Bromine-terminated SiNPs were synthesized following an adapted procedure by Kauzlarich and co-workers by means of oxidation of Mg2Si with bromine.17a Capping these bromine-terminated SiNPs with 3-butenylmagnesium bromide yielded alkene-terminated SiNPs (SiNP-ene, 1). The resulting SiNPs were purified by size-exclusion chromatography (SEC) with ethyl acetate as eluens to yield per reaction batch SiNPs in 30 mg quantities as an orange waxy material.
The size of the SiNP core was determined by TEM. Figure 1 shows a typical TEM image of 1; the observed particle size is (2.4±0.5) nm, with a radius-based polydispersity (PDI) of 1.12. (see the Supporting Information) The alkene-terminated SiNPs have a slightly smaller radius than observed earlier for butyl-terminated SiNPs.48 This is likely due to the use of SEC instead of silica chromatography as part of the purification procedure. SEC is based on hydrodynamic radius, and may result in a different size distribution than that obtained by a predominantly polarity-based separation.
SiNP-ene (1) was characterized by NMR spectroscopy. Figure 2 (top) shows an 1H NMR spectrum in which the SiCH2 protons result in a signal at δ=0.87 ppm (Figure 2 a), whereas the double bond protons are observed at δ=4.96 (Figure 2 c) and δ=5.79 ppm (Figure 2 d). The observed signal broadening is most likely due to the many different chemical environments of the protons, caused by attachment to different surface sites of the butylene chains. Moreover, due to substantial bromination of the reaction solvent, the attachment of linear and branched octyl chains as well as octyl bromides is also observed (δ=3.5–4.5 ppm). Diffusion-ordered NMR spectroscopy (DOSY; see the Supporting Information) revealed that the alkene moieties as well as these brominated species are attached to the SiNPs, since they have the same relatively low diffusion coefficient. The low diffusion coefficient (as compared to solvent) indicates that the signal stems from a relatively large moiety, that is, an SiNP. The amount of (brominated) octyl chains was determined by 1H NMR spectroscopy to be butene/octyl=2.78 (see the Supporting Information). The 13C NMR spectrum (Figure 2, bottom) shows signals that correspond to alkene moieties at δ=112 ppm (Figure 2 b) and δ=136 ppm (Figure 2 c), whereas the alkyl chain appears as multiple signals in the δ=21–31 ppm region (Figure 2 a). The assignment of these signals was confirmed by 2D COSY and 2D heteronuclear single quantum coherence (HSQC) spectra (see the Supporting Information). In the COSY spectrum, protons b and c were confirmed to be positioned on neighboring carbon atoms, furthermore, the 1H–13C HSQC confirms the assignment of carbon atoms a, b, and c in the 13C NMR spectrum, due to coupling with the corresponding signals in the 1H NMR spectrum. A broad signal at δ=130 ppm (Figure 2 d) in the 13C NMR spectrum indicates the presence of internal alkenes, most likely resulting from elimination reactions that take place on the aforementioned brominated octyl chains due to the Grignard reagent acting as a base.
SiNP-ene (1) was further characterized by X-ray photoelectron spectroscopy (XPS) to obtain information about the elemental composition. The wide-scan XPS spectrum revealed the presence of silicon (21.1 %), carbon (61.7 %), bromine (0.4 %), and oxygen (16.9 %). The oxygen in the sample is most likely largely due to environmental entrapment in the deposited SiNPs within the XPS and not as silicon oxide, since the silicon narrow-scan spectrum displays only a single signal at 101.8 eV, with a full width at half-maximum (FWHM) of 1.3 eV (Figure 4, right), which is near-identical to what is observed under similar instrumental conditions for the base peak of Si at 99.4 eV with a FWHM of 0.5 eV in a silicon(111) wafer. The shift in binding energy from 99.4 to 101.8 eV is thus at least partially caused by charging effects, which would indeed be expected for a small object that is well surrounded by an electrically insulating organic shell. Moreover, XPS analysis of samples of intentionally oxidized SiNPs resulted in two signals in the silicon region, the first at 101.8 eV corresponding to silicon, and a second signal at 104.9 eV for silicon oxide (see the Supporting Information). The peak shift is in line with the observation that the binding energy shifts to higher levels with decreasing nanoparticle size relative to bulk silicon.49 The narrow scan of the C1s region (Figure 4, left) reveals two types of carbon: carbon-bound carbon at 285.0 eV, and a minor fraction (14 %) of carbon atoms bound to electronegative elements such as Br or O (broad shoulder around 287 eV). The precise identity of this peak can, however, not be deduced from XPS analysis alone.
The UV/Vis absorption spectrum of 1 (Figure 5, left) was recorded. However, the spectrum does not display a maximum, it only shows a gradually higher absorption for decreasing wavelengths, as was observed before for this type of SiNPs.17a, 48 From these samples, the absorption coefficient was determined to be 0.14 (mg mL−1)−1 cm−1 at 300 nm and 0.035 (mg mL−1)−1 cm−1 at 350 nm. The extinction coefficient was determined in (mg mL−1)−1 cm−1, since the exact molecular mass of the SiNP is unknown. Based on a molecular weight of 10 000 g mol−1—estimated from TEM measurements in combination with a dense packing of butylene chains—approximate molar extinction coefficients of 3.0×102 M−1 cm−1 at 350 nm and 1.4×103 M−1 cm−1 at 300 nm were calculated. These values are lower than observed for the 1.5 nm SiNPs prepared by other wet-chemical methods (9.4×103 M−1 cm−1; 1.1 (mg mL−1)−1 cm−1)16a or for those reported for 1.5 nm QDs of II–VI semiconductors, such as CdSe and CdS (7×104 M−1 cm−1).50 However, bulk silicon has an indirect bandgap, which makes optical transitions without the assistance of a phonon inefficient. The optical transition in an SiNP is very efficient in comparison to the transitions in bulk silicon. In the nanometer size regime, the distinction between indirect and direct bandgaps is blurred.51
The fluorescence emission spectrum is strongly dependent on the excitation wavelength. Figure 5 (right) depicts fluorescence emission spectra of 1 at several excitation wavelengths that range from 350 to 450 nm. The spectra are corrected for the UV/Vis absorption at each excitation wavelength. The relatively highest fluorescence emission is observed at 525 nm at an excitation wavelength of 430 nm. This is distinctly redshifted as compared to butyl-terminated SiNPs described previously,17a which showed an emission maximum at 390 nm (obtained for irradiation at 340 nm). For SiNPs with a gradual increase in UV/Vis absorption at shorter wavelengths, correction for UV/Vis absorption results in a redshifted fluorescence emission maximum.
The fluorescence quantum yield (QY) was determined using a comparative method23, 52 at λexc=350, 366, and 496 nm. These dyes have a fluorescence emission that broadly overlaps the fluorescence emission of the SiNPs. The QYs are (1.5±0.2) % and (1.8±0.3) % for λexc=350 nm and 366 nm, respectively. The QY increases further with increasing excitation wavelength; the highest QY is observed at an excitation wavelength of 496 nm, and is (7.1±1.2) %. This is about 2 % higher than observed for highly similar butyl-terminated SiNPs,48 but still considerably smaller than the highest QY reported for SiNPs (60 %).53 It has been suggested that a high oxidation grade of the SiNPs results in lower quantum yields,54 however, such high degrees of oxidation are absent in this case as follows from IR and XPS data. The larger size (2.5 nm in this case) is therefore a more likely cause.
Time-resolved fluorescence spectroscopy was performed, and the amplitude-weighed fluorescence lifetime (see the Supporting Information) was determined to be 3.40 ns, which is similar to the lifetimes observed for butyl-terminated SiNPs.48 These results indicate that the optical properties of the SiNPs are not significantly influenced by the change in coating from butyl to butylene.
Thiol-Ene Functionalization of Alkene-Terminated SiNPs
With the alkene moieties present on the surface of the SiNPs, functionalization by means of thiol-ene chemistry is possible. Functional groups may be attached, which make the SiNPs suitable for applications that require specific labels. The SiNPs were mixed with an excess amount of the respective thiol and 0.2 equiv (with respect to the thiol) of 2,2-dimethoxy-2-phenylacetophenone (DMPA) as photoinitiator or 4,4′-azobis(4-cyanovaleric acid) (ACVA) as thermal initiator. The mixture was exposed to UV light (DMPA, 365 nm) or heated to 80 °C (ACVA) while stirring under ambient atmosphere. After 1 h, 1H NMR spectroscopy revealed complete conversion of the alkene moieties, and the functionalized SiNPs were purified by means of SEC. A number of thiols with different functionalities was selected: thiolacetic acid (SiNP-TAA, 2) and 2-mercaptoethanol (SiNP-OH, 3)—which provide for a functional group that may be used in further conjugation reactions—as well as thiolated triethylene glycol monomethyl ether (SiNP-EO3, 4), to render the SiNPs water-soluble and biocompatible. Furthermore, carboxylic acid terminated thiols with three different spacer lengths (no spacer (6), an EO4 spacer (7), and a PEG3000 spacer (8)) were coupled to the SiNPs to obtain functional groups available for bioconjugation on the surface of the SiNPs (Scheme 1).
The functionalized SiNPs 2, 3, 4, 6, 7, were purified by means of SEC, whereas pegylated SiNPs (8) were purified with a Amicon Ultra 3K nominal weight cutoff (NWCO) centrifugal units. To obtain SiNP-TGA (6), direct coupling of thioglycolic acid is possible and results in full conversion by means of the thiol-ene reaction. However, it appeared that the SiNP-TGA (6) obtained in this manner could not be purified. Therefore, the ethyl ester-protected SiNP-TGAEE (5) was hydrolyzed with potassium tert-butoxide (KOtBu) to obtain thioglycolic acid-modified SiNPs (SiNP-TGA, 6). The sample was neutralized and purified by evaporation of the solvents.
In the 1H NMR spectrum of purified SiNP-TAA (2) (Figure 6), no terminal alkenes are observed. Furthermore, a new signal at δ=2.86 ppm (Figure 6 d) corresponds to the newly formed thioester; signals that correspond to the CH2 next to the thioester appear at δ=1.57 ppm (Figure 6 c), whereas the signal of the methyl group next to the carbonyl appears at δ=2.32 ppm (Figure 6 e).
The 1H NMR spectrum of SiNP-TGAEE (5) (Figure 7) displays characteristic signals of the newly formed thioether at δ=2.62 ppm (Figure 7 d) (CH2-CH2-S) and δ=3.19 ppm (Figure 7 e) (S-CH2-COOEt). The signals of the ethyl ester moiety were observed at δ=4.17 ppm (Figure 7 f) and δ=1.26 ppm (Figure 7 g), which correspond to the CH2 and CH3, respectively. After hydrolysis with KOtBu, these signals are no longer present. The NMR spectra of the thiol-ene-modified SiNPs were assigned with 2D COSY NMR spectroscopy, and in all cases confirm quantitative coupling of the indicated thiols to the SiNPs, since no terminal alkenes are observed after modification, whereas the intensities of the new signals fit accordingly. 1H NMR spectra of other modified SiNPs can be found in the Supporting Information.
Infrared spectroscopy was performed on the SiNPs to confirm the presence of the characteristic groups and to determine whether the functionalized SiNPs retained their low oxidation grade (Figure 8). The CCH2 antisymmetric stretch signal, which appeared at 3077 cm−1 in the case of SiNP-ene, was absent in the spectra of the functionalized SiNPs. All functionalized SiNPs display the typical symmetric and antisymmetric stretch vibrations for CH2 bonds. Signals for the attached functional groups are readily observed. SiNP-TAA (2), for example, displays a distinct carbonyl stretch signal at 1690 cm−1, which confirms the presence of a thioester. SiNP-OH (3) displays a broad OH stretch vibration at 3354 cm−1, whereas SiNP-EO3 (4) displays characteristic ether bonds at 1108 cm−1. The spectra of SiNP-TGAEE (5), SiNP-TGA (6), and SiNP-EO4-COOH (7) display distinct carbonyl stretch signals at 1732 cm−1. The signal is, however, hardly observed in the case of SiNP-PEG3000-COOH (8), most likely due to the large amount of ether groups (approximately 170 per chain, at 1103 cm−1) relative to other characteristic groups present in the sample. The presence of ether groups also results in broadening of the signals for the CH2 symmetric and antisymmetric stretch (they appear as a single peak at 2880 cm−1). In the IR spectrum of SiNP-EO4-COOH (7) the characteristic ether bonds are observed at 1108 cm−1. Importantly, the signal that corresponds to SiO (1000–1100 cm−1) continues to be weak in all spectra, despite the high polarity of the SiO bond. This indicates that even upon performing the radical-based thiol-ene coupling the low degree of oxidation of the Si core is maintained, thus providing further proof of the passivating nature of the organic coating.
Figure 8. FTIR spectra of SiNPs 2, 4, and 7. FTIR spectra of SiNP 3, 5, 6, and 8 can be found in the Supporting Information.
Download figure to PowerPoint
The SiNPs were further characterized by using XPS to obtain an elemental composition of the SiNPs, as well as characterizing the different functional groups. A thin film of SiNP-TAA (2) was cast onto a gold surface and analyzed. The XPS narrow scan spectrum of the C1s region of 2 was fit with four components, of which the main peak at 285.0 eV is assigned to carbon-bound carbon atoms (Figure 9, left). The intensity of this signal is relatively high (78 % of the total carbon content), due to the earlier mentioned side reactions that occur, thereby leading to substantial attachment of alkyl chains (see the Supporting Information). The signals for the CO (288.0 eV) and CS (286.4 eV) carbon atoms integrate to a 1:1 ratio. The signal at 289.3 eV (2 % of the C1s total area) may arise from slight oxidation of the CC bond. The XPS narrow scan of the Si2p region (Figure 9, center) reveals only a single signal at 101.8 eV, similar to that observed for SiNP-ene (1). The sulfur region shows a clear signal of a single type of sulfur, which was fit with a single spin–orbital doublet (Figure 9, right), in agreement with the thioether that results from the attachment step. XPS spectra of other modified SiNPs can be found in the Supporting Information.
To determine whether the SiNPs retain their optical properties after thiol-ene functionalization, the UV/Vis absorption spectra were recorded (Figure 10, left). As evidenced by the spectra, with normalized absorption at 320 nm, the absorption is not significantly affected by the modification of the SiNPs. This confirms that the Si core remains unaltered under thiol-ene coupling conditions. In addition, fluorescence emission spectra were recorded. In Figure 10 (right), the UV/Vis absorption-normalized fluorescence emission spectra of SiNPs excited at 430 nm are shown, at which excitation wavelengths is obtained. The fluorescence emission maxima do not shift significantly for the modified SiNPs and remain at 525 nm. The signal intensity, however, does change to some degree. A slight shift in maximum is observed when comparing SiNP-PEG3000-COOH (8) and the other functionalized SiNPs. This is likely as a result of the size-based purification process of 8, which may alter the size distribution of the SiNPs. Since the fluorescence emission of SiNPs is size-dependent, changes in the size distribution consequently result in changes in the fluorescence emission maxima.15b This shift is not expected to result from the thiol-ene conjugation, since fluorescence originates from the Si core, with only a small electronic coupling to the covering organic layer, which is unlikely to extend to the distance at which the conjugation reaction takes place.55
Figure 10. UV/Vis absorption spectra (left) and fluorescence emission (right) of SiNPs 1, 2, and 4. Spectra of SiNPs 3, 5, 6, 7, and 8 can be found in the Supporting Information.
Download figure to PowerPoint
Besides steady-state fluorescence spectroscopy, time-resolved fluorescence emission was studied as well. The Supporting Information summarizes the amplitude-weighted averages of the fluorescence lifetimes of the functional SiNPs excited at 372 nm. The results are all in the same time range (≈4 ns), thereby further confirming that thiol-ene modification does not significantly affect the SiNPs.
To obtain biofunctional SiNPs with possible applications in biosensors, single-strand deoxyribonucleic acid (ssDNA) with a length of 100 bases with a 3′-amino modification was coupled to the SiNPs by using 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide and N-hydroxysuccinimide (EDC/NHS) chemistry (Scheme 2). Samples were purified with 50 k NWCO Amicon centrifugal filter units. Control experiments revealed that both uncoupled ssDNA and uncoupled SiNPs readily pass through 50 k NWCO filters, as evidenced by UV/Vis absorption spectroscopy.
After coupling of the ssDNA and subsequent purification to obtain ssDNA-functionalized SiNPs, the amount of DNA in the NP sample was determined by UV/Vis absorption measurements to determine its concentration. The complementary 3′-Atto 488 dye-modified ssDNA strand was added (1.5 equiv) and incubated at 80 °C for 1 min, after which the sample was allowed to cool to room temperature followed by cooling on ice. Excess amounts of complementary ssDNA were removed by filtration over a 50 k NWCO Amicon centrifugal filter unit. This resulted in double-strand (ds) DNA-modified SiNPs SiNP-TGA-dsDNA (9), SiNP-EO4-dsDNA (10), and SiNP-PEG-dsDNA (11) (Figure 11).
Figure 11. Schematic representation of SiNPs conjugated to a dsDNA strand; complementary strand contains a 3′-Atto 488 dye. (Dimensions scaled, spacers in blue.)
Download figure to PowerPoint
To confirm coupling of the DNA strands to the SiNPs, gel electrophoresis on an agarose gel was performed. Figure 12 shows the agarose gel after a run time of 1 h. Lane 1 contains the reference ladder; the numbers correspond to the number of base pairs (bp) in the respective band. Lane 2 contains SiNP-TGA-dsDNA (9); lane 3 SiNP-EO4-dsDNA (10); and lane 4 contains SiNP-PEG-dsDNA (11). Lane 5 is a reference lane that contains uncoupled dsDNA (100 bp) with the same sequence as the strands coupled to the NPs. The band in lane 5 runs slightly lower than the 100 bp ladder reference (Figure 12, arrow I). No DNA of this length is observed in lanes 3 and 4, thus indicating successful purification of the conjugated SiNPs. The heights of the bands in lanes 3 and 4 correspond to approximately 300 bp (Figure 12, arrow II). The SiNPs to which the DNA are coupled are of a specific size, which may influence the height of the band slightly. As such, the bands are proposed to indicate the presence of two to three DNA strands per NP for SiNP-EO4-dsDNA (10) and SiNP-PEG-dsDNA (11). A slightly less distinct band at the same height is observed in lane 2, which indicates that in the SiNP-TGA-dsDNA (9) sample, up to three strands are also attached to the SiNP. The clearer band in lane 2 just above the 100 bp level (Figure 12, arrow III) most likely corresponds to SiNP-TGA-dsDNA (9) with only a single dsDNA strand attached. Tailing of the bands may be due to the range of NP sizes present in the samples (due to a different number of polymeric chains attached), or different numbers or orientations of DNA strands on the SiNPs, thereby resulting in differently sized DNA-functional particles.
Figure 12. Agarose gel with dsDNA-modified SiNPs. Lane 1: 1 kb+ ladder; lane 2: SiNP (9); lane 3: SiNP (10); lane 4: SiNP (11); lane 5: uncoupled 100 bp dsDNA.
Download figure to PowerPoint
UV/Vis absorption and fluorescence emission data of the DNA-modified samples with different spacer lengths unfortunately do not display characteristics that allow for differentiation. The UV/Vis absorption spectrum of the pegylated SiNPs conjugated to ssDNA is shown in Figure 13 (left). However, the main absorption observed originates from the DNA. This points to attachment of the DNA strands to the SiNP, since otherwise these would have been removed in the purification process. The absorption of the SiNP core is still observed when ssDNA is conjugated to the particle, albeit only as a minor contribution, since the extinction coefficient of ssDNA (9.8×105 M−1 cm−1, for this particular sequence) is approximately 1000-fold higher than that of the SiNPs at 350 nm (≈3×102 M−1 cm−1). Furthermore, the fluorescence emission of the SiNPs with ssDNA is still observed, although again as a relatively small signal (Figure 13, right). Fluorescence emission contributions from the SiNP core as well as the UV absorption of the ssDNA moiety, indicate that both the SiNPs and ssDNA are present in the sample.
Figure 13. Left: normalized UV/Vis absorption of SiNP-PEG-ssDNA and dsDNA (11). Right: normalized fluorescence emission of SiNP-PEG-ssDNA (λexc=430 nm) and SiNP-PEG-dsDNA (11) (λexc=501 nm).
Download figure to PowerPoint
The UV absorption spectrum of the SiNPs that contain dsDNA (9, 10, 11) is shown in Figure 13 (left). The spectrum is dominated by the absorption of dsDNA and additionally the Atto 488 dye. This is similar to the ssDNA SiNP samples due to the high extinction coefficient for dsDNA (1.6×106 M−1 cm−1, for this particular sequence). The fluorescence emission of the Atto 488 dye fully dominates the fluorescence spectra of the samples that contain dsDNA-conjugated SiNPs (Figure 13, right).
The fluorescence contribution of the Atto dye indicates that the complementary strand is indeed hybridized to the DNA strand present on the SiNP. The dominance of the Atto 488 dye in the fluorescence spectra with dsDNA is due to the rather low quantum yield (QY) of the SiNPs (7 %), whereas the QY of the Atto 488 dye is 80 %. Figure 13 (right) shows the UV/Vis absorption and fluorescence emission of SiNP-dsDNA. The spectra are dominated by the absorption and emission of the coupled dsDNA strands and the Atto 488 dye. This indicates that the fluorescent dye is indeed coupled to the SiNPs through the complementary DNA strand, since all uncoupled DNA strands are removed by purification as observed by gel electrophoresis.