In Situ Tracking of Colloidally Stable and Ordered Assemblies of Gold Nanorods

Solution-phase self-assembly of anisotropic nanoparticles into complex 2D and 3D assemblies is one of the most promising strategies toward obtaining nanoparticle-based materials and devices with unique optical properties at the macroscale. However, controlling this process with single-particle precision is highly demanding, mostly due to insufficient understanding of the self-assembly process at the nanoscale. We report the use of in situ environmental scanning transmission electron microscopy (WetSTEM), combined with UV/vis spectroscopy, small-angle X-ray diffraction (SAXRD) and multiscale modeling, to draw a detailed picture of the dynamics of vertically aligned assemblies of gold nanorods. Detailed understanding of the self-assembly/disassembly mechanisms is obtained from real-time observations, which provide direct evidence of the colloidal stability of side-to-side nanorod clusters. Structural details and the forces governing the disassembly process are revealed with single particle resolution as well as in bulk samples, by combined experimental and theoretical modeling. In particular, this study provides unique information on the evolution of the orientational order of nanorods within side-to-side 2D assemblies and shows that both electrostatic (at the nanoscale) and thermal (in bulk) stimuli can be used to drive the process. These results not only give insight into the interactions between nanorods and the stability of their assemblies, thereby assisting the design of ordered, anisotropic nanomaterials but also broaden the available toolbox for in situ tracking of nanoparticle behavior at the single-particle level.


Supplementary Note 1: Reversibility of the assembly/disassembly process
To test the reversibility of assembly/disassembly process we performed UV/Vis analysis of MUDOL-functionalized gold nanorods in the assembled/dispersed states. First, assembled NRs were prepared. Disassembly of NRs was achieved by heating and ultrasounds. The latter was used to increase the kinetics of the process, as explained in the manuscript. UV/Vis analysis of the redispersed material ( Figure S2) revealed plasmonic band maxima characteristic to well dispersed AuNR@MUDOL. We also analyzed the influence the number of cycles on the disassembly performance. These measurements revealed that in consecutive measurements the longitudal plasmonic band exhibits redshift and decrease of intensity (4 nm and 20%, respectively, after 3 cycles). These changes can be attributed e.g. to lowering of the disassembly efficiency (partial clustering of the sample) and/or irreversible absorption of NRs on the glass vial and/or partial degradation of the material.

Supplementary Note 2: Electron beam induced thermal and charging effects
To estimate the electron beam induced thermal effect during STEM measurements we turned to used an equation known from the literature S1,S2 : where ∆ -estimated temperature change Ibbeam current dE/dxthe energy loss rate per electron thermal conductivity a, d -estimated dimensions of the Field of View. Based on the conditions we used in our experiments the effects should be below 1K.

Supplementary Note 3. MD-simulated stabilization of neutral small, partially submerged AuNR@MUDOL
To understand the self-assembly of AuNRs, seven neutral small AuNR@MUDOL were simulated in a 400×400×200 Å 3 water box. AuNRs were placed at a center-to-center distance of ~8 nm from each other in a hexagonal pattern and were partially submerged. MD simulation was performed in an NVT ensemble at T = 300 K. The system was equilibrated for 15 ns when the distance between the AuNRs stopped changing (for 2 ns) and the AuNRs were mostly submerged in water, as shown in Fig. S5a. More precisely, AuNRs tips, from one side, were exposed to vacuum (above water surface), while from the other side tips were fully submerged.
This system provided a starting point for the MD simulations of thermally and charge driven disassembly, which was done by transferring the equilibrated AuNRs cluster to water or glycerol, with an additional top layer of solvent. view of the small, partially exposed NRs in water at 25 °C equilibrated system at 15 ns simulation. NRs, ligands, and water are shown in yellow, grey, and blue, respectively.

AuNR@MUDOL submerged in glycerol
We modeled the disassembly of seven AuNR@MUDOL submerged in glycerol. We took the self-assembled system in Fig. S5 and submerged them in 320 x 320 x 280 Å 3 glycerol box. We did not observe the disassembly process in a given simulation time (~15 ns) in contrast to analogous simulation performed at 150 °C (Fig. 5c,d in the main text).

Supplementary Note 5. MD-simulated charge-driven disassembly of small AuNR@MUDOL
Here, we modeled the disassembly of seven charged (20 e and 50 e each) AuNR@MUDOL partially and fully submerged in water at 25 °C. We started with the self-assembled system as shown in Fig. S5 and charged the AuNRs with 20 e. Within 30 ns simulation time, we did not observe the disassembly of the AuNRs (Fig. S7a-b). For a fully submerged system, we took the final configuration from a previous simulation (Fig. S7a) and added a 5 nm layer of water, to cover the exposed AuNR@MUDOL part. Then, we continued the simulation for another 15 ns, which still did not show any disassembly of the AuNRs (Fig. S7c-d). This methodology allowed us to shorten the total simulation time required. When the self-assembled system in Fig. S5 was taken and charged with 50 e, the AuNRs disassembled within 30 ns, regardless of whether the AuNR@MUDOL were partially exposed ( Fig. S7e-f) or fully submerged in water (Fig. 5f,g).
The highly charged AuNRs, which were partially submerged (tips from one side exposed above the water surface) separated in a tilted manner (Fig. S7e-f), with the top parts always closer to each other than the lower parts, since their ligands couple stronger in a vacuum, in contrast to the fully submerged system AuNRs which exhibit a trembling behavior observed for experimental cases (Fig. 5g,h).

Supplementary Note 6. MD-simulated temperature driven disassembly of small AuNR@MUDOL
Here, we modeled the disassembly of seven neutral AuNR@MUDOL partially and fully submerged in water at 100 °C. When the neutral system as shown in Fig. S5 was equilibrated for 35 ns at 100 °C (Fig. S8a), the center-to-center distance between AuNRs changed from ~3.0 to 4.3 nm, as shown in Fig. S8b. Even though the system seems to gradually disassemble, the AuNRs stay close to each other, keeping the orientational and hexagonal order for a longer period than in the case of higher temperatures (simulation performed at 150 °C in a bulk glycerol, Fig. 5d,e). To model the disassembly of seven AuNR@MUDOL submerged in water, we took the final configuration of the system shown in Fig. S8a and added a 5 nm layer of water to cover the exposed AuNR@MUDOL surface, so that AuNRs were fully submerged. We ran the simulation in NPT ensemble for the 15 ns and we observed that the center-to-center distance between AuNRs changed from ~4.3 to 4.5 nm, as shown in Fig. S8c-d. This methodology allowed us to shorten the total simulation time required. With these simulations we confirm that thermal effects can help in the disassembly process.  Movie S1 Figure 3b WetSTEM recording of an AuNR@MUDOL vertical aggregate, showing a single nanoparticle detaching from the aggregate.
Movie S2 Figure 3d WetSTEM recording of an AuNR@MUDOL vertical aggregate, showing an example of AuNR cluster displacement.
Movie S3 Figure