Silicon nanocrystals are intriguing materials for biomedical imaging applications because of their unique optical properties and biological compatibility. We report a new surface functionalization route to synthesize biological buffer soluble and colloidally stable silicon nanocrystals, which is enabled by surface boron doping. Harnessing the distinctive Lewis acidic boron surface sites, postsynthetic modifications of plasma synthesized boron doped nanocrystals were carried out with polyethylene glycol (PEG-OH) ligands in dimethyl sulfoxide under photochemical conditions. The influence of PEG concentration, PEG molecular weight, and boron doping percentage on the nanocrystal solubility in a biological buffer has been investigated. The boron doping facilitates the surface functionalization via two probable pathways, by providing excellent initial dispersiblity in polar solvents and providing available acidic boron surface sites for bonding. These boron doped silicon nanocrystals have nearly identical absorption features as intrinsic silicon nanocrystals, indicating that they are promising candidates for biological imaging applications.
Semiconductor nanocrystals have become a forefront functional material for biomedical imaging applications due to their unique optical properties.1–6 To date, considerable emphasis has been given to cadmium containing nanocrystals, particularly cadmium chalcogenides, as biological chromophores due to their high quantum yield, narrow emission spectrum, and high photostability.1,7–9 However, concerns have been raised about the toxicity associated with cadmium containing compounds in biological systems, which might limit their use in such applications.7,10
Si nanocrystals, being relatively nontoxic, abundant, and biocompatible, are considered as a worthy alternative to cadmium compounds for bioimaging applications.11–20 Fabrication of colloidally stable water soluble Si nanocrystals is challenging as they are hydrophobic and mainly dispersible in nonpolar organic solvents through hydrosilylation with nonpolar alkyl ligands.21–25 Surface functionalization of Si nanocrystals with hydrophilic ligands is essential to use them in a biological medium. A variety of approaches have been used to synthesize hydrophilic Si nanocrystals including postsynthetic surface modifications with hydrophilic ligands containing carboxylic acids,26–28 amines,29–31 hydroxyl groups,29,32 hydrophilic polymers,33–35 encapsulation of hydrophobic Si nanocrystals in different carrier molecules,16,18,19 and through direct nonthermal plasma methods.36 The viability of silicon nanocrystals for imaging applications depends on solubility and colloidal stability in aqueous biological media. However, pertaining optical properties of Si nanocrystals in an aqueous medium seems to be challenging due to their high susceptibility to the water-mediated surface oxidation.16
Addressing the issue of detrimental surface oxidation, herein, we report a simple one-pot approach for synthesizing water soluble and biocompatible Si nanocrystals with significantly reduced surface oxidation enabled by surface boron doping. It has been reported that boron doped (B-doped) Si nanocrystals exhibit a significantly reduced rate of surface oxidation and are stable up to 57 days without appreciable Si-O content.37 According to the Cabrera-Mott theory of oxidation of silicon,38 electron tunneling can occur from the Fermi level of the Si core through the surface oxide forming negative ions of surface adsorbed oxygens. Boron doping is anticipated to lower the Fermi level, which is consistent with the reduced surface oxidation due to slow electron tunneling.
Using polyethylene glycol (PEG-OH) as the surface ligand and harnessing the advantage of Lewis acidic boron surface sites, we explored the synthesis of the buffer soluble and colloidally stable B-doped Si nanocrystals and studied the influence of unique boron surface sites in surface passivation.
Silicon nanocrystals are synthesized in a continuous-flow, low-pressure plasma reactor from an argon-silane (SiH4) gas mixture using a modified literature procedure.37 Diborane (B2H6) was introduced as a dopant to the gas mixture. The reactor schematic is illustrated in Fig. 1 and consists of a 1-in. outer diameter borosilicate glass tube through which the reactant gases are flown. Typical flow rates are 50 standard cubic centimeters per minute (sccm) of argon, 0.54 sccm of silane, and 0.3 sccm of diborane diluted in hydrogen (10:90). Radio frequency (rf) power is applied at 110 W (nominal) and 13.56 MHz to ring electrodes to form B-doped silicon nanocrystals in the plasma. The nanocrystals are collected directly onto glass substrates and transferred air-free to a nitrogen-purged glovebox using a push-rod assembly.
B-doped Si nanocrystal functionalization
A 5-min deposition (∼2 mg) of B-SiNCs was first dispersed in 2 ml of an appropriate solvent [dimethyl sulfoxide (DMSO), ethanol, chloroform, and acetonitrile were used for testing purposes] and sonicated for 5 min before introducing the appropriate amount of the polyethylene glycol (PEG-OH) ligand. The mixture was transferred to a quartz bulb and sonicated for another 2 min before initiating magnetic stirring while being exposed to ultraviolet (UV) irradiation for 2 h. Following UV irradiation, solvents were evaporated using a continuous nitrogen flow and a Schlenk line vacuum pump for over 18+ h. Dried and isolated nanocrystals were introduced to deionized (DI) water or Tyrode buffer and shaken by hand until thoroughly mixed.
The B-doped Si nanocrystals were characterized with dynamic light scattering (DLS), Fourier-transform infrared (FTIR) spectroscopy, and UV-visible (UV-Vis) absorption. DLS measurements were performed with a Microtrac NanoFlex particle size analyzer at room temperature. FTIR spectroscopy was performed on a Bruker Alpha spectrometer with the attenuated total reflection (ATR) single reflective module in a nitrogen filled glovebox. Samples were dropcast onto the ATR crystal and 20 scans were taken for each measurement at 2 cm−1 resolution. UV-Vis absorption spectra were taken with a Cary 5000 UV–vis spectrophotometer.
RESULTS AND DISCUSSION
Freestanding, 7 nm, 10% boron-doped Si nanocrystals were synthesized using a nonthermal plasma reactor, decomposing SiH4 in the presence of diborane (B2H6). Previous work has showed that these nanocrystals consist of a diamond cubic Si crystalline core and three-fold coordinated acidic B surface sites.39 The presence of Lewis acidic B surface sites in doped nanocrystals offers a unique opportunity for surface functionalization exploiting the acid-base interactions. In addition, these doped nanoparticles can be readily disperse in a range of polar solvents, forming spontaneous inorganic nanoparticle solutions.39 Utilizing the advantage of distinctive surface sites, we studied the synthesis of biocompatible and buffer-soluble B-doped Si nanocrystals using PEG-OH as the surface ligand. PEG-OH was chosen as the surface ligand due to its excellent compatibility with biological environments and nontoxic nature.
As initial dispersibility is critical for surface functionalization, several solvents were tested with 10% B-doped silicon nanocrystals. As illustrated in Fig. 2(a), doped particles were dispersed in dimethyl sulfoxide (DMSO), ethanol, chloroform, and acetonitrile and solvodynamic diameters were measured using DLS. Solvodynamic dimensions suggest that B-doped particles show better dispersibility in DMSO compared to other solvents. Thus, DMSO is used as the base solvent for our synthetic procedure. The basic reaction studied in this paper consists of injecting the PEG-OH ligand into a solution of B-doped nanocrystals in DMSO and UV-irradiating the mixture for 2 h while stirring. Surface-modified particles were isolated by evaporating the solvent. Tyrode buffer was introduced to the particles and the hydrodynamic diameter was evaluated. UV irradiated PEG-modified nanocrystals exhibit a hydrodynamic diameter of 23.3 ± 6.0 nm in Tyrode buffer, indicating that a solvent corona is formed by the solvent-surface ligand interactions. TEM images of surface functionalized nanocrystals reveal that B-doped Si nanocrystals are individually passivated and do not show encapsulation of multiple nanocrystals together (Fig. 3). To evaluate the impact of UV irradiation, a similar reaction was carried out for 2 h without UV-irradiation. A measured hydrodynamic diameter of 204 ± 75 nm in Tyrode buffer implies that UV-irradiation is essential for an efficient surface functionalization reaction [Fig. 2(b)]. They are colloidally stable for 8 h in the buffer without any apparent particle agglomeration or particle precipitation with a hydrodynamic diameter of 26.3 ± 5.3 nm [Fig. 2(c)].
In order to determine the base solvent effect, we substituted DMSO with ethanol and, as shown in Fig. 2(d), our results suggest that the presence of DMSO is critical for surface functionalization. PEG-modified particles show a hydrodynamic diameter of 200 ± 21 nm indicating significant particle agglomeration. Carrying out the reaction in two other solvents, acetonitrile and chloroform, also yields extremely poor solubility in Tyrode buffer, resulting in instant particle precipitation.
PEG concentration and molecular weight play a crucial role in surface properties and particle behavior in solvents. PEG-modified B-doped Si nanocrystals tend to agglomerate due to van der Waals forces when they offset the steric repulsion and electrostatic forces provided by surface ligands.40 Thus, it is important to assess the optimal PEG ligand concentration and PEG molecular weight and length to achieve the optimal solubility and stability. Typically, nanocrystals show excellent solubility with higher ligand concentrations and longer ligand chains due to enhanced steric repulsion. The effect of ligand concentration and molecular weight was monitored by evaluating hydrodynamic diameters of PEG-modified nanocrystals with varying DMSO:PEG volume ratios and PEG molar masses. Solubility of the B-doped Si nanocrystals improves with increased PEG concentration, as shown in Fig. 4(a). As expected, solubility of doped particles increases with the longer PEG chains and increasing molecular weights. There is no significant difference in hydrodynamic diameters in Tyrode buffer with 400, 600, and 1000 g/mol PEG ligands, but the smaller ligand, 200 g/mol, demonstrated substantial particle agglomeration with a hydrodynamic diameter of 417 ± 219 nm [Fig. 4(b)]. However, it is interesting to observe that PEG200 modified Si nanocrystals exhibit better dispersibility in DI water with a hydrodynamic diameter of 21.5 ± 5.8 nm. Other PEG-modified Si nanocrystals exhibit similar dispersibility in DI water as in Tyrode buffer [Fig. 4(d)].
To elucidate the effect of surface boron percentage on the solubility, we performed similar reactions with both undoped and B-doped Si nanocrystals. As shown in Fig. 4(c), 10% doped Si nanocrystals exhibit better dispersibility in Tyrode buffer than 5% doped nanocrystals. It should be noted that the fractional dopant amount was deduced by the gas flow rates rather than the measured boron percentage. Furthermore, undoped bare Si or 0% B-doped Si nanocrystals demonstrate poor dispersibility with a higher degree of particle agglomeration. It confirms that under the reaction conditions we studied, boron doping is essential for the optimal surface functionalization that leads to the enhanced solubility in Tyrode buffer. However, it is unclear how the presence of B surface sites contribute to the improved solubility. One possibility is the effective surface passivation by PEG ligands via Lewis acid-base interactions between PEG-OH and surface B sites or having an excellent initial dispersibility through forming a complex between DMSO molecules and surface boron sites. However, undoped bare Si has a relatively poor dispersibility in DMSO initially, which might influence the reaction.
To have a better understanding of the effect of distinctive surface properties in doped nanocrystals, surface characterization was performed using FTIR. Figure 5(a) shows the FTIR spectra of pristine and PEG-modified B-doped Si nanocrystals. Pristine B-doped Si nanocrystals exhibit the characteristic Si-H peaks at ∼2200–2000 cm−1 and B-H and B-O peaks at 2500, 1800, and 1400 cm−1,41 indicating the presence of the dopant on the nanocrystal surface. The FTIR spectrum of PEG-modified Si nanoparticles confirms the surface functionalization with PEG ligands, but highly concentrated PEG molecules dominate the nanocrystal surface suppressing the surface peaks of Si and B. However, it is not unusual to see PEG ligands suppressing the characteristic Si-H peaks in Si nanocrystals after surface functionalization as shown in a previous report.35
There are a number of feasible surface interactions between PEG-OH ligands, B-doped Si nanoparticle surfaces, and DMSO molecules. As previous work has shown, DMSO can form a donor-acceptor complex with Lewis acidic B sites forming a B–O bond that facilitates the enhanced initial dispersiblity of particles in DMSO. Surface Si, with very high affinity to O atoms, could very well form a strong Si–O bond attaching surface ligands to the nanoparticle surface. On the contrary, Lewis basic PEG-OH could react with available Lewis acidic boron surface sites. Furthermore, PEG-OH ligands could form hydrogen bonds with DMSO solvent molecules. There is a notable shift in the O-H peak from 3430 to 3367 cm−1 corresponding to the hydrogen bonding between DMSO and PEG-OH. Variable amounts of DMSO in PEG reveal varying degrees of the red shift, which confirm that the OH peak shift could be mainly due to DMSO-PEG hydrogen bonding [Fig. 5(b)]. Since surface functionalization without UV irradiation leads to agglomeration, this hydrogen bonding cannot be directly attributed to the solubility of PEG modified nanoparticles in Tyrode buffer. Currently, it is difficult to deduce the nature of the ligand bonding to the nanoparticle surface. Here, we propose two probable pathways for the surface ligand attachment: (1) Under UV-irradiation, homolytic cleavage of Si-H and RO-H can generate silyl and alkoxy radicals, which could react to form Si–O–R bonds passivating the nanoparticle surface. (2) Lewis acidic boron surface sites could react with Lewis basic PEG-OH to form the B–O–R bond.
The existence of limited boron surface sites compared to surface Si atoms and preferential bond affinity of Si-O led us to postulate that PEG-OH might predominantly attach to the nanoparticle surface through surface silicon sites. However, it is possible that both pathways could contribute to the surface functionalization. As discussed earlier, undoped bare Si forms large agglomerates in Tyrode buffer indicating poor surface passivation. Thus, initial dispersibility of nanocrystals in DMSO seems to be greatly influencing the surface passivation, and boron surface sites are critical in the synthesis of PEG-modified biocompatible Si nanocrystals.
To investigate the role of the terminal -OH group in surface passivating, we substituted PEG-OH with the thiolated PEG (PEG-SH) ligand and repeated the experiment. Figure 6 illustrates that under similar conditions, PEG-SH passivated B-doped Si nanocrystals show poor dispersibility leading to particle agglomeration in Tyrode buffer. Even though we observed an optically clear solution post reaction, particles were crashed instantaneously when introduced to the buffer. This observation corroborates that the -OH terminal group is crucial for dispersibility in the buffer.
As optical properties are pivotal for bio-imaging applications, absorption and PL properties of surface-modified nanocrystals were investigated. While PEG modified doped particles show negligible PL in both DMSO and buffer, the absorption spectrum is almost identical to the undoped bare Si nanocrystals (Fig. 7). Quenching of PL in B doped Si nanoparticles was observed in the literature and ascribed to the boron induced defect states on the nanoparticle surface.42 It has been reported that PL quenching of B-doped nanoparticles increased with increased nanoparticle size, and theoretical calculations showed that B concentrations exceeding 1% could substantially quench the photoluminescence.43 Near-identical absorption features could be attributed to the limited boron surface sites on the nanoparticle surface. Nonetheless, even with limited PL properties PEGylated B-doped Si nanocrystals are promising candidates for imaging applications as absorption related contrast agents in various bioimaging modalities.44 However, prior cytotoxicity assessments should be performed to rule out any possible toxicity that might arise from dopants.
B-doped Si nanocrystals, synthesized via a nonthermal plasma technique, with unique surface properties were exploited for the synthesis of bio-compatible PEG-grafted nanocrystals. The results revealed that PEG-grafted doped particles are nearly monodispersed in Tyrode buffer for 8 h. The optimization of nanoparticle characteristics for different PEG concentration, PEG molecular weight, and boron doping percentage has proven that colloidally stable PEG-modified particles can be synthesized in different reaction combinations. We postulated that Si–O bond formation could be mainly contributing to an effective surface passivation, but boron surface sites are critical for the reaction as initial dispersiblity plays a key role in successful surface passivation. With near identical absorption features as bare Si, PEG-modified B-doped Si particles can be utilized for the bio-imaging applications while providing the oxidation resistance.
This work of was supported by the National Institutes of Health under Award No. R01DA045549. Portions of this work were conducted in the Minnesota Nano Center, which is supported by the National Science Foundation through the National Nano Coordinated Infrastructure (NNCI) Network under Award No. NNCI-154220.