Single-Photon Emitting Arrays by Capillary Assembly of Colloidal Semiconductor CdSe/CdS/SiO2 Nanocrystals

The controlled placement of colloidal semiconductor nanocrystals (NCs) onto planar surfaces is crucial for scalable fabrication of single-photon emitters on-chip, which are critical elements of optical quantum computing, communication, and encryption. The positioning of colloidal semiconductor NCs such as metal chalcogenides or perovskites is still challenging, as it requires a nonaggressive fabrication process to preserve the optical properties of the NCs. In this work, periodic arrays of 2500 nanoholes are patterned by electron beam lithography in a poly(methyl methacrylate) (PMMA) thin film on indium tin oxide/glass substrates. Colloidal core/shell CdSe/CdS NCs, functionalized with a SiO2 capping layer to increase their size and facilitate deposition into 100 nm holes, are trapped with a close to optimal Poisson distribution into the PMMA nanoholes via a capillary assembly method. The resulting arrays of NCs contain hundreds of single-photon emitters each. We believe this work paves the way to an affordable, fast, and practical method for the fabrication of nanodevices, such as single-photon-emitting light-emitting diodes based on colloidal semiconductor NCs.


Additional morphological data and analysis
In Figure S1 we show the size histograms of bare CdSe/Cds NCs (pink bars) and CdSe/Cds covered by the SiO2 shell (red bars). Upon SiO2 shelling, the average NC size increases from 11±5 nm to 35±3 nm.

Colloidal suspension dropcast and capillary assembly process discussion
In Figure S2 we show a basic sketch of the principle of capillary assembly. The evaporation of the solvent of the colloidal droplet drags the particle to the meniscus due to a chemical gradient and subsequent convective solvent flow created by a faster evaporation rate at the border of the droplet.
At the meniscus interface with air and the substrate, the colloidal particles experience two main forces: i) a force parallel to the substrate that drags the particles towards the center of the droplet (due to the diminishing size of the droplet upon evaporation) and ii) a normal force with respect to the substrate which pushes particles against the solid surface. The balance between these two forces is a function of the colloidal solution contact angle (CA) with the substrate. Its value is critical for a successful capillary assembly (putting the particles inside pre-patterned features) or convective assembly (creating monolayers or 2D structures). [1][2][3] There are more complex particles flows inside the droplet (such as Marangoni flows) and many more relevant parameters for the 3 control of the process but this discussion goes beyond the scope of this work. In the main manuscript, we addressed relevant literature that the interested reader can refer to. 4,5 Figure S2 -Simple sketch of forces and particle flows involved in the drying of a droplet of colloidal solution.
In our experiments, we kept fixed the solid content inside the solution used for the dropcast to about 0.01 %, into a range that is reported in literature as a reasonable value for metal and polymeric sub-100 nm nanoparticles to favor the creation of a nanoparticles accumulation zone at the meniscus interface that avoids the hindering of the assembly by Brownian motion. 3 Preliminary optimization studies led us to choose a drying temperature of 45 °C to avoid coffee ring effects appearing at lower temperatures, thus preventing particles to confine and localize themselves on the outer border of the drying droplet, outside the hole patterned areas. 4,5 Higher temperatures led to an excessively fast meniscus drying speed, faster than reasonable ranges reported in literature and to chaotic dried patterns. We focused then on the exploration of the CA experimental parameter to optimize as much as possible the NCs hole filling efficiency.

RCA (°) LCA (°)
Untreated PMMA 70 ± 1 70 ± 1 O2 plasma (100 W -10 s) 42 ± 5 42 ± 5 largely dominates and capillary assembly is not achieved (Fig. S3c). The droplet dries into a small concentrated dot, a well-known dried colloidal drop shape in literature when the solution-surface interaction is hydrophobic. By lowering the CA with a plasma treatment to about 42° (Table S1, Figure S3b) the droplet dries over a much larger, coffee stain free area. Some holes are filled with NCs (also single NCs) but we observed very poor filling efficiency, and multiple stray NCs and agglomerates outside the holes are present (Fig. S3d) because of an excess of normal force. The standard deviation of the O2 treated PMMA CAs is very high (Table S1); the treatment is not homogenous over the sample surface and often quite asymmetric droplets are found. Moreover, this CA value is not very tunable with O2 plasma treatment and time, and it etches the PMMA resist leading to even more scarce overall reproducibility of the experiments. We then opted to tune the CA by adding a varying ethanol volume percentage to the NC aqueous solution. As shown in The presence of ethanol surely complicates and possibly increases the phenomena behind the droplet drying process (e.g. changing the Marangoni flows inside the drying solution), but it certainly proved to be a versatile way to tune the crucial CA parameter and achieve a capillary assembly with almost maximum efficiency allowed by the nanoholes size: a Poisson distribution.
In Fig. S4a,d we show how adding 5% ethanol leads to results similar to pure water where the CA is too high and no NC assembly takes place. In Fig. S4b,e we show how, by lowering the CA via addition of 15% of ethanol to the water solution, the normal force exerted on the NCs on the droplet is able to trap them in the nanoholes. The drag force is still sufficient to have a clean surface. In Fig. S4c,f instead the amount of ethanol of 25% is too high, normal force dominates leading to the random deposition of 2D assemblies of NCs.

Additional SEM images of the nanoholes arrays
In Figures S5 and S6 we show additional SEM images of the NCs filled nanoholes array. Figure   S6 shows how filling by multiple NCs can lead to clusters only partially filling the holes.

Optical characterization of a single NC
The intensity-time trace in Figure S7a shows flickering as opposed to telegraphic blinking  Figure S8 shows a refined measurement of the saturation excitation intensity for a SPE with g(2)(t = 0) = 0.29. The saturation curve ranges from 0 to 50 nW (117 W/cm 2 ) in 2.5 nW steps. The fit confirms the saturation intensity to be Is = 54 W/cm 2 .

Comparative emission spectra of SPE with different NCs in the nanohole
To identify the nature of the PL, an emission spectrum from nanoholes #1 and #3 was measured and fitted with a Lorentzian function yielding FWHM of 90 and 100 meV, respectively ( Figure   S9). This confirms that both SPEs originate from the decay of a single two-level system, most likely the exciton, regardless of the number of NCs present in the nanohole.