Silicon quantum dot - poly(methyl methacrylate) nanocomposites with

7 days ago - Silicon quantum dots with indirect bandgap photoluminescence are promising luminophores for large-area luminescent solar concentrators ...
0 downloads 0 Views 1MB Size
Download PDF
Subscriber access provided by University of Winnipeg Library

Article

Silicon quantum dot - poly(methyl methacrylate) nanocomposites with reduced light scattering for luminescent solar concentrators Samantha Hill, Ryan Connell, Colin Peterson, Jon Hollinger, Marc A. Hillmyer, Uwe R. Kortshagen, and Vivian E. Ferry ACS Photonics, Just Accepted Manuscript • DOI: 10.1021/acsphotonics.8b01346 • Publication Date (Web): 04 Dec 2018 Downloaded from http://pubs.acs.org on December 5, 2018

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 29 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 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Photonics

Silicon quantum dot - poly(methyl methacrylate) nanocomposites with reduced light scattering for luminescent solar concentrators Samantha K.E. Hill*1, Ryan Connell*2, Colin Peterson3, Jon Hollinger3, Marc A. Hillmyer3, Uwe Kortshagen†1, Vivian E. Ferry†2 1Department

of Mechanical Engineering, University of Minnesota, Minneapolis, MN

2Department

of Chemical Engineering & Materials Science, University of Minnesota,

Minneapolis, MN 3Department

of Chemistry, University of Minnesota, Minneapolis, MN

*Denotes equal contributions †Corresponding

authors: [email protected], [email protected]

Abstract Silicon quantum dots with indirect bandgap photoluminescence are promising luminophores for large-area luminescent solar concentrators (LSCs). However, if commercially viable devices are to be achieved, silicon quantum dots must be dispersed within functional, light-guiding matrices such as acrylic slabs without losing their high photoluminescent quantum yield or succumbing to light-scattering agglomeration. With a goal of limiting scattering and producing functional LSC materials, we study silicon quantum dot / poly(methyl methacrylate), or PMMA, bulk polymerized composites. Monte Carlo modeling predicts that scattering losses are significant for large-area silicon quantum dot LSCs unless the characteristic scattering length is at least as large as the LSC side length. We compare the effect of particle ligand choice on the nanocomposites, using particle loadings ranging from 0.06 to 0.50 weight percent. We find that methyl 10-undecenoate functionalized silicon quantum dots in PMMA composites exhibit low levels of particle agglomeration, and thus light scattering, as compared to analogous alkane-capped silicon quantum dots. As a result, these ester-Si / PMMA composites show an improvement in light guiding compared to the alkane-Si composites, which is beneficial for future LSC applications. Keywords: silicon quantum dot, poly(methyl methacrylate), luminescent solar concentrator, light scattering, nanocomposite

1 ACS Paragon Plus Environment

ACS Photonics 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 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 29

Luminescent solar concentrators, or LSCs, were first proposed in 1976 as a cost-saving alternative to traditional photovoltaics1,2. These devices consist of luminophores embedded in a transparent dielectric material that acts as a waveguide. Incident sunlight is absorbed and reemitted by the luminophores, and the waveguide concentrates the luminophore photoluminescence onto a small-area solar cell, thereby boosting the power output. LSCs are typically fabricated as luminophore-doped polymer slabs3,4 or thin films on glass5,6. As electrode-less light harvesting devices, LSCs can serve as transparent windows for building integrated photovoltaics3–6. Such transparent solar harvesters hold promise to invisibly integrate within the urban environment, circumventing spatial, functional, and aesthetic limitations of opaque photovoltaics while still providing renewable power generation at its point-of-use7–11. Despite the simplicity of the LSC concept and compelling applications, widespread use of this technology has been hindered by the modest efficiencies of demonstrated devices12. Several loss mechanisms combine to limit LSC performance, including limitations in the waveguide’s ability to trap light (escape cone losses13,14) or transmit light (reflection15, scattering16,17, and parasitic absorption losses18,19). More significant losses also arise from the luminophore due to limited absorption profiles, non-unity photoluminescent quantum yields (PLQYs), reabsorption of trapped light, and scattering12,20. However, a recent resurgence in LSC research has generated new material platforms and photonic designs for combating these losses, and LSC efficiencies are on the rise7,12,20–26. The use of brightly photoluminescent quantum dots (QDs) with broadband absorption and tunable optical spectra has been a major factor in achieving these advances20. Importantly, several different types of QDs have been shown capable of drastically reducing the reabsorption losses that previously limited the maximum size of LSCs. These QDs exhibit minimal overlap between their absorption and photoluminescence spectra and include heterostructured QDs27–30, ternary I-III-VI2 QDs3,6,31,32, impurity doped QDs30,33,34, and indirect bandgap silicon QDs4. This work focuses on silicon-based QDs which are especially well-suited for window LSC applications due to their indirect, near-infrared emission7,35, low toxicity36,37, and widespread elemental abundance4. However, since Si QDs have only recently been proposed for use in LSCs, little work has been completed to incorporate them into polymer matrices. Ideally, methods should be developed to incorporate Si QDs in a wide library of polymers while maintaining their high

2 ACS Paragon Plus Environment

Page 3 of 29 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 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Photonics

PLQY and suppressing particle agglomeration that leads to scattering losses38. As we demonstrate in this paper, scattering losses can significantly affect the performance of large-area Si QD LSCs. Limited reports indicate that Si QDs can withstand bulk polymerization in polymers including poly(methyl methacrylate), or PMMA39,40, poly(dimethyl siloxane)41, thiol-ene polymers42, silicone elastomers43, and polystyrene44, as well as in silica aerogels45. However, only one report provides significant evidence for low scattering cross sections in bulk polymerized Si QD / polymer composites, wherein Si QDs were incorporated within flexible slabs of poly(lauryl methacrylate) crosslinked with ethylene glycol dimethacrylate4. In terms of building-integrated LSC applications, however, PMMA and polycarbonate have been noted as desirable host polymers due to their favorable mechanical properties and relatively low parasitic absorption18,19. Herein, we focus on incorporating Si QDs into PMMA, a commodity thermoplastic with high clarity and light transmittance coupled with good strength, stability, and scratch resistance46. PMMA is popular for optical and structural applications alike and is less expensive than polycarbonate, making it an interesting polymer for Si QD encasement. In this study, we prepare and examine the optical properties of Si QD / PMMA composite slabs. We determine the effect of Si QD concentration on light scattering properties by examining loading fractions ranging from 0.06 to 0.50 weight percent (wt%). Using a variety of spectroscopic techniques, we also examine the effects of two different Si QD surface ligands on the transmission and light scattering properties of these composites. The studied ligands include a long alkane (1dodecene) and an ester-functionalized alkane (methyl 10-undecenoate), which has not previously been studied within the PMMA system. We find that the more polar surface moieties enable a single-particle dispersion of Si QDs in methyl methacrylate, or MMA, which translates to reduced light scattering by Si QDs in PMMA. Attenuation measurements demonstrate the impact scattering has on the waveguide losses of the composite slabs. These low scattering ester-capped Si QD / PMMA composites hold promise for realizing efficient, non-toxic, and large-area LSCs. Results & discussion Scattering in Si QD LSCs To illustrate the importance of decreasing scattering losses in Si QD-based LSC devices, we employed a ray-tracing Monte Carlo model as described in the Supporting Information to predict concentration factors under different scenarios. The concentration factor is an LSC figure of merit 3 ACS Paragon Plus Environment

ACS Photonics 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 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 29

that depends on device geometry and light concentration performance. It is defined as the geometric gain of the device (collection face area / concentration edge area) multiplied by the fraction of incident light that is successfully concentrated to the collection edge. In Figure 1, we show some concentration factor predictions, comparing results from 50% (a) or 100% (b) PLQY Si QDs with varying scattering probabilities. In all cases, the LSCs are modeled to be 3 mm thick

Figure 1: Simulated effect of scattering on Si QD LSC performance. The top row presents the concentration factor for square LSCs of side lengths ranging from 2 cm (red) to 1 m (purple) with a constant 3 mm thickness as a function of the scattering length to LSC side length ratio for a 0.1 wt% loading of Si QDs with (a) 50% or (b) 100% photoluminescence quantum yield. The bottom row shows the fraction of absorbed photons that are either collected by the solar cell at the edge of the concentrator or lost to various pathways for a 1 m2, 3 mm thick LSC with a 0.1 wt% Si QD loading and either (c) 50% or (d) 100% photoluminescence quantum yield. Photons that pass through the concentrator and are not absorbed are not shown here.

4 ACS Paragon Plus Environment

Page 5 of 29 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 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Photonics

as this is the approximate thickness of common window glass. The 50% PLQY model represents the currently achievable case based on quantum yield measurements from our fabricated Si QD / PMMA composite slabs. The incoming sunlight is modeled as normal incidence for these results, which provides the lower limit for the concentration factor since the addition of diffuse sunlight increases the concentration factor at relevant concentrator tilts (Supporting Information S2-1). An isotropic scattering probability is included in the model and described in terms of a characteristic scattering length. This is defined as the distance light travels such that the fraction 1

of incident light scattered is 𝑒 , where the probability of light scattering follows the equation:

𝑃𝑠 = 𝑒

―

𝑙 𝑙𝑠

where 𝑃𝑠 is the probability of scattering, 𝑙 is the distance traveled in meters, and 𝑙𝑠 is the scattering length in meters. Figures 1(a) and 1(b) show the predicted concentration factors for LSCs with a 3 mm thickness, a loading fraction of 0.1 wt%, different lateral sizes, and different scattering lengths. Scattering can originate from various sources, including individual nanocrystals, aggregates, and inhomogeneities within the polymer16,17. In some cases, scattering can be beneficial as either sunlight or luminescent light can be scattered directly into the solar cell. However, this tends to only be useful for small concentrators. As the lateral size increases, the effect of scattering becomes detrimental since it disrupts the propagation of luminescent light toward the edge and increases escape cone losses. In general, for large Si QD LSCs, significant concentration factor gains can be achieved by increasing the scattering length / LSC size ratio from 0.01 to 1. Some further improvement in the concentration factor is observed for ratios of up to ~100 for LSCs with side lengths greater than ~30 cm. In the case of perfectly dispersed ~3 nm diameter Si QDs at a loading of 0.10 wt%, a scattering length of 51 m is expected, which far exceeds the length scale required for window applications (Supporting Information S4-3). Figure 1(c) and (d) compare the fraction of absorbed photons that are either collected at the edge (optical efficiency, also called optical quantum efficiency) or lost via various mechanisms for a 1 m2 concentrator. The figures present the outcome only for photons that are absorbed at least once in the LSC; sunlight that passes through the concentrator and is neither absorbed nor collected 5 ACS Paragon Plus Environment

ACS Photonics 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 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 6 of 29

is taken out of this calculation in order to show the loss pathways clearly (see the Supporting Information). The category “absorbed loss” represents light that is lost non-radiatively after an absorption event. Reabsorption loss, in contrast, corresponds to light that experiences at least two absorption events before also being lost non-radiatively. Scattered loss includes emitted photons that exit out of the top or bottom face of the concentrator after one or more scattering events, while escape cone loss includes photons exiting the same way without any scattering event, but possibly after a reabsorption and reemission event. While reabsorption is expected to be small in these concentrators, the large path lengths of these 1 m x 1 m concentrators can lead to a significant amount of reabsorption. This leads to increased reabsorption and escape cone losses for 50% PLQY Si QDs (Fig 1(c)), as well as increased escape cone losses for 100% PLQY Si QDs (Fig 1(d)). If no reabsorption was assumed, then the predicted escape cone losses would approach ~25% for large scattering lengths as expected for a single photoluminescence event (Supporting Information S3-2). These figures show that scattering is one of the most significant loss mechanisms that limits the performance of these LSCs. In the case of 50% PLQY Si QDs, the largest loss mechanism is absorbed loss deriving from the non-unity quantum yield. For scattering lengths less than ~0.1 m, light scattering is the second most dominant loss mechanism. This loss does not become negligible until the scattering length exceeds 1 m, which is the side length of the modeled LSC. Similar trends are exhibited in the 100% PLQY case, wherein scattering is the primary loss mechanism for scattering lengths below ~0.1 m for 1 m2 LSCs. These results indicate that long scattering lengths are required for efficient, large-area LSCs. Strategies that reduce scattering losses may be more important than either reducing reabsorption through luminophore design or reducing coupling of luminescent light to the escape cone. Nanocomposite fabrication To fabricate Si QD / PMMA composites, we used Si QDs produced in a nonthermal plasma according to previously published methods47. The Si QD size was chosen here to optimize the PLQY and to minimize absorption losses related to C-H overtones in the polymer matrix4. 525 particles were individually measured from transmission electron microscopy (TEM) images such as the one shown in Figure 2(a) to yield the size distribution shown in Figure 2(b). These highly spherical nanocrystals had an arithmetic average core diameter of 2.90 nm with a standard 6 ACS Paragon Plus Environment

Page 7 of 29 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 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Photonics

deviation of 0.64 nm. The average particle size was also estimated to be 3.1 nm by a Scherrer fit to the broadening of the X-ray diffraction pattern48 (Supporting Information S4-4). While the Si QD cores prove to be monodisperse and very small, the core diameter does not necessarily capture the entirety of the average particle dimension when the particles are suspended in a solution or polymer matrix. Long organic ligands are typically added to the Si QD cores to both improve their quantum yield and aid their dispersion in solutions, which slightly increases the particle diameter. These Si QDs were functionalized with one of two different ligands: methyl 10undecenoate (referred to as ester-Si QDs) or 1-dodecene (referred to as alkane-Si QDs), as depicted in Figure 2(c). Methyl 10-undecenoate is not a commonly reported ligand for Si QDs, but enables good suspensions in polar solvents49. In contrast, 1-dodecene is frequently reported in the literature to enable single-dot dispersions of Si QDs in nonpolar solvents50. In either case, the ligands were grafted to the Si QD surfaces by thermal hydrosilylation, a reaction that is readily applicable to other alkene or alkyne chemistries50–52. Fourier transform infrared spectroscopy is consistent with the attachment of the ligands through their carbon-carbon double bond and not via the ester end

Figure 2: (a) ADF-STEM image of the Si QDs used in this study. Inset: higher magnification image of an individual QD showing Si crystal planes. (b) The Si QD core size distribution as measured from ADF-STEM images such as the one shown in (a). (c) The Si QD / PMMA nanocomposite fabrication scheme, wherein the Si QDs were functionalized with either 1-dodecene or methyl 10-undecenoate prior to mixing with methyl methacrylate. The nanocomposite is then formed via radically initiated polymerization using AIBN as the thermal initiator.

7 ACS Paragon Plus Environment

ACS Photonics 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 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 29

group in the case of methyl 10-undecenoate (Supporting Information S5-5). To our knowledge, 1-dodecene is the only ligand previously used to functionalize Si QDs prior to their incorporation into PMMA through bulk polymerization40, which is why we consider 1-dodecene as a ligand here. We neglect the hydride terminated Si QD case in this work since as-produced H-Si QDs from plasmas tend to exhibit very low PLQY until heated above the boiling point of MMA or for prolonged periods53,54. We consider methyl 10-undecenoate as a ligand because the terminal ester group is similar in structure to the repeating unit of PMMA, which may aid in particle dispersion. After functionalization, the Si QDs were dried under nitrogen flow and washed of excess ligands before being dispersed into MMA, the precursor to PMMA. We found that the methyl 10undecenoate functionalized particles readily disperse in MMA, while the 1-dodecene functionalized particles create a suspension only after sonication. By inspection, the ester-Si QD dispersion appears highly transparent and non-scattering, while the alkane-Si QD dispersion is obviously highly scattering (Supporting Information S6-6). To quantifiably assess these dispersions, we estimated the hydrodynamic size distribution of the suspended particles using

8 ACS Paragon Plus Environment Figure 3: Number-weighted size distributions estimated by dynamic light scattering of (a) ester-Si QDs and (b) alkaneSi QDs suspended in methyl methacrylate at concentrations ranging from 0.06 to 0.50 weight percent.

Page 9 of 29 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 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Photonics

dynamic light scattering (DLS). Figure 3 shows the number-weighted size distributions for the two different particle suspensions in MMA, assuming a spherical particle model. An average hydrodynamic diameter of about 4.4 nm and narrow size distribution is estimated for the ester-Si QDs in MMA. No significant concentration effect is present over the range measured (0.06 – 0.50 wt%). Compared to the TEM and XRD size estimations, the DLS distribution is in good agreement and indicates a primarily single-particle suspension of ester-Si QDs in MMA. The DLS results do not indicate a monodisperse suspension of alkane-Si QDs in MMA, however. Because the fitting algorithm assumes a spherical particle model and the agglomerates formed of spherical cores cannot be expected to also be spherical, the data presented for the alkaneSi QDs should not be accepted as an accurate portrayal of the complicated size distribution of an agglomerated suspension. However, the large difference in the size distributions between the two particle systems is strong evidence of differing suspension characteristics. Despite equal Si core diameters and similar ligand lengths, the DLS measurement indicates that the alkane-Si QDs form a suspension that behaves as though it has an average hydrodynamic diameter at least an order of magnitude larger than their ester-capped equivalents. This suggests that the 1-dodecene functionalized Si particles do not form a single particle dispersion in MMA, agglomerating instead to form the observed scattering suspension. This conclusion is supported by the observation that when left undisturbed, some precipitate forms from the concentrated alkane-Si QD suspensions in MMA but not from the ester-Si QD suspensions.

9 ACS Paragon Plus Environment

ACS Photonics 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 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 29

Figure 4: (a) The photoluminescence (solid) and absorption (dashed) spectra of the ester-Si QDs (blue) and alkane-Si QDs (red) dispersed in chloroform. (b) The photoluminescence quantum yield of the ester-Si QDs (blue) and alkane-Si QDs (red) in chloroform, in MMA, in partially polymerized MMA, in PMMA, and in PMMA following the final annealing. (c) The photoluminescence quantum yield of ester-Si QD (blue) and alkane-Si QD (red) PMMA composites as a function of time when stored in a nitrogen purged glovebox (closed symbol) or in air (open symbol).

Bulk PMMA slabs doped with Si QDs were prepared using azobisisobutyronitrile (AIBN) as the radical initiator, which we expect to fully react during the polymerization and leave byproducts with ultraviolet absorption strengths 2 to 3 orders of magnitude lower than that of the Si QDs. The presence of Si QDs appears to have little impact on the PMMA properties. The glass transition temperature and molar mass is similar between undoped PMMA samples and those doped with ester-Si QDs or alkane-Si QDs, as shown in the Supporting Information. The absorption and photoluminescence of the Si QDs are also unaffected by ligand choice or the polymerization

process.

Figure

4(a)

shows

measurements

of

the

absorption

and

photoluminescence spectra of the Si QDs in chloroform after surface functionalization. Both sets of Si QDs exhibit broad photoluminescence, with a spectral peak around 838 nm. Additionally, Figure 4(b) shows the PLQY of both sets of QDs taken at various stages during the polymerization process. The PLQY is approximately 50% before, during, and after polymerization is complete. A thorough review of the errors associated with the PLQY measurement method used in this study is offered by Resch-Genger’s group, who estimate total errors of ±8% in the averaged, absolute value of the PLQY, given good linearity and reproducibility in the detector response55. As the Si QD emission extends to the near-infrared, where detection capabilities are limited, this error estimate is a minimum. 10 ACS Paragon Plus Environment

Page 11 of 29 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 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Photonics

While the Si QDs are robust in the presence of AIBN and polymer annealing temperatures, PMMA proves insufficient as an oxidation barrier. Figure 4(c) presents the PLQY behavior of the composites over time stored in either a nitrogen-atmosphere glovebox or exposed to air. Stable PLQY is exhibited over months of time by the composites stored in nitrogen. However, once the samples are exposed to air, the PLQY noticeably drops by more than 10% in a few days. The oxidation of Si QDs is a well-known issue but is self-limiting56,57. The PLQY decay appears to plateau after about a week of exposure to air, wherein the samples stabilize to a reduced PLQY. These results are in agreement with the Si QD / PMMA composites studied by Sychugov’s group40. Should higher PLQYs be required, Si QD / PMMA nanocomposites would benefit from air and vapor barrier layers, like those typically used in the photovoltaic and organic light emitting devices industry58.

Figure 5: Visible light photographs of ester-Si QD / PMMA composites (left side) and alkane-Si QD / PMMA composites (right side) under (a) sunlight illumination and (b) ultraviolet illumination at particle loadings from 0 to 0.50 weight percent. Each disk is 2.5 cm in diameter.

To study the effects of particle loading in the Si QD / PMMA system, composites were made with 0.06, 0.10, 0.15, 0.25, and 0.50 wt% of either ester-Si QDs or alkane-Si QDs. Visible photographs of the 2.5 cm diameter, 4.4 mm thick samples under sunlight and UV light are presented in Figure 5. The samples are illuminated from below in the UV-lit photograph and a UV filter was used to render only the visible red photoluminescence. Reflection of this photoluminescence on the glass face of the UV lamp is visible. However, a reduced crispness of 11 ACS Paragon Plus Environment

ACS Photonics 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 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 29

text laid behind the samples is observed in Figure 5 for alkane-Si QD samples compared to their ester-Si QD counterparts. This is a sign of increased light scattering. Signs of scattering are not significantly visible until a loading of 0.25 wt% for the ester-Si QD samples while it is already obvious in the lowest concentration sample (0.06 wt%) for the alkane-Si QD samples. For both sample sets, transparency visibly decreases at high concentrations. Specular reflection measurements show similar behavior across all samples, ranging from roughly 2% to 6% (Supporting Information S8-8). This indicates that the surface roughness is comparable across the samples. Nanocomposite scattering properties To quantitatively characterize the light scattering of these slabs, transmission and total reflection spectra were measured from 400 to 1600 nm, as shown in Figure 6. The transmission spectra are corrected for the transmission of an undoped PMMA slab, which was used as the 100% transmission reference. Compared to the PMMA reference slab, transmission decreases either because of Si QD absorption, which occurs for wavelengths less than 800 nm (see Figure 4(a)), or scattering, which can occur over the entire wavelength range. Since Si QD absorption is expected to be nearly zero above 800 nm, we focus initially on the transmission of the near-infrared (NIR) light to discuss the scattering properties of the Si QD / PMMA slabs. For every Si QD / PMMA slab we fabricated, light transmission in the NIR region of the spectrum for the alkane-Si QD sample is significantly lower than the corresponding esterSi QD sample. We expect that this is due to Si QD agglomeration facilitated by the chosen surface species. Agglomeration leads to an increased number of strong scattering centers based on the volume dependence of scattering. Such scattering sites can both forward and backward scatter the incident light within the concentrator, which will decrease the transmittance measured by the detector positioned 180 degrees from the incident light beam. The ester-Si QDs also retain high optical transmission in the PMMA matrix for much higher loading fractions than the alkane-Si QDs. The ester-Si QDs retain high transmission, near 100% for loading fractions up to 0.15 wt%. The alkane-Si QDs, in contrast, do not have complete transmission in the NIR for even the lowest loading fraction of 0.06 wt%. In addition, the ester-Si QDs with a loading fraction of 0.25 wt% still show a greater NIR transmission than the alkane-Si

12 ACS Paragon Plus Environment

Page 13 of 29 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 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Photonics

QDs with a loading fraction of 0.06 wt%. These are promising results for designs that require high absorptivity from the composite combined with minimal agglomeration of the Si nanoparticles. The total reflection spectra in Figure 6(b) are used to assess the extent of scattering at wavelengths shorter than 800 nm. Scattering is important in this spectral range, as the Si QDs exhibit broad photoluminescence from 650 – 950 nm. Total reflection measurements detect contributions from specular reflection, diffuse reflection, as well as back scattering. The total reflection of our undoped PMMA slab is shown for reference in solid black line in Figure 6(b). For all of the nanocomposite samples, the total reflection spectra exhibit a common shape, featuring a maximum in the range of ~580 – 700 nm, depending on the loading fraction. Because

Figure 6: (a) Transmission at normal incidence and (b) total reflection of Si QD / PMMA composites using either alkane-Si QD (dashed line) or ester-Si QDs (solid line) at weight percentages ranging from 0.06 wt% (purple) to 0.50 wt% (red). Transmission is corrected by the PMMA reference while the total reflection is not, but includes the total reflection of the PMMA reference plotted in solid black line.

13 ACS Paragon Plus Environment

ACS Photonics 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 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 14 of 29

the loading fractions of Si QDs are low, the nanocomposite is expected to exhibit a refractive index and corresponding reflectivity following that of the PMMA reference. As a result, we conclude that the high reflection from the nanocomposites in the visible region is due to backscattering from agglomerated Si QDs. At shorter wavelengths, the Si QD absorption (see Figure 4(a)) starts to dominate over the scattering cross section, leading to the minimal reflection at 400 nm for all doped samples. The reflection minima present in all samples in the near-infrared correspond to absorption overtones from the polymer matrix (Supporting Information S8-7). For samples of the same loading fraction, the reflection data shows that ester-Si / PMMA slabs have reduced reflection relative to the alkane-Si / PMMA slabs. Since the slab surfaces exhibit similar specular reflection (Supporting Information S8-8), we conclude that the large differences in reflection are driven by agglomeration-induced scattering. Additionally, for PMMA slabs with 0.06 and 0.10 wt% ester-Si QDs, no extra reflection was measured compared to the undoped PMMA reference. These results suggest that good dispersions of the ester-Si QDs in PMMA are achieved at loading fractions of 0.10 wt% or less. Some additional reflection is present for the 0.15 wt% ester-Si QD / PMMA sample, which was not obvious in the NIR transmission. The increased sensitivity of the total reflection measurement to scattering at shorter wavelengths demonstrates the complementary nature of these two data sets. Along with these beneficial optical properties for LSCs, the ester-Si QD / PMMA composites at loading fractions up to 0.10 wt% also have minimal haze over the visible spectrum, making them useful for window applications (Supporting Information). Finally, we measured the attenuation of photoluminescence intensity as a function of excitation distance from the edge, with these results shown in Figure 7. These measurements probe the influence of light scattering on the trapped luminescent photons. In this experiment, a 405 nm laser was rastered in 0.5 mm steps across the face of the composite material. An integrating sphere was coupled to the edge to collect the Si QD photoluminescence, the spectra of which were fit to a Gaussian distribution for each excitation position. Representative spatial maps of these fits to the photoluminescence spectra as a function of excitation distance away from the edge are shown in Figure 7(a) and (b) for ester-Si and alkane-Si / PMMA samples, respectively, at a loading fraction of 0.10 wt%. Similar figures for each loading fraction are shown in Supporting Information

14 ACS Paragon Plus Environment

Page 15 of 29 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 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Photonics

Figures S11-9, S12-10, and S13-11. No significant peak shift is observed due to the Si QDs’ low reabsorption effects over the measured path lengths. In Figure 7c, the intensity at the wavelength of the peak of the photoluminescence spectrum at each excitation distance was divided by the intensity of the peak of the photoluminescence spectrum at the point where the laser is incident closest to the integrating sphere. The overall shape of the attenuation of the photoluminescence as a function of excitation position derives from the reduction in the solid angle of the collection port as the excitation spot moves farther away, as well as waveguide losses such as reabsorption and/or scattering. We expect the minor reabsorption effects to be mostly unaffected by the choice of capping ligand, but to increase with increasing Si QD loading. We also expect to observe changes in the scattering based on the agglomeration state for the different Si QD surfaces and loading fractions. Ester-Si QD / PMMA samples for loading fractions up to 0.10 wt% and alkane-Si QD / PMMA samples for loading fractions up to 0.06 wt% exhibit similar attenuation of the photoluminescence as the excitation distance from the collection edge increases to 2 cm. The differences between the ligand treatments are more apparent at higher loading fractions, where the ester-Si/ PMMA samples exhibit effective collection of photoluminescence from longer distances. The difference between the two sample sets increases with loading fraction until the highest concentration sample at 0.50 wt%. The quality of the fit in this case is reduced, however, since

Figure 7: The attenuation of emitted light in Si QD / PMMA composites excited by a 405 nm laser rastered from 0 to 20 mm from the collection edge. The attenuation map of (a) ester-Si QD and (b) alkane-Si QD / PMMA composites at 0.10 wt% particle loadings. (c) Peak emission intensity of attenuating light through ester-Si QD / PMMA slabs (solid lines) and alkane-Si QD / PMMA slabs (dashed) for particle loading fractions from 0.06 wt% (purple) to 0.25 wt% (orange). A calculation of the dependence of the solid angle for attenuation is shown in black.

15 ACS Paragon Plus Environment

ACS Photonics 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 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 16 of 29

transmission through the slab is nearly zero after a distance of 4 mm. Because of this, we cannot characterize the attenuation effectively out to 20 mm and do not include the 0.50 wt% loading fraction in this analysis. Based on these results, we conclude that ester-Si QDs result in longer photoluminescence propagation lengths and are more useful for light concentrating applications when using PMMA as the polymer matrix. Outlook for LSCs Using a combination of the same Monte Carlo models as discussed in Figure 1 and modifying a quality factor analysis found in Ref. 59, we make predictions about the limiting performance and sizes of these concentrators. Assuming that the Si nanocrystals are perfectly dispersed so that the only scattering derives from the isolated nanocrystal scattering limit, we find that among the nanocrystal concentrations we experimentally studied, a loading fraction of 0.10 wt% leads to the highest concentration factor limits for both 50% and 100% quantum yield cases in 3 mm thick devices. The concentration factor limit is estimated to be 21x for an 11 m concentrator edge length with 50% PLQY, and 103x for a 26 m concentrator edge length with 100% PLQY (Supporting Information S15-12). If we instead consider devices scaled for a residential window (3 mm thick slab, 1 m edge length), then the optimal loading for a nonscattering LSC is ~0.06 wt% with a PLQY of 50%, and ~0.1 wt% for QDs for a PLQY of 100% (Supporting Information S16-13). We also note that these 3 mm thick optimized devices have high visible light transmission, slightly exceeding our experimental, approximately 4 mm thick slabs as shown in Figures 5 & 6. From these predictions, we expect that it is important to have a good dispersion of Si QDs in PMMA for loading fractions up to ~0.1 wt% for large LSC applications. However, the difference in the quality of dispersions between ester-Si and alkane-Si composites is still useful at higher loading fractions as other composite applications may have different loading requirements due to different polymer thicknesses. The characterization of these composites indicates that the ester-Si QD / PMMA composites are well-suited for LSC applications, whereas the alkane-Si / PMMA composites are not. At a loading fraction of 0.10 wt%, the ester-Si QD / PMMA composites have transmission near 100% in the near infrared when adjusted for the extinction of a PMMA reference, have total reflection properties that mimic the blank PMMA reference, and exhibit improved photoluminescence propagation lengths when compared to the alkane-Si QD / PMMA composites. 16 ACS Paragon Plus Environment

Page 17 of 29 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 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Photonics

The alkane-Si QD / PMMA composites exhibit significant scattering that lead to increased diffuse reflection, incomplete transmission of unabsorbed light, and interrupt propagation of luminescent light toward the edge of the concentrator across the range of loading fractions that are appropriate for large-area LSCs. Conclusions In this work, we investigate the optical properties of Si QD / PMMA composites formed via free radical polymerization. We find that Si QDs protected by ligands attached via hydrosilylation exhibit excellent stability in the presence of the thermal initiator, AIBN. The nanocomposites have long term stability in their photoluminescence when exposed to air, although the quantum yield experiences an initial decay of roughly 20% from the initial values. We conclude that PMMA alone is not sufficient for protecting the oxidation-prone Si QDs. Other oxygen barriers will be required for applications demanding the highest quantum yields. For window LSC applications, this may take the form of inserting the Si QD LSC as a third pane inside a double pane window7, which are frequently filled with argon. Because our simulations show that scattering losses can significantly reduce Si QD LSC performances, we also establish a strategy for improving the dispersion of the Si QDs in PMMA compared to previously published methods. We find that using a ligand with some structural similarity to the repeating polymer unit can facilitate the QD dispersion within the polymer matrix. In the Si QD / PMMA system, the ester group in methyl 10-undecenoate is more similar to MMA than an analogous unfunctionalized alkane. Correspondingly, ester-Si QDs exhibit improved dispersion relative to alkane-Si QDs. The fine dispersion of ester-Si QDs within the monomer translates into a bulk polymer composite with lower scattering losses even at higher loading fractions as compared to alkane-Si QD nanocomposites. The loading of Si QDs is not limitless, however, as increased scattering is observable for even the ester-Si QD / PMMA composites at and above 0.15 wt% Si. Based on concentration factor limit and Monte Carlo models, we predict that a loading fraction up to 0.1 wt% is necessary to achieve high concentration factors for windowscale concentrators made with 3 mm thick polymer sheets. Therefore, we conclude that ester-Si / PMMA composites show good promise for low-scattering window-scale LSCs while alkane-Si / PMMA composites do not due to significant scattering losses.

17 ACS Paragon Plus Environment

ACS Photonics 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 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 18 of 29

Methods Si QD synthesis Silicon quantum dots were synthesized in a nonthermal plasma from silane as described previously47. For these experiments, 30 standard cubic centimeters (sccm) of Ar were flowed with 14 sccm of 5% silane in helium through a ¼” borosilicate glass tube which expanded to 1” with an injection point for 100 sccm of hydrogen added for surface passivation purposes. Power was applied nominally at 45-50W through an impedance matching network to two concentric electrodes placed 4 cm above the tube expansion at a frequency of 13.56 MHz. The visible plasma length after the expansion point in the tube was maintained around 2.6 cm. The plasma pressure was maintained at 1.4 – 1.5 Torr via an adjustable orifice located downstream of the plasma, which also served to accelerate the nanocrystals for collection by impaction upon glass substrates60. Si QD functionalization Ligands were grafted to the Si QDs to improve photoluminescence and stabilize them in solutions. We compared two different ligands: a nonpolar ligand, 1-dodecene (TCI Chemicals #D0974), and a bifunctional ligand with an ester end group, methyl 10-undecenoate (Millipore Sigma #115126). To carry out functionalization, the as-produced Si QD powder was transferred from the plasma reactor via a load-lock system into a nitrogen filled glovebox maintained with