pH-Induced transformation of ligated Au-25 to brighter Au-23 nanoclusters

Thiolate-protected gold nanoclusters have recently attracted considerable attention due to their size-dependent luminescence characterized by a long lifetime and large Stokes shift. However, the optimization of nanocluster properties such as the luminescence quantum yield is still a challenge. We report here transformation of Au 25 Capt 18 (Capt labels captopril) nanoclusters occurring at low pH and yielding a product with much increased luminescence quantum yield which we have identified as Au 23 Capt 17 . We applied a simple method of treatment with HCl to accomplish this transformation and we characterized the absorption and emission of the newly-created nanoclusters as well as their morphology. Based on DFT calculations we show which Au nanocluster size transformations can lead to the highly luminescent species such as Au 23 Capt 17 .


Introduction
2][3][4][5][6][7] Au NCs, containing a few up to several hundreds of gold atoms in the core surrounded by various ligands exhibit size-dependent luminescence, in the visible and/or NIR range of wavelengths . 8,9 he presence of emission in the NIR (in the biological transparency window region) is essential for applications of gold nanoclusters in biology and medicine.As most of biomolecules have low absorption coefficients in this range, a risk of irritation of normal metabolism is limited and the autofluorescence of the biological material is minimized as well as deep penetration (>1 cm) of the tissue is possible. 10Although strategies based on silver doping 11 or rigidifying the ligand shell have been proposed to enhance the photoluminescence quantum yield of the clusters both in the linear 12 and nonlinear optical regime 13 , it is usually rather low, and has been found to be strongly cluster size dependent. 8The ligands may play an important role in fluorescence enhancement, 14 and both the ligands and the nanocluster size may determine the activity and biodistribution in context of applications in vivo.Therefore, a precise choice of the stabilizing ligands is required, as well as the control of the quantum cluster size with the "atomic precision".
Major synthetic breakthroughs have been achieved using two systematic approaches: size-focusing methodology 15 and ligand exchange-induced size/structure transformation (LEIST) methodology. 16However, a transformation consisting in removing one or two metal atoms and/or ligands from a stable nanocluster size remains challenging.Core etching 17 or using reductive conditions 18 have permitted size conversion of AuNCs from Au25 to Au22 and Au23.Recently, an elegant strategy using a two-step metal exchange method has allowed for the "resection" of two surface gold atoms leading to the transformation from Au23 to Au21 NCs 19 .Another sizecontrolled synthesis of aqueous nanoclusters can be obtained through pH control.Xie and co-workers achieved a size-tunable synthesis of Au10-12, Au15, Au18, and Au25 nanoclusters, under pH adjustment utilizing carbon monoxide (CO) as a mild reducing agent. 20 this contribution, we report making of highly fluorescent NCs by size transformation of well-characterized Au25 stabilized by

Results
The captopril protected gold (AuCapt) nanoclusters were synthesized according to a literature procedure 21 and dispersed in water (pH=7).They exhibited absorption bands at 800, 675, 550, 505, 445and 400 nm (Fig. 1a), similar to the Au25(PET)18 (where PET -phenylethanethiol) clusters reported by Jin and co-workers. 22After transferring the nanoclusters to buffers with pH ranging from 2 to 10, no change has been initially observed in the absorption spectra.However, after several hours the absorption spectra for NCs in pH=2 lost the features characteristic of Au25 NCs i.e. absorption bands at 800, 670 nm and 450 nm gradually decreased.Additionally, a small shoulder at 550 nm developed into a band (Fig. 1a).For pH=3 the leading features of the absorption spectrum were also partly lost, but for other pH values (between 4 to 10) no changes occurred (see SI Fig. S1).
The photoluminescence (PL) spectra measured upon treatment with low pH revealed enhancement in the nanoclusters emission with a maximum at 700 nm (when excited at 550 nm, corresponding to Stokes shift of 0.12 eV).In order to compare the PL intensities, the data were normalized with respect to the same absorbance at the excitation wavelength.The PL intensity of NCs at pH = 2 was found to be an order of magnitude higher than that reported for nanoclusters at pH = 7 (Fig. 1b).The PL quantum yield (QY) of the nanoclusters was calculated by comparing their emission with fluorescence of a standard which was Oxazine 170. 23uCapt NCs at pH= 7 have QY = 0.37% whereas at pH = 2 the QY increased ~10 times, up to 3.9%.This dramatic increase of the brightness of the nanoclusters needs therefore to be understood in terms of the possible structural changes.One of the possibilities considered was the formation of Au(I)-Capt which is an intermediate in the synthesis of Au25Capt18.Since Au(I) complexes are known to fluoresce 24 , it might be assumed that the observed emission enhancement is connected with the presence of a significant amount of this Au(I)-Capt polymer.To investigate the potential impact of the polymer on emission of the NC solution at pH=2, we recorded separately its emission spectrum.The fluorescence was observed in the range between 400 and 660 nm with the maximum at 465 nm, which does not provide explanation of the emission enhancement of the nanoclusters (see SI Fig. S2).
To clarify if the observed changes arise due the presence of protonation or to chloride ions in the buffer solution, the neutral and concentrated solution of AuCapt was dissolved in HCl and HNO3, both corresponding to pH = 2.The samples emission and absorption were monitored over 24 hours, as well as after 72 hours.The comparison of the absorption and emission spectra is presented in Figs.S3-S4, which show that a restructuration of the spectra has occurred.The process was identical in the case of samples with the pH = 2 buffer and with HCl, and was completed in 5 hours.In contrast, in the sample with HNO3, different features are observed (SI Fig. S3).Please do not adjust margins Please do not adjust margins The luminescence decays were measured and fitted with double exponential function (Fig. 2).The PL lifetimes were found to depend only slightly on the pH, in case of the sample in pH = 7, they were τ1 = 264 (0.09) and τ2 = 1778 ns (0.91) with the average 1753 ns.For the sample in pH = 2, τ1 = 207 (0.18) and τ2 = 1516 ns (0.82), giving average of 1462 ns.The corresponding lifetime values are presented in Table 1.# average lifetime <τ> = (a1τ1 2 +a2τ2 2 )/(a1τ1+a2τ2) In order to test the reversibility of changes induced by low pH, we dried and redissolved the samples in a buffer of pH = 7.In the case of AuCapt at pH = 2, the absorption and emission spectra were preserved (SI Fig. S5).However, an increase of the PL lifetime was observed, and the value of the long component changed from ~1500 ns to ~1750 ns, which is a value similar to the lifetime of AuNC at pH = 7 (see Fig. 2).For AuCapt at pH = 7 the lifetimes did not change after drying and redissolving the sample.
Since the absorption of NCs at pH=2 did not unambiguously correspond to any reported absorption spectrum in the literature 25 , we checked whether we obtained a mixture of the products.For this reason the polyacrylamide gel electrophoresis (PAGE) was conducted and two separated bands were observed (see Fig. 3a).In order to identify the responsible species, the ESI-MS was carried out for the desalted product mixture as well as for AuCapt before the low pH treatment (Fig. 3b and c, and Fig. S6).The corresponding ESI deconvolution spectrum leads to Au25Capt18 before and after the low pH treatment leads to several peaks with smaller masses, which correspond to a mixture of Au24Capt16and Au23Capt16 and Au23Capt17 and Au22Capt17.
The separated NCs, extracted from a gel, were characterized by UV/Vis spectroscopy (Fig. S7).The absorption spectrum of upper fraction presents two bands at 460 and 575 nm.Similarly, the band of lower fraction was located at 575 nm with a shoulder at 460 nm.The emission spectra of separated fractions were blue-shifted by 6 nm with respect to the one for NCs in pH=2.However, more important findings are that the lower fraction has much higher luminescence quantum yield equal to ~5 % , when compared to the upper fraction which is ~0.1 %.

Discussion
In our experiments, the originally synthesized nanoclusters exhibited the absorption spectra with all the bands characteristic for Au25(SR)18.Based on the absorption spectra and ESI-MS results, we can confirm the formation of Au25(Capt)18 nanoclusters, with no need for post-synthesis separation.
Au25 PL decay times reported in the literature depend on the emission wavelength and stabilizing ligands.Link et al. reported biexponential decays for AuSG NCs, with lifetime values ranging from 177 ns/1380 ns for emission at 700 nm to 341 ns/1480 ns for emission at 800 nm. 26Wu et al. obtained single PL lifetimes of about 500-700ns for Au25 with -SC2H4Ph, -SC12H25, and -SC6H13 ligands as well as two PL lifetimes of 247 ns/1240 ns for Au25SG18 NCs. 27These data are in a good agreement with our results of PL decays.AuCapt functionalized clusters presented lifetimes of 210 ns/1790 ns (table 1), similar to those for other ligands with electron-rich groups.
It has been shown in some reports that gold nanoclusters exhibit good stability under pH although results on the influence of pH on PL intensity evolution differ considerably 28,29,30 Our work shows that after an incubation with HCl (pH = 2), the nanoclusters transform into species with increased luminescence.Based on the MS analysis and on the agreement This journal is © The Royal Society of Chemistry 20xx Please do not adjust margins Please do not adjust margins between experimental and theoretical absorption spectra for Au23L17 (see Fig. 4), we recognized the transformation as being from Au25(SR)18 to Au23Capt17 (Au23).The proposed composition of Au23SR17 was reported in the literature as possibly emitting bright red PL. 31 In this contribution, we show that application of low pH in the presence of Cl -ions is a new method of transformation from one stable size of nanoclusters to another, with the nanoclusters being produced with the same ligand, in water.On the contrary, previous protocols required thermal treatment and high excess of the thiol that substitute the outgoing ligand. 16Moreover, the incoming ligand was required to be significantly different from the original one.In early work simple ions like Cl -or OH -were used to stabilize the nanoclusters, but they exhibited low stability in drying process.Other studies showed the charge conversion of [Au25(SCH2CH2Ph)18] 0 clusters into anionic form in the presence of halide ions. 32In our studies comparison of HCl and HNO3 treatment shows that the presence of Cl -ions is important for the size transformation.On the other hand, application of nitric acid leads to the protonation of clusters.
Relative intensities of 800 nm/630 nm and 450 nm/400 nm absorption peaks might be connected with the change of the charge of ligated Au25 nanoclusters.High intensity of 800 nm and 450 nm peaks, as observed in our samples before treatment, is characteristic for anionic clusters. 33After HNO3 treatment the changes in absorption spectra shown in Fig. S3a are characteristic for the neutral cluster [Au25SR18] 0 . 33A similar absorption spectrum is observed at pH=3 with HCl.
Thus, one can suggest that the acidic treatment leads to at least two-step transformation process: The first stage could be explained on the basis of the pKa value for carboxylic group of captopril, which is equal to 3.35 ± 0.04 (for the structure of ligand see Fig. S9) 34 .The protonation of NCs occurs in pH ~ 3 and lower.
The degradation of NCs in the presence of halides was reported before by Zu et al. 32 In that case two types of salts were used -NaX and TOAX (where X -halide), but only NaX caused decomposition.Also in our work chloride ions played crucial role in the second stage of the transformation process.Although Au25(Capt)18 were stabilized with TOA + during whole process, the dramatically low pH could shift the equilibrium in our system.In pH = 2, TOA + had limited stabilizing activity on protonated NCs and as a result partial decomposition of nanoparticles with Cl -anions was possible.A Cl -ion, due to its higher electronegativity, acts as an acceptor of electrons and, when in contact with Au surface, electronic charge transfer from gold ions to chloride ions occurs.Additionally, the protonation of carboxylic groups decreases the electron density in the outer layer of NCs and thus chloride -gold interactions become more prominent.On the other hand, NO3 -has never One notes that the pH-induced transformation described here, has some similarity with the recently reported results from Jin et al.where [Au25(SePh)18] -nanoclusters were transformed into the [Au23(SePh)16] -nanocluster in the presence of NaBH4. 18In fact our computational results show that the transformation from Au23L16 to Au23L17 is more favorable than from Au25L18 to Au23L17.In the former case, addition of ligand to Au23L16 which is also stable produces the most stable Au23L17 with structure containing core of 12 Au atoms as shown in Fig. 4a), for which the calculated absorption spectrum is in a good agreement with experimental findings.In contrast, the structure of Au23L17 derived from Au25L18 by removing two gold atoms and one ligand is higher in energy and the TDDFT calculated absorption spectrum is not in agreement with experimental findings as shown in Fig. 4b).Although the core contains 12 atoms, its structure is different with respect to the one obtained starting from Au23L16.Since the excitation within the core is responsible for absorption spectra, its structural properties are responsible Please do not adjust margins Please do not adjust margins for the leading features.The above results show that the removal of the gold atoms and ligand versus addition of a ligand cause a reconstruction of the core which is responsible for the optical properties including fluorescence.

Conclusions
We presented here a new approach to transform Au nanoclusters of one size into another.In our simple method a low pH treatment (with HCl) was successfully applied for the nanoclusters transformation from Au25Capt18 to very fluorescent Au23Capt17.We monitored the changes of nanocluster absorption and emission spectra in the course of the transformation and measured the fluorescence decay times of the substrates and products.We observed enhancement of the fluorescence by up to 13 times when the nanocluster size changed.In fact, changes of optical properties by changing number of Au atoms and ligands occur due to structural changes, as shown by our theoretical findings.The final nanoclusters were stable after the transition from low to neutral pH.Because Au25 are the most easily obtained nanoclusters, our findings about their pH sensitivity and size transformation with corresponding fluorescence enhancement provide significant information useful for their further applications.Moreover, our approach gives a relatively simple procedure for the preparation of highly fluorescent Au23 nanoclusters from Au25.It may also be possible to obtain other sizes by using other monodispersed nanoclusters as a substrate.
Instrumentation and protocols.High-resolution transmission electron microscopy (TEM) was conducted with a FEI Tecnai G2 20 X-TWIN.TEM images of AuCapt nanoclusters in pH = 2, 3 and 7 are presented in Fig. S8 in SI.The absorption and fluorescence measurements were recorded on a JASCO V-670 spectrophotometer and a Hitachi F-4500 spectrofluorometer, respectively.Excited state lifetimes were measured with an Edinburgh Instruments FLS980 Fluorescence Spectrometer.All samples were excited at 379 nm with a PicoQuant pulsed laser and emission was detected at 720 nm using time-correlated single-photon counting (TCSPC).The experimental data was fitted with two exponential decays.
Evolution of the gold NC size without and with acid treatment was determined by electrospray ionization on a commercial quadrupole time-of-flight (micro-qTOF, Bruker-Daltonics, Bremen, Germany, mass resolution 10 000).AuCapt NCs without acid treatment and the same sample after the treatment (HCl, pH = 2) were prepared to a final concentration approximately of 50μM in water.The samples were analysed in negative ion mode: each data point was the summation of spectra over 5 min.External calibration was carried out with a set of synthetic peptides.We used our Au25Capt18 NCs to optimize the mass spectrometer and define the measurement uncertainties.Due to the huge amount of sodium adducts observed in ESI-MS, we decided to perform desalting procedure before MS analysis.The AuNC was dissolved in 1mL of methanol (or water).Then 1mL of glacial acetic acid was added and precipitation was induced by adding THF.Then the sample was centrifuged and dried.
A multiplicative correlation algorithm (MCA) was used to estimate the mass of nanoparticles from the mass-to-charge spectra produced by electrospray ionization mass spectrometry. 35,36 paration of AuCapt Clusters.Captopril-protected Au25 clusters were prepared using a modified Kumar one-pot procedure. 21riefly, TOABr (0.23 mmol, 126.8 mg) was dissolved in 10 mL of methanolic solution of HAuCl4•3H2O (0.02 M) and stirred vigorously.After 20 min, 1 mmol of captopril (217.2 mg) was dissolved in 5 mL of methanol and rapidly added into the reaction mixture under stirring.After 30 min, 5 mL of aqueous solution of NaBH4 (0.4 M), cooled to 4°C, was rapidly poured to the reaction mixture under vigorous stirring.The solution was allowed to react for 8 h in an ice bath.The suspension was centrifuged to remove unreacted, insoluble Au(I):SR polymers.The supernatant was collected and concentrated in water bath at 45°C.The clusters were precipitated with ethanol and repeatedly washed with methanol and ethanol.The final precipitate was dried in 45°C.

Polyacrylamide Gel Electrophoresis (PAGE)
The electrophoresis in a discontinuous gel system was carried out by using a PROTEAN II XL Cell slab gel electrophoresis unit (BIO-RAD).The separating and stacking gels were prepared from acrylamide monomers with the final concentrations of 18.3 %T; 4.2 %C (separating) and 7.5 %T; 2.6 %C (stacking), where %T -total monomer (acrylamide + bisacrylamide) concentration and %Cconcentration of the cross-linking.The clusters were dissolved in a 5% (v/v) glycerol/water solution.The electrophoresis was carried out for 13 h at a constant voltage mode (150 V) at 4°C.Finally, the separating gel was cut into bands and the particular fractions were eluted with water.

Figure 1 .
Figure 1.Absorption (a) and emission (b) spectra of AuCapt measured after several hours of incubation in buffers of various pH.The samples were excited with 550 nm.captopril (Capt) ligand occurring on treatment with low pH.The transformation of Au25 NCs into Au23 NCs is confirmed by electrospray ionization mass spectrometry and qualitative theoretical explanation based on DFT calculations.

Figure 3 .
Figure 3. a) Polyacrylamide gel electrophoresis of the product allowed to obtain two bands (left side), where the upper one is very similar to the basic sample Au25 (right side).ESI-MS analysis showed a detailed composition of the product.Well defined Au25Capt18 was used as control.Deconvoluted deconvoluted ESI-MS spectrum of AuCapt NCs before (b) and after pH=2 treatment (c).The corresponding ESI-MS spectra are given in Fig. S6 in SI.

Figure 4 .Scheme 1 .
Figure 4. Comparison between experimental (red line) and TDDFT calculated apsorption spectra (blue lines) for a) lowest energy structure of Au23Capt17 obtained by adding one ligand to Au23Capt16 and b) for higher energy structure (by 0.5eV) of Au23Capt17 obtained by removing two Au atoms and one ligand from Au25Capt18.Captopril ligands are replaced by -SCH3 groups in the calculations, for DFT calculations pbe0 functional and def2svp AO basis have been used.

Table 1 .
PL lifetimes and QY of AuCapt in pH = 7 and 2 and after treatment (drying and redissolving in pH=7 buffer).