IR → NIR UC to vis → UV UC.[1−8] The Yb3+–Er3+ ion couple is one of the most efficient UC couples showing efficient conversion of NIR (∼980 nm) to green and red light.[7,9] The intensity ratio of green and red emission can be tuned by varying experimental parameters, such as host matrix, dopant concentration, excitation power, dispersion medium, and size of NCs.[3] Trivalent Er3+ ion without a sensitizer also shows NIR-to-green UC, although less efficient, by excitation at 980 nm.[10] Also, an efficient IR (1520 nm)-to-NIR (980 nm) conversion has been reported for Er3+.[11]

In spite of the great potential of UC materials, applications are often hampered by the low quantum yield (QY), especially in NCs. The UC QY in NCs is typically below a few percent, in spite of approaches aimed at enhancing the efficiency, including surface passivation, shell growth, broadband sensitization, and photonic and plasmonic engineering.[12] Recently reported strategies of incorporating high dopant concentrations combined with increased excitation densities have shown superior UC brightness. The explanation is the enhanced absorption (related to higher concentrations of absorbing ions) and the quadratic dependence of UC rates on excitation power and dopant concentration. All of these factors cause the UC to rapidly gain importance over other processes, such as quenching and IR emission from intermediate levels that do not have a quadratic power/concentration dependence.[13,14] The extreme brightness of UC in highly doped UC NCs under high power density excitation is promising, but it is crucial for many applications to reduce the excitation power to prevent heating and allow for safe operation conditions. Therefore, it is important to quantify and understand the underlying additional (concentration) quenching mechanisms in highly doped UC NCs to realize high UC brightness under reduced excitation densities. Recently, progress has been made in reducing excitation powers for single-NC experiments by optimizing Yb and Er concentrations and single UC NC imaging was reported in highly doped NaYF4 core–shell NCs at excitation powers below 10 W/cm2.[15,16]

The efficiency of UC NCs is lower than that in bulk materials. One reason is the surface-related quenching by defects, surface ligands, and surrounding solvent molecules with high-energy vibrational modes.[17,18] These quenching mechanisms are especially pronounced for dopants on or near the NC surface. Coating of an inert (undoped) shell can effectively isolate the optically active lanthanide ions from the surface and surrounding energy vibrations to reduce surface quenching.[19−22] Recently, our group has established a model that accurately models solvent quenching for isolated lanthanide ions in UC NCs.[23] In this model, low dopant concentrations of Er3+ or Yb3+ were used to quantitatively describe and understand the quenching processes involved for individual dopant ions inside the NC. However, since UC relies on multistep energy transfer (ET), high dopant concentrations are crucial to realize efficient ET between lanthanide neighbors to enhance the UC efficiency. On the other hand, the high density of dopants opens additional channels for quenching and thus efficiency loss by concentration quenching via nonradiative processes, such as energy migration and cross-relaxation.[24−27] The trade-off between efficiency loss by concentration quenching and efficiency gain by ET UC determines the maximum UC QY. In addition, the higher absorption strength for higher dopant concentrations results in higher brightness, even at reduced QYs.

In spite of the various studies on UC in highly doped NCs, the role of concentration quenching in UC NCs is still not well understood and is not easy to model. To gain insight into the role of quenching processes at high dopant concentrations, here, we present a systematic investigation on the concentration dependence of UC dynamics of respective Er3+ and Yb3+ excited states for core and core–shell NaYF4 NCs doped with 1–100% Er3+ or Yb3+. Luminescence spectra and lifetime measurements under direct excitation in the emitting level are reported and analyzed to quantify concentration quenching for individual levels involved in UC processes. The results demonstrate that concentration quenching varies for different emitting levels and reveal how it is suppressed in core–shell NCs, e.g., the Er3+4I11/2 NIR level is highly susceptible to concentration quenching even in core–shell NCs, whereas for Yb3+ in core–shell NCs, concentration quenching is very limited. Finally, the role of solvent quenching was investigated and reveals the role of UC quenching in NCs due to coupling with high-energy vibrations of the solvent. The present results give insight on the role of ET processes in the quenching of emission for energy levels of Er3+ and Yb3+ involved in IR to vis UC in NCs and provide design rules for highly doped NCs with improved UC quantum efficiencies.

Results and Discussion

Concentration Quenching for UC NCs

Characterization of NCs

The Yb3+–Er3+ ion couple is widely used to upconvert IR radiation through ground state absorption by two Yb3+ ions followed by a two-step ET to an Er3+ neighbor. NaYF4:Yb,Er is an efficient UC material that serves as a model system both for bulk (microcrystalline) and nanocrystalline materials. In the present study, we analyze NaYF4 core NCs that are singly doped with 1–100% Er3+ or Yb3+. In addition, we investigate the influence of an inert 5 nm NaYF4 shell coated on a selection of concentrations (1, 10, and 60%) of Er3+- or Yb3+-doped core NCs. These three concentrations are selected to represent NCs with limited interaction between dopant ions (low concentration, 1%), ET between neighboring ions but with no energy migration (intermediate concentration, 10%), and a concentration above the percolation point allowing for 3D energy migration (high concentration, 60%). This allows to distinguish between two types of energy transfer: energy migration (involving energy transfer among dopants to the same excited state of identical neighboring ions), which in principle does not lead to quenching or shortening of the lifetime because the excited-state population of the level involved does not change. However, it can induce quenching by diffusion (multiple ET steps) of the excited state from ions in the center (no quenching sites nearby) to identical ions at the surface (close to quenching sites) where quenching occurs. This will result in a faster decay time for high doping concentrations, typically above 10% because multistep energy transfer is need. The second type of energy transfer is cross-relaxation, which depopulates the excited state by partial energy transfer to a neighboring ion. This directly quenches the excited state and shortens the decay time and is already observed at lower doping concentrations.

Details on synthesis are in the Supporting Information (SI). First, we focus on the Er3+-doped NCs. Figure a,b shows the representative transmission electron microscopy (TEM) images for monodisperse NaYF4:1% Er3+ core (25 × 18 nm) and NaYF4:1% Er3+@NaYF4 core–shell (35 × 28 nm) NCs. At higher Er3+ doping concentrations, the size and morphology of NCs are maintained (Figure S1). The 5 nm thick inert shell helps to suppress but will not fully eliminate surface quenching.[22,23,28]

Figure 1

TEM images of (a) NaYF4:1% Er3+ core and (b) NaYF4:1% Er3+@NaYF4 core–shell NCs (scale bar = 50 nm). (c) Excitation spectrum of NaYF4:1% Er3+@NaYF4 core–shell NCs for 1522 nm emission and emission spectra of NaYF4:X% Er3+@NaYF4 core–shell NCs (X = 1, 10, and 60) for 520 nm excitation. (d) Luminescence decay curves of 4I13/2 emission excited at 520 nm for NaYF4:X% Er3+ core NCs (X = 1–100). (e) Schematic illustration of interaction between Er3+ ions for low and high concentrations of Er3+. (f) Energy-level diagram of Er3+ showing emission (colored arrows) from levels studied in this work.

First, the optical properties were analyzed by recording luminescence spectra. Figure c shows the emission and excitation spectra for Er3+-doped core–shell NCs. The excitation lines at 377, 489, 520, 540, and 654 nm are assigned to characteristic Er3+ transitions from the 4I15/2 ground state to the 4G11/2, 4F7/2, 2H11/2, 4S3/2, and 4F9/2 states. The emission spectrum in the lower panel of Figure c recorded for 2H11/2 excitation (520 nm) shows that, in addition to emission in the visible (green 4S3/2 and red 4F9/2 emission, not shown), the NCs emit in the IR at 840 (4I9/2 → 4I15/2 and 4S3/2 → 4I13/2), 976 (4I11/2 → 4I15/2), and 1522 nm (4I13/2 → 4I15/2). It is clearly observed in the emission spectra that upon increasing the Er3+ concentration, the relative emission intensities from levels higher than 4I13/2 decrease by concentration quenching. In core-only NCs, emission from each excited state, including the 4I13/2 state, is quenched for higher dopant concentrations because of stronger surface-related quenching. Quenching of emission can be accurately probed by luminescence decay measurements. With a fixed radiative decay rate, the presence of additional nonradiative (quenching) pathways will result in a faster decay. As an example, in Figure d, the luminescence decay curves of the 4I13/2 emission are shown for increasing Er3+ concentrations in core-only NCs. Significant quenching is observed, especially for concentrations higher than 10%, when long-range energy migration becomes possible. To reduce quenching, an inert shell[18,21] is grown on the selection of cores with 1, 10, and 60% Er3+. The actual Er3+ concentrations incorporated in the NCs are verified by inductively coupled plasma (ICP) analysis. The results in Table S1 show that these concentrations in the NCs are close to the nominally added concentrations. In Figure e, the ET processes for the two extreme concentrations are schematically indicated (1%, limited interaction/no ET between Er3+ ions and 60%, energy migration over a network of multiple Er3+ ions).

Luminescence Decay Dynamics for Er3+ IR-Emitting Levels

Luminescence decay curves provide important information on luminescence quenching processes. In the present study, we focus on the 4S3/2, 4F9/2, 4I11/2, and 4I13/2 excited states, which are all involved in UC processes (Figure f). The solvent used for all measurements (unless otherwise specified) is cyclohexane. We first consider the concentration dependence of the luminescence decay for the IR-emitting 4I13/2 and 4I11/2 excited states, which serve as intermediate levels for IR → NIR and NIR → vis UC.[29−31]Figure shows the 4I13/2 and 4I11/2 decay dynamics after direct excitation in the emitting level for core and core–shell NCs with 1, 10, and 60% Er3+ in the core. There is a deviation from single-exponential decay, especially for Er3+ emission in core-only NCs. The initial fast decay is explained by faster quenching for near-surface Er3+ ions. Growth of a 5 nm protective shell reduces the initial fast decay. To obtain a better insight into the quenching of the emission, the luminescence decay is quantified by a lifetime obtained from single-exponential fitting and also by an average lifetime defined by ∑0tI(t)/∑0I(t), where I(t) is the emission intensity at time t. The average lifetimes (indicated in the figures) are close to the values obtained from a single-exponential fitting. For core and core–shell NCs, the same trend is observed: the 4I13/2 and 4I11/2 emission lifetimes decrease as Er3+ concentration increases due to concentration quenching. However, the reduction in lifetime is much stronger in the core-only NCs, especially for high Er3+ concentrations. There is also a significant difference between concentration quenching observed for the 4I13/2 and 4I11/2 levels. Both emissions are strongly quenched in the core-only NCs. On the basis of the reduction in 4I13/2 and 4I11/2 lifetimes from over 1 ms (1% Er3+) to ∼80 μs (60% Er3+) in the core-only NCs, it is evident that there is strong concentration quenching, giving rise to a quantum yield below 10% in core-only NCs. For the 1522 nm emission from the 4I13/2 level, the lifetime increases upon growth of an inert shell and a lifetime of ∼8 ms is observed, which is close to the radiative lifetime. The slight reduction from 8.51 (1% Er3+) to 7.34 ms (60% Er3+) reveals that the (concentration) quenching of the 4I13/2 level is almost exclusively surface-related and can be successfully suppressed by shell growth. As a result, the absolute intensity of the 4I13/2 emission increases in highly Er-doped core–shell NCs for excitation in higher-energy levels (4I11/2, 4F9/2, or 4S3/2) as relaxation to the 4I13/2 level is enhanced by cross-relaxation and energy migration whereas the 4I13/2 emission is not quenched upon increasing the Er-doping concentration.

Figure 2

Luminescence decay curves of Er3+4I11/2 ((a) core, (b) core–shell, and (c) corresponding energy-level diagram) and 4I13/2 ((d) core, (e) core–shell, and (f) corresponding energy-level diagram) emission by direct excitation at 960 and 1500 nm for NaYF4:X% Er3+ core and NaYF4:X% Er3+@NaYF4 core–shell NCs (X = 1, 10, and 60 in red, green, and blue color).

The situation is different for the 4I11/2 level. Shell growth reduces quenching, but still the decay time significantly changes in core–shell NCs upon raising the Er3+ concentration from 1 (2.25 ms) to 10 (1.56 ms) to 60% (721 μs). This difference indicates that for the 4I11/2 level, in addition to surface quenching, additional quenching pathways contribute. In our recent paper,[23] we suggested a role of the high-energy (3300–3500 cm–1) O–H vibrations, which are resonant with the 4I11/2 to 4I13/2 energy gap. It is hard to prevent the incorporation of OH– groups on F– lattice sites in view of their chemical similarity. The present observations are consistent with the presence of OH– quenching centers inside the NC core. At higher Er3+ concentrations, energy migration to OH– centers inside the core will occur, which reduces the luminescence efficiency and gives rise to faster luminescence decay even when quenching by surface sites is reduced by growth of an inert shell. For the emission from the 4I13/2 level (with a larger ∼6500 cm–1 energy gap to the ground state), quenching occurs through multiphonon relaxation at the surface, which can be effectively eliminated by shell growth, making the 4I13/2 emission highly efficient (close to 100% QY) even at high (60%) Er3+ concentrations.

For comparison, we also measured the decay dynamics for 4I13/2 and 4I11/2 emissions after indirect excitation at 520 nm (4I15/2 → 2H11/2, Figure S2). Because the 4S3/2/2H11/2 state acts as an energy reservoir that feeds the IR-emitting states through multiphonon relaxation, radiative decay, and cross-relaxation processes, a rise time can be observed. The rise time becomes shorter for higher Er3+ concentrations, resulting from faster ET and cross-relaxation.[32] Qualitatively, the influence of Er3+ concentration and shell growth on decay times after indirect (Figure S2) and direct (Figure ) excitation is the same. However, the absolute values of the lifetimes are different and longer lifetimes are measured after indirect excitation and even can be longer than the radiative lifetime.[33] This makes it difficult to make a meaningful comparison of lifetimes and gain insight into the intrinsic luminescence quenching for the levels of Er3+ and Yb3+ involved in the UC. This illustrates that it is cumbersome to retrieve quantitative information on luminescence quenching for specific energy levels following excitation in higher (feeding) energy levels and shows the importance of analyzing luminescence decay curves following direct excitation in the emitting state.

Luminescence Decay Dynamics for Er3+ Vis-Emitting Levels

After ET UC, the higher-energy 4S3/2 and 4F9/2 states are populated and emit green and red UC emissions. It is also interesting to investigate efficiency losses related to concentration quenching for these emitting levels. Figure shows the 4S3/2 and 4F9/2 decay dynamics as a function of Er3+ concentration under direct excitation. As for the IR-emitting levels, concentration quenching shortens the lifetime and the inert shell reduces the effect of concentration quenching. However, there are clear differences between the two levels. Upon increasing the Er3+ concentration, a much stronger quenching is observed for the green-emitting 4S3/2 level than for the red-emitting 4F9/2 level. This is explained by cross-relaxation[34] of 4S3/2 + 4I15/2 → 4I9/2 + 4I13/2 as an additional 4S3/2 decay channel that becomes available at higher Er3+ concentrations and causes a more than 10-fold decrease in lifetime upon raising the Er3+ concentration from 1 to 10% (345–24 μs). In the literature, energy migration was suggested to be the mechanism of concentration quenching of the 4S3/2 emission, excluding cross-relaxation.[33] The present results show pronounced quenching of the 4S3/2 emission both in core and core–shell NCs already for 10% Er3+ where energy migration is limited. Efficient cross-relaxation quenching for Er3+4S3/2 emission has been reported, both in bulk and nanocrystalline materials.[35] The strong contribution to cross-relaxation quenching is confirmed by lifetime measurements of the green Er3+ emission in the core–shell NCs. Shell growth reduces the quenching, but still a strong decrease of the decay time is observed from 436 (1% Er3+) to 125 μs (10% Er3+). This is a clear signature of cross-relaxation quenching and not surface-related quenching.

Figure 3

Luminescence decay curves of Er3+4S3/2 ((a) core; (b) core–shell; (c) corresponding energy-level diagram) and 4F9/2 ((d) core; (e) core–shell; (f) corresponding energy-level diagram) emissions by direct excitations of 520 and 640 nm for NaYF4:X% Er3+ core and NaYF4:X% Er3+@NaYF4 core–shell NCs (X = 1, 10, 60 in red, green, and blue color, respectively).

The luminescence decay curves for the red 4F9/2 emission (Figure d,e) in core-only NCs show only limited quenching at 10% doping (decrease in lifetime from 252 to 170 μs). This is consistent with the fact that there is no resonant cross-relaxation path for the 4F9/2 level. At 60% Er3+, a stronger decrease in lifetime is observed (53 μs) that can be explained by energy migration to the surface where quenching occurs. Upon shell growth, surface quenching is suppressed and the 4F9/2 emission lifetimes lengthen to values close to those in the 1% sample, showing that concentration quenching for the 4F9/2 emission is very limited up to very high (60%) Er3+ concentrations in core–shell NCs.

On the basis of the results above, important information on the concentration quenching in highly Er3+-doped NaYF4 NCs can be obtained. Under low excitation powers, strong concentration quenching is observed for all emitting levels (4S3/2, 4F9/2, 4I11/2, and 4I13/2) in core-only NCs with 60% Er3+. The quenching for the 4S3/2 (545 nm) and 4I11/2 (976 nm) emission is stronger and is not only caused by surface quenching but also by cross-relaxation (4S3/2) and O–H vibrations (4I11/2) inside the NC core. Growth of an inert shell cannot eliminate these quenching pathways. For the 4F9/2 (654 nm) and 4I13/2 (1522 nm) levels, the dominant mechanism for concentration quenching is energy migration to the surface where multiphonon relaxation and possibly also surface defect states quench the emission. Growth of an inert shell (5 nm thickness here) can effectively eliminate these quenching processes and efficient red (4F9/2) or IR (4I13/2) emission is observed up to the highest Er3+ concentrations.

To analyze the concentration quenching for UC more quantitatively, we determined lifetime ratios for core and core–shell NCs for the three different Er3+ concentrations (1, 10, and 60%), as summarized in Table . The lifetime ratios for 1% Er3+-doped core–shell and core NCs (1st row) are a measure for how well the 5 nm shell shields the respective levels from quenching by high-energy vibrations of the solvent/ligand. The ratios are similar (around 1.3) for the four different levels. This indicates that vibrational quenching is an active quenching mechanism in the 1%-doped NCs and that shell growth is effective in reducing the losses for all emitting levels. This is consistent with previous reports.[36−38] The role of ET and concentration quenching processes is quantified in the 2nd to 4th row where ratios of emission lifetimes for core–shell NCs are tabulated for different doping concentrations. Shell growth limits surface quenching, and the ratios reflect the role of quenching processes inside the NC core. The ratio of emission lifetimes in the core–shell NCs with 1 and 10% Er3+ is the highest for the 4S3/2 level (3.49 vs 1.06–1.44 for other levels). This is explained by efficient 4S3/2 cross-relaxation quenching between Er3+ neighbors that is effective at moderate Er3+ concentrations, well below the percolation point for energy migration. The ratio of the lifetimes for the 1/60% core–shell samples provides insight into the role of long-distance energy migration. The relatively large value for the 4I11/2 emission (3.12 vs ∼1.3 for the 4F9/2 and 4I13/2 levels) shows that energy migration to quenching sites (possibly OH–) is important for quenching of the 4I11/2 emission in 60% Er3+-doped NCs. The fourth row gives the ratio of lifetimes in 10% over 60% Er3+-doped core–shell NCs and signifies the role of (long-range) energy migration vs (short range) cross-relaxation ET quenching. The high number for the 4I11/2 level reflects that for this level, quenching through energy migration is important whereas for the 4S3/2 level, short-range cross-relaxation is the dominant quenching process. For the 4F9/2 and 4I13/2 levels, the lifetime ratios in the core–shell NCs are between 1.1 and 1.4 for all Er3+ concentrations as quenching processes in the NC core are inefficient. Both the 4F9/2 and 4I13/2 levels are efficiently luminescing up to high Er3+ concentrations for core–shell NCs. Finally, in the fifth row, the ratios of the radiative decay time and the decay time in 1%-doped core–shell NCs are tabulated. The intrinsic radiative lifetime τ0 is determined taking into account the correction for local field effects[39−42] (see the SI for details). The ratio of the radiative and observed lifetimes is the highest for the 4I11/2 state (3.8), more than twice the ratios obtained for the 4S3/2, 4F9/2, and 4I13/2 states. The relative strong quenching of the 4I11/2 emission in low-doped core–shell NCs has been observed before and was attributed to efficient quenching by O–H vibrations resonant with the 4I11/2 to 4I13/2 energy gap.[38,43] Recently, confirmation for this hypothesis was reported: NaYF4:Er,Yb NCs with superior UC QYs, close to the highest QYs obtained in bulk NaYF4:Er,Yb, were realized by using a synthesis method in which all reactants containing OH groups (such as water and MeOH) were excluded.[44]

Table 1

Ratios of Emission Lifetimes for Various Levels of Er3+ (4S3/2, 4F9/2, 4I11/2, and 4I13/2) and Yb3+ (2F5/2) Doped in Core-Only (τc) and Core–Shell (τc–s) NCs with Different Concentrations (1, 10, or 60%) of Er3+ or Yb3+

	4S3/2	4F9/2	4I11/2	4I13/2	2F5/2
τc–s/1%/τc/1%	1.26	1.29	1.38	1.22	1.08
τc–s/1%/τc–s/10%		1.19	1.44	1.06	1.07
τc–s/1%/τc–s/60%		1.38	3.12	1.22	2.92
τc–s/10%/τc–s/60%		1.17	2.16	1.16	2.72
τ0/τc–s/1%	1.66	1.81	3.80	1.06	1.10

The present results on lifetime ratios in NaYF4 core–shell NCs with 1, 10, and 60% Er3+ reveal that emission from the 4I11/2 and the 4S3/2 levels is strongly quenched at higher Er3+ concentrations, in spite of an inert NaYF4 shell. The 4I11/2 emission is sensitive to quenching sites inside the core that can be effectively reached by energy migration at elevated Er3+ concentrations. This will give rise to a significant reduction of UC efficiency, especially for the highest dopant concentration (60% Er3+) at low excitation densities. Because of the resonance of the Er3+ 4I11/2 and Yb3+2F5/2 levels, energy migration to defect sites in co-doped Yb3+–Er3+ NCs with a high total Yb + Er concentration can also lead to strongly reduced UC efficiencies. Emission from the 4S3/2 level is quenched by effective cross-relaxation, already at Er3+ concentration of 10%. Emission from the 4F9/2 and 4I13/2 levels remains efficient (QY > 50%) up to the highest Er3+ concentration (60%) in core–shell NCs. This also indicates that IR (1.5 μm)-to-NIR (1000 nm) UC from the 4I13/2 level can be more efficient than the more commonly used IR (980 nm)-to-visible (545 nm) UC.[25]

Luminescence Decay Dynamics for Yb3+ IR-Emitting Level

To further understand and optimize the UC efficiency in highly doped core–shell NCs under low excitation densities, also lifetime measurements were done for Yb3+-doped NCs. The Yb3+ ion has a single (2F5/2) excited state and is commonly used as a sensitizer to realize brighter UC emission. Energy transfer from Yb3+ to Er3+ feeds the UC emission, and it has been established that UC QYs correlate with the lifetime of the Yb3+ excited state.[37] Faster relaxation of the Yb3+ emission through energy migration to quenching sites reduces the UC efficiency. Modeling of energy migration is complex, but rate equation models have been able to reproduce excited-state dynamics in concentrated UC NCs.[37] In Figure S3, 2F5/2 emission decay curves are displayed as a function of Yb3+ concentration for NaYF4:X% Yb3+ core-only NCs (X = 1–100). The lifetime shortens with increasing Yb3+ concentration, which can be explained by concentration quenching related to energy migration to surface quenching sites at elevated Yb3+ concentrations. To investigate the influence of shell growth, Figure displays the luminescence decay curves for the Yb3+ emission in core-only and core–shell NaYF4 NCs with 1, 10, and 60% Yb3+. The lifetime ratios are also included in Table . For core-only NCs, a rapid decrease in lifetime is observed, especially in the 60%-doped NC. After shell growth, 2F5/2 emission decay time is similar for the 1 and 10% doped NCs and close to the radiative decay time (1.93 ms, see the SI for details). Upon doping with 60%, the decay time drops to 603 μs corresponding to a QY of ∼30%. The high efficiency and limited quenching of the Yb3+2F5/2 emission can be understood from the large energy separation (∼10 000 cm–1) between 2F5/2 excited state and 2F7/2 ground state which limits multiphonon relaxation. The present results indicate that in spite of this large gap, concentration quenching does occur at concentrations above 10%, which can cause a reduction in UC efficiency at low excitation densities in highly Yb3+-doped NCs reported in the literature.[45,46] Optimizing the Er concentration to enhance the Yb3+–Er3+ energy transfer rate can provide a more efficient competing UC channel and reduce the role of concentration quenching. Indeed, in a recent study, it was found that for low excitation powers in concentrated Yb NCs (NaYb1–ErF4), the optimum Er concentration is 8% (x = 0.08), significantly higher than the typically used Er3+ concentration of 2%.[15]

Figure 4

Luminescence decay curves of Yb3+2F5/2 emission for direct excitation at 960 nm for (a) NaYF4:X% Yb3+ core and (b) NaYF4:X% Yb3+@NaYF4 core–shell NCs (X = 1, 10, 60 in red, green, and blue color) and (c) corresponding energy-level diagram.

Solvent Quenching for UC NCs

Er3+ IR-Emitting Levels

In addition to dopant concentration, also the solvents in which the NCs are dispersed affect the UC efficiency. In a recent paper,[23] Plwe demonstrated how by coupling with high-energy vibrations of surrounding solvent molecules, the various energy levels involved in the UC process are quenched. Shell growth strongly reduces the quenching by increasing the distance between the optically active centers and the high-energy vibrations.[22,23] To further understand the role of solvent quenching, the luminescence decay dynamics of Er3+ and Yb3+ states were investigated for NaYF4:1% Er3+@NaYF4 and NaYF4:1% Yb3+@NaYF4 core–shell NCs in three solvents: cyclohexane (with 2900–3000 cm–1 of C–H vibrations from surrounding solvent molecules and surface ligands), H2O (3200–3700 cm–1 of O–H vibrations), and heavy water D2O (chemically the same as water but with much lower vibrational energies 2200–2500 cm–1 for O–D vibrations). Low doping concentrations were used to prevent interference with concentration quenching. Hydrophobic NCs passivated with oleate ligands are dispersed in cyclohexane. Acid treatment (see the SI for details) removes the oleate ligands and allows NCs to dissolve in deionized water H2O and heavy water D2O.[47,48]Figure a,b shows the decay curves of IR 4I11/2 and 4I13/2 emission for NaYF4:1% Er3+@NaYF4 core–shell NCs in the three solvents. The decay time decreases in the order of D2O, cyclohexane, and H2O. When considering the influence of local field effects (n(cyclohexane) = 1.43 and n(H2O/D2O) = 1.33), the radiative lifetimes are 8.54 and 10.26 ms for 4I11/2 emission in cyclohexane and H2O/D2O and 9.06 and 11.01 ms for 4I13/2 emission in cyclohexane and H2O/D2O, respectively.

Figure 5

Luminescence decay curves of (a) Er3+4I11/2 and (b) 4I13/2 emission by direct excitations of 960 and 1500 nm and (c) emission spectra of 520 nm excitation for NaYF4:1% Er3+@NaYF4 core–shell NCs in heavy water, cyclohexane, and deionized water.

The lifetimes observed for the 4I11/2 and 4I13/2 emission are similar for the NCs in cyclohexane and D2O, indicating similar QY for IR emission in the two solvents. For the 4I13/2 level, the lifetimes in the core–shell NCs are close to the radiative lifetimes in D2O and cyclohexane. The energy gap to the ground state of ∼6500 cm–1 requires more than two vibrations in D2O and cyclohexane, and clearly the three-phonon relaxation rate for the Er3+ ions in core–shell NCs is much lower than the radiative decay rate and the efficiency of the 4I13/2 level is close to 100% in cyclohexane and D2O. For the NCs in water, the two-vibrational overtone of the O–H vibrations is resonant with the 6500 cm–1 energy gap. Even with the 5 nm protective shell, two-phonon relaxation can compete with radiative decay, giving rise to a faster (209 μs) decay of the 4I13/2 emission in H2O. The 4I11/2–4I13/2 energy gap of ∼3500 cm–1 is small enough to allow the two-phonon relaxation process in D2O and cyclohexane, and this gives rise to a shortening of the lifetime to 2.2–2.6 ms for the 4I11/2 emission of Er3+ in the core–shell NCs (radiative decay time 8.54 ms). As a single O–H vibration can bridge the energy gap to the 4I13/2 level, the 4I11/2 emission is very sensitive to the presence of O–H vibrations. In H2O, the emission decay time is 60 μs and the 4I11/2 emission is strongly quenched by phonon relaxation, in spite of the 5 nm inert shell. The sensitivity to water is further illustrated in Figure S4 where the 4I11/2 decay dynamics is shown for gradual H2O addition to NaYF4:1% Er3+@NaYF4 core–shell NCs in D2O. For small amounts of H2O, strong quenching of the 4I11/2 emission is observed. The emission spectra upon excitation in the 2H11/2 level at 520 nm shown in Figure c provide further evidence for the solvent effect. Taking the 4S3/2–4I13/2/4I9/2–4I15/2 emission at 840 nm as a reference, the emission at 976 and 1522 nm (4I11/2, 4I13/2 → 4I15/2) is strongly quenched in H2O but is equally strong for the NCs in D2O and cyclohexane.

Er3+ Vis-Emitting Levels

We also investigated the role of vibrationally induced quenching in water for the visible UC emission (Figure S5). The emission lifetimes are similar for the 4S3/2 and 4F9/2 levels in cyclohexane and D2O (436 and 471 μs for the 4S3/2 level and 325 and 379 μs for the 4F9/2 level). In H2O, the lifetimes become faster (241 and 262 μs for 4S3/2 and 4F9/2 emission). The decrease in lifetime by changing the solvent from D2O to H2O is 49% for the 4S3/2 emission and 31% for the 4F9/2 emission. The fact that the O–H vibrational energy is closer in resonance to the energy separation between 4S3/2 and 4F9/2 states (∼3200 cm–1) than for 4F9/2 to 4I9/2 energy gap (∼2800 cm–1) can explain the stronger quenching for 4S3/2 state in H2O. This result is in agreement with the observation that changing to H2O as solvent can enhance red emission at the expense of green emission upon a 980 nm excitation.[47] Previously, an unexpected longer lifetime for red 4F9/2 emission was reported for Er3+-doped NCs in H2O compared to that for D2O under 380 nm excitation in the 4G11/2 state.[49] This was interpreted as an increased 4F9/2 population through multiphonon relaxation from higher-energy levels. Here, lifetime measurement under direct excitation (Figure S5b) clearly demonstrates more pronounced quenching for the red-emitting 4F9/2 level by O–H vibrations (emission lifetime 262 μs) in comparison to O–D vibrations (379 μs), in line with what is expected based on higher vibrational energy of O–H vibrations. Note that all experiments for solvent quenching were conducted on core–shell NCs. Much stronger vibrational quenching is observed for core-only NCs.

The present analysis of luminescence decay times under direct excitation for various energy levels of Er3+ in core–shell NCs dispersed in different solvents reveals a clear difference for the role of the high-energy O–H vibrations of water. Both IR levels (4I11/2 and 4I13/2) that serve as intermediate levels in UC processes show significant quenching by ET to the high-energy O–H vibrations, even in core–shell NCs. The presently used shell thickness of 5 nm is not sufficient to protect the IR-emitting levels from quenching by ET to the fundamental O–H vibration (3500 cm–1) for the 4I11/2 level or the two-phonon mode for the 4I13/2 level. The reason for the strong quenching is a combination of a high oscillator strength for O–H vibrations (favoring ET as the Förster-type ET rate is proportional to the acceptor oscillator strength) and a long radiative lifetime for the IR-emitting levels (∼8 ms), which makes it harder for the slow radiative decay to compete with ET. To enhance UC efficiencies in aqueous media, a thicker inert shell will be beneficial. For the visible UC emission in the green and red spectral region, the transition from apolar aliphatic solvents to water induces additional nonradiative decay but the effect is less than a factor of 2.

Yb3+ IR-Emitting Level

In Er3+,Yb3+ co-doped UC NCs, quenching of the 980 nm NIR emission from Yb3+ in different solvents is known to reduce the UC efficiency. To investigate the role of Yb3+ concentration quenching, we measured the solvent dependence of 2F5/2 decay for singly doped NaYF4:1% Yb3+@NaYF4 core–shell NCs (Figure S6). The decay times for the NIR Yb3+ emission are very similar independent of the solvent, indicating a limited contribution of water vibrations in quenching the ∼980 nm IR emission.[23,38,49] The absence of solvent quenching is not unexpected in view of a large energy gap between 2F5/2 and 2F7/2 states of 10 000 cm–1 that requires three-phonon relaxation in water (vs four phonons in organic solvents). For the core–shell NCs, the minimum distance of 5 nm imposed by the inert shell is sufficient to virtually eliminate multiphonon quenching. The small decrease in the luminescence decay time of the 2F5/2 emission from 2.10 in D2O to 1.98 ms in H2O can be ascribed to some multiphonon quenching in H2O, and the quenching rate can be estimated to be ∼30 s–1. Note that water quenching is present in core-only NCs,[43] Yb3+ complexes,[50] and silica-encapsulated NCs,[49] where Yb3+ ions are directly coupled to water molecules and ligands with high-energy vibrational modes. However, doping in crystalline dielectric media with a 5 nm shell passivation gives rise to efficient Yb3+ emission and sensitization of ET UC. The quenching of 980 nm NIR emission in Er3+,Yb3+ co-doped NCs in water is primarily due to rapid ET between the 2F5/2 level of Yb3+ and the resonant 4I11/2 level of Er3+ and quenching of the 4I11/2 level by O–H vibrations, as discussed above.

Conclusions

There is a trend toward higher doping concentrations for both Yb3+ and Er3+ to achieve higher brightness of UC NCs. High brightness and superior quantum yields of UC emission have been reported under high power density excitation for these highly doped NCs. Insight into concentration and solvent quenching processes is however lacking and requires studies at low excitation powers where UC rates are low and decay dynamics is dominated by nonradiative and radiative decay rates of excited states. Here, we have presented a systematic investigation of decay dynamics for Er3+ and Yb3+ in core and core–shell NCs under direct excitation in various emitting levels involved in UC processes for a wide range of concentrations (1–100% doping).

Core-only NCs suffer from strong concentration quenching. Fast migration to surface ions that are strongly quenched by vibrational coupling with nearby high-energy vibrations of solvent and capping molecules can explain the pronounced concentration quenching. For core–shell NCs, surface quenching is suppressed but still concentration quenching is observed. A detailed analysis on decay behavior following direct excitation in the emitting levels demonstrates that both the 4S3/2 (green) and 4I11/2 (NIR) levels are strongly quenched upon raising the Er3+ concentration. For the 4S3/2 level, cross-relaxation has been identified as the main quenching mechanism, whereas for the 4I11/2 level, energy migration to defects/impurities in the core (possibly OH–) is responsible for the observed concentration quenching. The IR-emitting 4I13/2 and the red-emitting 4F9/2 levels do not show strong concentration quenching, and their emission remains efficient up to high (60%) doping concentrations in core–shell NCs. Also for Yb3+, limited quenching is observed in highly (60%) Yb3+-doped core–shell NCs.

The role of solvent vibrations was investigated by studying decay dynamics for emission from various levels of Er3+ and for the NIR emission of Yb3+ in core–shell NCs. The results show a much stronger quenching for the IR-emitting levels 4I11/2 and 4I13/2 (intermediate levels in the UC process) than for the 4S3/2 and 4F9/2 levels responsible for the green and red UC emission. Our work provides a better understanding and deeper insight on quenching pathways in highly doped UC NCs and provides important information on design rules for efficient UC under lower power densities that are required in many applications.