Multimodal microscopy-based identification of surface nanobubbles

Journal of Colloid and Interface Science 547 (2019) 162–170

Contents lists available at ScienceDirect

Journal of Colloid and Interface Science journal homepage: www.elsevier.com/locate/jcis

Multimodal microscopy-based identification of surface nanobubbles Nicole Hain, Stephan Handschuh-Wang, Daniel Wesner, Sergey I. Druzhinin, Holger Schönherr ⇑ Physical Chemistry I & Research Center of Micro and Nanochemistry and Engineering (Cl), Department of Chemistry and Biology, School of Science and Technology, University of Siegen, Adolf-Reichwein-Str. 2, 57076 Siegen, Germany

g r a p h i c a l a b s t r a c t

a r t i c l e

i n f o

Article history: Received 26 January 2019 Revised 24 March 2019 Accepted 25 March 2019 Available online 26 March 2019 Keywords: Surface nanobubbles AFM FLIM Combined AFM-FLIM Fluorescence lifetime

a b s t r a c t Hypothesis: Surface nanobubbles, which were controversially discussed in the literature, promise a number of outstanding applications, and their presence may hamper nanoscale processes at solid-aqueous interfaces. A most crucial and yet unsolved question is the rapid and conclusive identification of gas-filled (surface) nanobubbles. We hypothesize that surface nanobubbles and oil nanodroplets can be conclusively differentiated in co-localization experiments with atomic force microscopy (AFM) and time-resolved fluorescence microscopy by localizing tracer fluorophores and analyzing their fluorescence lifetimes. Experiments: Combined AFM and fluorescence lifetime imaging microscopy (FLIM) were conducted to localize the various interfaces labelled by the reporter dye rhodamine 6G (Rh6G). The dependence of the fluorescence lifetime of Rh6G on its local environment was determined for air/water, water/glass and polysiloxane/water interfaces under different conditions. Findings: In in situ co-localization experiments, surface nanobubbles labeled with Rh6G were probed by AFM with high spatial resolution and were differentiated from polysiloxane droplets as well as contamination originating from lubricant-coated syringe needles owing to the characteristic short fluorescence lifetime of Rh6G at the gas/water interface observed in FLIM. In particular, this approach lends itself to conclusively identify and rapidly differentiate these gas-filled entities from adsorbed contamination, such as siloxane-based oil nanodroplets. Ó 2019 Elsevier Inc. All rights reserved.

1. Introduction

⇑ Corresponding author. E-mail address: [email protected] (H. Schönherr). https://doi.org/10.1016/j.jcis.2019.03.084 0021-9797/Ó 2019 Elsevier Inc. All rights reserved.

The importance of microbubbles (MBs) in liquids is evident in many natural processes and engineering applications; for example, they are essential for understanding the nucleation mecha-

N. Hain et al. / Journal of Colloid and Interface Science 547 (2019) 162–170

nisms in cavitation/boiling [1]. They are also considered critical in the fields of materials [2], sonoreactors [3] and waste treatment [4], among others. Stabilized microbubbles have also found their way into medical applications, e.g. as contrast agents in ultrasound imaging [5], and for drug delivery [6]. Unstabilized MBs are strikingly expected to become smaller [7], forming nanobubbles (NBs) in an accelerating manner until they dissolve and finally disappear within several ls, due to the high Laplace pressure. However, the opposite has been reported [7–9], which contradicts estimates of the diffusive gas transport based on the Epstein and Plesset model [10]. Although early data suggests that stable bulk nanobubbles (BNBs) may exist [11–14], it is certainly correct to conclude that the mechanisms that may explain their stability are not well understood or quantified in a consistent way. Moreover, there is no rigorous evidence to distinguish them from nanodroplets. The localization and characterization of surface nanobubbles [15], which were initially proposed by Parker et al. to explain the long-range hydrophobic attraction and peculiar force-distance curves observed in their surface forces apparatus experiments between hydrophobic surfaces in water [16], was for many years only possible by atomic force microscopy (AFM) [17–20]. According to several reports surface nanobubbles are relevant for applications and technology development. Examples include froth flotation, [21–25] protein adsorption to surfaces [26–28], patterning polymer surfaces [29], medicine [30] as well as cleaning of contaminated surfaces [27,31,32]. Since AFM provides, albeit only if applied correctly [33–35], at most the contour of the nanoscale objects and a characteristic mechanical and adhesive response of the transient nanoscale contact formed between the objects and the AFM tip, no conclusive evidence for the gaseous nature of surface nanobubbles was reported until recently. Thanks to the pioneering work of Lohse and Zhang, a conclusive stabilizing mechanism via surface pinning as well as the possible shift of the equilibrium due to additional gas from an enrichment layer near the solid wall on the one hand [36–38], and an explanation for the unusually small experimentally deduced contact angles on the other hand, have been provided [15,36]. From an experimental point of view, the identification of the gaseous content of surface nanobubbles as well as the differentiation from nanoscale oil droplets has proven to be much more challenging. However, in view of alarming reports on siloxane oil nanodroplets that show similar appearance in AFM micrographs [39,40], it is of utmost importance to establish robust and versatile approaches to selectively identify the nanoscale features’ content. The same holds true for bulk nanobubbles. To date only very few reports provide information about the content of surfaces nanobubbles. When carbon dioxide saturated water was used to form nanobubbles on hydrophobic surfaces, it was shown that gaseous CO2 was present near the surface using attenuated total internal reflection ATR-IR spectroscopy [41,42]. Similar data was reported for n-butane gas [43], however, conventional ATR-FTIR spectroscopy does not provide any spatial resolution and hence no localization. The disappearance of alleged surface nanobubbles during degassing, as followed by AFM, provides indirect evidence for the gas filling [44], and was in fact utilized to differentiate nanobubbles from siloxane oil nanodroplets [39]. The drops, which e.g. originate from lubricant coated syringe needles, exhibit a virtually identical topography and it is likely that many studies reported in the literature addressed unknowingly oil nanodroplets [39]. The first optical microscopy evidence of nanobubbles was reported by Karpitschka et al., who observed nanobubble nucleation on glass non-invasively using interference-enhanced reflection microscopy and could show by co-localized AFM analysis that the features observed optically correspond to elevated

163

nanoscale spherical caps [45]. In addition, Chan and Ohl provided complementary optical evidence of surface nanobubbles using total internal reflection fluorescence microscopy (TIRF) with dye-labeled nanobubbles [46]. Later they were able to identify nanobubbles in a microfluidic channel [47]. All these optical techniques have diffraction limited lateral resolution, exceeding typical size of NBs, and they do not provide evidence for the gas content. Latter challenge was addressed by Hain et al., who used Rh6G as label in time resolved confocal fluorescence microscopy [48]. Similarly, Seo et al. exploited changes in emission wavelength and intensity of a surface immobilized dansyl reporter dye, which were observed in fluorescence microscopy and were found to be co-localized with spherical cap-shaped objects observed by AFM [49]. The diffraction limit can be overcome in the analysis of nanobubbles by transmission electron microscopy (TEM) [50] or scanning transmission soft X-ray microscopy (STXM) [51] carried out in special ultrathin graphene or silicon nitride windows as well as by optical surface plasmon resonance microscopy [52] and super-resolution iPAINT microscopy [53]. Of all these techniques, only STXM may provide chemical information, but remains limited in scope due to the special constraints of the window material and the need for a synchrotron facility. A versatile approach to characterize surface nanobubbles should provide access to the corresponding metrology data, it should be independent of the surface chemistry of the underlying substrate and should be able to distinguish gas-filled bubbles from the abovementioned oil-based nanodroplets. In addition, a sufficiently large data set should be attainable to afford appropriate statistics. As will be discussed here, we report on an expanded minimally invasive approach based on co-localized atomic force microscopy (AFM) and fluorescence lifetime imaging microscopy (FLIM) to differentiate gas-filled surface nanobubbles from nanoscale contaminant oil droplets (Fig. 1). In this generalizable approach the aqueous phase and various interfaces are labeled with the reporter dye rhodamine 6G (Rh6G) and via time-resolved fluorescence measurements the gas/water interface of surface nanobubbles is mapped spatially, conclusively identified and more importantly differentiated from siloxane-based oil nanodroplets.

Fig. 1. Schematic of the combined AFM-FLIM experiment, in which a dye-labeled nanobubble on a transparent glass substrate is probed from the top by AFM and from below by a confocal fluorescence microscope. The color code of the dye molecules codes for the different fluorescence lifetimes of the dye molecules in different nanoenvironments, which are determined by time correlated single photon counting (TCSPC) [48].

164

N. Hain et al. / Journal of Colloid and Interface Science 547 (2019) 162–170

2. Experimental section

2.4. Fitting procedure

2.1. Materials

The fluorescence decay times and amplitudes of Rh6G in aqueous solution and at its air-, glass- and siloxane oil-interfaces were obtained with a home-made application using a least-squares method [54], by minimization of the global target function  2 PP S ¼ j i wij yij  yeij . Here the square of the difference between

Water (Millipore Direct Q 8 system, Millipore, Schwalbach, Germany) and ethanol (EMSUREÒ ACS, ISO, Reag. Ph. EUR, Merck Millipore, Darmstadt, Germany) solutions of rhodamine 6G (Rh6G,
95%, Fluka, Germany) were used in the experiments. The MilliQ water posseses a resistivity of 18.0 MX cm and a surface tension of 72.69 ± 0.07 mN/m, determined by the Wilhelmi plate method on a DCAT 11EC tensiometer (DataPhysics Instruments GmbH, Filderstadt, Germany).

2.2. Preparation of the surfaces For confocal measurements, square borosilicate cover slips (20 mm  20 mm) with a thickness of 0.13–0.16 mm (Gerhard Menzel GmbH, Braunschweig, Germany) were used, while for the combined FLIM/AFM experiments round 35 mm diameter borosilicate cover slips with a thickness of 0.13–0.16 mm (Gerhard Menzel GmbH, Braunschweig, Germany) were employed. The cover slips were washed with chloroform (ROTIPURAN 99% p.a., ROTH, Karlsruhe, Germany), rinsed with ethanol and Milli-Q water, cleaned with Piranha (sulfuric acid/hydrogen peroxide) solution (3:1 v/v), for 1 min, rinsed again with Milli-Q water and then with pure ethanol. Caution: Piranha solution should be prepared, used and discarded with extreme caution! It has been reported to detonate unexpectedly. All contact with organic matter must be avoided. Surface nanobubbles were nucleated by carrying out the ethanol–water exchange employing two plastic Eppendorf pipettes (Eppendorf Research plus, 100 lL–1000 lL, Hamburg, Germany) or Piranhacleaned glass Pasteur pipettes to nucleate on the pre-cleaned glass, as reported before [48]. All measurements in aqueous media were conducted at 23 °C with a freshly prepared 850 nM solution of the dye in water. To mimic PDMS contaminations, drops of SYLGARD 184 SILICONE ELASTOMER KIT (Dow Corning GmbH, Wiesbaden, Germany) were placed with disposable glass Pasteur pipette (150 mm, VWR, Darmstadt, Germany) on the cleaned cover slides in the confocal cell. Then aqueous R6G was pipetted in the confocal cell. For the degassing experiments, the solutions were degassed for 30 min at the pressure of 80 mbar at 20 °C. During pumping down (Type MZ 2C, Vacuubrand, Germany), the solution were continuously sonicated (Bandelin Sonorex, Rk 100H, Berlin, Germany).

2.3. Fluorescence lifetime imaging microscopy (FLIM) [48] The confocal fluorescence microscopy images and timecorrelated single photon counting (TCSPC) data of nanobubbles and nanodroplets were measured at room temperature using a confocal laser scanning fluorescence microscope (PicoQuant, Berlin, Germany) composed of a Microtime 200 main optical unit, a FCU II fiber coupling unit, a PicoHarp 300 data acquisition module and an OLYMPUS IX-71 microscope frame (Olympus, Hamburg, Germany). Data were collected through a water immersion 60 objective (UPlanSApo 60/1.20 N.A., Olympus, Hamburg, Germany). The fluorescence of Rh6G was excited at 485 nm with a rate of 20 MHz by a LDH-D-C-485 pulsed laser (PicoQuant, Berlin, Germany), keeping the excitation power low to reduce photobleaching of the fluorophore. The fluorescence of Rh6G was registered with a PD1CTC Single-Photon Avalanche Diode detector (Micro Photon Devices, Bolzano, Italy) at a time resolution of 16 ps per channel. Images in the horizontal xy- or in the vertical xz-plane were recorded with a piezo XYZ-scanner in an area up to 80  80 mm2 with up to 512  512 pixel resolution.

calculated (y) and experimental (ye ) fluorescence intensities is multiplied with the weight (w). The values P   R ti x yij ¼ 1 dx, which were calculated as a pðt  xÞ a exp  k jk s k

convolution of the instrument response function (IRF) p, measured as the reflection of the excitation light from a glass/air interface with a sum of 1–3 exponents (amplitude a and decay time s). In such a global fit both a and s are varied, but the decay times s1, s2 and s3 for each decay are fitted globally for all data sets, i.e. the resulting values for the decay times s1, s2 and s3 are the same for all response functions. The index j refers to the response functions, usually
10, i counts the experimental points for each response function, usually
2500, and superscript ‘e’ indicates that the quantity is an experimental value. For the fluorescence decay the factor w can be written as: w–1 = ye. The number of exponential terms required for a certain fluorescence decay was determined by comparison of the residuals and the autocorrelation plots for the fits with different numbers of exponential terms. The lowest number of such terms giving flat residuals and autocorrelation plots was considered as an appropriate fit of the fluorescence decay. For the different fit conditions or data sets, a rather small variation of decay times of
0.05 ns was found. The fluorescence lifetime of Rh6G in water measured in different cells and on different days lies in the narrow range between 3.85 and 3.93 ns. Fluorescence lifetime images (FLIM) were calculated from a set of amplitude ak -images with fixed component decay times s determined from the global fit of the TCSPC data together with those for Rh6G in aqueous solution and at the model interfaces. 2.5. Combined AFM-FLIM measurements An atomic force microscope (MFP-3D-Bio, Asylum Research Inc., Santa Barbara, CA) equipped with a Closed Fluid Cell (Asylum Research) was used to measure AFM data in liquid. The baseplate of the AFM comprising the xy-scanner is tailor-made for the OLYMPUS IX-71 frame that holds the optics (see above). The AFM data were captured at 0.5 Hertz scan rate with a resolution of 512 points and lines each and the setpoint was chosen at 87–90% of the free tapping amplitude of 50 nm. Cleaning the cantilevers was done by UVozone treatment before the measurements (ProCleaner, UVOTECH Systems, Inc., Walnut Creek, CA). The nucleation of nanobubbles in the AFM liquid cell was performed by ethanol-water exchange using glass Pasteur pipettes before mounting the AFM-head. The cantilever holder sealed the liquid cell so that degassed aqueous solution of Rh6G could be pumped through the cell for experiments for the analysis of the stability of the nanobubbles. For the combined AFM-FLIM measurements the alignment of the AFM tip and the focus of the laser from the confocal microscope was done utilizing the cameras of the AFM and of the optical microscope and by monitoring the backscattering, while scanning the AFM tip with the laser beam. The measurements by AFM and FLIM were performed using the individual control units in a sequential manner to avoid interferences. 3. Results and discussion 3.1. Localization of Rh6G at different interfaces The time-resolved fluorescence microscopy approach to study Rh6G-labeled surface nanobubbles requires the determination of

N. Hain et al. / Journal of Colloid and Interface Science 547 (2019) 162–170

the fluorescence lifetimes of the reporter dye in the three anticipated nanoenvironments: (a) in bulk aqueous solution, (b) at the water/glass interface and (c) at the air/water interface. All measurements were carried out at a concentration of 850 nM, which did not markedly alter the surface tension of the solution (c = 71.72 ± 0.15 mN/m for 850 nM aqueous Rh6G compared to c = 72.69 ± 0.07 mN/m for Milli Q water). Both interfaces could be reliably located in the confocal microscope for an approx. 20 lm thin layer of an aqueous Rh6G solution by moving the focus from the top air/water interface stepwise down to the water/glass interface and simultaneously recording the fluorescence emission. The corresponding xz-image is shown in Fig. 2 together with an intensity plot. The fluorescence intensity of Rh6G is much higher at these interfaces than in bulk solution. Especially, at the air/water interface it is enhanced compared to the fluorescence Rh6G in bulk water, probably because of dye enrichment (surface excess) at the interfaces. Hence, these interfaces may be confidently identified as bright areas in the confocal fluorescence intensity images. When the size of a fluorescent object exceeds the spatial resolution of

-10

850 nM Rh6G in water Air / Water x, µm -10

10

0

10

20

30

40 Int

40

0

Water / Glass

20

z, µm

z, µm

0

10 20

20

0

30

30

0

5

10

Fluorescence intensity, kcounts

15

Fig. 2. Fluorescence intensity z-profile of a thin
20 mm layer of Rh6G aqueous solution on a fused quartz surface. The fluorescence intensity was integrated over the horizontal x-axis (pixel resolution in x- and z-direction: 157 nm; excitation wavelength 485 nm). The peaks correspond to the top air/water (z = 0) and bottom water/glass (z = 20.4 mm) interfaces, as indicated. Inset: a part of the confocal fluorescence intensity xz-image with these interfaces employed for calculation of the intensity profile.

165

confocal optical microscopy, the analysis of the relative intensity of the areas helps to decide whether such an area is associated with the water/glass or the air/water interface, respectively. But this simple approach cannot be employed for surface nanobubbles, which possess characteristic dimension well below the spatial resolution of confocal optical microscopy. 3.2. Lifetime of Rh6G at different interfaces Subsequently, TCSPC data acquired in bulk solution and at the two interfaces was analyzed. In bulk aqueous solution the fluorescence of Rh6G decays single-exponentially (Fig. 3a), whereas its fluorescence at the two interfaces decays double-exponentially (Fig. 3b,c). The decay time s1 observed for Rh6G in water, 3.83 ns, agrees well with the literature (s = 4.08 ns) [55] and is attributed to molecularly dissolved Rh6G in water. The shorter decay time of Rh6G in the present work could be a consequence of a certain ethanol residue remaining in solution after the ethanol-water exchange as well as different oxygen concentration. In addition, Magde et al. excited Rh6G at 400 nm, which is close to the minimum of absorption spectrum, where the effect of dye contaminations might have a substantial contribution to the Rh6G fluorescence decay [55]). These fits over the entire decay curves, including the excitation pulse, are of high quality, as the values of v2 are close to 1.0 and the weighted deviation r and auto-correlation function (A-C plot at Dt – 0) are uniformly and featureless distributed around zero. For the double-exponential decays shown in panels (b) and (c), in addition to the decay times s2 equal to 3.03 ns for the dye located at the water/glass interface and 0.66 ns for the dye located at the air/water interface, the same lifetime s1 of 3.83 ns is present, which is attributed to the extension of the confocal voxel above or below the corresponding interface into the Rh6G-containing solution. The second decay component possesses in both cases a significantly shorter decay time, indicating dynamic fluorescence quenching of the dye at the interface. This fact in combination with the high fluorescence intensity of Rh6G at the interface (Fig. 2) suggests that this quenching may be concentration quenching. The quenching effect is overcompensated by enrichment of the dye at the interface. In general, the fluorescence lifetime is known to depend on the orientation of the dye molecules at the dielectric/dielectric interface [56]. Also, the fluorescence intensity of Rh6G increases with pH stronger at the air/water interface than in bulk solution [57].

Fig. 3. Semi-log plot of the TCSPC fluorescence decays (black dots) of 850 nM Rh6G and IRF function (gray line) together with calculated fits (red line). (a) Single-exponential decay in bulk water, (b) double-exponential decay at the water/glass and (c) at the air/water interfaces. In the panels, the weighted residuals r and the autocorrelation function (A-C) plots are shown. In the panels the values of decay times s and their amplitudes a as well as v2 are also indicated. The excitation wavelength was 485 nm and the time resolution 16 ps. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

166

N. Hain et al. / Journal of Colloid and Interface Science 547 (2019) 162–170

These effects together with dye adsorption and concentration quenching [58,59] may lead to observable effects of brightening interfaces having short fluorescence decay times. In subsequent experiments we fix the Rh6G concentration to 850 nM, which affords a sufficiently high and, importantly, also constant concentration in solution, while achieving equilibrium coverages at the water/glass and air/water interfaces, respectively.

3.3. FLIM of Rh6G labeled nanobubbles Due to the clear separation of the lifetimes for the three nanoenvironments and in particular the very short lifetime for Rh6G at the air/water interface, the decay times offer a handle to identify and localize the hidden glass/water interface and the alleged gas/water interface of surface nanobubbles in confocal

Fig. 4. (a) Fluorescence intensity image of nanobubbles nucleated by ethanol-water exchange in the presence of 850 nM Rh6G, FLIM amplitude maps of image (a) for: (b) short-lived, 0.66 ns, and (c) long-lived, 3.83 and 3.03 ns, fluorescence decay components. Fluorescence decays of the nanobubbles area (d) marked with red circle and of the background area (e) marked with green rectangle in panel (a). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

N. Hain et al. / Journal of Colloid and Interface Science 547 (2019) 162–170

microscopy. Surface nanobubbles on Piranha-cleaned glass were nucleated using the ethanol-water exchange procedure, which is a well-established method to obtain nanobubbles at various surfaces. The fluorescence intensity image shown in Fig. 4a corresponds very well to the literature data [47,48]. Bright spots of high fluorescence intensity are discernible, which differ in size and intensity. As shown above (Fig. 2), we can conclude from this that they are probably NBs, the larger features very likely exceed the length scale set by the optical diffraction limit. The data from the red circle, comprising a feature of high fluorescence intensity, and the green rectangle, where practically no fluorescence is discernible, were analyzed analogously to the data shown above (Fig. 4d, e). The decay curves of these two areas are

167

triple-exponential and possess the same decay constants as the decays in Fig. 3. Hence, the marker dye is present in the three nanoenvironments as well. A direct consequence of the observation of the short-lived component of 0.66 ns with substantial amplitude a (Fig. 4d) is that the brightly fluorescent feature in Fig. 4b apparently comprises a gas/water interface. The contribution of such short-lived component is much reduced for the other area (green rectangle in Fig. 4a), see Fig. 4e. This means that a Rh6G labeled gas/water interface of the nanobubbles with substantial lower number density is present. Due to the optical diffraction limit we cannot resolve agglomerates of smallest structures that are present e.g. in AFM images. We attribute these tiny dark gray features, which are more clearly

Fig. 5. (a) AFM height image, (b) confocal fluorescence intensity image and (c) fluorescence triple-exponential decay of surface nanobubbles nucleated with ethanol-water exchange of 850 nM Rh6G solutions; (d) AFM height image, (e) confocal fluorescence intensity image and (f) fluorescence triple-exponential decay of surface nanodroplets originating from a disposable needle in aqueous 850 nM Rh6G solution; (g) confocal fluorescence intensity image and (h) double-exponential fluorescence decay of PDMS droplets in aqueous 850 nM Rh6G solution.

168

N. Hain et al. / Journal of Colloid and Interface Science 547 (2019) 162–170

visible in the FLIM image in Fig. 4b compared to the intensity image in Fig. 4a, tentatively to agglomerates of small nanobubbles. The lower (>3 times) fluorescence intensity of the short-lived component could be caused by low surface coverage of glass by submersed water/air interface due to the small nanobubble number density. The FLIM images in Fig. 4b and 4c show the spatial distribution of the 0.66 ns component and of both of the longer components, respectively. A comparison of panel (a) with panels (b) and (c) shows that the brightly fluorescent features observed in (a) are co-localized with the shortest lifetime component. In addition, some barely visible features present in intensity image 4a are more clearly recognizable in FLIM image 4b and are attributed to surface nanobubbles. Consequently, panel (c) exhibits dark areas at these locations. From these observations it is very probable that the ethanol-water exchange resulted in the nucleation of surface nanobubbles that can be differentiated from the glass substrate (i) by the bright fluorescence of enriched Rh6G and more selectively (ii) by the characteristic decay constant for the reporter dye. 3.4. Surface nanobubbles and nanodroplets analyzed by combined AFM and FLIM The diagnostic power of the time resolved fluorescence microscopy thus allows one to map the presence of Rh6G at air/water interfaces with diffraction limited resolution. To conclusively link the observed features to surface nanobubbles, the same area was analyzed by FLIM as well as intermittent contact mode AFM (Fig. 5a–c). The elevated features observed by AFM are fully consistent with a multitude of related reports in the literature and afford the dimensions of nanoobjects imaged. The fluorescence data in panel (b) is of lower resolution, but it is clear that the two microscopy images show complementary information. The decay curve in panel (c), furthermore, provides irrefutable evidence for the

substantial presence of the 0.66 ns component, attributed to surface nanobubbles. The diagnostic power of the combined AFM-FLIM approach is further demonstrated by analyzing a sample that was prepared analogously using a lubricant coated disposable medical syringe needle (Fig. 5d–f). The AFM image in panel (d) shows a discontinuous layer of elevated nanoscale features, which are also imaged in panel (e) as bright areas, when labeled by Rh6G. Strikingly, the decay curve is triple-exponential, but does not show any sign for the 0.66 ns component. By contrast, the two major components that are found in addition to the minor 3.83 ns (Fig. 3a) component of Rh6G in water are 3.26 ns and 1.12 ns. Hence, the material that is seen in AFM and fluorescence intensity is not comprising an air/ water interface. Therefore, the features observed by AFM cannot be gas-filled nanobubbles, but they are nanodroplets. The longer of these decay components was found in separate blank experiments (Fig. 5g, h) to coincide with the lifetime of Rh6G on the surface of liquid non-crosslinked poly(dimethyl siloxane) (PDMS) nanodroplets, which has been previously identified as contaminant from lubricant coated disposable medical syringe needles [39]. While we were unable to image the comparatively large PDMS drops under water by AFM in a non-destructive manner, such data can be obtained under water, if the droplets are smaller [60]. Additional XPS data confirmed the presence of silicon in dried samples (on HOPG), which is also consistent with the work of Berkelaar et al. [39] (Fig. S1, Supporting Information). In the presence of a large amount of oil droplets most of the glass surface can be covered by oil, which will change the adsorption of Rh6G on this interface and its fluorescence lifetime there. In addition, the weak emission of Rh6G on glass with a lifetime of 3.03 ns can be masked by the strong emission corresponding to droplets, where a similar lifetime of 3.26 ns was detected. Thus, on the basis of the AFM-FLIM data shown in Fig. 5, the contaminant from syringe needles could be (i) mapped with high

Fig. 6. (a), (c) Fluorescence intensity image, and (b), (d) corresponding decay curves of surface nanobubbles captured after (a), (b) 0, and (c), (d) 4.0 h of flushing with degassed aqueous 850 nM Rh6G solution. The surface nanobubbles were nucleated on a glass surface by the ethanol-water exchange with 850 nM Rh6G solutions. (e)–(h): Confocal fluorescence images of a separate degassing experiment with surface nanobubbles, nucleated by ethanol-water exchange of 850 nM Rh6G solutions, after (e) 0, (f) 0.5, (g) 2.5 and (h) 4.0 h of flushing of aqueous 850 nM Rh6G solutions with a flow rate of 500 mL/min. The flushing solution was degassed by reducing the pressure down to 10 mbar for 30 min while agitating by ultrasonication. Time equals zero corresponds to a blank in gas saturated water.

N. Hain et al. / Journal of Colloid and Interface Science 547 (2019) 162–170

spatial resolution, (ii) differentiated from air-filled surface nanobubbles and (iii) in parts attributed to siloxane oil material. The material responsible for the shorter lifetime of Rh6G of 1.12 ns is currently unknown. 3.5. Degassing experiments Finally, the effect of degassing on the surface nanobubbles identified above was followed by AFM-FLIM in situ measurements [39]. As shown in Fig. 6, the brightly fluorescing Rh6G-labeled nanobubbles became less intense and vanished after 4 h flushing with degassed aqueous 850 nM Rh6G solution at a flow rate of 500 mL/ min. Also, by AFM the decrease in nanobubble size was detected and both nanobubble width and height decreased significantly (Fig. S2). These observations are again fully in line with the work of Berkelaar et al. [39]. The apparent absence of the 3.83 ns decay component in Fig. 6b can also be explained by the shrinking of the probed bulk Rh6G solution volume, e.g. when the water/glass interface spontaneously drifts up in the microscope. Parallel to the reduction in fluorescence intensity, the contribution of the 0.66 ns component in the triple-exponential decay curves also decreased. This can be attributed to the disappearance of gas/water interface area due to the gas that has diffused into the gas depleted water phase. However, since this short-lived component is still required to obtain a high-quality fit at t = 4 h, where no bright spots are visible anymore in the intensity image in Fig. 6c, we conclude that FLIM certainly possesses a higher resolving ability compared to conventional confocal fluorescence microscopy. In this approach only intensity is monitored and the background intensity of Rh6G adsorbed on the glass makes it difficult to detect very small nanobubbles that may still exist after several hours of degassing. 4. Conclusions We showed that an expanded combined AFM-FLIM approach [48] using a tracer dye is a versatile approach, based on a quantitative measurement, for the detection, identification and investigation of surface nanobubbles at solid/aqueous interfaces and in particular to differentiate gas-filled surface nanobubbles from oil nanodroplets with high spatial resolution. The approach is based on the different fluorescence decay times of a marker dye at the liquid/glass interface and the gas/liquid interface of surface nanobubbles, which can be distinguished from such decay time of the dye in bulk aqueous phase. The fluorescence lifetime of Rh6G observed on the bubbles (0.66 ns) is in agreement with the one found at the non-curved air/water interface, confirming that surface nanobubbles nucleated by the ethanol-water exchange indeed contain gas. In particular, the approach enabled the differentiation of surface nanobubbles from siloxane oil contamination adsorbed at the solid/water interface. This work expands the previous attempts to confirm the gaseous nature of surface nanobubbles with higher resolution [42,43,45,46] and to distinguish them from oil nanodroplets [39,40,60]. Compared to fluorescence-based methods with surface immobilized reporter dyes [49], this method analyzes the bubble – air interface and is therefore probing the interface area that is more relevant for bubble interactions. It also affords an extensive quantitative measure instead of fluorescence intensity, which may vary due to photobleaching [46,49]. Compared to very recent SPR based approaches [52], the method is fully complementary, but adds simultaneously the power of AFM as a standard tool for the investigation of nanobubbles [15,20]. Having this new tool in hands, the formation, dynamics and localization of gas-filled nanobubbles can be experimentally addressed in a conclusive manner, which will be beneficial in rapid

169

screening of surfaces, in which nanobubbles mask surface-attached functionalities or contribute to adsorption and adhesive phenomena. In these future studies nanobubbles and nanodroplets can be conclusively differentiated, hence also systems comprising both species can be addressed. Acknowledgment The authors would like to thank Dr.-Ing. Martin Pätzold (Computer Graphics and Multimedia Systems Group, University of Siegen) for his contributions to design the graphics in Fig. 1 and gratefully acknowledge financial support from the Deutsche Forschungsgemeinschaft (DFG grant no. INST 221/87-1 FUGG), the European Research Council (ERC grant to HS, ERC grant agreement No. 279202) and the University of Siegen. Appendix A. Supplementary material Supplementary data to this article can be found online at https://doi.org/10.1016/j.jcis.2019.03.084. References [1] K.A. Mørch, Cavitation inception from bubble nuclei, Interf. Focus 5 (2015) 20150006. [2] I. Tzanakis, G.S.B. Lebon, D.G. Eskin, K.A. Pericleous, Characterisation of the ultrasonic acoustic spectrum and pressure field in aluminium melt with an advanced cavitometer, J. Mater. Process. Technol. 229 (2016) 582–586. [3] D. Fernandez Rivas, S. Kuhn, Synergy of microfluidics and ultrasound: Process intensification challenges and opportunities, Top. Curr. Chem. 374 (2016) 70. [4] T. Temesgen, T.T. Bui, M. Han, T. Kim, H. Park, Micro and nanobubble technologies as a new horizon for water-treatment techniques: a review, Adv. Colloid Interf. Sci. 246 (2017) 40–51. [5] J. Lux, A.M. Vezeridis, K. Hoyt, S.R. Adams, A.M. Armstrong, S.R. Sirsi, R.F. Mattrey, Thrombin-activatable microbubbles as potential ultrasound contrast agents for the detection of acute thrombosis, ACS Appl. Mater. Interf. 9 (2017) 37587–37596. [6] R. De Luca, G. Silvani, C. Scognamiglio, G. Sinibaldi, G. Peruzzi, M. Chinappi, M. F. Kiani, C.M. Casciola, Towards cavitation-enhanced permeability in blood vessel on a chip, AIP Conf. Proc. 1873 (2017) 020010. [7] M. Alheshibri, J. Qian, M. Jehannin, V.S.J. Craig, A history of nanobubbles, Langmuir 32 (2016) 11086–11100. [8] B.D. Johnson, R.C. Cooke, Generation of stabilized microbubbles in seawater, Science 213 (1981) 209–211. [9] P. Attard, The stability of nanobubbles, Eur. Phys. J. Spec. Top. (2013) 1–22. [10] P.S. Epstein, M.S. Plesset, On the stability of gas bubbles in liquid-gas solutions, J. Chem. Phys. 18 (1950) 1505–1509. [11] K. Kikuchi, I. Aoi, O. Takeo, T. Yoshinori, S. Yasuhiro, O. Zempachi, Concentration determination of oxygen nanobubbles in electrolyzed water, J. Colloid Interf. Sci. 329 (2009) 306–309. [12] H. Tsuge, Micro- and Nanobubbles: Fundamentals and Applications, Pan Stanford Publishing Pte. Ltd., Singapore, 2014. [13] N. Nirmalkar, A.W. Pacek, M. Barigou, On the existence and stability of bulk nanobubbles, Langmuir 34 (2018) 10964–10973. [14] S.H. Oh, J.-M. Kim, Generation and stability of bulk nanobubbles, Langmuir 33 (2017) 3818–3823. [15] D. Lohse, X.H. Zhang, Surface nanobubbles and nanodroplets, Rev. Mod. Phys. 87 (2015) 981–1035. [16] J.L. Parker, P.M. Claesson, P. Attard, Bubbles, cavities and the long-ranged attraction between hydrophobic surfaces, J. Phys. Chem. 98 (1994) 8468–8480. [17] S.T. Lou, Z.Q. Ouyang, Y. Zhang, X.J. Li, J. Hu, M.Q. Li, F.J. Yang, Nanobubbles on solid surface imaged by atomic force microscopy, J. Vac. Sci. Technol. B 18 (2000) 2573–2575. [18] N. Ishida, T. Inoue, M. Miyahara, K. Higashitani, Nanobubbles on a hydrophobic surface in water observed by tapping-mode atomic force microscopy, Langmuir 16 (2000) 6377–6380. [19] J.W. Tyrrell, P. Attard, Images of nanobubbles on hydrophobic surfaces and their interactions, Phys. Rev. Lett. 87 (2001) 176104. [20] B.M. Borkent, S. de Beer, F. Mugele, D. Lohse, On the shape of surface nanobubbles, Langmuir 26 (2010) 260–268. [21] N. Mishchuk, J. Ralston, D. Fornasiero, Influence of very small bubbles on particle/bubble heterocoagulation, J. Colloid Interf. Sci. 301 (2006) 168–175. [22] S. Calgaroto, A. Azevedo, J. Rubio, Flotation of quartz particles assisted by nanobubbles, Int. J. Mineral Proc. Suppl. C 137 (2015) 64–70. [23] R. Etchepare, H. Oliveira, A. Azevedo, J. Rubio, Separation of emulsified crude oil in saline water by dissolved air flotation with micro and nanobubbles, Sep. Purif. Techn. Suppl. C 186 (2017) 326–332. [24] C. Li, K. Zhen, Y. Hao, H. Zhang, Effect of dissolved gases in natural water on the flotation behavior of coal, Fuel 233 (2018) 604–609.

170

N. Hain et al. / Journal of Colloid and Interface Science 547 (2019) 162–170

[25] A. Hassanzadeh, M. Firouzi, B. Albijanic, M.S. Celik, A review on determination of particle–bubble encounter using analytical, experimental and numerical methods, Miner. Eng. 122 (2018) 296–311. [26] Z.H. Wu, X.H. Zhang, X.D. Zhang, J.L. Sun, Y.M. Dong, J. Hu, In situ AFM observation of BSA adsorption on HOPG with nanobubble, Chin. Sci. Bull. 52 (2007) 1913–1919. [27] G.M. Liu, V.S.J. Craig, Improved cleaning of hydrophilic protein-coated surfaces using the combination of nanobubbles and SDS, ACS Appl. Mater. Interf. 1 (2009) 481–487. [28] V. Kolivoska, M. Gal, M. Hromadova, S. Lachmanova, H. Tarabkova, P. Janda, L. Pospisil, A.M. Turonova, Bovine serum albumin film as a template for controlled nanopancake and nanobubble formation: In situ atomic force microscopy and nanolithography study, Colloids Surf. B 94 (2012) 213–219. [29] P. Janda, Nanobubble-assisted nanopatterning, in: J. Rodriguez-Hernandez, C. Drummond (Eds.), Polymer Surfaces in Motion: Unconventional Patterning Methods, Springer, Heidelberg, Germany, 2015, pp. 273–285. [30] P.K. Shende, D. Desai, R.S. Gaud, Role of Solid-gas interface of nanobubbles for therapeutic applications, Crit. Rev. Therap. Drug Carrier Syst. 35 (2018) 469– 494. [31] G.M. Liu, Z.H. Wu, V.S.J. Craig, Cleaning of protein-coated surfaces using nanobubbles – an investigation using a quartz crystal microbalance, J. Phys. Chem. C 112 (2008) 16748–16753. [32] S. Yang, A. Duisterwinkel, Removal of nanoparticles from plain and patterned surfaces using nanobubbles, Langmuir 27 (2011) 11430–11435. [33] W. Walczyk, H. Schönherr, Closer look at the effect of AFM imaging conditions on the apparent dimensions of surface nanobubbles, Langmuir 29 (2013) 620– 632. [34] W. Walczyk, N. Hain, H. Schönherr, Hydrodynamic effects of the tip movement on surface nanobubbles: a combined tapping mode, lift mode and force volume mode AFM study, Soft Matter 10 (2014) 5945–5954. [35] W. Walczyk, H. Schönherr, Hydrodynamic effects of the tip movement on surface nanobubbles: a combined tapping mode, lift mode and force volume mode AFM study, Langmuir 30 (2014) 11955–11965. [36] D. Lohse, X. Zhang, Pinning and gas oversaturation imply stable single surface nanobubbles, Phys. Rev. E 91 (2015) 031003(R). [37] S.M. Dammer, D. Lohse, Gas enrichment at liquid-wall interfaces, Phys. Rev. Lett. 96 (2006) 206101. [38] B.H. Tan, H. An, H.C.D. Ohl, Surface nanobubbles are stabilized by hydrophobic attraction, Phys. Rev. Lett. 120 (2018) 164502. [39] R.P. Berkelaar, E. Dietrich, G.A.M. Kip, E.S. Kooij, H.J.W. Zandvliet, D. Lohse, Exposing nanobubble-like objects to a degassed environment, Soft Matter 10 (2014) 4947–4955. [40] H. An, G. Liu, V.S.J. Craig, Wetting of nanophases: nanobubbles, nanodroplets and micropancakes on hydrophobic surfaces, Adv. Colloid Interf. Sci. 222 (2015) 9–17. [41] X.H. Zhang, A. Quinn, W. Ducker, Nanobubbles at the interface between water and a hydrophobic solid, Langmuir 24 (2008) 4756–4764.

[42] X.H. Zhang, A. Khan, W. Ducker, A nanoscale gas state, Phys. Rev. Lett. 98 (2007) 136101–136105. [43] J.D. Miller, Y. Hu, S. Veeramasuneni, Y. Lu, In situ detection of butane gas at a hydrophobic silicon surface, Colloids Surf. A 154 (1999) 137–147. [44] X.H. Zhang, X.J. Li, N. Maeda, J. Hu, Removal of induced nanobubbles from water/graphite interfaces by partial degassing, Langmuir 22 (2006) 9238– 9243. [45] S. Karpitschka, E. Dietrich, J.R.T. Seddon, H.J.W. Zandvliet, D. Lohse, H. Riegler, Nonintrusive optical visualization of surface nanobubbles, Phys. Rev. Lett. 109 (2012) 066102. [46] C.U. Chan, C.-D. Ohl, Total-internal-reflection-fluorescence microscopy for the study of nanobubble dynamics, Phys. Rev. Lett. 109 (2012) 174501. [47] C.U. Chan, L. Chen, M. Arora, C.-D. Ohl, Collapse of surface nanobubbles, Phys. Rev. Lett. 114 (2015) 114505. [48] N. Hain, D. Wesner, S.I. Druzhinin, H. Schönherr, Surface nanobubbles studied by time-resolved fluorescence microscopy methods combined with AFM: the impact of surface treatment on nanobubble nucleation, Langmuir 32 (2016) 11155–11163. [49] D. Seo, S.R. German, T.L. Mega, W.A. Ducker, Phase state of interfacial nanobubbles, J. Phys. Chem. C 119 (2015) 14262–14266. [50] C. Li, A.-M. Zhang, S. Wang, P. Cui, Formation and coalescence of nanobubbles under controlled gas concentration and species, AIP Adv. 8 (2018) 015104. [51] L. Zhang, B. Zhao, L. Xue, Z. Guo, Y. Dong, H. Fang, R. Tai, J. Hu, Imaging interfacial micro- and nano-bubbles by scanning transmission soft X-ray microscopy, J. Synchr. Rad. 20 (2013) 413–418. [52] Y.M. Fang, Z.M. Li, J.J. Jiang, H.Y. Chen, N.J. Tao, W. Wang, Intermittent photocatalytic activity of single CdS nanoparticles, PNAS 114 (2017) 10566–10571. [53] A. Aloi, N. Vilanova, L. Albertazzi, I.K. Voets, iPAINT: a general approach tailored to image the topology of interfaces with nanometer resolution, Nanoscale 8 (2016) 8712–8716. [54] A. Grinvald, I.Z. Steinberg, On the analysis of fluorescence decay kinetics by the method of least-squares, Anal. Biochem. 59 (1974) 583–598. [55] D. Magde, G.E. Rojas, P.G. Seybold, Solvent dependence of the fluorescence lifetimes of xanthene dyes, Photochem. Photobiol. 70 (1999) 737–744. [56] M. Kreiter, M. Prummer, B. Hecht, U.P. Wild, Orientation dependence of fluorescence lifetimes near an interface, J. Chem. Phys. 117 (2002) 9430–9435. [57] M. Savarese, A. Aliberti, I. de Santo, E. Battista, F. Causa, P.A. Netti, N. Rega, Fluorescence lifetimes and quantum yields of rhodamine derivatives: new insights from theory and experiment, J. Phys. Chem. A 116 (2012) 7491–7497. [58] F.L. Arbeloa, P.R. Ojeda, I.L. Arbeloa, Dimerization and trimerization of rhodamine 6G in aqueous solution, J. Chem. Soc., Faraday Trans. 2 (84) (1988) 1903–1912. [59] A. Penzkofer, Y. Lu, Fluorescence quenching of rhodamine 6G in methanol at high concentration, Chem. Phys. 103 (1986) 399–405. [60] H. Hillborg, N. Tomczak, A. Olàh, H. Schönherr, G.J. Vancso, Nanoscale hydrophobic recovery: a chemical force microscopy study of UV/ozone – treated crosslinked poly(dimethylsiloxane), Langmuir 20 (2004) 785–794.