Integrated atomic force microscopy and x-ray irradiation for in situ characterization of radiation-induced processes

Understanding reactions induced by ionizing radiation at material interfaces represents a challenge in many fields, including atmospheric chemistry,^1–6 nuclear reactor material design, radioactive waste repositories,^7–9 dosimetry,¹⁰ medical devices,^11,12 and space science engineering.⁵ Developing a fundamental understanding of these systems is best served by in situ observations because of the transient nature of radiolytic products. However, reproducing the environmental and ionizing radiation conditions in an experimental setting during analysis often proves difficult. Recently, in situ holders for electron microscopy, and combined synchrotron-based microscopy and spectroscopy, have been employed to observed radiation-induced phenomena in situ.^13–20 For instance, the formation and growth of uranyl peroxide induced by radiolysis produced by a scanning electron microscope e⁻ beam has been observed using an in situ liquid cell, while polymer morphology and identity have been measured by co-located atomic force microscopy (AFM) and synchrotron-based scanning transmission x-ray microcopy; the lipid membrane structure, morphology, nano-mechanics, and subsequent beam damage have been interrogated by a similarly collocated AFM and synchrotron-based x-ray reflectivity.^17,19,20 However, in most cases, the radiation environment and time scales achieved do not represent those of the reaction of interest. Additionally, the radiation being used is often necessary for the measurement itself, which imposes limitations on the type of data that can be acquired and complicates its interpretation. For example, typical dose rates for in situ liquid cell transmission electron microscopy and x-ray microscopy are in the range of 5 × 10^6 21 and 10⁴ Gy/h,¹³ respectively, far exceeding dose rates found even in high-level waste, which is estimated to result in 2 × 10⁶ kGy over 10 000 years.²² New insights will be made possible by developing techniques that can interrogate dynamic phenomenon in a radiation environment that is decoupled from the measurement probe, particularly under flexible environmental conditions and at high spatial resolution.

For this purpose, we have designed and developed a novel (AFM), hereafter termed the “rad-AFM,” that couples high-resolution imaging with in situ x-ray irradiation capabilities. The present paper overviews the instrument and illustrates its use in two case studies. First, the adhesion between a silicon AFM tip and a mica substrate is measured in water before, during, and after irradiation. A reversible decrease in adhesive force was observed the moment the interface was exposed to x rays and increased in magnitude over several minutes. In the second illustrative system, we examine the radiation-induced growth of KNO₃ on a KBr(100) surface catalyzed by air radiolysis products. In situ characterization under 18 kGy/h irradiation reveals surface structural evolution with nanometer resolution. Topography and lateral force imaging provided unique insight into the dynamic physiochemical changes of the KBr substrate and the nucleation and growth behavior of KNO₃ crystallites. These in situ observations provide a basis for better understanding the consequences of ionizing radiation at interfaces under controlled environmental conditions.

Our rad-AFM is an adaptation of the commercial instrument by Asylum, model MFP-3D. The system is housed in a home-built ¾ in. thick aluminum enclosure [Fig. 1(a-1)] that was designed for radiation shielding based on the penetration depth of 20 kV x rays and modeled using Monte Carlo N-Particle Transport software (Los Alamos National Laboratory, https://mcnp.lanl.gov/). The aluminum shield is interlocked such that the doors must be closed and either the scan head or a beam stop is in place above the sample before the x-ray source can be operated. The system within consists of an optical camera, aluminum enclosure, scan head, sample stage, and x-ray source [Figs. 1(a) and 1(b-2-5), respectively]. The AFM sits on a Hertzan AVI-200 active vibrational control unit, and the entire system fits within a Hertzan acoustic enclosure. The Moxtek MAGPRO 60 kV, 12 W x-ray source is mounted under the sample and AFM without making contact to reduce noise. The x-ray source has been modified with a copper cooling jacket nose cone [Figs. 1(a) and 1(b-7)], which is in series with a Rosewill PB120 central processing unit (CPU) liquid cooler [Fig. 1(a-6)], which replaces the original cooling fan. These heatsinks are connected to a Thermo Fisher Accel 250 LC recirculating chiller. This system ensures that the x-ray source is sufficiently cooled while removing heat from the sample with a minimal amount of noise introduced to the instrument. With this system in place, 1 ml of water placed in the sample cell above the x-ray source, at a room temperature of 21.9 °C, cools to 20.8 °C with the chiller on (set to 14.5 °C) and increases to 21.8 °C with the x-ray source on at 12 W, resulting in an overall ΔT of +0.1 °C. In comparison, if the stock fan is used, mechanical vibration introduces considerable noise and the sample cell heats to 36 °C during operation, introducing considerable drift. The x-ray assembly is mounted on an XYZ stage [Fig. 1(a-8)] such that the location of the x-ray spot may be chosen relative to the sample and the spot size and intensity adjusted by the sample/x-ray separation. The sample cell sits above the x-ray source on the sample stage connected to the scan head by Viton bellows sealed with Viton O-rings [Fig. 1(b)]. This allows the AFM head to move relative to the sample while still ensuring a gas/liquid tight seal. Samples are mounted on a clear 0.25 mm thick sheet of polyethylene terephthalate (PET) that is also sealed with Viton O-rings. This substrate reduces attenuation of x rays while providing adequate support and visual confirmation of the x-ray position. In this configuration, the tip may freely scan the sample while controlled gas or liquid is flowed through the sealed sample cell (see gas inlet and outlet) while under x-ray irradiation from below.

A fully automated mass flow controller apparatus was employed to control the relative humidity and flow rate through the fluid cell for experiments performed in the gas phase. This system was composed of two mass flow controllers (Alicat Scientific) and a humidity sensor (Vaisala). One mass flow controller was dedicated to anhydrous gas, while the other controlled the flow of gas that was nearly saturated with H₂O by flow through a semipermeable nafion tube (Permapure) immersed in degassed water. The combined gas streams flowed over the humidity sensor. The ratio of the flow rate of the two mass flow controllers was controlled in real-time using custom software that utilized feedback from the humidity sensor, providing constant flow at a target relative humidity.

The dose rate estimate of this configuration was measured by irradiating thin radiochromic films (HD-V2 Gafchromic film, International Specialty Products, Wayne, NJ, USA) at low power and extrapolating to full power. Upon irradiation, this film “self-develops,” changing color in an approximately linear relation to radiation dose [Fig. 2(a)]. An image of each film’s irradiated spot is obtained with a flatbed scanner (model 10 000, Hewlett Packard), analyzed using ImageJ software, and compared to a dose response curve from a set of films irradiated to a known dose using ¹³⁷Cs (662 keV). A three-dimensional profile of the true dose distribution (in Gy) received across the x-ray beam area, as well as the physical size of the spot, is then produced at each output and used to determine the dose rate at maximum power (12 W) [Figs. 2(b) and 2(c)]. The x-ray source output is a cone with an angle of 46° with a maximum power of 12 W and can operate with a beam energy from 4 to 60 kV and current between 0 and 1 mA. However, we have limited our system to a maximum voltage of 20 kV, thereby limiting the current to 600 μA at max power. For the initial experiments described here, the x-ray was operated at these peak values and a dose rate of 17.7 kGy/h was measured with a 3 mm x-ray source/sample distance.

AFM-based force curve measurements have been employed to directly measure adhesion in many systems to better understand chemical interactions,^23,24 protein structure,^25,26 particle–particle interactions,²³ material coatings,²⁷ and live cells.^28,29 Adhesion is an important factor that can be used to understand and predict the bonding and aggregation behavior of these materials and organisms. For instance, adhesion plays a prominent role in particle growth by attachment, aggregation (reversible), agglomeration (irreversible), and mobility in a given media such as soil. Perhaps the simplest interaction that can be studied is that between a bare standard AFM tip and a model atomically flat material, such as mica, in solution. However, even in this routinely deployed experimental system, complex physical and mechanical interactions between the tip material, substrate, and solution constituents are revealed by trends in adhesion determined by maximum loading force, the force loading rate, dwell time, salt type and concentration, pH, and hydration of cations present.^30,31 The typical applicability of classical theories such as Derjaguin, Landau, Verwey, and Overbeek (DLVO) and its extensions to such data, which mainly focus on the net electrostatic and van der Waals interactions, emphasizes the sensitivity of adhesion to solution conditions, which regulate, among other things, the electrostatic potential at both the tip and mica surfaces. Because x-ray irradiation at our achievable dose rates can be expected to strongly and reversibly influence speciation and surface charge by solution radiolysis, we sought to explore such effects by attempting to detect changes in adhesion during irradiation. Because the radiolysis products are transient species present only when x-ray irradiation is present, the experiment is ideal for the rad-AFM. Therefore, such measurements cannot be performed ex situ.

The x-ray response was measured by mapping the adhesion force between a standard silicon AFM tip and mica in water before, during, and after irradiation. A single crystal of mica (10 × 10 × 0.25 mm³, International Crystal Laboratories, NJ, USA) was freshly cleaved and fixed in place on the PET substrate using crystal bond and placed in a fluid cell with 2 ml of 18 MΩ-cm purity water (NANOpure Diamond, Barnstead). Force mapping was conducted using a new silicon nitride AFM tip (MSNL-D, Bruker Nano Inc.) with a setpoint force of 3 nN to produce a 32 × 32 pixel² map [Figs. 3(a) and 3(b)]. Each force map consists of an array of force curves measured by approaching and retracting the AFM tip at each point in the array. A value for a given feature of each force curve is then determined and displayed as a pixel with a color associated with the magnitude of that value. In the case of an adhesion map, the measured value is the pull-off force required before the cantilever deflection returns to zero [Fig. 3(c)]. Initially, an area-averaged adhesion value of 655 ± 65 pN was measured across the mica surface. After establishing this baseline, the x-ray source was turned on at full power for 6.5 min. The adhesion force immediately began to decrease until an average value of 267 ± 26 pN was measured. The x-ray source was then turned off, and the adhesion began to increase back to the baseline [Fig. 3(a)]. The scan direction was reversed (to reduce time between scans) for a second scan over the same region, while the adhesion force continued to increase until the baseline was re-established (over the same region that previously showed the lowest adhesion force while the x-ray source was on). The finding shows that the influence of the x-ray dose on the tip–mica adhesion force is reversible. The x-ray source was then turned on again and this process repeated, resulting in a similar drop in adhesion [Fig. 3(b)]. Previous studies concluded the adhered interfacial water layer on a charged tip (or particle) and a mica surface provides the primary basis for adhesion under our conditions.³² If so, the short-lived radiation-induced drop in adhesion demonstrated by these measurements could arise from disruption of these near-surface water layers by local accumulation of radiolytic products. Local heating effects cannot be ruled out, although little change in bulk water temperature was measured (∼0.1 °C).

Alkali halide salts irradiated in air undergo complex reactions that result in the formation of a nitrate of the alkali metal on the surface, as shown by infrared spectroscopy, x-ray diffraction, and chemical analysis.^33–36 This process is initiated by the irradiation of air, which produces a wide range of short-lived, highly reactive intermediate nitrogen- and oxygen-containing species, ions, and radicals.^37–44 Of the dozens of observed reactions, the most relevant are given by reactions (1)–(4) to produce NO₂, which is the most important radiolytic precursor for the formation of the NO₃⁻ ion as given by reaction (5),^{34,36,45–47}

where Me = alkali metal and X = halide.

In reaction (5), the halogen from the lattice desorbs into the gas phase in the form of volatile nitrosyl halide, while the alkali halides are converted into alkali nitrates. Sub-micron crystallites attributed to these radiation-induced nitrates have been observed using electron microscopy.⁴⁸ However, the evolution of the substrate and nitrate formation, including adsorption/desorption, surface migration, reactions, dissolution, and finally nucleation and growth processes, is poorly understood. To gain insight into this process, potassium bromide (KBr) (100) surfaces were characterized with the rad-AFM both during and after x-ray irradiation. This study builds on our recent prior work using the rad-AFM that established the importance of air concentration (which moderates the supply of N₂) and humidity on the surface transformation process, along with detailed ex situ characterization of the KNO₃ surface product phase.⁴⁹ Here, we focus in more detail on observations that relate to KBr surface changes, KNO₃ nucleation, and other instrumental details related to imaging conditions.

KBr single crystals (10 × 10 × 50 mm³, International Crystal Laboratories, NJ, USA) were freshly cleaved to 4 × 4 × 0.5 mm³ and fixed in place on the PET substrate by heating a small piece of crystal bond (Crystalbond 509, SPI Supplies) until liquid, affixing the KBr crystal, and cooling on an aluminum block. The fluid cell was then sealed and placed in the AFM instrument for imaging. The x-ray source/sample separation was minimized to 3 mm and aligned on one corner of the KBr crystal such that one-fourth of the beam was absorbed by the crystal, ensuring that the gas in the cell would be sufficiently irradiated. Air with a relative humidity of 60% was flowed through the sample cell for 1 h, and then gas flow was turned off and the system sealed before irradiation. Samples were imaged in contact/lateral force microscopy (LFM) mode with a soft cantilever (MSNL-D, 0.01 n/M, Bruker) to achieve a low imaging force of ∼0.6 nN in order to reduce tip-induced changes in nucleation and growth. LFM was employed to provide additional insights into the surface structure/chemistry. When compared to tapping mode imaging, no damage or consistent differences in morphology were evident (Ref. 49, No. 107). Images were processed using Gwyddion (v 2.55, http://gwyddion.net/) to plane flattened and row aligned (median of differences or polynomial 4) after masking crystallites. Time-lapse videos were made by creating stacks in ImageJ (1.47v, http://imagej.nih.gov/ij) and aligning by common fiducial features (plugin: NMS_fixTranslation_ver1.ijm, 2014, Nicholas M. Schneider).

Freshly cleaved KBr surfaces consist of large flat terraces separated by steps, as shown by AFM topography [Fig. 4(a)]. Initially, the KBr surface is pristine; however, over time in ambient air, the surface appears to roughen and imaging resolution is reduced [Figs. 4(b) and 4(c)]. Given enough time or after irradiation Fig. 5, the surface “clears” and step edges and surfaces begin to once again be clearly defined [Fig. 4(d)]. In addition to topography, LFM measures the tilt of the AFM cantilever and is captured simultaneously. This technique is sensitive to local surface features that cause changes in relative friction between the tip and the surface, including surface chemistry (e.g., hydrophobic/hydrophilic moieties, ionic species, etc.) or topological features (e.g., surface roughness, step edges, etc). Initially, step edges appear as sharp (bright) lines in the lateral channel caused by an abrupt height change [Figs. 4(e) and 4(f)]. However, over time, regions of higher friction expand first from step edges (bright triangular and “noisy” features) and eventually within terraces (bright bands) [Figs. 4(g) and 4(h)]. The high-friction regions (bright dots) that appear within terraces correspond to small pits in topography [Figs. 4(d) and 4(h)]. Overall, this dynamic behavior is indicative of the dissolution and reconstruction process that takes place as atmospheric water adsorbs to the surface first at step edges and later within terraces, evidenced by pitting. It thus seems likely that the observed “clearing” of initially poor imaging conditions results from greater tip–surface interactions that hinder imaging quality as compared to an initially dry or, after sufficient air exposure, hydrated surface.

After x-ray irradiation, ex situ AFM reveals KNO₃ crystallites formed on the surface that adopt a triangular and scalloped shape (Fig. 5), as observed previously by others^20,28,29 as well as in our prior rad-AFM study.⁴⁹ However, many additional morphologies were also observed, including parallelograms and needle-like and branched structures.⁴⁹ The identity of KNO₃ was confirmed by infrared spectroscopy, nano-secondary ionization mass spectroscopy (nano-SIMS), and micro-x-ray diffraction (μ-XRD). μ-XRD analysis identified the KNO₃ crystallites as the γ phase oriented epitaxially with the KBr {001} surface. The γ-KNO₃ phase exhibits ferroelectric and ferromagnetic behavior that is well suited for use in nonvolatile random access memories, ferroelectric random access memories, and dynamic random access memories.⁵⁰ Thus, our discovery of a radiation-induced synthesis pathway for γ-KNO₃ may be of interest to diverse fields.

Differences in crystallite density, morphology, and step edge affinity were observed between regions nearest and farthest from the center of x-ray irradiation. In the region nearest the x-ray, most crystallites were triangular and scalloped ranging from 22 to 46 nm in height with a surface contact area of 0.13–0.44 μm². A density of 0.48 crystallites/μm² was observed, with 19.5% of crystallites growing on/from step edges [Fig. 5(a)]. In the region farthest from the x-ray source, crystallites were rod-like and ranged from 6.2 to 19.7 nm in height, with a surface contact area of 0.09–0.20 μm² and a density of 0.12 crystallites/μm² with 81.3% associated with step edges [Figs. 5(d) and 5(h)]. These differences suggest dose-rate-dependent diffusion rates of radiolytically produced species. In both cases, crystallites were typically oriented along two perpendicular planes.

At higher resolution, three distinct species were observed in the region nearest the x-ray source. The large triangular crystals [Figs. 5(c) and 5(g-1)] were identified as KNO₃ by nano-SIMS, which indicated that regions of this crystallite shape were high in NO₃⁻ and depleted in Br⁻ concentrations.⁴⁹ In the region surrounding each of these larger crystals, a depression in the KBr substrate is formed due to increased dissolution caused by a local depletion of KNO₃ near growing crystals.⁴⁹ The compositions of the two additional surface products are yet to be identified. One is characterized by a height of ∼10.5 nm, a diameter of 105 nm, and a polycrystalline appearance in lateral force [Figs. 5(c) and 5(g-2)]. The other exhibits a height of ∼15.5 nm, a diameter of 65 nm, and a uniform circular region of very low friction in lateral force [Figs. 5(c) and 5(g-3)].

In situ AFM characterization of a KBr crystal was accomplished at full x-ray power under continuous imaging over a period of 20 h. The KBr crystal was first equilibrated by flowing 60% RH air through the sample cell for 1 h before gas flow was stopped and irradiation was conducted in a static volume of humid air. In the first several hours, few surface features aside from strep edges were observed in topography. However, changes in lateral force contrast that first initiated at the step edges in a manner similar to the unirradiated KBr [Figs. 4(e)–4(h)] were observed during this time frame [Figs. 6(a) and 6(b), Multimedia view]. After nearly 8 h of irradiation, crystallites first formed on the KBr surface. This long induction period, which has not been previously reported, is likely due to the accumulation of nitrate species from a cascade of low-efficiency reactions. Once nucleation was initiated, the number of nuclei increased rapidly until, after 11 h of irradiation, no further nucleation was observed [Fig. 6(c)]. Three crystallite populations were identified that were distinct in their size, morphology, and lateral force contrast. The largest crystallites have been positively identified by nano-SIMS as KNO₃.⁴⁹ The smaller crystallites were not large enough to confirm their composition with the analytical methods used. The average KNO₃ crystallite height and total surface coverage increased at a steady rate that slowly began to taper after about 13 h [Figs. 6(d) and 6(e)]. After nearly 19 h of irradiation, individual crystallites ranged from 27 to 49 nm in height with a surface contact area of 0.05–0.17 μm².

We have developed a first-of-a-kind rad-AFM that allows nanoscale visualization and characterization of radiation-induced reactions at material interfaces arising from x-ray irradiation under controlled environmental conditions in gas or liquid. Modifications to the commercial instrumental setup to integrate the x-ray source were done in such a way as to ensure low noise and heat transfer away from the sample stage, yielding stable high-resolution imaging without sample heating. After hundreds of hours of in situ irradiation, no adverse effects on the AFM scanner, electronics, or optical viewing components were apparent. The instrument enables exploration of radiation effects at interfaces in situ, which opens doors into both resolving ambiguities associated with ex situ studies and performing measurements that cannot be done ex situ. As an example of the latter, the rad-AFM enabled first measurements of solution radiolysis effects on the adhesive force between the AFM tip and a mica surface. Analysis of KBr irradiation under atmospheric conditions expanded our insight into the dynamic processes involved in the restructuring of the KBr substrate and subsequent KNO₃ nucleation and growth under environmentally controlled conditions. Additional instrumental design work is currently under way to incorporate an electron beam as an alternative radiation source into this system, which will provide higher local irradiation dose rates and enable direct comparison to electron microscopy observations. In contrast to existing experimental approaches, this new instrument provides a robust and flexible platform for in situ observation of radiation effects at material interfaces without tying the radiation type and dose to the imaging mechanism, which should greatly aid in advancing our understanding of these complex processes across a wide range of scientific contexts.

Development of the rad-AFM instrument was supported as part of the Laboratory Directed Research and Development Nuclear Processing Science Initiative (NPSI) at Pacific Northwest National Laboratory (PNNL). K.M.R. acknowledges support for his role in data interpretation and manuscript development as part of IDREAM (Interfacial Dynamics in Radioactive Environments and Materials), an Energy Frontier Research Center funded by the U.S. Department of Energy (DOE), Office of Science (SC), Basic Energy Sciences (BES) program. A portion of the research was performed in the Environmental Molecular Sciences Laboratory (EMSL), a national scientific user facility sponsored by the DOE Biological and Environmental Research program and located at PNNL. PNNL is a multiprogram national laboratory operated for DOE by the Battelle Memorial Institute under Contract No. DE‐AC05‐76RL0‐1830. We gratefully thank Sue B. Clark and Reid A. Peterson for their support of the rad-AFM development, Woody Buckner for calculating radiation shielding requirements, and James Ewing for instrument drawings.

The authors declare that they have no known competing interests that have influenced the work reported in this paper.

The data that support the findings of this study are available from the corresponding author upon reasonable request.