Thermodynamic Driving Force in the Spontaneous Formation of

Letter Cite This: Nano Lett. 2018, 18, 1888−1895

pubs.acs.org/NanoLett

Thermodynamic Driving Force in the Spontaneous Formation of Inorganic Nanoparticle Solutions Lance M. Wheeler,*,† Nicolaas J. Kramer, and Uwe R. Kortshagen* Department of Mechanical Engineering, University of Minnesota, Minneapolis, Minnesota 55455, United States S Supporting Information *

ABSTRACT: Nanoparticles are the bridge between the molecular and the macroscopic worlds. The growing number of commercial applications for nanoparticles spans from consumer products to new frontiers of medicine and next-generation optoelectronic technology. They are most commonly deployed in the form of a colloid, or “ink”, which are formulated with solvents, surfactants, and electrolytes to kinetically prevent the solid particulate phase from reaching the thermodynamically favored state of separate solid and liquid phases. In this work, we theoretically determine the thermodynamic requirements for forming a single-phase solution of spherical particles and engineer a model system to experimentally demonstrate the spontaneous formation of solutions composed of only solvent and bare inorganic nanoparticles. We show molecular interactions at the nanoparticle interface are the driving force in highconcentration nanoparticle solutions. The work establishes a regime where inorganic nanoparticles behave as molecular solutes as opposed to kinetically stable colloids, which has far-reaching implications for the future design and deployment of nanomaterial technologies. KEYWORDS: Silicon, ligand-free, quantum dot, semiconductor nanocrystal, colloidal stability

B

electrical double layer is a kinetic barrier to NP agglomeration. However, there are two main issues with this evaluation: (i) DLVO describes lyophobic particlesan underlying assumption is solvent molecules do not favorably interact with the particle surface. The solvent is considered a continuous dielectric medium. (ii) Even with extensions to the DLVO model that incorporate molecular interactions,21 the theory describes kinetic stability af ter colloid formationformation is not spontaneous; it requires the addition of energy well above kT to the system (typically sonication) to drive the solid phase temporarily into the liquid phase. Here we show thermodynamically stable solutions of Si NPs with engineered surfaces spontaneously form and develop a model that shows molecular interactions at the NP surface are the driving force for solution formation. Figure 1a shows still-frame images extracted from Movie S1, Supporting Information, of dimethyl sulfoxide (DMSO) being added to 7 nm (±20%) silicon NP powder to spontaneously form a single phase. The process resembles solvating a molecular solid into a liquid (e.g., sugar in water). Shaking the vial expedites solution formation, especially for high concentrations (>20 mg mL−1), but it is not necessary. It is clear, from the transmission electron microscopy (TEM) image in Figure 1b, the NPs have gone from an agglomerated solid phase of NPs to a solution of singly isolated NPs exhibiting a solvodynamic diameter of 7−10 nm observed by dynamic light

are inorganic particles are typically observed to form a metastable two-phase dispersion after the addition of mechanical energy to transfer the particles from the thermodynamically favored agglomerated state into the liquid phase. Particles may remain isolated in the liquid phase for a finite time due to a kinetic barrier in the form of an electrical double layer1 or steric repulsions due to long-chain organic molecules.2 In contrast, lyophilic colloids spontaneously form a single-phase solution (not dispersion) when solute is brought into contact with solvent. In the absence of changes to chemistry or temperature, a solution is stable indefinitely.3 Beyond typical molecular systems, this behavior has only been observed for organic systems such as polymers,4 proteins,5 and microemulsions.6 Recent reports of spontaneous solution formation of carbon nanotubes7−10 and graphene11 have demonstrated true thermodynamic solution behavior from exciting new organic nanomaterials. The possibility of thermodynamic solutions of bare inorganic particles is often reflexively ignored, but there is no a priori reason to argue against their existence.12−14 Thermodynamic behavior of ligand-functionalized nanoparticles (NPs) has been recognized, but the thermodynamics are dominated by the ligand interactions with solvent molecules.15 Spontaneous transfer of NPs from a nonpolar solvent to a polar one has been shown for a number of materials, including metallic,16−18 metal oxide,17 and semiconducting16−20 NPs following ionic ligand functionalization. The exchange of long-chain ligands for short, labile ones that are typically balanced with a counterion in solution has led many researchers to believe kinetic models like DLVO theory1 describe these systems, in which an © 2018 American Chemical Society

Received: December 9, 2017 Revised: February 20, 2018 Published: February 26, 2018 1888

DOI: 10.1021/acs.nanolett.7b05187 Nano Lett. 2018, 18, 1888−1895

Letter

Nano Letters

section for details). NPs produced without B2H6 do not spontaneously form solutions.23 Figure 2a is a diagram illustrating the structure of boronated Si NP that we determined using X-ray photoelectron spectroscopy (XPS) and attenuated total reflectance Fourier transform infrared (ATR-FTIR) spectroscopy. Figure 2b shows the B 1s core-level XPS spectrum of the as-synthesized Si NPs. The spectrum is deconvolved into four peaks wherein the broad peak centered at 188 eV (cyan) is Lewis acidic three-coordinate boron29 that resides at the Si NP surface with a 2pz orbital available for bonding (Figure 2a). Peak integration of the XPS spectrum reveals ∼60% of the boron is on the Si NP surface. First-principles calculations show B atoms to energetically favor surface segregation as opposed to substitutional incorporation.30,31 High-energy X-ray diffraction measurements coupled to atomic pair distribution function analysis and reverse Monte Carlo simulations have also shown B to reside at the Si NP surface.32 The orange peak at 185 eV corresponds to fourcoordinate boron that is incorporated into the core of the NP, and the two smaller high binding energy peaks (magenta) are four-coordinate B bonded to oxygen atoms at the Si NP surface.29 The diamond cubic crystalline core and the disordered acidic B surface are observed in the transmission electron microscope (TEM) image of Figure 2c. Our previous work showed donor−acceptor (Lewis acid− base) solvent interaction were critical in achieving solutions of Si NPs with chlorinated surfaces.23 Interactions with the acidic B surface of the NPs are demonstrated using ATR-FTIR. In Figure 2d, n-methylpyrrolidone (NMP), which features a spectrally isolated carbonyl stretch vibration at 1684 cm−1, is added to Si NPs residing on an ATR crystal and allowed to evaporate over the course of ∼60 min. At early times, the spectrum is dominated by features corresponding to free NMP molecules in solution (purple spectrum). As the solution evaporates, and the relative Si NP concentration increases, the carbonyl group of NMP shifts from 1684 cm−1 to reach a final state (red spectrum) with a broader peak centered at 1647 cm−1, which is an indication of electron density transfer from the Lewis basic CO group to a Lewis acidic B group of the Si NP surface to form a donor−acceptor complex.33 A peak emerges at 1329 cm−1 that we assign to the *Si3BO vibration of the carbonyl group interacting with B at the Si NP surface.

Figure 1. Spontaneous NP solution formation. (a) Still-frame images from Movie S1, Supporting Information, showing the spontaneous formation of a NP solution after the addition of DMSO to yield a ∼10 mg mL−1 solution. (b) TEM image shows the initial agglomerated state of Si NPs before the addition of solvent. The scale bar is 10 nm. (c) DLS data showing Si NPs form a solution of individually isolated particles after the addition of DMSO. (d) Still-frame images from Movie S1, Supporting Information, showing water added to Si NPs to yield a 2-phase system.

scattering (DLS) in Figure 1c. These solutions have remained as a single-phase for over three years. Different behavior is observed for other solvents. The still-frame images in Figure 1d show Si NPs remain in an isolated second phase after H2O addition. Evidence in previous work suggests NP solution formation is strongly dependent on NP surface chemistry and choice of solvent.20,22−27 Here we control NP surface chemistry using a nonthermal plasma reactor.28 Silane (SiH4) is decomposed in the presence of diborane (B2H6) to produce silicon NPs with boron terminating the surface (see the Experimental Methods

Figure 2. Boronated Si NP surface. (a) Diagram illustrating the structure of boronated Si NPs. (b) XPS B 1s spectrum of Si NPs as they are produced from the plasma reaction. Surface B (cyan) accounts for ∼60% of the boron in the sample. (c) TEM image of a Si NP with a crystalline core and acidic surface. The scale bar is 3 nm. (d) FTIR spectra of a Si NP solution in NMP as NMP evaporates to form a Si NP film, which clearly illustrates strong donor−acceptor bonds formed between the carbonyl group of NMP and the acidic B sites at the NP surface. Spectra are taken continuously over the course of 60 min. Spectra are chosen at varied intervals to show the evolution. 1889

DOI: 10.1021/acs.nanolett.7b05187 Nano Lett. 2018, 18, 1888−1895

Letter

Nano Letters

Figure 3. Thermodynamics of NP solution formation. (a) Thermodynamic process describing ΔGsolv for NP solution formation. Two phases are considered: the solvent phase composed of hard spheres of diameter σs (blue) and the NP phase composed of hard spheres of diameter σNP (red). (b) ΔGsolv as a function of the number fraction of NPs in solution, xNP. Each set of lines represents sphere diameter ratios, σNP/σs = 2000 (red), 1000 (orange), 200 (yellow), 100 (green), 20 (blue). Each line is calculated by varying Δgs−NP from 10kT (leftmost thick line) to −25kT (rightmost thick line) in increments of 5kT (thin lines). (c, d) Each component of the thermodynamic process is plotted as a function of diameter ratio, σNP/σs for xNP = 1 × 10−6 (c) and xNP = 1 × 10−4 (d). Positive contributions are red, whereas negative contributions are shown in blue (ΔGs−NP) and green (ΔGmix) and correspond to the steps labeled in part a. ΔGs−NP (blue) is evaluated at Δgs−NP = −25kT. The black curves are ΔGsolv with labels that correspond to Δgs−NP values. Unlabeled black curves are calculated in 5kT increments. In all calculations (b−d), ΔGs−s is evaluated at Δgs−s = 10kT, ΔGcav = ΔGcav,s + ΔGcav,NP, and ΔGNP−NP is evaluated at ΔgNP−NP = 10kT, and the sphere number density is η = 0.48.

ΔGcav,NP, and loss of cohesion between particles, ΔGNP−NP, are energetically unfavorable processes. (III) The NP is transferred to the solvent phase, which is described by the Gibbs energy of mixing, ΔGmix, and the new interactions between NP−solvent and solvent−NP, ΔGNP−s and ΔGs−NP, respectively, to yield an expression for ΔGsolv analogous to regular solution theory and similar to those used to effectively model micelle formation35 and fullerene solvation:36

However, boron may also exist on the surface with a terminal hydride (*Si2BH). Boron in this state also serves as a Lewis acidic surface site that will interact with the carbonyl group. Spontaneous solution formation is a thermodynamic effect. Here we derive a thermodynamic model that predicts the observed NP solution behavior by extending the thermodynamic process considered for molecular solvation34 to a statistical mechanical system composed of two adjacent phases of hard spheres (Figure 3a). The solvent phase is composed of hard spheres of diameter σs (blue), and the NP phase is composed of hard spheres of diameter σNP (red). The transfer of a NP from the NP phase into the solvent phase is described by the Gibbs energy of solvation, ΔGsolv, and must be less than, or equal to, zero for a single-phase solution to spontaneously form: ⎧≤ 0, spontaneous solution ΔGsolv = ⎨ ⎩> 0, two phases

ΔGsolv = ΔGmix + ΔGcav,NP + ΔGcav,s + ΔGs − s + ΔG NP − NP + ΔGs − NP + ΔG NP − s

where ΔGmix = ΔGideal + ΔGex. ΔGideal is the conventional ideal term, ΔGideal = xi ln xi + xj ln xj, and ΔGex is the excess Gibbs energy based on the Boublik−Mansoori−Carnahan−Starling− Leland equation of state (BMCSL EoS).37 We describe the Gibbs energy of cavity formation in both phases, ΔGcav,i (where i is the solvent, s, or nanoparticle, NP), with a derivative of scaled particle theory38,39 developed by Matyushov and Ladanyi,40 which is based on the BMCSL EoS. The interaction terms, ΔGi−j (where i and j are the solvent, s, or nanoparticle, NP), are defined as the number of interactions at the cavity interface in hard sphere system i, zi, times the strength of molecular interaction, Δgi−j, between species i at the j interface:

⎪

⎪

(2)

(1)

Figure 3a shows the three thermodynamic steps considered to determine ΔGsolv: (I) A cavity the diameter of a NP, σNP, is formed in the solvent phase, which gives rise to two unfavorable processesthe Gibbs energy of cavity formation, ΔGcav,s, and loss of interaction between solvent molecules, ΔGs−s. (II) A cavity the diameter of a NP, σNP, is formed in the NP phase. The Gibbs energy of cavity formation in the NP phase,

ΔGi − j/NkT = x iz j 1890

Δg i − j kT

, where N = Ni + Nj is the total number DOI: 10.1021/acs.nanolett.7b05187 Nano Lett. 2018, 18, 1888−1895

Letter

Nano Letters

to the extensive data available.42 Δgs−s is approximated as Gibbs energy of vaporization (Δgs−s ≈ ΔGvap), which is a measure of the strength of intermolecular interactions. We identify a number of solvents that provide high NP concentrations. For 7 nm (±20%) NPs, xNP = 1 × 10−4 is equivalent to ∼4% volume fraction or 100 mg mL−1 without any stabilizing ligand or additive. The maximum number fraction is presented in Figure 4 where each solvent is represented by a blue bubble and a number that corresponds to the solvent listed in Table 1.

of spheres, xi is the number fraction of i, and zi is determined by integrating over the radial distribution function at contact over a spherical shell volume at the cavity interface (Figure S1). Details on the mathematical framework developed to describe this process are included in the Supporting Information. ΔGsolv is plotted as a function of the number fraction of NPs in solution, xNP, in Figure 3b. Each series of curves are for a particular diameter ratio (σNP/σs = 20 (blue), 100 (green), 200 (yellow), 1000 (orange), and 2000 (red)). Each curve increases Δgs−NP from −25kT to 10kT, keeping Δgs−s constant. It is clear the maximum theoretical concentration and thermodynamic driving force (magnitude of −ΔGsolv) are strongly dependent on NP size. At large sizes (σNP/σs > 200), small, but finite, solubilities are predicted but with a driving force only slightly above the thermal energy (∼2kT). The Gibbs energy of mixing, ΔGmix, is the dominant contributor in these systems. This is clear from Figure 3c, which plots each thermodynamic contribution as a function of diameter ratio at a concentration of xNP = 1 × 10−6. Higher concentrations are limited to smaller-diameter solutes (NPs). At high absolute values of Δgs−NP (Δgs−NP < −15kT), NPs are even “miscible” for all xNP. Rather than an entropic driving force, there is a shift to molecular interactions as the dominant driving force for small-diameter NPs. This is illustrated in Figure 3d, which shows the individual contributions to ΔGsolv at high concentration (xNP = 1 × 10−4). In this regime, eq 2 is reduced to Δg + Δgs − NP ΔGcav,s ΔGsolv ΔGmix ≈ + + x NPzs s − s kT NkT NkT NkT (3)

Figure 4. Evaluation of NP solubility. (a) Saturated NP concentration as a function of ΔGvap + ΔGd−a for a select number of solvents. Bubble diameter reflects relative solvent diameter. Red curves are δ ΔGsolv (Δgs − NP , σNP/σ s) = 0 as a function of Δgs−NP for diameter δx

Equation 3 shows the Gibbs energy of solvation is determined by the Gibbs energy of mixing, the Gibbs energy of cavity formation in the solvent, and competitive intermolecular interactions: The energy cost of removing interactions from the cavity interface, ΔGs−s, is in direct competition with the favorable interactions at the cavity interface after the NP is transferred to the solvent, ΔGs−NP. Δgs−NP must be negative and greater than Δgs−s in order for spontaneous solution formation to occur. The importance of competitive molecular interactions has been similarly described in theories of polymer solvation.41 We evaluate eq 3 by adding solvents to B-terminated Si NPs and experimentally measuring the saturated NP concentration (xNP). Experimentally determined xNP is plotted as a function of ΔGvap + ΔGd−a, where ΔGvap is the Gibbs energy of vaporization, and ΔGd−a is the Gibbs energy of donor− acceptor bond formation. In principle, the NPs interact with solvent molecules through van der Waals forces (ion−dipole, dipole−dipole or dipole−induced dipole) as well as donor− acceptor (Lewis acid−base) interactions. For example, dielectric constant was found to not be a good indicator of solubility, and no solvents with dielectric constant below 12 showed NP solubility even with significant basicity (e.g. triethylamine, pyridine). Steric hinderence of the Lewis basic group likely plays a larger role than in the solvents discussed below and is an important topic for future work. Here, we assume donor−acceptor interactions dominate and set the Gibbs energy of solvent-NP interactions equal to the Gibbs energy of donor-acceptor interactions, Δgs−NP ≈ ΔGd−a, because we experimentally observe these interactions, and they are typically significantly stronger than other noncovalent interactions. We apply a = SbCl5 as the standard acceptor due

NP

ratios of 15, 20, and 25 (red text). White and gray number labels indicate solvent, which correspond to data in Table 1.

The derivative of the Gibbs energy of solvation with respect δ to xNP is set to zero ( δx ΔGsolv = 0) to determine theoretically NP

predicted saturated xNP values. This function is superposed onto the experimental data in Figure 3 as red curves. Diameter ratios of 15, 20, and 25 are shown to accommodate the spread of data. For reference, the diameter ratio of an 8 nm NP in DMSO (σs ≈ 0.4 nm) is 20. The simple hard sphere model is nicely correlated to the experimental data and validates molecular interactions as the thermodynamic driving force for spontaneous solution formation in this size regime. Discrepancies observed are easily rationalized due to the oversimplifying assumptions to the complex system. Considering the long-standing difficulty with predicting solubility of even molecular systems, the model is surprisingly effective at reproducing the experimental data. It is interesting to consider the physics captured by our thermodynamic model may be interpreted in a number of ways. Models describing the thermodynamics of micelle and microemulsion systems consider similar contributions to the Gibbs energy of the colloidal system.46 Entropy drives the system into a thermodynamic solution when the interaction of surfactant molecules sufficiently reduces interfacial tension (surface energy) at the oil/water interface. Decreasing interface energy is analogous to the solvent−NP interactions; this may be viewed as reduction in surface tension. Implicit solvation 1891

DOI: 10.1021/acs.nanolett.7b05187 Nano Lett. 2018, 18, 1888−1895

Letter

Nano Letters Table 1. Solvent Data Compiled for Figure 4 ΔHvapa solvent f

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 a

dimethyl sulfoxide dimethyacetamide N,N-dimethylformamide N-methylformamide N-methyl-2-pyrrolidone cyclohexanone acetophenone 2-pentanone benzonitrile 2-butanone acetone methanol formamide n-butanol butanenitrile propionitrile isobutyl ketone ethanol water 1,4-dioxane acetonitrile phosphorus oxychloride chloroform nitromethane 1,2-dichlorobenzene hexane toluene

−1

kJ mol 52.9 50.2 46.9 54.4 55.3 45.1 55.4 36.1 49.1 34.4 29.0 37.5 60.2 52.4 39.2 35.3 37.7 42.4 40.7 38.6 33.0 35.1 31.4 38.3 48.5 31.0 38.1

ΔSvapb −1

JK

ΔGvapc −1

−1

mol

kJ mol

114.4 114.6 110.1 119.4 116.4 105.1 116.6 96.6 106.5 97.5 88.1 110.8 124.6 134.0 100.5 95.6 96.6 120.5 108.9 103.3 92.9 92.6 93.9 102.6 107.0 90.9 99.2

Data compiled from ref 43. bCalculated using the relation ΔSvap =

19.3 16.6 14.6 19.4 21.2 14.3 21.2 7.8 17.9 5.8 3.2 5.0 23.7 13.1 9.7 7.3 9.4 7.0 8.7 8.4 5.8 7.9 3.9 8.2 17.1 4.4 9.0 ΔH vap Tboil

ΔHd−ad −1

kJ mol

−124.7 −116.3 −111.3 −113.0 −114.2 −75.3 −62.8 −62.8 −49.8 −72.8 −71.1 −79.5 −100.4 −81.6 −69.5 −67.4 −66.9 −80.3 −75.3 −61.9 −59.0 −49.0 −16.7 −11.3 −12.6 0.0 −0.4

ΔGd−ae

(ΔGvap + ΔGd−a)

kJ mol−1

kJ mol−1

kT

−63.4 −59.9g −51.2 −52.1 −52.8 −31.7 −24.9 −24.9 −12.2g −30.3 −27.7g −34.0 −45.3 −35.1 −28.5 −27.4 −27.1 −34.4 −30.0g −24.4 −22.8 −4.0g 0.1 3.1 2.4 9.2 9.0

−44.0 −43.3 −36.6 −32.7 −31.6 −17.4 −3.6 −17.0 5.7 −24.5 −24.5 −29.0 −21.6 −22.0 −18.8 −20.1 −17.8 −27.4 −21.3 −16.0 −17.1 3.9 4.0 11.3 19.5 13.6 18.0

−17.8 −17.5 −14.8 −13.2 −12.8 −7.0 −1.5 −6.9 2.3 −9.9 −9.9 −11.7 −8.7 −8.9 −7.6 −8.1 −7.2 −11.1 −8.6 −6.5 −6.9 1.6 1.6 4.6 7.9 5.5 7.3

g

, where Tboil is the boiling point of the solvent. cΔGvap = ΔHvap − TΔSvap

where T = 20 °C is assumed. dThis data is for SbCl5 as the reference acceptor (ΔHd−SbCl5). This value is traditionally known as the donor number, DN.44 eDue to the scarcity of experimental data, most ΔGd−a values are obtained from a linear fit to ΔGd−a vs ΔHd−a of available data in ref 45: ΔGd−a = −0.54 × Hd−a + 9.23. fNumbers correspond to labels in Figure 4. gIndicates experimental data is used.

Si NPs can be diluted in water after being solubilized in DMSO to yield a biocompatible solution of Si NPs. Figure 4a is a photograph of four additional cosolvent solutions of NPs prepared in the same way: H2O is added to NPs solvated in NMP, nitromethane (NM) to NPs solvated in acetophenone (APh), 1,2-dichlorobenzene (DCB) to NPs solvated in dimethylacetamide (DMA), and acetonitrile (ACN) to NPs solvated in DMSO at 50% by volume to yield stable NP solutions. A similar observation was briefly reported previously by Dong et al., who noticed a variety of metallic and semiconducting NPs will solubilize in water only if DMF is included as well.17 We believe cosolvent NP solutions are enabled by selective NP solvation and a decrease in Δgs−s. Moreover, it is likely a large driving force is needed to overcome any kinetic effects in the system. A smaller driving force is needed to mix additional solvents after the NP solution is formed. Evidence for selective solvation is provided in Figure 5. Si NPs in NMP are diluted with water from 0% up to 94% by volume and left at ambient conditions for 2 weeks. These solutions were cast onto an ATR crystal and allowed to evaporate. The hydrocarbon (ν(CHx) = 1684 cm−1) and carbonyl (ν(CO) = 1684 cm−1) stretching modes of NMP, highlighted with dashed lines, are observed even at high volume fractions of water. Note the ν(CO) stretch is obscured by the bending vibration of water at ∼1644 cm−1 at volume fractions of 50% and 94%. Interestingly, the

models are also heavily employed to describe the solvation of complex biomolecules like proteins.47 In these models, one considers the Gibbs energy change to transfer a solute from the gas phase to the solvent phase. The solvent is treated as a structureless continuum dielectric medium described by the dielectric constant. The solute is treated as a collection of partial charges in a cavity and is also described by a dielectric constant. In our work, the dielectric constant of bulk silicon (11.7) may be reduced due to the size of the crystal48,49 as well as the charge at the NP surface. The addition of electron density to the NP by the formation of donor−acceptor bonds with solvent molecules at the NP surface increases charge in the NP, thus reducing the dielectric constant. This would increase the magnitude of the Gibbs energy of transfer and make solvation more thermodynamically favorable. Similar to our model, this insight suggests conductive materials, such as metals, or materials with acidic surface sites, such as metal oxides and chalcogenides, should display similar NP solution behavior to the Si NPs studied here. The observed solvent-specific solubility of Si NPs limits the number of solvents that can be practically applied. However, we are able to formulate an array of cosolvent mixtures that produce NP solutions in solvents that do not yield solubility alone. This is a strategy often used in molecular systems50 to, for instance, enhance aqueous solubility of drugs51 or better solubilize proteins.52 Movie S1, Supporting Information, shows 1892

DOI: 10.1021/acs.nanolett.7b05187 Nano Lett. 2018, 18, 1888−1895

Letter

Nano Letters

ring electrodes secured to it. Typical flow rates were 30−50 standard cubic centimeters per minute (sccm) of argon, 0.4− 0.6 sccm of silane, and 0−2 sccm of diborane diluted in hydrogen (10%). A 13.56 MHz radiofrequency power at 110− 130 W was supplied to the ring electrodes to strike a plasma and yield crystalline Si NPs. A rectangular nozzle controlled the gas pressure in the plasma region by restricting the flow. Adjusting the width of the nozzle opening allowed for the pressure to change independently of the gas flow. This method was used to produce nanocrystals at reactor pressures ranging from 90 to 150 Pa. As a result, the nanoparticle size can be precisely controlled between 5 and 15 nm. The 7.5 nm Si NPs were used throughout this study. Si NP Collection and Solution Formation. Powder samples of silicon nanocrystals were collected directly from the gas phase by impacting them onto a substrate mounted onto a manual feed-through located inside the reactor. A rectangular nozzle was placed in between the deposition and plasma region to accelerate the particles and impact them directly onto the substrate. The substrate was then retracted into a portable loadlock and transferred air-free to a N2-purged glovebox for further processing. Si NPs were weighed and solvated in various solvents. XPS. XPS spectra were acquired on a Surface Science Laboratories, Inc., SSX-100 XPS with a monochromatic Al Kα X-ray source. An X-ray power of 200 W with a 1 × 1 mm2 spot size was used. Si NP samples were prepared by directly impacting a