Pickering Bubbles as Dual-Modality Ultrasound and Photoacoustic Contrast Agents

Abstract

Microbubbles (MBs) stabilized by particle surfactants (i.e., Pickering bubbles) have better thermodynamic stability compared to MBs stabilized by small molecules as a result of steric hindrance against coalescence, higher diffusion resistance, and higher particle desorption energy. In addition, the use of particles to stabilize MBs that are typically used as an ultrasound (US) contrast agent can also introduce photoacoustic (PA) properties, thus enabling a highly effective dual-modality US and PA contrast agent. Here, we report the use of partially reduced and functionalized graphene oxide as the sole surfactant to stabilize perfluorocarbon gas bubbles in the preparation of a dual-modality US and PA agent, with high contrast in both imaging modes and without the need for small-molecule or polymer additives. This approach offers an increase in loading of the PA agent without destabilization and increased thickness of the MB shell compared to traditional systems, in which the focus is on adding a PA agent to existing MB formulations.

Graphical Abstract

INTRODUCTION

Emulsions stabilized by particle surfactants have shown favorable thermodynamic stability compared to emulsions stabilized by small molecules. This work explores, for the first time, the use of solid surfactants for stabilization of a gas core and their application as an ultrasound (US) and photoacoustic (PA) imaging contrast agent. The use of particles to stabilize fluid–fluid interfaces has garnered much attention because of higher stability against coalescence compared to droplets stabilized by small-molecule surfactants.^1,2 The presence of particles at the interface of two immiscible liquids provides steric hindrance against coalescence by providing a physical barrier between the two liquids, with the energy required for a particle to desorb (i.e., six orders of magnitude of energy is required to remove the solid surfactant from the oil–water interface) from the interface significantly higher than that of small molecules, such as phospholipids.^3–5 Indeed, Pickering emulsions (those which utilize particle surfactants) have been prepared using water–oil, oil–oil, ionic liquid–water, and ionic liquid–oil systems.^1,2,6–10 A variety of different particles have been used as surfactants in Pickering emulsions, including clay nanosheets, modified silica particles, metal oxide particles, chitosan, carbon nanotubes, hydroxyapatite, and graphene oxide (GO).^11–17 Particle surfactants can impart conductivity, magnetic susceptibility, PA properties, and so forth to the droplets and emulsions, depending on the chemical composition and size of the particles used.^18–20

Particle-stabilized emulsions and foams are prevalent in industrial applications such as personal care, food, and oil and mineral processing; in contrast to Pickering emulsions, the preparation and application of Pickering bubbles or foams are underdeveloped.^21–23 Pickering bubbles are especially attractive for biomedical applications, but require that suitable particle compositions and fluid–gas interfaces are identified.^24–26 Murray et al. produced Pickering bubbles of air stabilized in pure water using partially hydrophobized quasispherical silica nanoparticles (diameter of 20 nm) to stabilize the air–water interface.²⁷ Alternatively, Gauckler et al. utilized surface-modified metal oxide particles (e.g., Al₂O₃ and ZrO₂) to stabilize air/water foams, producing bubbles that did not undergo rapid disproportionation, drainage, and coalescence.²⁸ Although these limited examples illustrate the ability to form Pickering bubbles, advances in the types of particles that can be used to stabilize these systems are required to fully explore and exploit their distinct properties.^24–26

Of the potential particle surfactants, GO nanosheets are particularly attractive for stabilization of bubbles: the particles’ aspect ratio can be controlled, the surface energy can be modified via physical or covalent functionalization, and the nanosheets are multifunctional. Moreover, when GO nanosheets are adsorbed at an interface, they have highly confined rotation and require a large amount of energy to be desorbed, owing to the high aspect ratio (i.e., they are expected to lie parallel to the interface, overlapping side to side and face to face).¹⁷ GO has been previously utilized in biosensors, stem cell differentiation, cancer treatment, gene and drug delivery, biological imaging, and photothermal therapy.^31–33 Whereas the use of GO nanosheets to stabilize gas bubbles has received little attention, this platform may be attractive for biomedical applications because it exploits known benefits of acoustics and photoacoustics in one platform. This multifunctional construct can be beneficial for diagnostic and therapeutic applications. For example, the bubbles can serve as highly sensitive dual-modality contrast agents for US and PA imaging (as demonstrated below). They can also be used for high-efficiency ultrasound-guided therapy, as the particle stabilizers can be covalently modified to load therapeutic compounds and deliver them on demand, while taking advantage of thermal and acoustic bioeffects to increase the therapeutic effect. Furthermore, much like Pickering emulsions, Pickering bubbles have distinct stability and opportunities to produce multifunctional and stimuli-responsive structures, which are not possible with small-molecule or polymer surfactants.^26,34–37

Dual-modality US and PA imaging has been used to acquire both anatomical structure of tissues and the corresponding tissue optical absorption.^38,39 US imaging operates by sending high-frequency sound from a piezoelectric transducer through tissue, which then scatters the sound back to the transducer.^40–42 PA imaging, on the other hand, operates by sending nonionizing electromagnetic waves into biological tissue and observing acoustic waves generated from the thermoelastic expansion of tissue structures containing chromophores.^43–45 These chromophores can be endogenous, for example, blood, melanin, and water, or exogenous, for example, nanoparticles and dyes.^46–51 US imaging is among the most widely used and safest modalities for visualizing organs and PA methods are of the fastest growing ones for biomedical imaging.⁵² The development of materials and structures that can be used for both US and PA imaging may help advance imaging capabilities beyond current systems.

Several approaches to prepare dual-modality US and PA contrast agents have been reported, yet nearly all load a PA agent (such as an organic small molecule dye) on the surface or within the membrane of the US contrast agents (UCAs).^53–55 UCAs are typically microbubbles (MBs) 1–10 μm in diameter, containing hydrophobic gases, such as sulfur hexafluoride (SF₆) and octafluoropropane (C₃F₈), which are stabilized by a polymer, protein, or phospholipids.^56–60 For example, Moon et al. prepared a dual-modality US and PA agent from a phospholipid mixture with a small fraction of the phospholipid conjugated with a porphyrin, a well-known PA agent.⁶¹ Das et al. generated polymer-stabilized nitrogen MBs loaded with the PA agents methylene blue and black ink, using a flow-focusing junction-based microfluidic device.⁶² More recently, Toumia et al. loaded the surface of a polyvinyl alcohol-stabilized air bubble with graphene, using the nanosheets as the PA agent.⁶³ These approaches are limited by the amount of PA agents that can be loaded onto the surface of the UCA bubbles; attempts to increase PA agent loading lead to destabilized or an increase in thickness of the membrane such that a high peak negative pressure may be required to elicit the desired nonlinear activity. As such, an ideal system for a dual US-PA imaging system can be found in UCAs, in which the hydrophobic gas bubble is stabilized solely by PA agents.²⁶

Herein, we show for the first time that bubbles of C₃F₈ stabilized solely by functionalized and partially reduced GO nanosheets behave as dual-modality US and PA agents, with high contrast in both imaging modes and without the need for small molecules or polymers.

RESULTS AND DISCUSSION

To access these particle surfactants, the surface energy of the GO nanosheets was tuned by functionalization and partial reduction with an alkylamine (e.g., 3-aminopentane, C5). As illustrated in Scheme 1, the Pickering bubbles were prepared by dissolution of functionalized GO in water and subsequent activation by amalgamation in the presence of perfluorocarbon gas (C₃F₈). We compare the use of “normal” GO nanosheets (nGO-C5, 500–2000 nm in diameter) and “small” GO nanosheets (sGO-C5, 50–200 nm in diameter), illustrating that both types of alkylated and partially reduced nanosheets can be used to produce bubbles 2–3 μm in diameter, with the quality and stability observed by optical microscopy. Of note, unmodified nanosheets did not lead to the formation of bubbles. The US and PA response of GO-stabilized bubbles was studied, revealing strong imaging contrast for both techniques. This particle-only approach to the preparation of stable gas bubbles and their use in advanced imaging techniques illustrate an advancement in understanding the impact of interfacial activity of particles and a new direction forward in the preparation of multifunctional composite structures.

Scheme 1. Preparation of nGO-C5 or sGO-C5 via Alkylation and Probe Sonication and Their Use as the Sole Surfactant for Pickering Bubbles for Dual-Modality US and PA Contrast Imaging^a.

^aScale bar: 1 mm.

To prepare GO-based surfactants for the stabilization of Pickering bubbles for use in dual-modality US and PA imaging, the surface energy and size of the nanosheets should be considered, such that the nanosheets assemble at the C₃F₈-water interface.²⁶ GO nanosheets are ~1 nm thick and can have diameters of nanometers to microns, and the oxygen-containing functionalities can be modified through simple chemical reactions.²⁹ The flake size (i.e., nanosheet width) dictates how well the GO nanosheets assemble at interfaces and will impact how the bubbles respond to mechanical oscillation (i.e., in response to ultrasound and/or thermal expansion due to absorption of light). As such, we evaluated two different diameters of GO nanosheets: the as-prepared “normal” GO (nGO, 500–2000 nm wide) and “small” GO nanosheets (sGO, 50–200 nm wide), the latter of which was prepared by probe sonication of nGO. The as-prepared GO nanosheets are exceptionally hydrophilic and highly dispersible in water because of alcohol functionalities on the basal plane and carboxylic acids/carboxylates along the edges.^29,64,65 As such, chemical modification or flocculation is required for GO nanosheets to assemble at fluid–fluid interfaces. Building on the previous work from our lab, we modified nGO and sGO nanosheets with a number of different amines (see the Experimental Section for details) and determined that modification with 3-aminopentane yielded the highest concentration of stable C₃F₈ Pickering bubbles.^9,64,66,67

As shown in Scheme 1, GO nanosheets were prepared by oxidation of graphite using potassium permanganate and sulfuric acid, as previously reported, and both nGO and sGO were modified with 1-ethylpropylamine (C5–NH₂) via 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide (EDC) coupling, yielding nGO-C5 and sGO-C5, respectively.³⁰ The successful preparation and functionalization of GO were confirmed via Fourier transform infrared spectroscopy (FTIR) and X-ray photoelectron spectroscopy (XPS). As the characterization data for the functionalization of sGO and nGO are similar, only the data for sGO are discussed here, and all data for nGO are available in the Supporting Information (Figures S1 and S2). Figure 1a compares the FTIR spectra for sGO and sGO-C5; the sGO spectrum shows the expected peaks for highly oxidized GO, while the sGO-C5 spectrum contains an additional peak at ~1650 cm^-1, indicative of the amide C=O functionality, and the signal at ~2970 cm^-1 indicates the alkyl C–H stretching.^30,64,65 These differences in the FTIR spectra support successful covalent modification and preparation of sGO-C5. Characterization of the nanosheets by XPS further supports the functionalization (Figure 1b); for sGO, the typical carbon, oxygen, and sulfur peaks are present, whereas the spectrum for sGO-C5 contains an additional peak for nitrogen. Conversion of sGO to sGO-C5 results in an increase in the N/C ratio from 0 to 0.08 and an increase in the C/O ratio from 1.58 to 2.11 (Table S1).^17,68,69 Carbon 1s high-resolution XPS data reveal that the change in the C/O ratio is attributed to a decrease in the C–O/C–C ratio, as shown in Figure 1c; this change is not only attributed to the presence of the alkyl chain but also suggests partial reduction of the carbon frame work of the GO nanosheet, well known to occur upon covalent modification.^65,70 Reduction of the nanosheet is also supported by the ultraviolet–visible (UV–vis) absorption data; comparison of the spectra of aqueous solutions of sGO and sGO-C5 in Figure 1d reveals that upon functionalization, an increase in absorbance at 229 nm occurs, attributed to ketones, dienes, and π-π* transitions of C=C.^71,72 Atomic force microscopy (AFM) was also utilized to confirm the successful preparation of both sGO (Figure 1e) and nGO (Figure S2a), showing a sheet thickness of ~1 nm and smaller flake sizes of sGO relative to nGO. Functionalization and partial reduction of sGO and nGO did not result in an increase in sheet thickness, as sGO-C5 and nGO-C5 are ~1 nm thick (Figures 1f and S2b).^29,64,68

Figure 1.

Spectroscopy and topography of sGO and sGO-C5. FT-IR spectroscopy (a), survey (b), carbon (c) high-resolution X-ray photoelectron spectroscopy, ultraviolet–visible spectroscopy (d), and atomic force microscopy images of sGO (e) and sGO-C5 (f).

To form Pickering bubbles with only the GO-based surfactant, an aqueous suspension (2 mg/mL) of the nanosheets (sGO, nGO, sGO-C5, or nGO-C5) was mechanically agitated in the presence of C3F8 gas using a commercially available device (VialMix, Lantheus), which is used to activate the clinical MB formulation DEFINITY prior to use. We found that no bubbles formed when only nGO, sGO, or C5–NH₂ was present (Figure S3) or if nGO and sGO were only stirred with C5–NH₂ (without a coupling reagent, Figure S4). Thus, the as-prepared GO cannot serve as a particle surfactant for the water–C₃F₈ system, regardless of the nanosheet diameter. This is likely because the nanosheets are too hydrophilic and thus remain dispersed in the water.^73–75 However, upon covalent modification of GO with C5–NH₂, the surface energy of the nanosheets is modified, such that they stabilize C₃F₈ bubbles without any cosurfactant. Activation by amalgamation did lead to bubbles of C₃F₈ in water when nGO-C5 or sGO-C5 was present; after activation, a milky solution with a small amount of foam was formed (Figure S5c).

Figure 2 shows the optical images of the Pickering bubbles and histogram of the bubble diameter at time 0 and 60 min at 2 mg/mL concentration of sGO-C5 (Figure 2a–c) and nGO-C5 (Figure 2d–f). The concentration of bubbles produced from sGO-C5 or nGO-C5 was optimized at 2 mg/mL (see Figure S6 for the optimization). sGO-C5 produces slightly smaller bubbles than nGO-C5 with median diameters of 3.06 and 3.41 μm, respectively. Previous reports suggest that smaller bubbles are produced if more energy is applied to the system or a more compatible surfactant is used.^76–78 Because both sGO-C5 and nGO-C5 were prepared with the same parameters, we hypothesize that the smaller bubbles produced from sGO-C5 imply that sGO-C5 is a better surfactant for C₃F₈ and water than nGO-C5. Figures 2b,e, and S7 show the optical images of the Pickering bubbles left undisturbed under the microscope and imaged after 60 min. Both the sGO-C5 and nGO-C5 bubbles remain intact after, suggesting that the nanosheets provide an excellent barrier, preventing escape of the C₃F₈ gas or coalescence of the bubbles. The average size measurement (number weighted) using dynamic light scattering (Figure S8) shows that nGO-C5 bubbles and sGO-C5 bubbles have a hydrodynamic diameter of 0.953 and 0.636 μm, respectively. In addition, the average size measurement using a Coulter counter (Figure S9) shows 0.891 and 1.00 μm and a concentration of 2.85 × 10⁵ and 2.96 × 10⁵ bubbles/mL for nGO-C5 bubbles and sGO-C5 bubbles, respectively. The difference between the measured average diameter through optical microscopy and Coulter counter and DLS is brought about, in part, by the detection limit of optical microscopy. GO nanosheets have been well-reported to have high gas-barrier properties by themselves or as a nanofiller to different polymer matrices; thus, C₃F₈ molecules can likely only escape at defect areas between the nanosheets tiled at the C₃F₈–water interface.^79–82

Figure 2.

Optical microscopy images and bubble size histograms of sGO-C5 bubbles (a–c) and nGO-C5 bubbles (d–f) at t = 0 min and t = 60 min.

After the verification of successful fabrication of Pickering bubbles, their applicability as UCAs was evaluated by exposing the bubble solution to an acoustic field using a commercial US imaging system (CPS 7.3 MHz, MI = 0.1). The bubble solution was placed in an agarose phantom (Figure S10), and the backscattered US signal was collected and analyzed. Figure 3a shows the nonlinear Cadence contrast pulse sequencing (CPS) images for water, dispersions of each type of nanosheets, and bubbles made with nGO-C5 and sGO-C5. CPS imaging uses a proprietary pulse sequence (i.e., a combination of pulse and amplitude modulation) that is designed to recognize and process the unique nonlinear fundamental and high-order harmonic signals generated by UCAs.^83–85 As expected, degassed water and dispersions of nGO-C5 and sGO-C5 do not show any nonlinear acoustic activity, as illustrated by dark areas (i.e., lack of signal), nor do solutions of sGO and nGO (Figure S11). In contrast, Pickering bubbles stabilized by nGO-C5 and sGO-C5 display strong signals, with sGO-C5 bubbles having a slightly brighter signal compared to nGO-C5 bubbles. Analysis of the frequency spectrum of the backscattered US for these two samples reveals the nonlinear nature of the oscillation and higher average intensity for sGO-C5 bubbles relative to nGO-C5 bubbles. Figure 3b shows the intensity at the fundamental oscillation centered at 7 MHz and strong harmonic oscillation centered at 14 MHz. Further analysis of the frequency spectrum (Figure 3c) shows that the sGO-C5 bubbles have an average signal intensity of 37.1 ± 0.8 dB, while nGO-C5 bubbles have an average signal intensity of 34.5 ± 0.4 dB. To assess bubble in vitro stability under constant insonation, solutions of sGO-C5 bubbles and nGO-C5 bubbles were placed in an agarose phantom with a narrow channel (Figure S12) and exposed to US (12 MHz, MI 0.1, 0.2 frames per second) using a clinical system in the contrast harmonic imaging (CHI) mode. The raw echo power was recorded and plotted with time, as shown in Figure 3d: sGO-C5 bubbles show a significantly slower decay compared to nGO-C5 bubbles (in vitro half life of 101 ± 18 and 54 ± 15 s, respectively), implying that the smaller flakes are more stable to US mechanical perturbations compared to larger flakes.

Figure 3.

Ultrasound scattering from water and solutions of nGO-C5, sGO-C5, nGO-C5 bubbles, and sGO-C5 bubbles. Representative US CPS (Siemens S3000) images (a), frequency spectrum (b), and average US intensity (c) of nGO-C5 bubbles, sGO-C5 bubbles, and controls. The time–intensity curve of nGO-C5 bubbles and sGO-C5 bubbles exposed to US (Toshiba AplioXG SSA-790A, CHI 12 MHz, MI = 0.1, 1 fps) for 600 s (d). Statistical significance is denoted by *, and NS means the difference is not statistically significant.

The PA response of nGO-C5 bubbles and sGO-C5 bubbles was then tested by exposing the solutions to a 680 nm laser and collecting the generated acoustic signal with a transducer operating at 21 MHz central frequency with 13–24 MHz bandwidth (Figures S13 and S14). We also studied the PA response of a solution of black microbeads (2.94 μm diameter), nGO solution, sGO solution, and commercially available UCAs (DEFINITY) as controls. Note that nGO solution and sGO solution samples were subjected to similar C₃F₈ gas exchange and activation via amalgamation as nGO-C5 bubbles and sGO-C5 bubbles, respectively.

Figure 4a,b show that among the six samples tested, black microbeads (channel i) had the highest PA amplitude signal; however, this could be attributed to high local concentration of microbeads settling within the vessels. Furthermore, the signal from the black microbeads did not fully delineate the entire vessel, as evidenced by the strong signal at the bottom of the vessel. Note that the flat, horizontal reflection band at a depth of ~12 mm in the image is an aliasing effect produced from the waves bouncing off the base of the phantom holder. Vessels with nGO-C5 bubbles (channel iv) and sGO-C5 bubbles (channel v) also had high PA amplitude signals. The functionalized GO samples produced PA signals 9 dB greater compared to the ones produced by nonfunctionalized GO (channels ii and iii), as is seen in Figure 4c. The reduction of GO during functionalization imparts higher optical absorption at 680 nm as shown in UV–vis measurements (Figures 1d and S1d). Additionally, C₃F₈ gas cores enable the emission of secondary acoustic waves, which are produced by acoustic scattering of the primary PA waves by neighboring GO bubbles. This phenomenon, known as multiple scattering, is well-established in UCA research.^86–88 The low PA signal from nGO solution and sGO solution is attributed to their lower optical absorption (i.e., as they are not reduced) and low bubble yield. Finally, the PA signal from MBs made from commercially available DEFINITY (channel vi) appears indistinguishable from noise levels in Figure 4a,b.

Figure 4.

PA and US response of nGO-C5 bubbles, sGO-C5 bubbles, and controls. Representative PA images before (a) and after (b) thresholding. B-mode US images (d) and contrast-enhanced US (CEUS) images (e) of (i) black microbeads (2.94 μm polystyrene black beads), (ii) nGO-solution, (iii) sGO-solution, (iv) nGO-C5-bubbles, (v) sGO-C5-bubbles, and (vi) DEFINITY (3 μL/mL). The average vessel PA signal of vessel contents (c); error bars represent standard deviation between samples (n = 8). The US image in (d) is a logarithmically compressed image normalized to the maximum value of the RF data. Additionally, the nGO solution and sGO solution samples were subjected to similar C₃F₈ gas exchange and activation via amalgamation as nGO-C5 bubbles and sGO-C5 bubbles, respectively.

To ensure that the PA signals were measured from bubbles, B-mode US and CEUS images were also taken. nGO solution and sGO solution produce weak US signals in comparison to DEFINITY MBs. This is attributed to low bubble yield observed in light microscopy images of nGO solution and sGO solution after activation via amalgamation. In comparison, solutions of nGO-C5 bubbles and sGO-C5 bubbles produce intense US signals, which is indicative of highly scattering gaseous cores. nGO-C5 bubbles and sGO-C5 bubbles had an average maximum US signal of 6 and 10 dB respectively higher than that of DEFINITY MBs diluted to 3 μL/mL when averaging over five locations within a vessel using the data in Figure 4d. The US and CEUS images (Figure 4e) show higher signal amplitude from nGO-C5 bubbles and sGO-C5 bubbles than from diluted DEFINITY MBs over the course of the experiment, where measurements were collected less than 1 h after activation of the MBs. This was consistent for different cross-sections of the phantom vessels. The width of sGO-C5 and nGO-C5 is between 50–200 and 500–2000 nm, respectively, and their height is ~1 nm, which do not scatter sound efficiently at these acoustic frequencies. Because of the low scattering from the sGO-C5 and nGO-C5 themselves, the amplitude of the US from the vessels containing GO is attributed to sGO-C5 bubbles and nGO-C5 bubbles. Solutions of sGO-C5 bubbles and nGO-C5 bubbles generate intense PA power because of the highly absorbing nature of the reduced GO shell and their compressible C₃F₈ gaseous core.

CONCLUSIONS

In summary, dual-modality US and PA agents were prepared using Pickering bubbles of perfluoropropane stabilized solely by modified GO nanosheets. Two different sizes of GO nanosheets were covalently modified by coupling with an alkylamine, and the chemical composition was characterized by FTIR and XPS and the size and thickness were confirmed by AFM; the partial reduction of the nanosheets was confirmed by UV–vis spectroscopy. Pickering bubbles of C₃F₈ stabilized by only nGO-C5 or sGO-C5 were prepared using a standard formulation technique and both systems were shown to be stable up to 1 h when left unagitated and exposed to the environment with minimal change in the size and concentration. In contrast, unmodified nanosheets did not lead to bubble formation, thus highlighting the power for tailored nanosheet modification. Pickering bubbles stabilized by sGO-C5 or nGO-C5 are more acoustically active than solutions of unmodified GO after activation via amalgamation, with sGO-C5 bubbles showing higher US response (37.1 ± 0.8 dB vs 34.5 ± 0.4 dB) and better stability (in vitro half-life of 101 ± 18 s vs 54 ± 15 s) than nGO-C5 bubbles with the tested parameters. In addition, solutions of sGO-C5 bubbles and nGO-C5 bubbles have higher PA response as compared to sGO solution and nGO solution. Future studies will include increasing the concentration of fabricated Pickering bubbles to match the concentration of lipid- or polymer-stabilized bubbles. The preparation of stable Pickering gas bubbles establishes a base platform for more advanced dual-mode imaging, theranostics, and photothermal therapy.

EXPERIMENTAL SECTION

Materials and Instrumentation.

All reagents were purchased from commercial suppliers and used as received. Graphite flakes, N,N-dimethylformamide (DMF), 1-ethylpropylamine, agarose, and tetramethylethylenediamine (TEMED) were purchased from Sigma-Aldrich. 1-Ethyl-3-(3-dimethylaminopropyl)carbodiimide hydrochloride (EDC·HCl 98%) was ordered from Acros Organics. Octafluoropropane (C₃F₈) was purchased from Electronic Fluorocarbons, LLC, PA. Black polystyrene microbeads were purchased from Polysciences in Warrington, PA. DEFINITY was purchased from Lantheus Medical Imaging in Billerica, MA. Acrylamide and bis-acrylamide were purchased from Fisher in Mississauga, ON. Fire-polished borosilicate rods were purchased from Sutter in Novato, CA.

The drying process was accomplished with a freeze dryer (HRFD-SBL, Harvest Right). Optical images were taken with an OMAX optical microscope with an 18.0 MP USB digital camera. Probe sonication was completed with a Sonics Vibra-Cell VCX 750 ultrasonic processor. Centrifugation was completed using a SORVALL Evolution RC centrifuge (15,000 RCF) or an Eppendorf Minispin centrifuge with an F-45–12-11 rotor (12,100 RCF). FTIR spectra were obtained using an Agilent Cary 630 FTIR in the ATR mode and diamond/ZnSe crystal. UV–vis experiments were performed with Agilent Cary 5000 UV–vis–NIR in a quartz cuvette (path length: 10 mm). AFM was performed on an NX-10 Park system in the tapping mode. XPS (survey and high-resolution scans) was performed using a PHI Versaprobe 5000 X-ray photoelectron spectrometer with Al Kα radiation with 100 μm spot size and was referenced to internal SiO₂.

Preparation of nGO.

GO was synthesized from graphite flakes following a reported method.^29,89 Briefly, graphite flakes (1.0 g) were magnetically stirred in concentrated H₂SO₄ (134 mL) at room temperature. Then, KMnO₄ (1.0 g, 0.063 mol) was slowly added to the suspension. The mixture appeared dark green and was stirred at 25 °C for 24 h; the addition of an equivalent amount of KMnO₄ was repeated three more times every 24 h, until a total of 4 g of KMnO₄ was added. At the end of the reaction, the mixture was purple and had become more viscous. The mixture was added to ice water (0.70 L), followed by the slow addition of an aqueous H₂O₂ solution (30%), until the pink color changed to bright yellow, indicating quenching of excess KMnO₄. Finally, centrifugation using a SORVALL Evolution RC centrifuge (15,000 RCF) led to the isolation of a yellow-brown solid, and the supernatant was discarded. The pellet was washed repeatedly with 2-propanol until the supernatant had a neutral pH, and then, the solid was dried under reduced pressure at room temperature. The dry solid was blended into a powder. To make nGO solutions in water, the as-prepared GO powder was dispersed in Milli-Q water at a concentration of 2.0 mg/mL by mild stirring (i.e., without sonication, as this can lead to a decrease in nanosheet diameter).

Preparation of sGO.

A total of 30 mL of the as-prepared nGO aqueous suspension was added to a beaker with an ice bath and probe sonicated (300 W) for 5 h. The processed GO solution was then centrifuged using an Eppendorf Minispin centrifuge with an F-45–12-11 rotor (12,100 RCF) for 1 h. The collected supernatant was freeze-dried for 30 h to remove all water, yielding sGO. To make sGO solutions in water, the isolated sGO powder was dispersed in Milli-Q water at a concentration of 2.0 mg/mL by bath sonication.

Preparation of 3-Aminopentane (C5)-Functionalized GO (nGO-C5 and sGO-C5).

3-Aminopentane-functionalized GO (nGO-C5 and sGO-C5) was prepared following a reported method.³⁰ nGO or sGO (20 mg) was first dispersed in DMF (40 mL); then, EDC· HCl (40 mg) and 1-ethylpropylamine (20 mg) were added. The mixture was stirred vigorously at room temperature for 24 h. After this time, a dark brown powder was isolated by centrifugation, washed with ethyl acetate and methanol, and dried under reduced pressure at room temperature. UV–vis measurements were done with the following sample concentrations: nGO—0.2 mg/mL, sGO—0.1 mg/mL, nGO-C5—0.3 mg/mL, sGO-C5—0.2 mg/mL, and C₅NH₂—0.01 mg/mL. All the UV–vis spectra are normalized with respect to the highest peak.

Preparation of sGO-C5 Bubbles and nGO-C5 Bubbles.

nGO C5 or sGO-C5 was dispersed in water to achieve nGO-C5 or sGO-C5 solutions, with a concentration of 2.0 mg/mL. 1 mL of each solution was transferred to a 3 mL headspace vial, capped with a rubber septum and aluminum seal and sealed with a vial crimper. Air was manually removed with a 30 mL syringe and was replaced by injecting octafluoropropane gas. The self-assembly of Pickering bubbles was driven by mechanical shaking with a VialMix shaker (Bristol-Myers Squibb Medical Imaging Inc., N. Billerica, MA) for 45 s. sGO-C5 bubble and nGO-C5 bubble samples were isolated by having the headspace vial inverted for 10 s before withdrawing 100 μL of the bubble solution from a fixed distance of 5 mm from the bottom with a 21G needle.

Optical Microscopy.

nGO-C5 bubble and sGO-C5 bubble samples were placed between two cover slides with another two cover slides (0.15 mm) as the spacer (Figure S15). The whole setup was observed under a microscope. Images were then acquired with an OMAX 18.0 MP USB digital camera at 10×, 40×, and 100× using ToupView. Images were processed with ImageJ (Figure S16).

Coulter Counter.

The size of the nGO-C5 bubbles or sGO-C5 bubbles was characterized using a Multisizer 4 particle analyzer (Beckman Coulter, Inc). 300 μL of nGO-C5 bubbles or sGO-C5 bubbles was diluted into 20 mL of isotone. 100 μL of the diluted sample was used for measurement.

Dynamic Light Scattering.

The size of the nGO-C5 bubbles or sGO-C5 bubbles was characterized using Litesizer 500 from Anton Paar with a 658 nm laser light source. All samples were diluted 100× prior to measurement and repeated for a total of three trials each.

Sample Preparation for Ultrasound Imaging (Siemens S3000) and in Vitro Stability (Toshiba AplioXG).

Nonlinear US imaging was carried out using a commercial clinical US system, Siemens S3000, in the research mode with an 18 MHz center frequency linear array transducer (18L6 HD). Images and raw RF data were acquired in the Cadence CPS mode with parameters set as CPS 7.3 MHz, MI = 0.1, MIF 0.06, 2D-0.5%, 0 dB/DR55, and CPS 0 dB. nGO-C5 bubble and sGO-C5 bubble samples were diluted (1 mL, 100× diluted) to avoid signal attenuation and were placed in an agarose phantom. The agarose phantom was composed of 1.5 wt % agarose in Milli-Q water (resistivity of ~18 Mω·cm) and heated in a microwave until the agarose is dissolved. The hot agarose solution was then poured into a mold, avoiding any trapped bubbles, and cooled down to obtain a phantom with the desired channel dimension. The power spectrum and average intensities were calculated from the RF data using MATLAB R2018a.

The time–intensity curve of sGO-C5 bubbles and nGO-C5 bubbles was determined by continuously exposing the diluted solution to US. sGO-C5 bubble and nGO-C5 bubble samples were diluted (400 μL, 100× dilution) with water and were placed in an agarose phantom with a thin channel (L × W × H = 22 × 1 × 10 mm). The thin channel was chosen to ensure that the diffusion of bubbles in and out of the US field is minimized and bubbles were continuously insonated. Nonlinear US imaging was done on an AplioXG SSA-790A clinical US imaging system (Toshiba Medical Imaging Systems, Otawara-Shi, Japan) with a 12 MHz center frequency linear array transducer (PLT-1204BT). Images were acquired in the CHI mode with parameters set as 65 dB dynamic range, 70 dB gain, imaging frame rate 0.2 fps, and MI of 0.1. Raw echo powers were acquired, and the intensity per frame is analyzed with a built-in software tool (CHI-Q). The in vitro half-life (in vitro t_1/2) was calculated with the formula for first-order decay.

Sample Preparation for Ultrasound and Photoacoustic Imaging (LZ-250, 21 MHz) and CEUS Imaging (LZ250, 18 MHz).

US and PA imaging was done using the Vevo LAZR 2100 system (FujiFilm VisualSonics) with an LZ250 transducer operating at 21 MHz central frequency (13–24 MHz bandwidth). The transducer is a 256-element linear array transducer coupled with a laser (λ = 680–970 nm) system. The transducer and laser have a focus at 11 mm imaging depth.

Polyacrylamide (10%) phantoms containing six 1 mm diameter vessels were prepared 1 day prior to imaging, using degassed, deionized (DDI) water. The deionized water was degassed using a SRDS-1000 water-degassing system (FUS Instruments, Toronto, ON). For each phantom, the polyacrylamide solution was prepared and poured into a 2 cm × 2 cm holder with six parallel fire-polished borosilicate vessels, each 1 mm in outer diameter. The six borosilicate vessels were carefully removed after polymerization of the polyacrylamide solution, and the phantom was removed from the holder for storage. In storage, phantoms were hydrated in DDI water at 4 °C for 24 h.

Prior to imaging, the vessels were filled with solutions of (i) 2.94 μm black polystyrene microbeads (diluted, 25 μL in 1 mL of Milli-Q water for a concentration of 3.70 × 10⁸ microbeads/mL calculated based on a known stock concentration), (ii) nGO solution, (iii) sGO solution, (iv) nGO-C5 bubbles (with an average concentration of 6.00 × 10⁷ MBs/mL), (v) sGO-C5 bubbles (with an average concentration of 6.23 × 10⁷ MBs/mL), and (vi) DEFINITY MBs (diluted, 3 μL in 1 mL of Milli-Q water for an average concentration of 3.69 × 10⁸ MBs/mL). MB concentrations were measured using a 30 μm aperture Coulter counter. Black polystyrene microbeads were used as a baseline for the PA signal, and DEFINITY was used as a baseline for the B-mode US imaging and nonlinear CEUS imaging. The phantom was placed in a phantom holder, and the vessels were sealed at the ends using glass slides. The sealed phantom-in-holder was placed in a DDI water bath for imaging. The center of the vessels was aligned at the laser-transducer focus. The transducer was oriented parallel to the base of the phantom. The agent-filled vessels were oriented orthogonally to the direction of the transmit wave path in order to image a full circular vessel cross-section.

Data Analysis for PA.

PA data from the six contrast agents were analyzed using MATLAB R2019a. The laser energy was recorded for each frame and used to normalize the PA images. To suppress noise, reverberation artifacts, and wing artifacts, thresholding was applied to the beamformed images. A mask was created of circles, with centers corresponding to the centers of the vessels and diameters matching the vessel diameter (as determined from the US image). The mask was multiplied by the original image to isolate the signal from each vessel. The average signal from a vessel was determined by taking the mean signal expressed by all pixels within a single vessel region. The average signal from each vessel was determined and averaged for all 25 frames and all eight cross-sections of the same vessel type.

B-mode US and PA RF data were acquired simultaneously when acquiring PA images. B-mode US images were collected at 4% transducer power (1.890 MPa) and PA images at 100% laser power (average fluence = 29.4 ± 4.6 [mJ/m²]) at a wavelength of 680 nm. 25 frames of US and PA images were acquired and were subsequently analyzed using MATLAB R2019a. CEUS imaging was performed at 18 MHz at 100% transmit power. The CEUS uses a nonlinear contrast mode based on amplitude and pulse inversion and nonlinear fundamental and subharmonic energy filtering. The gain for US imaging was set at 35 dB for each measurement, and the gain for PA imaging was set to 60 dB for each measurement.