Stimuli-Responsive Langmuir Films Composed of Nanoparticles Decorated with Poly(N-isopropyl acrylamide) (PNIPAM) at the Air/Water Interface

The nanotechnology shift from static toward stimuli-responsive systems is gaining momentum. We study adaptive and responsive Langmuir films at the air/water interface to facilitate the creation of two-dimensional (2D) complex systems. We verify the possibility of controlling the assembly of relatively large entities, i.e., nanoparticles with diameter around 90 nm, by inducing conformational changes within an about 5 nm poly(N-isopropyl acrylamide) (PNIPAM) capping layer. The system performs reversible switching between uniform and nonuniform states. The densely packed and uniform state is observed at a higher temperature, i.e., opposite to most phase transitions, where more ordered phases appear at lower temperatures. The induced nanoparticles’ conformational changes result in different properties of the interfacial monolayer, including various types of aggregation. The analysis of surface pressure at different temperatures and upon temperature changes, surface potential measurements, surface rheology experiments, Brewster angle microscopy (BAM), and scanning electron microscopy (SEM) observations are accompanied by calculations to discuss the principles of the nanoparticles’ self-assembly. Those findings provide guidelines for designing other adaptive 2D systems, such as programable membranes or optical interfacial devices.


S4
The analysis of DLS measurements of the diluted sample (1 mg·ml -1 ) revealed that the average size (by volume) was similar at both experimental temperatures (100 nm and 90 nm at 40 °C and 20 °C, respectively). However, the standard deviation of the sample at 40 °C was higher (35 nm vs. 31 nm). The expected change in the size of PSiFe due to changes in the conformation of PNIPAM was in the range from around 10% to around 20%, and it was not visible in DLS measurements, most likely due to the polydispersity of particles. SEM confirmed the polydispersity of the cores (Figure S1).
We do not treat those average sizes as differentiating compared to the other results and experiments. Our analysis of Langmuir isotherms proved a change in the size of PSiFe upon temperature changes. Also, the average distances between the surfaces of cores were much smaller for HT compared to LT (cf. Figure S11). However, when we analyzed the distances between the centers of the cores, we did not observe any significant differences between both temperatures due to the polydispersity of the cores.
The decrease of surface zeta potential (Table S1) due to the temperature change matched the theory of the stability of PSiFe nanoparticles. Stable colloid below LCST became less stable above the LCST of PNIPAM. In "closed" conformation, PNIPAM became more hydrophobic, thus having a smaller absolute value of zeta potential at HT (around -42 mV) than at LT (around -55 mV).

The stability of PSiFe films at the interface
Data for three consecutive compression/decompression cycles without changing the temperature regimes in between cycles (as complementary to Figure 1 d) in the main text) is shown in Figure S4. Isotherms at HT were reproducible (left). Cycles at LT (right) were poorly reproducible, which reflected the tendency of PSiFe to aggregate in this temperature regime. Figure S4. Three consecutive compression-decompression cycles of PSiFe films at HT (left) and LT (right). Isotherms at HT were reproducible. However, at LT, some irreversible aggregation occurred.
We judiciously chose 7.5 mN•m -1 to show the linear relationship between the deposited volume of PSiFe suspension and surface area (Figure 1). We did not follow the standard estimation of contact cross-sectional area (CCSA). This was done because the surface pressure increase started early but with a small slope. Even minuscule changes in inputs resulted in a considerable variation in CSSA. Analogical calculations were conducted for 0.5 mN•m -1 , 1.0 mN•m -1 , 2.5 mN•m -1 , and 5.0 mN•m -1 leading to the same conclusion.

S6
The compression isotherms were recalculated to unify the amount of sample deposited on the interface (Figure S5 a) b)). The three isotherms for both temperatures were close to each other.
However, adding consecutive portions of PSiFe resulted in a relatively small shift towards smaller surface area values. These deviations arose from consecutive compression/decompression cycles.
Upon compression, a small number of particles could drown into the subphase, or aggregates could be formed, which did not entirely spread upon decompression.

Surface potential and dipole moment
Surface potential was measured simultaneously with the surface pressure. The potential rises slowly and almost evenly at the HT with the surface pressure (Figure S6 a)). Close to the maximal compression of the film, a rapid increase in the surface potential was observed. This might be related to the artifact caused by the proximity of the barriers (having paramagnetic parts) to the Kelvin electrode. However, this sudden jump in surface potential was not visible at LT (Figure S6 b)).
S7 Figure S6. The surface potential (vivid colors) at high (a) and low-temperature regimes (b) and surface pressure isotherms of those compression-decompression cycles (pale colors).
We calculated the surface potential and dipole moment differences between temperature regimes based on the experimental data presented in Figure S6. These differences are shown in Figure 1 d) (main text). The area per nanoparticle was fixed according to the size of the nanoparticles, and the close-packed hexagonal structure of cross-sections was assumed. The dielectric permittivity (ε) of the layer changed linearly from 1 to 3.9 (the value for SiO2), consistent with the area per nanoparticle change. That was used to compensate for the unknown parameter of the monolayer permittivity, which changed in time upon the compression. The calculations were carried out to illustrate the dipole moment's difference. The positive value indicated that the dipole moment (perpendicular to the interface) of PSiFe was higher at LT.

Rheological measurements
The monotonic increase of surface compressional modulus was observed upon compression at both temperature regimes. The inflection point was not reached, suggesting we operated below the S8 collapse. This was important to explore the dynamic properties of the films without the risk of irreversible aggregation. for LT). We assumed that allowing the system to equilibrate longer (60 min) was enough, but it was visible in the Lissajous diagrams that the small drift was still present. At HT, it was most likely related to the evaporation of water at a relatively higher temperature. Small changes in water level reflected in the Lissajous diagrams. In other words, there was a background, and the depreciation discussed by Hilles et al. 1 and Morioka and Kawaguchi 2 was not intrinsic to the system.

S9
We performed analysis according to Hilles et al. 1 to prove that the oscillatory barrier method experiments were performed in the linear regime. In Figure S7,

The trapping energy
We assumed that PSiFe particles were small enough, so they were irreversibly trapped at the interface. Here, we provide both theoretical and experimental explanations for such an assumption.
Applying the same calculations to their system, the absolute values of the entrapment energy at 20 °C were approximately equal to 19 kT for 20° and 675 kT for 130°. 20° and 130° are extreme contact angles tested by the authors, while the optimum contact angle leading to the highest adsorption energy was equal to 90°1 0 .
In this method, other capillary forces are neglected. For particles smaller than 5-10 μm with Eötvös number not much less than 1, the deformations of the interface due to the gravity are negligible as well as lateral flotation 5,11 .

Experimental explanation
Consecutive addition of deposited sample caused an almost linear shift of the isotherms (Figure 1 a), b) and Figure S4 a)

S15
The origin of the temperature-induced shifting of the isotherms was the change in the ligand structure from "open" at LT to "closed" at HT. Additional analysis of SEM pictures clearly showed that the particles at LT were further apart compared to HT (Figure S11). The analysis was done using ImageJ, and the distance between closest neighbors was measured. All particles further away than 30 nm were neglected. and "open" (below LCST), respectively.
We also performed additional SEM imaging of Langmuir-Blodgett films transferred at HT ( Figure S12) and LT (Figure S13), both at low (5 mN•m -1 ) and high (25 mN•m -1 ) surface pressures. PSiFe particles were dispersed over the area at low surface pressure and HT (see inset in Figure S11). At low surface pressure, most PSiFe formed large islands at LT, which corresponded to condensed regions recorded in Figure S10. At high surface pressure, films at HT S16 were more uniform than LT (which was in line with observations at 9 mN•m -1 ). Both at HT and LT, multilayer formation was visible but without any sign of collapse at the surface pressure isotherm curves (cf. Figure 1 d)) in the main text). We assumed that transitioning from 2D to 3D was not a sharp event but rather a continuous process that started above some threshold value (larger than around 10 mN•m -1 and lower than 25 mN•m -1 ).

Magnetic properties
The NIMA Technology set was used to transfer the monolayer with or without the neodymium magnet. The silicon wafer was placed in the dipper and partially immersed in the subphase.
Depending on the variant, the neodymium magnet was partially submerged next to it, providing the constant and strong magnetic field (approx.. 0.17 T for LT and 0.12 T for HT in the proximity of the Si substrate) during the Langmuir monolayer creation and its transfer onto the Si substrate.

S18
PSiFe nanoparticles (500 μl) were deposited on the water surface at 40 °C using the glass rod and the Hamilton syringe. Then, the system was cooled down to the temperature of 20 °C. The The results were not consistent nor reproducible. In the number of attempts, we observed anisotropically orientated domains ( Figure S14).
It is crucial to underline that the precipitated PSiFe (i.e., after a month of storage) were responsive to the same magnets, and their movement was undeniably visible (the precipitate color was brownish). S19 Figure S14. The magnetic domains of PSiFe were observed during the evaporation of the nanoparticles' dispersion in the magnetic field or Langmuir-Blodgett films transferred in the external magnetic field. a) lower magnitude; b) and c) higher magnitude.

Isotherms of non-thermo-responsive materials
Films of non-thermoresponsive particles or amphiphiles compressed to the same surface pressure but at a higher temperature should have a larger surface area due to thermal expansion.
That is, the isotherm would be shifted to the right. Suppose a plateau or inflection on the isotherm indicates a phase transformation. In that case, Clapeyron's equation for phase transformations in 2D shows that if (dp/dT) is positive (only for water, ammonia, and a few metals is negative), then with increasing temperature, the pressure at which the transformation occurs is higher. The collapse occurs at lower surface pressure if the phase transformation is not there. The film collapses more readily at higher T due to thermal fluctuations. This was observed in the literature before 12 .
We performed a control experiment by compressing stearic acid at LT and HT (Figure S15). pressure values and close to collapse. In this regime, the size of the molecule, which did not change with temperature, was the limitation; ii) at lower surface pressure values, the isotherm was shifted to the right at the higher temperature due to thermal expansion.

The temperature gradient in the system does not cause instabilities
We performed computer simulation using the finite element method by employing Comsol software. We found that in the system, the temperature gradient was minimal (the largest we observed was 0.1 K between the bottom and the top of the trough). This was because it was not only the water heated up but also the air in the proximity of the surface. With such a small temperature gradient, we neglected the impact of convection on the observed phenomena.

S21
We simulated a PTFE block with 5 mm-thick walls and a 1 mm-thick bottom and heated/cooled from the bottom (40 °C/20 °C). The ambient temperature was set to 23 °C. We put the container into the large enclosure filled with air ( Figure S16, right upper panels). The system was equilibrated for 60 minutes (blue line) and 90 minutes (green line).
The most pronounced gradients were found when the trough was fully filled with water, i.e., the depth was 9 mm (Figure S16 Next, we calculated Rayleigh numbers for the system to establish if Rayleigh-Benard cells could form. The highest estimated number was around 1343 (for the situation depicted in Figure S16).
This result corresponded to 60 min of equilibration. Upon additional 30 minutes of equilibration, the Rayleigh number decreases to around 309. The obtained Rayleigh values were below the critical value, i.e., around 1700, above which convection cells appear 13 . Therefore, we neglected the small temperature gradient, as is generally accepted by Langmuir trough users.
The values used for calculating the Rayleigh numbers were: width of the trough 50 mm, depth of the water subphase 9 mm, and thickness of the PTFE layer between the heating element and water subphase 1 mm. Temperature-specific parameters were as follows: