Giant Light‐Harvesting in Dye‐Loaded Nanoparticles Enhanced by Blank Hydrophobic Salts

Light‐harvesting is a fundamental process in nature, which inspires researchers to develop artificial systems for photocatalysis, photovoltaics, and biosensing. A previously introduced light‐harvesting nanoantenna, based on polymeric nanoparticles (NPs) loaded with rhodamine dyes and bulky hydrophobic counterions, provides a record‐breaking antenna effect ≈1000. However, the high dye cooperativity of its thousands of encapsulated dyes causes energy losses by traces of self‐quenched dye aggregates. Here, it is found that these imperfections can be suppressed by blank hydrophobic salts (BHS) formed by the same bulky counterion (fluorinated tetraphenylborate) with an optically inactive cation, analogs of ionic liquids. The presence of BHS increases twofold the fluorescence quantum yields and fluorescence lifetimes of NPs and suppresses their fluorescence blinking. This study assumes that BHS provides an excess of bulky counterions that excludes traces of dye aggregates. As a result, an efficient Forster resonance energy transfer (FRET) is achieved from 40 000 dye donors to a single acceptor within a 70 nm particle, leading to the antenna effect of 4800, which is by far the highest value reported to date. Using this nanoantenna, a single‐molecule detection of the FRET acceptor is realized at low excitation power using an RGB camera of a smartphone.


Introduction
8][9][10] The wide range of applications DOI: 10.1002/adom.202301671 of light-harvesting antennas (LHA) includes photosynthesis, [11,12] photocatalysis, [6,[13][14][15] single molecule tracking in biology, [16] or photo-electronic devices. [17]he basic concept of LHA relies on efficient energy collection and fast excitation energy transfer (EET), by densely packed chromophores, toward a single reaction center.[27][28][29][30][31][32] In such applications, the key parameters of the so-called "antennas" are their brightness, [33,34] energy transfer efficiency, and antenna effect (amplification of the acceptor signal).High brightness together with a strong antenna effect should enable single-molecule detection using simple detection devices.Thus, smartphone-based fluorescence detection has emerged as a promising field of study, [24,35] which we recently showed for the FRET-based nanoprobes at the single-particle level. [36]Moreover, a recent study enabled detection of single molecules by a smartphone using plasmonic nanoantennas. [37]However, it is noteworthy that no previous study has successfully demonstrated single-molecule detection in LHA materials using a smartphone camera.
Fluorescent dyes, displaying high brightness, are promising candidates for building light-harvesting materials.However, their fluorescence quantum yield (QY) is limited by aggregation-caused quenching (ACQ) upon incorporation into LHA. [38,39][53] In these organic systems, rational molecular design can minimize ACQ, which ensures their high fluorescence brightness and efficient energy transfer. [54]However, these approaches require long and fastidious synthesis.In this respect, a simple platform to build LHA is dye-loaded polymeric nanoparticles. [55]Cationic dyes, such as rhodamines and cyanines, which are among the brightest dyes available to date, [56] can be effectively encapsulated inside polymer NPs with minimized ACQ using bulky hydrophobic counterions (e.g., tetrakis (pentafluorophenyl) borate, F5-TPB) [55,[57][58][59] The anions are also known as weakly coordinating anions, which has been extensively used for catalysis, [59,60] ionic liquids, [61] lithium batteries, [62,63] and organic light-emitting diodes. [64]The hydrophobic counterions play the role of insulator preventing dyes from pi-stacking inside the polymeric matrix and maintaining a short dye-dye distance to ensure efficient and ultrafast EET. [65]Bulky anions have been used in form of neat salts in ion-associated materials. [66]Supramolecular counterions composed of small inorganic anions and bulky macrocycle (cyanostar) enable preparation of ultrabright materials (SMILES) based on cationic dyes with efficient energy transfer. [67,68]In polymeric NPs, the use of bulky counterions decreases drastically the ACQ, so that the dyes can be loaded into polymeric NPs at high concentrations while keeping nearly 50% quantum yield, [69] which yielded 20-40 nm NPs ≈100 times brighter than QDs. [65,69]Importantly, encapsulation of ≈10 000 rhodamine dyes (R18) with F5-TPB counterion in (e.g., poly (methyl methacrylate-co-methacrylic acid (PMMA-MA) NPs resulted in giant light-harvesting nanoantenna with an unprecedented antenna effect of 1150±150. [65]However, in this LHA, significant ACQ phenomena were still observed for high dye loading, [70] probably due to a small fraction of self-quenched dye aggregates.The latter could originate from traces of small hydrophilic counterions, unavoidably present in the synthetic dye ion pair (R18/F5-TPB) and/or entrapped from the buffer during particle preparation.In this case, one dark dyes' aggregate per 1000 emissive dyes would be sufficient to significantly quench the particle fluorescence, as shown in our multiple energy transfer studies. [57,65,69]Fighting ACQ in LHA materials constitutes a major challenge that could further enhance their performance.
In the present work, to minimize ACQ in LHA materials, and thus enhance their antenna effect, we developed an approach based on blank hydrophobic salts (BHS).These ion pairs are composed of the same bulky hydrophobic anion (F5-TPB) paired with optically inactive hydrophobic cation.We synthesized a number of hydrophobic cations based on aliphatic tetrabutyl ammonium, and aromatic alkyl pyridinium and imidazolium.These cations are close analogs of those used for the preparation of ionic liquids, [61,71] although here we were particularly interested in producing their hydrophobic salts.We found that loading the R18/F5-TPB dye with BHS salts into polymeric NPs significantly improved the fluorescence quantum yield and increased their fluorescence lifetime.This allowed preparation of NPs with extremely high dye loading and fluorescence brightness per volume (10 400 ± 100 M −1 cm −1 nm −3 ).The obtained nanoparticles (≈70 nm in size) loaded with ≈40 000 donor dyes undergo efficient FRET to a single acceptor, giving rise to an antenna effect of 4800 ± 300, which is by far the highest value achieved to date for LHA.

Results and Discussion
Charge-controlled nanoprecipitation using PMMA-MA polymer was used to obtain small dye-loaded nanoparticles (NPs). [55,65]ydrophobic rhodamine dye (R18) with bulky hydrophobic coun-terions (F5-TPB) was used for encapsulation into PMMA-MA NPs.F5-TPB plays a dual function: it prevents the formation of H-aggregates of R18 and ensures efficient encapsulation without dye leakage. [57]Nevertheless, according to the previous data, the increase in the dye loading from 1 to 30 wt.% (23% with respect to the total particle mass), the quantum yield (QY) of NPs decreased from 0.8 to 0.3. [70]Even though the QY values were still high, improving QY at high dye loading could further enhance particle brightness. [54]The observed self-quenching at higher dye loading can have several explanations.On the one hand, higher dye loading implies a lower dye-dye distance, which could increase the probability of imperfections with dye-dye aggregates with an improperly placed counterion.On the other hand, it could be associated with presence of small fractions of dye aggregates with traces of small inorganic counterions.Indeed, during nanoprecipitation in phosphate buffer, a small fraction of F5-TPB counterions can be replaced by some other anions present in buffer like phosphate, or chloride, etc.Moreover, one cannot exclude traces (≈0.1 mol.%) of inorganic counterions in R18/F5-TPB salt, which cannot be detected by standard characterization methods.Then, even a single dye-dye aggregate among thousands of encapsulate dyes would lead to an efficient energy transfer to this dark species, [57,65] thus decreasing the quantum yield.
To provide additional dye insulation against ACQ and minimize the eventual traces of other anions than F5-TPB within the NPs, one could think of adding an excess amount of bulky counterions (F5-TPB) to the system.However, the commercially available salts of these bulky anion are with potassium, sodium, or lithium cations, which are soluble in water and thus, may not be well encapsulated inside hydrophobic polymeric NPs.In order to generate hydrophobic salts of F5-TPB, similar to that of R18/F5-TPB, we replaced its potassium ions with bulky hydrophobic cations optically transparent in the visible region: tetrabutylammonium (TBA), dodecylimidazolium (DIMD) and hexadecyl 4dimylylaminopiridinium (HDMAP) (Figure 1), which present different sizes, hydrophobicity and aromaticity.Tetrabutylammonium (TBA) is a well-known hydrophobic cation, extensively used in phase transfer catalysis. [72]In case of DIMD and HDMAP, the charge is delocalized within the aromatic heterocycle, providing "soft" cationic characteristics similar to R18 dye, while long alkyl chains ensure their hydrophobicity.In addition, the introduction of an additional redox-active system in case of DIMD and HDMAP may also improve the dye photostability as it was previously shown for organic dyes. [73]The salts of F5-TPB with corresponding hydrophobic cations were synthesized by ion exchange reaction.These salts are analogs of ionic liquids, although according to our observations, they are present in form of solid amorphous materials.We name them as blank hydrophobic salts (BHS), highlighting hydrophobic and optically inactive nature of these salts.
Then, we formulated PMMA-MA NPs loaded with different amounts of R18/F5-TPB and studied the effect of added BHS.At low dye loading (5% with respect to the total mass) the effect of BHS (added at molar ratio 1:2 with respect to the dye) was negligible.Indeed, the sizes of NPs according to the dynamic light scattering (DLS) were ≈45-50 nm (Figure S1, Supporting Information) for formulations with and without BHS.The absorption and emission spectra were typical to that of rhodamine dye in a molecular form, being close to those in methanol (Figure S2, Supporting Information).Moreover, the fluorescence quantum yields (QY) of all studied NPs were high (0.75-0.83) and did not vary with the addition of BHS (Figure S3, Supporting Information).
We next prepared PMMA-MA NPs at 23% of dye loading (with respect to the total mass) without (NP23) and with BHS (NP23+TBA, NP23+DIMD and NP23+HDMAP).DLS suggested that the presence of BHS decreased the hydrodynamic diameter from 66 to 42-52 nm (Figure S1 and Table S1, Supporting Information).Transmission electron microscopy (TEM) showed the spherical shape of NPs (Figure 2).With this method, the reference NPs NP23 displayed an average diameter of 36 ± 5 nm, while the NPs prepared with the BHS had a significantly lower average diameter: NP23 + TBA (29 ± 4 nm), NP23+DIMD (25 ± 2 nm) and NP23+HDMAP23 (30 ± 3 nm) (Figures 2 and 3b; Table S1, Supporting Information).Thus, TEM revealed smaller size of NPs compared to DLS, clearly because TEM reflects the size of the polymer core, and confirms the tendency of forming smaller NPs in the presence of BHS.The use of BHS salts likely increased the number of nucleation sites during nanoprecipitation.Due to the law of mass conservation, a greater number of NPs formed should lead to smaller NPs.The absorption and fluorescence spectra remained unchanged upon the addition of BHS (Figure 3a).However, one should note higher short-wavelength shoulder in the absorption spectra compared to 5% loading.In line with previous data, [57] it indicates an increase in the dye aggregation, responsive for the dye self-quenching (ACQ).Indeed, the QY values decreased to 0.37 ± 0.03 compared to 0.75 ± 0.03 at 5% loading (Figure S3, Supporting Information), confirming the ACQ effect.Importantly, NPs with BHS displayed higher    S1, Supporting Information).We also checked different molar ratios of BHS with respect to R18/F5-TPB and found that the originally used 1:2 ratio was optimal to achieve high QY values (Figure S4, Supporting Information).Therefore, in all our next experiments we systemically used this fraction of BHS.This increase in QY was a clear indication that the use of additional F5-TPB counterions decreased the formation of selfquenched dye aggregates, which suggests that the excess of the BHS removes from the NPs core the traces of small counterions like chloride and phosphate, present in the aqueous medium.To corroborate this observation, time-resolved fluorescence spectroscopy of NPs was conducted.As expected, the average fluorescence lifetime increased nearly twofold in the presence BHS (Figure S5 and Table S2, Supporting Information), concomitantly with changes in the QY, confirming that the BHS inhibit formation of the dark states inside NPs.The total brightness (B) of NPs was calculated (expressed as extinction coefficient × number of dyes per particle × QY) and further divided by particle volume (brightness per volume B/V, Figure 3c; Table S1, Supporting Information), in order to compare the brightness of NPs of different size. [54]NP23 + DIMD and NP23 + TBA displayed the highest values, exceeding 6000 M −1 cm −1 nm −3 .To verify applicability BHS concept to other dyes, we tested a far-red dye from cyanine family, namely hydrophobic Cy5 derivative DiD with F5-TPB counterion.At high loading (23 wt.%) in PMMA-MA NPs, it showed rather low QY (0.04) due to significant ACQ (Figure S6, Supporting Information).The presence BHS significantly improved the QY values, with DIMD/F5-TPB and TBA/F5-TPB showing the best performance: 0.09 and 0.1, respectively (Figure S6, Supporting Information).HDMAP/F5-TPB was the least efficient for the QY enhancement of DiD/F5-TPB dye, similarl to the case of R18/F5-TPB dye.We assume that the highest hydrophobicity of HDMAP/F5-TPB compared to the other two BHS could compromise its proper dispersion in the PMMA-MA matrix of NPs and provide competition for the dye encapsulation.Therefore, the HDMAP/F5-TPB was discarded from the next studies The results presented above clearly demonstrated that BHS can minimize ACQ within the NPs, which prompted us to increase the dye loading to a value of 50% R18/F5-TPB with respect to total mass.NP50 was prepared only with R18/F5-TPB, while NP50 + TBA and NP50 + DIMD contained additional BHS: TBA/F5-TPB and DIMD/F5-TPB, respectively.Upon increase in the dye loading, we first observed that absorption and emission spectra were marginally modified as compared to 23% dye loading, although a small increase in the shortwavelength shoulder was observed in the absorption spectra (Figure 3d).Again, the addition of BHS yielded smaller NPs (30-32 nm vs 40 nm), characterized by a spherical shape (Figure S8, Supporting Information) with higher QY (0.38 vs 0.24) and lifetimes (1.06-1.09ns vs 0.66 ns) (Figure 3e; Figure S9 and Table S3).Interestingly, the discarded BHS HDMAP/F5-TPB provided smaller QY (0.30) and larger particle size by DLS (Table S1, Supporting Information), which confirmed its systematically lower performance.One should note that the presence of BHS decreases the dye concentration within polymeric NPs by only 10%, while the QY increases by 58%.The total particle brightness of these NPs was remarkably high for all NPs at 50% dye loading, ranging from 1.47 × 10 8 to 2.20 × 10 8 M −1 .cm−1 .Importantly, NP50+TBA and NP50+DIMD displayed improved B/V values as compared to NP50 (Figure 3f): they were ≈1.5 times higher for NP50+TBA (10 400 ± 150 M −1 cm −1 nm −3 ) and NP50+DIMD (10 400 ± 100 M −1 cm −1 nm −3 ) as compared to NP50 (6550 ± 100 M −1 cm −1 nm −3 ).The obtained B/V values are among the highest reported to date for nanoparticles, higher than previously reported dye-loaded polymeric NPs-based PEMA-MA polymer (9480 M −1 cm −1 nm −3 ), [31] being close to the brightest conjugated polymer NPs 11 300 (M −1 cm −1 nm −3 ). [74]he development of ultrabright NPs allowed single-particle detection using low-power excitation sources, as it was demonstrated earlier for quantum dots, [75] dye-assembled NPs, [76] and dye-loaded polymeric NPs. [65]Therefore, we next characterized our NPs at a single-particle level using wide-field microscopy (Figure 4).After deposition on the glass surface, PMMA-MA at 50% dye loading appeared as bright dots.They were much brighter than QD-605, because to achieve comparable signal, QD-605 require 100-fold higher excitation power density (Figure 4a,b).Indeed, quantitative analysis of the total experimental brightness yields >100-fold higher brightness for dyeloaded PMMA-MA NPs versus that of QD-605, namely 1900-2500 photons.s−1 versus only 7 ± 1 photons.s−1 (Table S4, Supporting Information), when normalized to the same excitation power (0.5 W cm −2 at 532 nm).This dramatic difference in the experimental brightness is expected considering that the estimated brightness of QD-605 (extinction coefficient × QY) is only ≈3 × 10 5 M −1 cm −1 (vs ≈2 × 10 8 M −1 cm −1 for our NPs).
In order to compare NPs of different sizes, their brightness per volume should be analyzed. [54]Similarly to the ensemble experiments, the addition of BHS improved the experimental B/V value obtained by single-particle microscopy, expressed in photon count per second per volume (photons.s−1 nm −3 ).The obtained values were 0.110 ± 0.009 and 0.176 ± 0.007 photons.s−1 nm −3 for NP50+TBA and NP50+DIMD, respectively, both displaying higher B/V than NP50 (0.061 ± 0.006) photons.s−1 nm −3 (Figure 4c; Table S1, Supporting Information).Their B/V values were ≈100-fold higher compared to that of QD-605 (1.67 ± 0.02 × 10 −3 photons.s−1 nm −3 for particle diameter of 20 nm), when compared at the same excitation power density (Table S4, Supporting Information).Interestingly, NP50 + DIMD were brighter than NP50 + TBA which was not the case for ensemble B/V experiments (Figure 4c).Such observation is likely related to the higher excitation power density used in single-particle experiments as compared to ensemble measurements (using spectrofluorometer).Higher power density can promote dyes photobleaching and/or formation of dark species and therefore decrease the particle brightness.
Recently, we reported that PMMA-MA NPs with high dye loading displayed intensity fluctuations between bright and less emissive states. [70]So, we next measured the fraction of time a single NP remains in the bright state (Figure 4d; Table S4, Supporting Information).Among the different compositions, NPs containing BHS, NP50 + DIMD, and NP50 + TBA spent more time in bright state, 72% and 61%, respectively, compared to control NPs (48% for no BHS), confirming that they inhibit formation of the dark states.We can speculate that BHS minimizes dye-dye aggregation in the excited state (i.e., excimer formation), eventually causing deem state, because the access of bulky hydrophobic counterion decreases the probably of pi-stacking of R18 dyes.Moreover, it was previously shown that the redox systems can inhibit OFF-sates by quenching radical and triplet dark states of dyes. [73]In this respect, the highest bright-state fraction for DIMD/F5-TPB suggests that imidazolium group of DIMD, known to display redox activity, [78,79] could play a role in the stabilization of the bright state.
Our previously reported giant light-harvesting nanoantennas of 63 nm diameter loaded with 23% (30 wt.% vs polymer) R18/F5-TPB, exhibited QY of 0.28 and antenna effect (AE) of ≈1000. [65]As shown above, NP50 + DIMD displayed improved QY (0.38) at even higher dye loading (50%), which should further enhance its light-harvesting properties.Therefore, their light-harvesting capabilities were next evaluated in the presence of FRET acceptor (DiD/F5-TPB, Figure 5a,b).To obtain larger nanoantenna particles, we prepared PMMA-MA NPs by nanoprecipitation at pH 6.4.Our previous works showed that lower pH decreases charge in PMMA-MA polymer because of its carboxylate protonation.The latter leads to larger particle sizes. [65]The diameter of obtained NPs (NP50-6.4) by TEM was 79 ± 4 nm, whereas in the presence of DIMD/F5-TPB, the corresponding NPs (NP50 + DIMD-6.4)exhibited a smaller average diameter (67 ± 5 nm) (Figure S10, Supporting Information).Moreover, the measured QY for NP50 + DIMD-6.4 was higher compared to that of NP50-6.4 (Table S5, Supporting Information).We next co-encapsulated different amounts of FRET acceptor dyes (DiD/F5-TPB), while keeping constant wt.% of the donor dye R18/F5-TPB, in order to achieve the following acceptor/donor ratios: 1:2000; 1:10 000 and 1:40 000.NPs without the acceptor dye were used as a reference.The increase in the acceptor/donor ratio increased the acceptor emission, accompanied by the drop in the donor emission (Figure 5c), typical for a FRET system.However, the rise of acceptor emission was faster than the drop of the donor emission, which indicates that FRET competes with non-radiative deactivation processes within the donor dyes in the particle.In the excitation spectra measured for each composition (Figure 5d), the intensity of the acceptor peak at 661 nm was much lower than that of the donor peak at 560 nm, indicating that acceptor emission was much stronger when nanoantenna (donor) was excited.From the excitation and emission spectra, we determined the AE of the FRET NPs (Figure 5e; Table S5, Supporting Information). [65,80]The measured AE values were the highest for NPs with the lowest acceptor/donor ratio 1:40 000, reaching 4800 ± 300.This amplification factor is by far the highest value reported to date for a light-harvesting system.Previously, the highest recorded value of the antenna effect was 1150 ± 150 (i.e., ≈4-fold lower than the present one), which was achieved for analogs NPs without BHS and at lower donor R18/F5-TPB dye loading (23%). [65]Based on the AE values, we estimated FRET efficiency in these nanoantennas (see Experimental Section).At 1:10 000 and 1:40 000 ratios, the FRET efficiency was 0.47 and 0.23, respectively (Figure 5e; Table S5, Supporting Information).The FRET efficiency was further characterized by time-resolved fluorescence spectroscopy.We found that the donor lifetime decreased strongly upon increase in the acceptor to donor ratio from 1:40 000 to 1:2000, confirming the efficient FRET from large ensemble of donors to few acceptors (Figure S11 and Table S6, Supporting Information).Considering that the particle size is ≈70 nm and that the Forster radius of R18-DiD pair [65] is 6 nm, the obtained significant FRET efficiency evidenced that excitation energy can be transported thought thousands of donor dyes over distances far beyond the Forster radius.This long-distance excitation energy transfer within the nanoantenna can explain the observed outstanding AE values.
Next, single-particle microscopy was used to record real-time two-color wide-field images of FRET NPs deposited on the glass surface.Already at the acceptor:donor ratio 1:2000, practically all detected particles exhibited predominant red emission (Figure 6).The decrease in this ratio led to a decrease in the fraction of red NPs on images, even though they could be clearly observed even for 1:40 000 ratio.At this low ratio, one would expect a single acceptor per particle, so that red emission implies very efficient energy transfer from ≈40 000 dyes to a single acceptor dye.This enables detection of acceptor emission at very low excitation power (0.016 W cm −2 ), far below those used in single-molecule detection.
As reported in literature, cyanine dyes are prone to photobleaching and thermal degradation. [81]In a widefield microscopy, where the excitation power is much higher than the one used in a spectrofluorometer, the photobleaching of DiD/F5-TPB limits the observation time of the acceptor emission.Therefore, a more photostable red acceptor (Atto647N) was used. [69]So, we have modified Atto647N with hydrophobic long chain and then replaced the counterion with bulky hydrophobic F5-TPB (ATTO647N-C18/F5-TPB). [32]Similarly, we next prepared FRET NPs with R18/F5-TPB as the donor and ATTO647N-C18/F5-TPB as the acceptor, which showed characteristic dual emission with the acceptor emission increasing at molar fraction of the acceptor (Figure S12, Supporting Information).According to the singleparticle microscopy, at 1:40 000 donor to acceptor ratio, the twocolor images showed green and red emitting NPs, while control NPs without the acceptor showed only green NPs (Figure 7a,b).Therefore, we can conclude that the green and red NPs correspond to particles without and with a single acceptor molecule.The red-emitting NPs showed a single-step drop in the emission intensity to the level of the background, which was accompanied by the increase in the donor emission (Figure 7c; Figure S13, Supporting Information).These observations confirm that a single acceptor dye collects the energy from a giant donor NP containing ≈40 000 dyes, which is clearly linked to the excitation energy transfer within the donor dyes that propagates over the whole particle.Previously, a similar phenomenon was reported by us for a giant light-harvesting nanoantenna with the acceptor:donor ratio of 1:10 000. [65]Here, the performance of the nanoantenna was further improved due to the use of the BHS, allowing efficient energy transfer from 40 000 donors to a single acceptor.
Owing to the high brightness of our FRET NPs and their giant antenna effect, the fluorescence signal of the acceptor can be amplified, so it should become possible to detect it with a smartphone camera and ultra-low excitation power.To this end, we placed a smartphone camera at the optical output of the wide-field microscope and used green and red channels of the RGB camera to detect donor and acceptor signals, respectively.We found that FRET NPs with 1:10 000 acceptor:donor ratio, exhibit a color change from red to green at the level of single particles (Figure 8a), which corresponds to just a few acceptor molecules.The intensity traces of the single FRET NPs showed an anti-correlation between the donor and acceptor emission, where the single-step bleaching of the acceptor was accompanied by the single-step light-up of the donor (Figure 8b).This anti-correlation suggested that we detected a single-molecule acceptor, undergoing a single-step bleaching, which led to the dequenching of the donor particle.One should note that these single-particle measurements were done using ultra-low excitation power density of ≈0.016 W cm −2 in a simple wide-field microscopy setup.Previously, single-molecule measurements by a smartphone camera were achieved using amplification by plasmonic nanoantennas. [37]However, it required TIRF excitation with much higher excitation power density that was 54 W cm −2 .Thus, with our light-harvesting nanoantenna, we are able to achieve single-molecule detection with a smartphone camera at much lower irradiance.

Conclusion
Bulky hydrophobic counterions became a powerful method to prevent aggregation-cased quenching in nanomaterials and enable preparation of ultrabright NPs for biosensing and bioimaging.In particular, they allow preparation of giant lightharvesting nanoantenna with record-breaking antenna effect due to high cooperativity of thousands of dyes encapsulated at high local concentration.Nevertheless, at extremely high loading, some fluorescence quenching was still observed for these nanoantennas, because even a single dye aggregate could be able to quench fluorescence of thousands of dyes within the particle.We hypothesized that this quenching could originate from small fraction of inorganic counterion trapped from buffer solution during nanoprecipitation of NPs or present in the encapsulated dye salt.To minimize the quenching effect, we propose to use blank hydrophobic salts (BHS) composed of optically inactive hydrophobic cation and bulky hydrophobic counterion in combination with dye/hydrophobic counterion pair.In particular, we synthesized several hydrophobic cations paired with bulky fluorinated tetraphenylborate and co-encapsulated them together with dye/bulky counterion salt inside polymer NPs.This strategy allowed us to improve fluorescence quantum yield of NPs at high dye loading while decreasing particle size.The improved quantum yield in the presence of BHS was observed for both rhodamine and cyanine dyes, indicating that the approach could be applied for different cationic dyes.As a consequence, the brightness per volume of the NP increased by a factor of 1.5 and the use of BHS allowed us to increase their loading up to 40% in total mass (100 wt.% vs polymer).This brightness increase was observed both in ensemble and single-particle experiments.Moreover, BHS drastically decreases single-particle blinking by increasing their fraction of the ON states.Thus, our data suggest that BHS prevents ACQ in the particle core, thus minimizing energy losses in the nanoantenna.The BHS effect could be explained by two mechanisms: i) improved insulation of encapsulated dyes and ii) exclusion of traces of small inorganic anions.
The light-harvesting properties of the obtained NP were evaluated by incorporating FRET acceptor dyes within the NP down to a single acceptor per NP.In the new FRET NPs, we reached the antenna effect of ≈4800, which is much higher than the one we previously reported. [65]On the basis of their improved excitation energy transfer capabilities, single particle experiments showed that emission from a single acceptor dye can be observed under low excitation power densities (i.e., 0.016 W cm −2 ) allowing single-molecule acceptor detection in single nanoantenna particles using a smartphone RGB camera.Overall, we show that BHS can minimize quenching processes in dye-loaded nanoparticles and boost performance of light-harvesting devices.We expect that our BHS approach could be extended to other LHA materials based on ionic dyes, which could include rhodamines [32,36] and cyanines, [55] SMILES approach [67,68] and ionic AIE dyes [48] as well as to dye-loaded materials of higher dimensionality, such as polymeric nanofibers and thin films. [82]Moreover, the achieved boost in the performance of light-harvesting nanoantennas opens the way to detection of single molecules using portable user-friendly devices.

Synthesis of fluorescent NPs:
The stock solution of PMMA-MA in acetonitrile (2 mg mL −1 ) were taken with different amount of R18/F5-TPB (5, 23 and 50% of dye with respect to the total mass of NPs).Then, desired BHS solutions in acetonitrile were mixed with R18/F5-TPB stock solutions in different molar ratio of dye:BHS.Then, 50 μL of these stock solutions were added quickly to 450 μL of 20 mm phosphate buffer at pH 7.4, r.t under vigorous shaking (thermomixer in 1150 rpm) using micropipette.Next, 200 μL of these NP suspensions were again quickly added to 800 μL with the same phosphate buffer (20 mm, pH 7.4) at r.t.
For the preparation of FRET NPs, different amounts of DiD/F5-TPB or ATTO647N-C18/F5-TPB in acetonitrile (1:2000, 1:10 000 and 1:40 000 acceptor:donor molar ratio) were added to the polymer stock solutions already containing R18/F5-TPB along with BHS in acetonitrile.Then, NPs were prepared using above-mentioned procedure.For the larger-sized NPs, phosphate buffer (20 mm, pH 6.4) was used only at the first dilution step.
Characterization of NPs: The hydrodynamic diameter measurements of the NPs were performed using dynamic light scattering (DLS) method on a Zetasizer Nano ZSP (Malvern Instruments).The averaged values of the diameters of NPs were calculated from three sets of results based on volume distribution analysis.
Transmission electron microscopy (TEM) was performed to check the absolute diameter and the shape of the NPs.Carbon-coated copperrhodium electron microscopy grids with a 300 mesh (Euromedex, France) were surface treated with a glow discharge in amylamine atmosphere (0.5 mbar, 4-5 mA, 30 s) in an Elmo glow discharge system (Cordouan Technologies, France).Then, 5 μL of the solution of NPs at 0.04 g L −1 were deposited on the grids and left for 3 min two times to get countable numbers of NPs on the grid.The grids were then treated for 2 min with a 2% uranyl acetate solution for staining.After that, the experiments were performed on a Philips CM120 transmission electron microscope equipped with a LaB6 filament and operating at 100 kV.Images were analyzed using Fiji software.
Absorption spectra were recorded on a Cary 5000 scan UV-vis spectrophotometer (Varian); and excitation and emission spectra were recorded on a FS5 Spectrofluoremeter (Edinburg Instruments) equipped with thermostated compartment.The fluorescence spectra the NPs were recorded at the excitation wavelength of 530 nm.The QYs of the R18/F5-TPB NPs were measured using rhodamine 101 in ethanol as a reference (QY = 1.00) and for DID/F5-TPB NPs DiD in MeOH (QY = 0.33) were used as standard. [57]olume normalized brightness was calculated using the following Equation ( 1): where n is number of dyes present inside the NP,  is the extinction coefficient of the dyes, and volume of NPs is the value obtained based on TEM measurements. [67]he antenna effect (AE) in the NPs were then calculated using the following Equation ( 2 , where I em D−FRET and I em D are the emission maxima of the donors of the NPs with and without acceptor, respectively. [65]he FRET efficiency (E) of the NPs was calculated from AE based on the following Equation ( 3): where n D and n A are the numbers of donors and acceptors, respectively, per particle,  D and  A are the extinction coefficient of donors and acceptors. [80]luorescence Lifetime Measurements: Time-resolved fluorescence measurements were performed using time-correlated single-photon counting (TCSPC) technique.Excitation pulses at 532 nm were generated by a supercontinuum laser (NKT Photonics SuperK Extreme) with 10 MHz repetition rate.The fluorescence decays were collected at 585 nm, using a polarizer set at magic angle and a 16 mm band-pass monochromator (Jobin Yvon).The single-photon events were detected with a micro-channel plate photomultiplier R3809U Hamamatsu, coupled to a pulse pre-amplifier HFAC (Becker-Hickl GmbH) and recorded on a time-correlated singlephoton counting board SPC-130 (Becker-Hickl GmbH).The instrumental response function (IRF) recorded with a polished aluminum reflector showed a full-width at half-maximum of 70 ps for 532 nm excitation.For this experiments NPs were directly used without any dilution.Time-resolved exponential decays were fitted by using the global fit procedure of Igor Pro (Wavemetrics).The fitting function was a sum of exponential decays (up to four components) convolved with a normalized Gaussian curve of standard deviation  standing for the temporal IRF and a Heavy side function.All emission decays were fitted using a weighting that corresponds to the standard deviation of the photon number squared root.
Single-Particle Measurements: First, the NPs were diluted 5000 and 1000 times for 23% loaded and 50 wt.%loaded particles with milli-Q water, respectively.Then, 20 μL of each NPs solution was evaporated on a clean glass slide surface (glass slides were initially cleaned using plasma cleaner followed by UV cleaning).Same method was also applied for the diluted solution of QD-605 solution.
Single particle fluorescence measurements were measured on a homemade wide-field set-up.Briefly, a cw 532 nm diode laser was focused onto the back focal plane of an air objective (40X, NA = 0.6) to excite the fluorescent nanoparticles.The laser beam was circularly polarized using a quarter waveplate, the power density was continuously adjusted (0.001-10 W cm −2 ) using the laser head controller and neutral density filters.Emission from the sample was spectrally filtered with the help of a single-edge dichroic mirror (532 nm, lasers BrightLine quad-edge super-resolution laser dichroic beamsplitter: Di03-R532-t1-25 × 36, Semrock) and notch filters (532 nm StopLine single-notch filters: NF03-532E-25, Semrock; in order to remove the scattered laser light), and then was imaged on an EM-CCD camera from Hamamatsu (ImagEM).Images were acquired using MicroManager and analyzed with ImageJ and a custom macro written in IgorPro (Wavemetrics).At least five image sequences were analyzed for each condition with 50-100 particles.
The excitation power was measured using a power meter positioned in the sample plane.The illuminated area was determined using a rhodamine solution, the obtained fluorescence spot was then fitted with a 2D-Gaussian with a FWHM of 120 μm from which it was possible to calculate power density.
The number of photons collected for each NPs using the following equation:

=
(signal − background) × conversion factor analog gain × EM gain × quantum efficiency (4)   where the conversion factor (electrons/counts) is 5.8, here; the analog gain was 1, the value for the EM gain depended on the particles and illumination power (800 in most cases); and the quantum efficiency of the detector is 90% (0.9) in the wavelength region of interest.Then, from the statistical distribution of the number of collected photons for each condition, the median value (med(x)) of the number of collected photons was calculated using the following formula: where x i represents the probability to observe a number of photons n i .This median value is considered as the total number of collected photons.Then the total brightness of NPs (in photons/s) were calculated by dividing the total number of collected photons by the total exposure time (30 s).To get the volume normalized brightness (photons s −1 nm −3 ), the total brightness of each NPs was divided by their average volume (nm 3 ), which is calculated from their average diameter.The fraction of time spent in the bright state was calculated from the intensity-time traces of each NPs using a threshold value corresponding to 50% of the maximum intensity.By counting the number of transitions between the bright and the dimer state it was possible to retrieve the number of fluctuations.Then the average value of all NPs in each condition was divided by the total exposure time to get the frequency of fluctuations (s −1 ) for each NP.
For single-particle FRET measurements, the LabTek chambers were incubated with 1 m KOH solution followed by extensive washing with milli-Q water.Then 200 μL of 10 mm MgCl 2 solution in milli-Q water and 200 μL of diluted NP solutions were added and mixed well in each chamber of LabTek.After 1 h of incubation, the chamber was washed two times with milli-Q water and then the surface of each chamber was fully covered by milli-Q water.
Single-particle measurements were performed in the epi-fluorescence mode using Nikon Ti-E inverted microscope with a 60x objective (Apo TIRF, oil, NA 1.49, Nikon).The excitation was provided by light-emitting diodes (SpectraX, Lumencor) at 550 nm.The 550 nm light power density was 0.016 W cm −2 at the sample level.To have two-color image detection, corresponding to R18/F5-TPB and Cy5 signals, W-VIEW GEMINI image splitting system (Hamamatsu) was used with dichroic 640 nm (Semrock FF640-FDi01-25 × 36).Single-particle analysis was performed using the Fiji software.
For smartphone-based measurements, an Samsung Galaxy Note 20 was used with an app (Halide developed by Ben Sandofsky) which allowed acquiring RAW images and controlling camera parameters (described previously [36,24] ).The images were recorded with a shutter exposure of 1/10s of a second and an ISO of 500.For data acquired with smartphone RGB camera the images were obtained by summing two image frames and then using an ImageJ.Images were recorded in three channels (red, green, and blue) and the ratios were calculated by dividing red channel intensity to the green channel intensity of same NPs and vice versa; while the intensity corresponds to the integral intensity recorded for both channels at the corresponding image.The 550 nm light power density was 0.016 W cm −2 at the sample level.
Calculation of the Detection Area by Smartphone Camera: To calculate the area of detection with the smartphone camera a calibration experiment was performed with variable line grating (R1L3S6P, Thor labs).The surface was imaged by a RGB camera of a smartphone Samsung Galaxy Note 20 (SM-N980F/DS).From there the pixel dimension was calculated and the smartphone images was 0.25 μm px −1 .The size of the images analyzed for smartphone-based FRET NPs experiments was of 200 px × 200 px, meaning that the area of detection was 50 μm × 50 μm.

Figure 3 .
Figure 3. Spectroscopic characterization and size of dye-loaded NPs.a,d) Absorption and emission spectra of PMMA-MA NPs containing 23% (upper panel) and 50% (lower panel) of R18/F5-TPB dye with respect to the total mass, prepared at pH 7.4 without (NP23 for 23% dye-loaded NPs and NP50 for 50% dye-loaded NPs) or with different BHS, where NPs prepared with TBA/F5-TPB, DIMD/F5-TPB and HDMAP/F5-TPB are noted as +TPB,+ DIMD, and +HDMAP, respectively.Here, for all NPs the molar ratio of dye:BHS is 2:1.b,e) Fluorescence quantum yield and size (diameter, measured using TEM) of these NPs.c,f) Brightness per volume (B/V) of these NPs.All errors are s.e.m. (n≥ 3).

Figure 4 .
Figure 4. Single-particle fluorescence evaluation of NPs.a) Single-particle fluorescence microscopic images of NPs containing 50% R18/F5-TPB with respect to the total mass, prepared at pH 7.4, without (NP50) or with different BHS, where NPs prepared with TBA/F5-TPB and DIMD/F5-TPB are noted as +TPB and +DIMD, respectively at illumination power density of 0.5 W cm −2 (532 nm); scale bar: 10 μm.b) Single-particle fluorescence microscopic image of quantum dot-605 (QD-605) at illumination power density of 50 W cm −2 (532 nm); scale bar: 10 μm.c) Brightness per volume (Exp.B/V) of single particles containing 50% R18/F5-TPB with respect to the total mass, prepared at pH 7.4 and with or without different BHS measured at illumination power density of 0.5 W cm −2 (532 nm) and QD-605 (considering diameter 20 nm[77]  ) measured at 100-fold higher power density: 50 W cm −2 (532 nm).d) Fraction (%) of time spent in the ON state by these NPs at the single-particle level (calculated by taking 50% of maximum fluorescence intensity as threshold).Here, for all NPs the molar ratio of dye:BHS is 2:1.All errors are s.e.m. (n≥ 3).

Figure 5 .
Figure 5. Steady-state fluorescence analysis of FRET NPs and the antenna effect.a) Chemical structure of the cationic dye (DiD) with F5-TPB counterion, used as FRET acceptor.b) Schematic representation of NPs with the donors, BHS and the acceptor inside the NPs.c) Emission spectra of FRET NPs containing 50% R18/F5-TPB with respect to the total mass and with DIMD/F5-TPB, where the molar ratio of dye:BHS is 2:1, prepared at pH 6.4 and with different acceptor dye (DiD) to donor dye (R18/F5-TPB) ratios (excitation at 530 nm).d) Excitation spectra of these FRET NPs.e) Antenna effect and the FRET efficiency of these NPs with different acceptor (DiD/F5-TPB) to donor (R18/F5-TPB) ratios.All errors are s.e.m. (n≥ 3).

Figure 7 .
Figure 7. Single-particle evaluation of FRET NPs.a,b) Wide-field fluorescence microscopy images of FRET NPs containing 50% R18/F5-TPB with respect to the total mass and with DIMD/F5-TPB, where the molar ratio of dye:BHS is 2:1, prepared at pH 6.4 and with different acceptor (ATTO647N-C18/F5-TPB) to donor (R18/F5-TPB) ratios (a, NPs without any acceptor; b, NPs with 1:40 000 acceptor:donor ratio).Both channels are represented at same intensity scale.Overlay images represent false color composite of the donor (green) and acceptor (red) channels; scale bar: 5 μm.c) The single-particle intensity traces of the donor (green) and acceptor (red).The excitation wavelength was 550 nm (power density 0.016 W cm −2 ).

Figure 8 .
Figure 8. Single-particle evaluation of FRET NPs with a RGB camera of a smartphone.a) Single-particle fluorescence ratiometric images (red/green ratio) of FRET NPs containing 50% R18/F5-TPB with respect to the total mass and with DIMD/F5-TPB where the molar ratio of dye:BHS is 2:1, prepared at pH 6.4 and with 1:10 000 acceptor (ATTO647N-C18/F5-TPB) to donor (R18/F5-TPB) ratio, taken by smartphone camera at different time.b) Singleparticle intensity traces of the donors (green) and acceptors (red) by a smartphone RGB camera.The excitation wavelength was 532 nm (power density 0.016 W cm −2 ).All the images were represented at same intensity scale.The images represent false color composite of the donor (green) and acceptor (red) channels recorded by the smartphone camera; scale bar: 5 μm.