One-Pot Gram-Scale Synthesis of Hydrogen-Terminated Silicon Nanoparticles.

Silicon nanoparticles (Si NPs) are highly attractive materials for typical quantum dots functions, such as in light-emitting and bioimaging applications, owing to silicon's intrinsic merits of minimal toxicity, low cost, high abundance, and easy and highly stable functionalization. Especially nonoxidized Si NPs with a covalently bound coating serve well in these respects, given the minimization of surface defects upon hydrosilylation of H-terminated Si NPs. However, to date, methods to obtain such H-terminated Si NPs are still not easy. Herein, we report a new synthetic method to produce size-tunable robust, highly crystalline H-terminated Si NPs (4-9 nm) using microwave irradiation within 5 min at temperatures between 25 and 200 °C and their further covalent functionalization. The key step to obtain highly fluorescent (quantum yield of 7-16%) green-red Si NPs in one simple step is the reduction of triethoxysilane and (+)-sodium l-ascorbate, yielding routinely ∼1 g of H-Si NPs via a highly scalable route in 5-15 min. Subsequent functionalization via hydrosilylation yielded Si NPs with an emission quantum yield of 12-14%. This approach can be used to easily produce high-quality H-Si NPs in gram-scale quantities, which brings the application of functionalized Si NPs significantly closer.

Introduction

Fluorescent semiconductor nanoparticles, also coined quantum dots, have received significant attention over the past decades from many fields, such as biology[1−3] as well as electronic and optoelectronic applications.[4−8] They exhibit superior resistance against photobleaching and photochemical degradation, and their size-dependent fluorescence in combination with broad excitation bands makes them extremely useful for multiplex fluorescence analysis and bioimaging techniques.[9] Among those, oxide-free silicon nanoparticles (Si NPs) are considered particularly promising.

Si NPs hold part of that promise for a wide range of applications thanks to their tunable optical and electronic properties as well as their bio/environmental compatibility[10−13] and low intrinsic toxicity.[14,15] With regard to the first, tuning of both the size and the surface coating has allowed to produce Si NPs with a range of fluorescence wavelengths from the UV to the red part of the spectrum. In combination with their low toxicity they are also viewed as an attractive nontoxic alternative for Cd-based quantum dots.[10] Using such Si NPs, prototype Si NP-based sensors,[16] solar cells,[17,18] and light-emitting diodes[19] have appeared.

Despite their obvious relevance, there is up to now no easy and efficient method to prepare Si NPs with well-defined and reproducible characteristics and with the possibility for a flexible functionalization in significant quantities. The first means that the Si NPs need to be crystalline. Crystalline Si NPs display more easily tunable photophysics, as at least the “bulk” of the NP is constant and highly ordered. The second implies that after preparation of the Si NP the monolayer or polymer coating that will be applied to make it e.g. water-soluble, or in contrast hydrophobic, still can be defined e.g. to be obtained via H-terminated Si NPs that allow an easy and stable surface modification via hydrosilylation reactions with terminal alkenes or alkynes. Such an approach would contrast with many current methods to prepare surface-functionalized Si NPs in which the coating is attached in one step, which means that for each coating a novel Si NP synthesis needs to be developed and performed. Finally, to be of any use—apart from for bioimaging, for which small amounts of highly fluorescent nanoparticles may already suffice—the method has to be scalable. Methods that can contribute to achieving this high-end goal are thus extremely desirable.

While a wide variety of robust preparation methods has been developed, including by our laboratories, that give H-terminated Si NPs of high quality, none of these methods are particularly easy to use. Specific examples include the reduction of hydrogen silsesquioxane (HSQ) by H2 at high temperature (1100–1400 °C) and HF etching,[20−23] the reduction of microemulsions of SiX4 (X = Cl, Br),[24−27] laser-driven pyrolysis of silanes,[28−30] and oxidation of Zintl phase Si sources (Na4Si4 + NH4Br) in DMF at 275 °C.[31,32] Despite that some of the methods give highly defined Si NPs that can be further functionalized, these synthetic methods typically require complicated procedures.

To tackle these problems, He’s group advanced a “bottom-up” strategy using APTES as silicon precursors and sodium ascorbate as a reducing agent, showing a fabrication strategy with simplified purification steps. This yielded a scalable method (up to 10 g) for Si NPs that could be used for long-term biological imaging.[33,34] However, the use of APTES as reactant implied that the resultant monolayer coated-Si NPs were always terminated with Si(CH2)3NH2 moieties, i.e., the possible functionalities on the surface of Si NP that are allowed by this method are determined by the available silanes. To further improve the flexibility of the approach, Si NP synthesis should ideally yield Si–H functionalities on the surface, so that follow-up hydrosilylation with terminal alkenes or alkynes can be used for surface functionalization. This method is not only highly robust but also—given the extremely wide range of available alkynes and alkenes—highly flexible.[35−38] Therefore, a solution-phase room temperature or facile microwave irradiation[31,39−41] approach to prepare H–Si NPs using routine equipment would be highly desirable.

With this in mind, we developed a novel solution-phase synthetic approach to produce H-terminated Si NPs (see Scheme ). Using microwave irradiation (building on the efforts of He’s group[11,33,34]), triethoxysilane (CH3CH2O)3Si–H as silicon source, and sodium ascorbate as reducing agent, we were able to quickly produce luminescent H–Si NPs (<5 min; diameter 4–9 nm; typically ∼0.3–1 g in one reaction workup via in situ growth at 25–200 °C in DMSO as solvent). The as-prepared H–Si NPs display an intense green-red fluorescence (QY up to 16%) with emission wavelengths up to ∼900 nm and excited-state lifetimes of 4–10 ns, while alkylated Si NPs could be easily formed from here by a simple one-pot, two-step microwave-assisted reaction (Scheme ).

Scheme 1

Microwave-Assisted Synthesis of H-Terminated Si NPs and Subsequently Alkylated and Alkenylated Si NPs

Experimental Section

Chemicals

All chemicals were used without further purification unless stated otherwise. Triethoxysilane (97%) was purchased from ABCR. Dimethyl sulfoxide (DMSO, anhydrous, ≥99.9%), (+)-sodium l-ascorbate (BioXtra, ≥99.0% (NT)), propargyl alcohol (99%), propargylamine (98%) 1-pentyne (99%), and 1-pentene (99%) were purchased from Sigma-Aldrich.

Synthesis of H–Si NPs in Microwave

Inside a Ar-filled glovebox (MBRAUN’s MB 20 G-LMF gas purifier with H2O and O2 values of <0.1 ppm), the Si NP precursor solution was prepared in a 30 mL microwave vessel with a stir bar by adding 0.20 g (1.0 mmol) of (+)-sodium l-ascorbate to 8.0 mL of anhydrous DMSO, and this solution was saturated with nitrogen gas. Next, 2.0 mL of triethoxysilane (10.8 mmol) was added by syringe, and a Teflon cap was placed over the vessel under a flow of nitrogen. The mixture was stirred for 2 min. Caution! DMSO is a strong microwave-absorbing solvent, which may result in pressure spikes under microwave powers that are too high. An automatic pressure measurement to shut down the microwave power upon reaching pressure thresholds is recommended (in our system: 30 bar). The vessel was taken out of the glovebox and placed in the microwave reactor (Anton Paar Monowave 300). The reaction mixture was allowed to react at a controlled temperature (in the range 25–200 °C; temperature control: ±3 °C, measured by an IR sensor, and a built-in camera (for changes in appearance)) under stirring (600 rpm) at 800 W maximum microwave power, at which ∼5 bar pressure was observed. The microwave automatically adjusted the power output to maintain the set temperature. At the end of the reaction, the vessel was automatically cooled to room temperature by flowing compressed air past the vessel. Next, the vessel was transferred to the glovebox and opened, and the reaction mixture was removed. Subsequently, it was filtered with a 0.1 μm pore size syringe filter to remove any large particulate residues from the solution, The resulting mixture was characterized to consist of partially oxidized H-terminated Si NPs (see text), and it was either characterized without further purification (e.g., for TEM and EDX), etched in HF, or used directly for subsequent hydrosilylation.

HF Etching for Oxide-Free H-Terminated Si NPs

To remove any surface-bound oxidation in the Si NPs, the resulting reaction product could be etched using HF, albeit at a strongly reduced yield of Si NPs. First, and prior to etching, any residual DMSO was removed under vacuum (10 mbar at 40 °C) to obtain dry, glassy dark brown flakes, which were subsequently etched by the well-established method by Veinot and co-workers.[20] In brief, the glassy flakes were grinded under ambient atmosphere using mortar and pestle to produce a very fine orange-brown powder. Next, 60 mg of this powder was transferred to a home-built N2-filled glovebox (0.1% O2) and in there to a Teflon beaker containing a stir bar, ethanol (1 mL), water (1 mL), and 49% aqueous HF (1 mL). (Caution! HF is highly corrosive; please consult a materials safety data sheet prior to handling HF.) This was stirred for 5 min, then 5 mL of toluene was added, and the Si NPs were extracted with the toluene. The Si NP-in-toluene suspension was then added to a capped centrifuge tube saturated with nitrogen, taken out of the glovebox, and centrifuged at 3000 rpm for 10 min. This yielded a clear solution with the NPs forming a powder at the bottom of the tube. These H-terminated oxide-free Si NPs were quickly characterized by IR, XPS, fluorescence, and TEM.

Hydrosilylation

For hydrosilylation reactions, the reaction mixture from the microwave reaction was, inside a glovebox, added to another 5 mL microwave reaction vessel, and 1-pentyne or 1-pentene (0.2 mL) was added via syringe to the solution under an inert atmosphere. A Teflon cap was placed over the reaction vessel, and this was then transferred from the glovebox and placed into the microwave. The reaction mixture was heated to 80 °C (for 1-alkynes) or 120 °C (for 1-alkenes) for 5 min under magnetic stirring (600 rpm). Upon completion of the microwave reaction, the reaction mixture was allowed to cool to room temperature and transferred to a 25 mL round-bottom flask. Unreacted 1-pentyne or 1-pentene was removed under vacuum (10 mbar) in a rotavap. The dark orange solutions containing alkyl/alkenyl-functionalized nanoparticles and DMSO were dispensed in 10 mL of 1:1 toluene and water. The solution was shaken vigorously to form a yellowish turbid solution, the two layers were allowed to separate, and the toluene layer was decanted. This step was repeated twice, after which the toluene fractions were combined and toluene was removed. For water-soluble modified Si NPs (after modification with propargylamine or propargyl alcohol) DMSO was evaporated overnight under 10 mbar vacuum at 80 °C; the resulting brown powder was dispersed in water, and sodium ascorbate remains were removed by dialysis (1 kDa dialysis tube) in water. Both alkyl-terminated and water-soluble functionalized Si NPs were finally filtered through a 0.2 μm polytetrafluoroethylene (PTFE) syringe filter and stored in vials in the glovebox for further use. The resulting modified Si NPs were analyzed by TEM, fluorescence, NMR, and XPS.

Characterization

Optical measurements were performed at room temperature under ambient air conditions. The samples as obtained from the synthesis were for optical measurements diluted in DMSO and stored in standard 1 × 1 cm2 quartz fluorescence cuvettes. UV spectra were recorded on a Varian Cary 50 UV–vis spectrophotometer. For steady-state fluorescence measurements, the concentrations were adjusted to OD λexcitation ≤ 0.1. All fluorescence measurements were performed on an Edinburgh Instruments FLS900 fluorescence spectrometer using the F900 software and instrumental count rate settings: start rate 2500 kHz, stop rate 2000 ± 50 Hz, time range 50 ns, and peak counts = 10000. Photoluminescence spectra were collected at 2.0 nm intervals with a 2 nm slit width. Data were fitted using the F900 program of Edinburgh Instruments. The system was equipped with a wavelength-tunable picosecond pulsed diode laser (PDL800-B PicoQuant), with the following specifications: laser power <5 mW, λem = 400–700 nm. The PL quantum yield was measured based on a relative method using Rhodamine B (in ethanol; ΦF = 68%)[42] as a reference. TEM analysis was performed using a Tecnai G2 F20 Super Twin TEM microscope at 200 kV with a LaB6 emitter. The microscope was fully equipped for analytical work with an energy-dispersive X-ray (EDX) detector with S-UTW window and a high-angle annular dark-field (HAADF) detector for scanning transmission electron microscopy (STEM) imaging. Unless stated otherwise, the STEM imaging and all analytical work were performed with a probe size of 1 nm, resulting in a beam current of about 0.5 nA. TEM images and selected area diffraction (SAD) patterns were collected using a GATAN US1000 2K HR 200 kV CCD camera. The HAADF-STEM EDX and CCD line traces were collected fully automatically using the Tecnai G2 user interface and processed with both Digital Micrograph software Version 2.3 and the Tecnai Imaging and Analysis (TIA) software Version 1.9.162. X-ray diffraction (XRD) was performed using an Inel Equinox 1000 powder diffractometer equipped with a CPS 180 detector (filtered Co Kα1 irradiation, 30 kV, 30 mA, λ = 1.789 Å, zero background spinning sample holder). Powder pattern analyses were processed using Match Crystal Impact software (v.1.11e) for phase identification (using both COD and ICSD databases), and IMADINEL XRD software (v.4.8) for graphical illustrations. All data were collected under the same conditions. FT-IR measurements were recorded with a Bruker Alpha-P FTIR diamond ATR spectrometer. XPS measurements were obtained with an ultrahigh-vacuum JPS-9200 photoelectron spectrometer (JEOL, Japan) operating at base pressures of 5 × 10–7 Torr. The sample was kept in the XPS prechamber until the pressure in there became <10–4 Torr to effect removal of traces of DMSO after Si NP synthesis. A standard dual-anode X-ray source (Al Kα, 1486.6 eV) was used to irradiate the sample surface (12 kV, 20 mA) at an 80° electron takeoff angle relative to the sample surface plane. The binding energies were calibrated on the hydrocarbon (CH2) peak with a binding energy of 284.8 eV.

Results and Discussion

Microwave Synthesis of H–Si NPs

The procedure to synthesize H-terminated and follow-up alkylated/alkenylated Si NPs is shown in Scheme . Using triethoxysilane as Si source and sodium ascorbate as reducing agent, fluorescent H-terminated Si NPs can be easily prepared under microwave irradiation in DMSO (5 min) (see Supporting Information Video S1 to see the appearance of the Si NPs). It has been reported that the cross-linked sol–gel polymer obtained from partial hydrolysis and condensation of triethoxysilane (hydrogen silsesquioxane, HSQ) does not show any fluorescent properties,[43] and Veinot and co-workers reported that only upon pyrolysis thereof at high temperature (>1000 °C) are fluorescent Si NPs obtained.[44−46] Therefore, it is of interest that after mixing triethoxysilane and sodium ascorbate in DMSO under stirring at room temperature (i.e., without microwave irradiation) the nearly colorless mixture spontaneously turned pale yellow within minutes and displayed an intense cyan emission under irradiation with UV light (365 nm), which reaches maximum fluorescence after 2 h (see Supporting Information Figure S1). Under similar experimental conditions, no fluorescent product is obtained if other common reductants and/or solvents were used (see Table , entries 1–5). This is due to the limited solubility of these strong reducing agents even in hot DMSO, which hampers the formation of nanoparticles. In contrast, entries 6 and 7, using sodium ascorbate (in DMF) and trisodium citrate (in DMSO) as respective reductants, yielded intensely blue fluorescent mixtures after 30 min microwave reaction at 160 °C only in sodium ascorbate (in DMF). From TEM images we can deduce the formation of Si NPs (see Figure S11) with sizes around 2–4 nm and well-resolved Si(111) lattice planes of ∼0.32 nm spacing. Using sodium ascorbate in DMF as solvent requires heating the solvent above its boiling point to obtain sufficient solubility and reactivity—the resulting fluorescence maxima are then 360 nm (excitation) and 458 nm (emission). In DMSO the reaction performs much more smoothly; already with the weak reductant trisodium citrate H–Si NPs are obtained, yielding 357 and 420 nm respectively as excitation and emission maximum, albeit that the overall fluorescence intensity is about 100× as low as in DMF with ascorbate. In contrast, a very strong fluorescence (about 100× as high as in DMF with ascorbate) was obtained in DMSO with ascorbate at 160 °C (entry 8), yielding H-terminated Si NPs with a strongly red-shifted fluorescence with excitation and emission maxima at 583 and 660 nm, respectively. Under these low-temperature conditions gram-scale H–Si NPs can be formed (6.32 g of triethoxysilane (38 mmol) yielded 1.04 g of H–Si NPs), which are difficult to obtain according to previously reported low-temperature solution-phase procedures.[47] In the gas phase and/or at high temperatures, such scale increase (0.1–10 g h–1) has been reported by e.g. Swihart et al. using laser pyrolysis of silane (SiH4) in a microwave plasma reactor albeit with lower size definition[48] and by Veinot (see above).[21] Another solution-phase scalable method to produce Si NPs has been reported by Kauzlarich and co-workers; however, macroscopic amounts of thus obtained Si NPs did not have detectable photoluminescence and required subsequent etching with HF solution to develop photoluminescence properties.[49]

Table 1

Formation of H–Si NPs under Various Reaction Conditions

entry	solvent	reducing agent	temp (°C)	time (min)	fluorescence
1	THF/toluene	LiAlH4	60–80	30	not fluorescent
2	DMSO	LiAlH4	130	30	not fluorescent
3	THF/toluene	NaBH4	60–80	30	not fluorescent
4	DMSO	NaBH4	130	30	not fluorescent
5	CH3CN	sodium ascorbate	160	30	not fluorescent
6	DMF	sodium ascorbate	160	30	360, 458
7	DMSO	trisodium citrate	160	30	357, 420
8	DMSO	sodium ascorbate	160	5	583, 660
9	DMSO		160	30	not fluorescent

IR Characterization of H–Si NPs

FTIR spectroscopy was used to qualitatively study the degree of Si–H bond formation on crystalline Si NPs (Figure ). The Si NPs as prepared using microwave irradiation, and still in the DMSO solution, yield Si–H stretching peaks with maxima at 2000–2180 cm–1 and Si–H bending vibrations at ∼900 cm–1 (Figure C). After removal of the DMSO under vacuum the spectrum in Figure D results. These data can be compared with typical IR spectra of crystalline H-terminated Si NPs, which reveal main signals attributed to Si–H (x = 1, 2, 3) at ∼2100 cm–1 and to Si–O bonds at ca. 1065 cm–1.[20,50−52] Silsesquioxane (HSiO1.5) has been shown to display characteristic and sharp Si–H stretching vibrations at ca. 2255 cm–1.[44,53] In addition, Figure B shows characteristic O3Si–H stretching modes for triethoxysilane mixed in DMSO solvent (i.e., without reducing agent) at 2192 cm–1. This comparison is therefore a strong indication of the formation of H–Si NPs in one mild step in the current, microwave-stimulated reaction.

Figure 1

Full-range FT-IR spectra (and zoomed-in expansions thereof from 1800–2400 and 600–1200 cm–1) of (A) DMSO, (B) triethoxysilane in DMSO, and (C) H–Si NPs as prepared in DMSO at 160 °C. (D) H–Si NPs after removal of DMSO. (E) H–Si NPs prepared via HF etching.

The Si–O region of the IR spectrum shows that, apart from intense Si–H stretching bands and weak Si–H bending vibrations, also Si–O bands can be observed, revealing partial oxide formation. To minimize this oxidation and confirm the assignment of the 2100 and 900 cm–1 bands as belonging to Si–H bonds, dried Si NPs were treated with aqueous HF[20] and extracted in toluene. As shown in Figure E, this indeed further reduced the formation of silicon oxide yielding H–Si NPs, although this etching step in itself has only a low yield (≤30%).

XPS, XRD, and TEM Analyses

Freshly prepared H–Si NPs were analyzed by XPS. The narrow-scan Si 2p spectrum (Figure A) shows a Si peak at 100.8 eV, which suggests a silicon core with surface oxidation. This XPS signature was obtained also for several series of Si NP samples that were prepared under a rigorously oxygen-free atmosphere, indicating that DMSO itself might be an oxygen source.[54,55] This surface oxidation can be removed by a small HF etching dip, after which nearly oxide-free Si NPs result (see also Figure A). These data compare well with that of oxide-embedded Si NPs obtained via Veinot’s silsesquioxane method, followed by HF etching, which yields XPS Si 2p peaks at 99.9 eV (for Si0) and a shoulder in the 100–101 eV range.[20,56−58] Because we aimed for a rapid, easy-to-use, high-yield synthesis, we consistently continued with H-terminated Si NPs that we interpret as displaying only surface oxidation, noting that—if needed—a simple HF treatment would nearly fully remove this. To substantiate the idea that thus-synthesized Si NPs display only surface oxidation, we treated them with an oxygen plasma for 5 min—this should yield nearly fully oxidized SiO2 NPs, and Si 2p XPS analysis indeed yields a spectrum with a single Si 2p peak at 103.1 eV.

Figure 2

(A) XPS Si 2p narrow scan of freshly prepared H–Si NPs obtained in DMSO at 160 °C (black; 100.8 eV) and of the same H–Si NPs after 5 min oxygen plasma (red; 102.7 eV) and HF etched H–Si NPs (blue; main peak at 99.8 eV and shoulder peak at 100.3 eV). All spectra were internally calibrated to the energy of the C 1s emission (284.8 eV) following the precedent for analysis of H–Si NPs. Powder X-ray diffraction pattern of Si NPs prepared at 25 °C (B) or 160 °C (C).

Furthermore, we studied the degree of crystallinity of the thus-prepared Si NPs. Powder XRD data of Si display three peaks at 2θ values (for Co Kα X-rays) of about 28°, 47°, and 56°, which correspond to diffraction from the Si(111), (220), and (311) lattice planes, respectively.[56,59] As shown in Figure C, Si NPs prepared at 160 °C display such reflections at 28.4°, 47.0°, and 56.3°. This indicates that the microwave reaction process can yield crystalline Si NPs. We can also estimate the NP size using Scherrer’s formula on the Si(111) peak.[60] This gives an average NP size of ca. 9.5 nm (peak center at 28.5°; fwhm (2θ) = 1.05°)) (see eq 1S in the Supporting Information). The TEM data (see below) show that this value is a slight overestimation. In addition, we observed the broad diffraction peak at ∼22.7°, which is thus representative for partially oxidized Si around a crystalline Si core. For H–Si NPs prepared at 25 °C, the XRD data are featureless (Figure B), indicating the amorphous nature of Si NPs prepared at around room temperature.[59]

TEM analysis (Figure a for NPs prepared at 160 °C in DMSO) of the as-prepared H–Si NPs displays the reaction product as rather monodisperse spherical particles, with a diameter that is highly dependent on the temperature used in the synthesis. This size increases with reaction temperature: for the NPs prepared at 25, 50, 120, 160, and 200 °C, sizes of 4 ± 1, 5 ± 1, 6 ± 1, 7 ± 1, and 9 ± 2 nm are obtained, respectively (see TEM images for several of these samples in Figure S9). In other words, the size of the H–Si NPs can be tuned in detail by the temperature in the microwave reaction. HRTEM data show a high crystallinity and well-resolved Si(111) lattice planes of ∼0.32 nm spacing (Figure b), confirming the nature and crystallinity of the as-prepared H–Si NPs.[61] The EDX data show an O/Si ratio of 0.15 for H–Si NPs obtained at 160 °C, but even for the smaller H–Si NPs obtained at 25 °C an O/Si ratio of only 0.22 was obtained. While these values are already rather low, they still reflect oxygen contributions from both trace amounts of DMSO remnants and atmospheric contamination. As a result, the TEM and EDX (see Figure S2) data confirm that the as-prepared H–Si NPs are highly crystalline and hardly oxidized. In contrast, XPS data show an O/Si ratio of 0.98 for H–Si NPs obtained at 160 °C (see Figure S12A), which we attribute to oxidation of these Si NPs which occurred upon the slow (overnight) removal of the DMSO at 10 mbar that was applied here and exposure to the ambient atmosphere—his surface-bound oxide can be removed by HF etching (see Figure ), after which the oxygen content of the Si NPs is <5% (Figure S12B).

Figure 3

(a) TEM image of H–Si NPs prepared at 160 °C. (b) Higher magnification image of H–Si NPs prepared at 160 °C revealing crystallinity with a well-resolved Si(111) lattice spacing of 0.32 nm. TEM size distribution histogram of Si NPs prepared (c) at 25 °C and (d) at 160 °C.

Optical Properties of the As-Prepared Si NPs

Sodium ascorbate is a strong enough reducing agent to allow this one-step synthesis of H–Si NPs to be performed at room temperature (all 25 °C data refer to thermal reactions without microwave irradiation) or under mild microwave conditions (50–170 °C). Figure presents the optical properties of the H–Si NPs as synthesized in DMSO solution at different temperatures ranging from 25 to 170 °C for 5 min. (The full plots of the excitation wavelength ependence of the emission are shown in Figure S3 for H–Si NPs synthesized at various temperatures under current study.) As shown in Figure A, the UV–vis spectra show a very high absorption in the UV, with convergence to 0 at ∼500 nm for the Si NPs prepared at 25–140 °C. In contrast, the absorption continues to about 700 nm for the Si NPs prepared at 150–170 °C, i.e., at temperatures clearly above the boiling point of triethoxysilane at atmospheric pressure (135 °C). Analogously, while the excitation maximum shifts slightly for Si NPs prepared at 25–140 °C, upon using higher temperatures, the maximum shifts rather drastically by almost 150 to ∼590 nm for the samples prepared at 160–170 °C. This also leads to exceedingly long fluorescence wavelengths, in which the emission continues to ∼900 nm (Figure C,D). As far as we know, this is the highest fluorescence wavelength reported to date for Si NPs prepared using microwave methods; e.g., He’s group reported propylamine-terminated Si NPs from microwave heating to 160 °C that display a fluorescence maximum at 460 nm,[33,34] while Wu et al. reported maxima of 445 nm for super bright N-[3-(trimethoxysilyl)propyl]ethylenediamine (DAMO)-terminated Si NPs prepared under microwave conditions.[62] The current data suggest that heating to a refluxing temperature is needed to obtain these Si NPs with strongly extended absorption and fluorescence, in accordance with their larger size (Figure C,D). Such strong microwave-induced temperature effects on the size and concomitantly absorption and fluorescence properties have also been reported by Kauzlarich et al. for the microwave-induced synthesis of Ge NPs[39] and by McLaurin and co-worker for InP NPs.[63] After etching in HF these H–Si NPs show (see Figure S8) an optical absorption and emission at lower wavelengths (e.g., excitation maximum at 372 nm and emission at 472 nm) as well as a shorter decay time (1.6 ns) compared to freshly prepared (but likely partially oxidized) H–Si NPs. It is well-known that during the etching process the nanoparticle optical properties are blue-shifted due to the reduction in diameter.[59−61]

Figure 4

Photophysical properties as a function of H–Si NPs preparation temperature: (A) UV–vis absorbance spectra, (B) fluorescence excitation spectra, and (C) fluorescence emission spectra. (D) Maximum excitation and emission wavelength plotted against Si NPs preparation temperature.

In addition, the fluorescence lifetimes were measured using single-photon counting equipment. In most of the literature it has been observed that the fluorescence lifetimes of Si NPs typically fall in the nanosecond-scale range,[62,64−67] although Veinot’s hydrogen silsesquioxane method yields Si NPs with lifetimes up to microseconds.[21]Figure shows the time-resolved decay of the fluorescence of freshly prepared H–Si NPs dispersed in DMSO, which typically falls in the 4–10 ns range. (All decay curves can be fitted properly by three exponents, but biexponential decay typically yields slightly higher residuals at shorter time scale (<1 ns) after the fits.) These fluorescence lifetimes depend strongly on the temperature at which the H–Si NPs were prepared. For NPs prepared at 25 °C the fluorescence decay yields a characteristic lifetime of ca. 6 ns (see Table ; experimental uncertainty typically ±1 ns). For the NPs prepared at 50–140 °C, a longer, overall lifetime of 9 ± 1 ns was observed. In contrast, for the H–Si NPs prepared at 150, 160, and 170 °C the lifetimes are appreciably smaller with values of 4.5, 4.1, and 3.9 ns, respectively. Typical spectral decays for these three populations (25, 50–140, and 150–170 °C) are presented inFigure ; the full data are given in Figure S4. Similar size effects, but then for organic monolayer-stabilized Si NPs, have been reported by Korgel and co-workers.[64]

Figure 5

Photoluminescence decay curves (λex and λem based on excitation and emission maxima as shown in Figure D).

Table 2

Overall Fluorescence Lifetimes and Quantum Yield of H–Si NPs Dependent on the Synthesis Temperature

temp (°C)	decay timea (ns)	quantum yieldb (%)
25	5.7	9
50	9.8	8
80	8.4	16
100	8.0	12
120	8.8	11
140	8.7	7
150	4.5	7
160	4.1	8
170	3.9	8

Three-exponential decay fit.

Standard deviation ±2%.

Finally, the fluorescence quantum yield was measured based on a relative method, using rhodamine B (in ethanol) as a reference with Φ = 68%.[42] While the data showed some sample-to-sample variation, typical quantum yields varied from 7 to 16%, with slightly lower values for the high-temperature NPs than for the mid-temperature ones. These values are comparable to those obtained by He’s group for propylamine-terminated Si NPs obtained via microwave synthesis (20–25%).[33]

Alkylated Si Nanoparticles

It is well-known that the Si–H bonds on H–Si NPs are highly photoreactive under UV irradiation and can undergo hydrosilylation reactions with terminal alkenes and alkynes to yield air-stable alkylated/alkenylated Si NPs[25,68−70] that allow detailed characterization.[60] Along these lines we thus reacted freshly prepared H–Si NPs with 1-pentyne and 1-pentene. For the hydrosilylation procedure (Scheme ), 1-pentene or 1-pentyne was added under an inert atmosphere to as-prepared H–Si NPs (synthesized at 160 °C), when still in DMSO. The reaction vessel was subsequently sealed and placed into the microwave reactor for a second short reaction. The yield of this reaction was typically almost quantitative, as e.g. 0.031 g of as-prepared Si NPs yielded 0.035 g of purified pentenylated Si NPs. A typical transmission electron microscopy (TEM) image and a high-resolution scanning TEM (HRTEM) image of pentene-modified H–Si NPs are shown in Figure .[61] The average particle diameter was found to be 6.5 ± 0.9 nm, as indicated in the histogram (Figure d), in line with the data shown in Figure d. The bright-field HRTEM image of the pentene-modified H–Si NPs resolved lattice fringes consistent with the Si(111) spacing of 0.325 nm and the Si(220) spacing of 0.192 nm for diamond-structured silicon (in Figure ); see Figure S15 for XRD data. In addition, TEM and electron diffraction data for propargyl amine-coated Si NPs are shown in Figure S10. In the latter, sharp Bragg reflections from the first two planes are indicated ((111) and (202)) from the diamond cubic structure.

Figure 6

TEM-based analysis (a–d) of 1-pentene-modified H–Si NPs: (a) Overview for size distribution measurements. (b) High-resolution image to obtain information about crystallinity. (c) Lattice spacing measurements from HR-TEM image, revealing a well-resolved Si(111) lattice spacing of 0.32 nm. (d) Size distribution of pentylated Si NPs. (e) Picture showing tunable hydrophobicity of the Si NPs upon introduction of surface molecules: vials 1 and 2 contain Si NPs coated with propargylamine (HC≡CCH2NH2) and propargyl alcohol (HC≡CCH2OH), respectively; vials 3 and 4 contain pentyne-coated and pentene-coated Si NPs, respectively. Each vial contains CHCl3 (lower layer) and H2O (upper layer) with a NP concentration of 2 mg/mL.

Proton nuclear magnetic resonance (1H NMR) spectroscopy provides an alternative method for interrogating surface-bonded moieties.[25,35,71−73]Figure S13 shows the 1H NMR spectrum of pentenylated Si NPs (in CDCl3) that were functionalized via thermally activated hydrosilylation with 1-pentyne. Thus, modified Si NPs show broad resonances arising from terminal methyl protons at about 0.8 ppm and a broad resonance at 1.2–1.6 ppm. Such broad structureless aliphatic CH signals are partially caused by the reduced rotation of the chains due to attachment to the NP that would increase the relaxation times and partially indicative of multiple, slightly different surface-bonded environments. In addition, there are no signals from the ethoxy groups that would be visible in the case of incomplete removal of that moiety. Despite the peak broadening, the integration ratio of the methyl (at 0.8 ppm) and other alkyl proton signals was determined to be 3:4.4, respectively. This ratio agrees well with that of 3:4.2 reported by Veinot and co-workers for pentenyl monolayers on Si NPs.[35] Like these authors, we also hardly observe any features belonging to the vinylic protons—only upon zooming in peaks at 4.2 and 5.5 ppm are observed, each with an intensity <1% of the alkyl peaks. This strong decrease in intensity can attributed to the strongly reduced rotation of the vinylic group as a result of the direct coupling to the Si surface and should not be taken to indicate any other complexity in binding modes, as sometimes observed on flat Si surfaces.[35,74]

Analogously, these H–Si NPs (prepared at 160 °C) were coated with propargylamine (HC≡CCH2NH2) or propargyl alcohol (HC≡CCH2OH) and purified; the optical characteristics in absorption and emission were measured in water. Interestingly, the fluorescence of such alkylated Si NPs was highly stable. For example, the fluorescence of 1-pentyne-coated Si NPs did only vary over <3% upon continuous irradiation and fluorescence measurements over 13 h (see the Supporting Information S5). Table presents the excitation and emission maxima of the four alkylated Si NPs under current study. These results strongly suggest that modification of these (6.5 nm) Si NPs with these functional monolayers does not shift the excitation or emission wavelengths appreciably. This results contrast with previous observations on smaller (<2 nm) Si NPs,[75] in which significant shifts were observed e.g. upon the presence of or the lengthening of the distance of an amine functional group attached at the end of an alkyl chain.

Table 3

Maximum Excitation and Emission Wavelengths, Overall Fluorescence Lifetimes, and Quantum Yields of H–Si NPs Prepared at 160 °C after Passivation with Functional 1-Alkene or 1-Alkyne Molecules

coating of Si NP	excitation wavelength (nm)	emission wavelength (nm)	decay timea (ns)	quantum yield (%)
Si∼CH=CH–CH2OH	562	650	3.7	14.3
Si∼CH=CH–CH2NH2	577	654	4.2	12.7
Si–C5H12	578	669	3.8	12.5
Si∼CH=CH–(CH2)2CH3	541	651	3.7	12.1

Data fitted by a three-exponential decay.

Similar observations were made for the emission lifetimes, which do not vary by less than 10% with variation of the functional group (see Table ). After modification these nanoparticles show higher quantum yield (12–14%) compared with H–Si NPs (8%). This enhanced quantum yield is likely due to our procedure with intermediate cleaning step, which removed DMSO and ascorbic acid. It is noteworthy that these quantum yields did not change appreciably upon storage of the Si NPs for over two months in water or dichloromethane, respectively. DMSO is a highly hydrophilic solvent and a mild oxidizing agent toward flat or hydrogen-terminated porous silicon.[54,55] Song et al. have shown that the rate of oxidation by DMSO is reduced in the presence of the radical scavenger,[54] which thus illuminates why our procedure works in the first place at all, as our reduction reagent sodium ascorbate is of course also a strong antioxidant scavenging radical.[76]

In summary, we developed a facile one-pot microwave-assisted synthesis of hydrogen-terminated silicon nanoparticles, which is easy and efficient and yields gram-scale quantities of product using standard synthesis setup and glassware at relatively low temperatures. Highly photoluminescent H–Si NPs can be obtained in as little as 5 min in routine microwave synthesis equipment; for 1 g of H-terminated Si NPs 15 min was required. Microwave-assisted heating produced H–Si NPs with a size from 4 to 9 nm, which can simply be selected via the reaction temperature in the microwave. The resulting Si NPs are brightly fluorescent, with emissions as long as 900 nm and quantum yields in DMSO in the 7–16% range. Follow-up hydrosilylation can be performed using alkene and alkyne reagents with various organic functionalities yielding air-stable Si NPs with tunable hydrophobicity/hydrophilicity and high fluorescence quantum yields (12–14%). This route is therefore a significant extension upon current solution-phase procedures to prepare functional Si NPs and displays significant potential for further investigations.