Evaluation of an X-ray-Excited Optical Microscope ... - ACS Publications

Article pubs.acs.org/ac

Evaluation of an X‑ray-Excited Optical Microscope for Chemical Imaging of Metal and Other Surfaces Pieter-Jan Sabbe,† Mark Dowsett,‡ Matthew Hand,‡ Rosie Grayburn,†,‡ Paul Thompson,§,∥ Wim Bras,⊥ and Annemie Adriaens*,† †

Department of Analytical Chemistry, Ghent University, Krijgslaan 281-S12, B-9000 Ghent, Belgium Department of Physics, University of Warwick, Coventry CV4 7AL, United Kingdom § XMaSThe UK CRG, ESRFThe European Synchrotron, CS40220, 38043 Grenoble Cedex 9, France ∥ Department of Physics, University of Liverpool, Liverpool L69 7ZX, United Kingdom ⊥ Netherlands Organisation for Scientific Research (NWO) DUBBLE@ESRF CS40220, 38043 Grenoble Cedex 9, France ‡

ABSTRACT: The application of a modular system for the nondestructive chemical imaging of metal and other surfaces is described using heritage metals as an example. The custombuilt X-ray-excited optical luminescence (XEOL) microscope, XEOM 1, images the chemical state and short-range atomic order of the top 200 nm of both amorphous and crystalline surfaces. A broad X-ray beam is used to illuminate large areas (up to 4 mm2) of the sample, and the resulting XEOL emission is collected simultaneously for each pixel by a chargecoupled device sensor to form an image. The input X-ray energy is incremented across a range typical for the X-ray absorption near-edge structure (XANES) and an image collected for each increment. The use of large-footprint beams combined with parallel detection allows the power density to be kept low and facilitates complete nondestructive XANES mapping on a reasonable time scale. In this study the microscope was evaluated by imaging copper surfaces with well-defined patterns of different corrosion products (cuprite Cu2O and nantokite CuCl). The images obtained show chemical contrast, and filtering the XEOL light allowed different corrosion products to be imaged separately. Absorption spectra extracted from software-selected regions of interest exhibit characteristic XANES fingerprints for the compounds present. Moreover, when the X-ray absorption edge positions were extracted from each spectrum, an oxidation state map of the sample could be compiled. The results show that this method allows one to obtain nondestructive and noninvasive information at the micrometer scale while using full-field imaging.

M

maps of large surfaces. To circumvent these problems, we have developed a method using X-ray-excited optical luminescence (XEOL) as an information carrier to construct chemical maps of metal surfaces. XEOL is the process of detecting photons from the optical region (ultraviolet to near-infrared) emitted when the sample absorbs X-rays typically originating from a tunable X-ray source such as a synchrotron.11,12 The photons detected by XEOL during X-ray bombardment carry a broad range of information.11 For example, the luminescent intensity can be modulated by the processes that give rise to the spectral features observed in the X-ray absorption near-edge structure (XANES) and the extended X-ray absorption fine structure (EXAFS) regions of conventional XAS. The resulting technique

onitoring the details of corrosion of metallic objects requires sophisticated tools; for example, synchrotron radiation is often used to study corrosion in real time.1−5 Over the past 20 years, X-ray techniques using synchrotron radiation as a probe, such as X-ray fluorescence,6,7 X-ray absorption spectroscopy (XAS),8 X-ray diffraction (XRD),5,6 and Fouriertransform infrared microspectroscopy,9,10 have been extensively used in corrosion studies. One example of research subjects to which this has been applied in a noninvasive manner is the chemical analysis and imaging of precious and rare heritage samples.1 The first tests of a new X-ray-excited optical microscopy tool (XEOM 1) whose design was stimulated by the need for nondestructive surface-specific chemical imaging of heritage metals are described here. Major problems with existing techniques include a lack of lateral resolution, surface specificity, and the time consumed in producing chemical © 2014 American Chemical Society

Received: September 2, 2014 Accepted: November 6, 2014 Published: November 6, 2014 11789

dx.doi.org/10.1021/ac503284r | Anal. Chem. 2014, 86, 11789−11796

Analytical Chemistry

Article

system. A major advantage is that a sample chamber filled with ambient air, helium (for transmission of low-energy inbound Xrays), or any other gas (for example, a corrosive gas mixture) or functioning as an electrochemical cell can be used. This means that patinas and corrosion layers, equivalent to those found on cultural heritage relevant metal objects, can be created in situ or maintained in their normal environment. Here, we report the first results obtained by XEOM 1, taken from samples with artificial corrosion layers relevant to copper. Cuprite (Cu2O) is the most common corrosion product of copper and its alloys and appears when fresh copper is exposed to clean damp air.31 Nantokite (CuCl) is one of the primary active corrosion agents that form on copper and its alloys. Its presence is often symptomatic of bronze disease and can degrade the long-term stability of an object.32,33 The test samples, described elsewhere in more detail,21 have been chosen since they exhibit a regular geometrical pattern, which allows the straightforward evaluation of the optical system and the determination of the lateral resolution.

of XEOL−XAS (or optically detected X-ray absorption spectroscopy, ODXAS) gives similar information regarding the chemical nature (e.g., oxidation state) and structural environment (e.g., bond length and coordination) of the absorbing atom.13 Moreover, spectral analysis or filtering of the light itself gives spectral information on the emitting states.14,15 The mean free path of visible photons is generally less than that of X-ray photons in a given material. Although the substrate contributes to the luminescent yield via secondary processes, e.g., via fluorescent X-rays exciting atoms in the superficial layer, XEOL emission is more surface-specific than conventional X-ray fluorescence. Whereas, for instance, the absorption length of 9 keV X-rays is 4 μm in copper, XEOL typically probes only the top 200 nm below the surface, which is an interesting region in the studies of metallic corrosion reactions. Elsewhere we have described the use of ODXAS to monitor electrochemical treatments of copper-based heritage alloys.14,16 However, corrosion processes do not generally evolve homogeneously over a metal surface.17 Therefore, lateral information, additional to that extracted from the volumeintegrated measurements, would be valuable. Although this can be done with a simple detector (such as the one described in ref 15) and an X-ray microprobe, we have shown that the exposure to the intense synchrotron beam can modify the surface composition of corrosion layers on copper faster than they can be imaged by this method.16 The modification of the sample due to exposure to intense X-ray beams is a general problem which is becoming more recognized. Less well ordered or amorphous materials appear to be prone to such modifications which are not always recognized as radiation-induced effects but sometimes take the disguise of cold crystal formation on stressed surfaces,18 crystallization nucleation,19 or changes in the valence state of metals.20To alleviate such problems, we have adopted a microscopy approach using a broad X-ray beam, broad-band imaging optics, and a pixelated charge-coupled device (CCD) as the detector.21 This allows the power density to which the sample will be exposed to be minimized while providing data with a similar statistical precision. The use of XEOL as an information source for chemical imaging is not novel and has been explored in the past. Most techniques used previously were designed to investigate ́ semiconductors and nanostructured materials. Martinez-Criado et al. described a system for beamline ID22 at the European Synchrotron Radiation Facility (ESRF) for scanning luminescence spectroscopy with a hard X-ray microbeam to image optical inhomogeneities in, e.g., GaN.22,23 Poolton et al. described the use of a portable system for the microimaging of X-ray absorption features via luminescent emission in solids (CLASSIX).24,25 Others have described the linking of established scanning techniques with the use of synchrotron radiation. In this area, for instance, Ishii et al. achieved nanometer-order lateral resolution by coupling a scanning capacitance microscopy module to the synchrotron beamline and detected the absorption fine structure by measuring the changes in sample capacitance that result from X-ray bombardment.26 Lacheri et al. designed a similar XAS scanning nearfield optical microscope with a tapered optical fiber probe to detect the XEOL of an irradiated sample. This approach allows element-specific profilometry to be performed next to chemical imaging.27−30 XEOM differs from these approaches because the sample is illuminated with a macroprobe X-ray beam; it is also distinctive in its use of a portable full-field optical imaging

■

EXPERIMENTAL SECTION Sample Preparation. Artificial oxide (predominantly Cu2O with a tenorite (CuO) impurity21) and CuCl layers were grown on transmission electron microscopy (TEM) sample holder grids (20 ± 3 μm thick, 3.05 mm diameter, 127 μm pitch; Athene Old 200, Agar Scientific). The TEM grids were used in combination with coupons of 99.9% pure copper (2 mm thick, 12.5 mm diameter; Goodfellow, Huntingdon, U.K.) and 9% (w/w) leaded bronze (2 mm thick, 12.5 mm diameter; IMMACO project34). Silver conductive paint (RS Components, Brussels, Belgium) was used as an adhesive between the edge of the grid and the underlying coupon. This resulted in the creation of well-defined regions of 90 × 90 μm (99.9% pure copper, holes in the mesh) and 90 × 35 μm (cuprite or nantokite, bars of the mesh) with copper in different states of corrosion. To obtain a clean metal surface, the coupons were first mechanically cleaned with P400 grit SiC abrasive paper (Buehler) and subsequently polished with a finer wet P1200 grit. The coupons were rinsed with deionized water and ultrasonically cleaned for 15 min in 2-propanol (SigmaAldrich). In the final stage, the coupons were manually polished using a microcloth (Buehler) with 1 μm deagglomerated α alumina particles (Buehler) suspended in water. The remaining adherent alumina particles were removed by rinsing and ultrasonic cleaning in 2-propanol for 15 min. The fragility of the TEM grids prevented them from being subjected to this procedure. Cuprite was grown on the copper coupons and TEM grids using a method described elsewhere in more detail.35 More specifically, it was grown on the copper coupon via heating in a reducing Bunsen burner flame until the coupon glowed bright cherry red. The coupon was then immediately removed from the flame. The sudden exposure of the hot surface to oxygen in the air resulted in a discoloration of the coupon to dark gray. A cuprite patina was produced on the grids by placing each one on a coupon and following a similar procedure. Without this precaution, the unsupported grid would melt upon direct contact with the flame. Nantokite layers were obtained by immersing a polished copper coupon for 1 h in a saturated CuCl2·H2O (Aldrich, >99%) solution followed by removal of the remaining salts by rinsing with deionized water. The grids were treated similarly 11790



Article

ray access. The first section can be replaced by an electrochemical cell and can be used for spectroelectrochemistry experiments.38 The second section, the optical column, holds four UV-grade fused silica lenses (Knight Optical Ltd., Harrietsham, U.K.) for focusing, magnification, and collimation of the XEOL signal onto a CCD chip. In the bandwidth of interest (250−1000 nm), the fused silica has a flat transmission in excess of 93% per lens dominated by surface reflection. However, because the optics are made from glass with only one refractive index, the column is not achromatic. The position of the focusing lens is stepper motor driven to allow remote controlled focusing. The end of the optical column is fitted with a holder for 25 mm diameter filters. Additive and subtractive color filters (Edmund Optics, Barrington, NJ), fluorescence filters (Knight Optical), and long- and short-pass filters (Edmund Optics) are used for the selection of a particular wavelength of the XEOL radiation. The final section can hold a 506 × 1024 pixel FLI microline ML 1109 CCD camera (Finger Lake Instruments, Lima, NY) or a 2048 × 2048 pixel Andor Ikon-L CCD detector (Andor, Belfast, U.K.). During the experiments, it is necessary to operate XEOM 1 remotely when the system is locked in the experimental hutch of an X-ray beamline. All operations are computer controlled from outside the experimental hutch through an electronic interface unit operated by control and acquisition software21 esaXAS (Mark Dowsett, EVA Surface Analysis, United Kingdom, and Matt Hand, University of Warwick, United Kingdom, 2013, 2014), which runs on a Windows laptop and communicates with the hardware through USB and U2351 multifunction ports (Agilent Technologies Inc., Santa Clara, CA). Beamline Settings. Synchrotron radiation experiments were carried out at the bending magnet beamlines BM28 (XMaS)39 and BM26A (DUBBLE)40 at the ESRF in Grenoble, France. An advantage of XEOM 1 is that no special microfocusing beamline optics is needed; consequently, setting up the instrument is quick and straightforward. At BM28, the χcircle of an 11-axis Huber 4-circle diffractometer was set at χ = 90° for mounting XEOM 1. This configuration allows the Xrays to hit the sample at 60° to the surface, resulting in a

by placing a drop of the saturated CuCl2·H2O solution on a new grid for 5 min with subsequent rinsing. Parts A and B of Figure 1 show pictures of the samples. Although we show

Figure 1. (a) Lead bronze coupon coated with an artificial nantokite layer. (b) Copper coupon coated with an artificial cuprite layer supporting a copper TEM sample holder grid coated with an artificial nantokite layer. (c) Sample composed of fluorescent paper and a copper TEM grid used for alignment and calibration of XEOM 1.

elsewhere36 that this process makes a mixture of CuCl and Cu2O due to a rapid reaction between the CuCl and the rinsing water, it should still provide more than sufficient nantokite for chemical differentiation. Powder samples of Cu2O (Fluka, >99% purity) and CuCl (Alfa Aesar, >99%) were used as high-purity materials for XEOL and XAS reference spectra. A Cu reference spectrum was recorded from a fresh polished copper coupon. XEOM 1. The design and construction of XEOM 1 is described fully elsewhere.21,37 It was manufactured using black acetal copolymer, shown to have a low intrinsic visible fluorescence under X-ray excitation, and comprises three main sections (Figure 2). The first consists of the sample stage surrounded by a number of ports at 10°, 30°, 45°, and 60°. The sample stage is the base for various tailored sample holders that allow coupons, grids, and powders to be placed precisely in the focal plane of the microscope optics. The ports are used for modular accessories such as an avalanche photodiode, a photomultiplier tube, and a webcam or for X-

Figure 2. 3D section view of the XEOM 1 microscopy system: (1) Andor Ikon-L camera with broad-band CCD, (2) Melles Griot optical rail, (3) filter, (4) single filter carrier, (5) projector lens, (6) remote focusing mechanism, (7) focusing/collimating lens, (8) beamline mounting, (9) X-rays (one of six possible ports), (10) aspherical doublet objective, (11) sample housing (may be replaced by electrochemical/environmental cell). 11791



Article

Figure 3. Graphical representation of the acquisition of an XEOM image stack with subsequent ROI analysis.

rectangular beam footprint of approximately 1 × 2 mm. At BM26A the microscope was mounted on a set of x−y and z stages and received primary X-rays through the 45° angle port. The X-rays travel, in both cases, in a vacuum until a few centimeters from the 45°/60° angle port entrance. Background light was minimized, and aluminum foil and black acetal tubes were used to prevent stray light from entering the microscope. A typical XEOM-acquired XEOL−XANES image stack consists of 100 or more images, each acquired with a monochromatic primary X-ray beam at some point in the energy range of 8960−9016 eV to span the Cu K-edge. Automation of data acquisition is obtained by synchronizing the operation of esaXAS with the monochromator and other beamline systems in one of two ways. Either esaXAS is slaved to the beamline using transistor−transistor logic (TTL) pulses derived from the beamline timing system (DUBBLE) or the monochromator stepping is slaved to the esaXAS clock using an output TTL pulse to trigger the step (XMaS). The acquisition time was sample dependent due to sampleto-sample variations in XEOL intensity. The acquisition time was chosen to produce data with acceptable statistics within a reasonable time; a typical measurement time was 2 h for the image set, but in certain cases this was extended to 7 h. The resulting spectra covered enough of the XANES region for comparison of its fingerprint with the corresponding reference spectra. Data Handling. All processing of the data acquired by XEOM 1 is carried out by the custom software package esaProject (Mark Dowsett, EVA Surface Analysis, 2014) . The data system for XEOM control and the custom image file format21 were designed to be complementary to esaProject.

Individual images or complete sets of images (image stacks) can be imported into the software. Images are shown as false-color intensity maps displaying the intensity value stored in each pixel. The code offers the capability to extract XEOL spectra from different regions of interest (ROI) or from the entire image (see the discussion below). The resulting spectra can be further processed by esaProject or exported in formats suitable for other software for XAFS data analysis, e.g., the Horae suite (i.e., Athena, Arthemis).41 An overview of the features that esaProject incorporates is given elsewhere21 and summarized below. Data Acquisition with XEOM 1. Figure 3 depicts the process of acquiring chemical data with XEOM 1. The X-ray beam is first aligned precisely with the area of the sample within the field of view. This is done by imaging a bare copper TEM grid fixed to a piece of fluorescent paper (Figure 1c) and moving the stages or goniometer on which XEOM 1 is mounted until the illuminated region is completely within the field of view of the camera. The clear pattern on the sample is generated by the shadow of the grid on the green fluorescent emission from the paper. Data collection generally proceeds as follows: The photon energy is stepped in 0.5−1 eV increments across the energy range of interest, dwelling on each step for a period slightly longer than the image acquisition time (typically ≥10 s). In contrast to the usual practice in XAS acquisition, where the photon energy step size and dwell time are varied in the preedge, edge, and postedge regions, the steps and data acquisition times are kept constant so that the background due to slowly decaying phosphorescent modes is smooth (the effect of varying the data acquisition times conventionally can be seen in 11792



Article

ref 16). The XEOM images may be delivered raw, and the normal CCD corrections for bias, flat field, and thermal noise can be applied by esaProject. Alternatively, this can be done by the acquisition software. After the entire image stack has been recorded, it is normalized to the beam monitor values for each image to correct for changes in the X-ray intensity and for monochromator glitches. From this point on, pixels are stored as binary 64 (8-byte double-precision floating point numbers) for speed and accuracy in onward computation. XEOL−XAS spectra are extracted through either pixel-by-pixel integration of the intensity over an entire image, via ROI analysis, or from individual columns of pixels orthogonal to the image stack. In the first case, a one-dimensional spectrum is obtained without any spatial resolution, similar to the data obtained by a photomultiplier tube in ODXAS experiments.14 In the other two cases, a degree of spatial resolution is obtained, depending on the size of the ROI or the binning of the images in the stack. During subsequent processing steps, chemical and structural information can be extracted from the XEOL−XAS spectra in the usual way for XAS with the proviso that the pre-edge background cannot always be subtracted meaningfully from the data.16 For example, oxidation state maps can be reconstructed from the absorption edge position in each column of pixels.

■

RESULTS AND DISCUSSION Imaging of a Homogeneous CuCl Layer on a Bronze Sample. Figure 4a displays the imaging characteristics of an image stack recorded from a leaded bronze coupon coated homogeneously with nantokite. A series of 102 images were acquired at X-ray beam energies from 8960 to 9010 eV using an exposure time of 25 s per point and no optical filter. The image stack displayed in Figure 4 is represented as a cuboid with the first image of the stack (at 8960 eV) as the top surface and the energy cross sections parallel to the x- and y-axes as the side faces. Illumination of the homogeneous nantokite layer results in the visualization of the beam intensity profile on the topsurface image. The beam follows a Gaussian profile with the highest intensity in the middle around ROI 1 (colored yellow) and fades out toward the edges of the image (purple/blue). This pattern is also observed on the cross section parallel to the x-axis through ROI 1 seen on the front face of the cuboid. Additionally, the absorption edge is clearly visible in this energy cross section with a sharp rise in intensity around 8979 eV. A similar trend is observed on the cross section parallel to the yaxis through ROI 1 at the right face of the cuboid. Figure 4b shows the normalized integrated XEOL intensity in the range of 8970−9000 eV for the total image and three selected ROIs. The data are a XANES fingerprint of a mixture of nantokite with a minor fraction of paratacamite (Cu2Cl(OH)3). Nantokite can be discerned by the presence of a characteristic “white line” (nomenclature historically used for a sharp peak occurring on the top of the edge30) at 8986 eV (B), resulting from the intra-atomic 1s → 4p Cu transition.42 The characteristic shoulder at 8984 eV (A) and the feature at 8991 eV (C) are also observed in the reference spectrum of pure nantokite. A part of the nantokite produced hydrolyzes rapidly with water from the rinsing procedure to form cuprite.36,43 Hence, some features characteristic of cuprite (e.g., shoulder A at 8984 eV and increased intensity around 8995 eV) are observed in the total spectrum. The inset gives the XEOL spectra of the reference powders for comparison.

Figure 4. (a) XEOM image stack acquired from a lead bronze coupon coated with nantokite. (b) Normalized XEOL−XANES spectra extracted from ROI 1, ROI 2, and ROI 3 and the total image. Inset: XEOL−XANES reference spectra of nantokite and paratacamite powder.

The areas selected for the ROI analysis are on the top-face image of the cuboid. The spectra extracted from the different ROIs can be superimposed to within the noise level, indicating there are no measurable changes among the different areas. Imaging of a CuCl Grid on a Cu2O Surface. Experiments prior to developing XEOM 1 revealed that optical filtering is advisible during the recording of XEOL data to alleviate statistical effects of background removal algorithms and to reduce pre-edge background levels.14 Significant chemical differentiation by imaging with XEOM 1 through different optical filters is demonstrated in Figure 5. Figure 5 displays two images of a nantokite-coated mesh on a cuprite coupon. The image in Figure 5a was acquired at 9000 eV (out of a series of 112 images collected at 8960−9016 eV) through a red fluorescence line filter (pass band 612−644 nm) with an acquisition time of 270 s per image. That in Figure 5b was acquired in a similar way, but with a green fluorescence line filter (pass band 513−556 nm) and with an acquisition time of 35 s. The green emission in Figure 5b is dominated by the coupon, with the mesh only shadowing the underlying surface. The emission is probably close to the 2.27 eV exciton (E0B) green line observed in room temperature photoluminescence from bulk cuprite by Park et al.,44 and the rather similar image seen through adjacent filters is consistent with the width of the 11793



Article

nantokite and cuprite. This is possibly due to strong red emission from the nantokite bars being distributed across the detector from regions outside the intended field of view and swamping the weak red emission from the cuprite. XEOM 1 is being modified to correct this problem. The pre-edge emission mainly results from inefficient excitation of L-level Auger transitions in copper, whereas at the edge itself the principal nonradiative electron mechanism switches to a KLL Auger transition.11 We have shown elsewhere that the pre-edge and postedge de-excitation channels can be different and excite different end states as distinguished by the color of the optical emission.16 Although the chemical resolution of the obtained images is not perfect, we used the data to test the concept of oxidation stage mapping (associating false color with the shift of an edge from a nominal position). The map in Figure 6 was compiled

Figure 6. Oxidation state map of a nantokite-coated grid on a cupritecoated substrate acquired through a red fluorescence line filter.

Figure 5. (a) Image at 9 keV from a cuprite-coated coupon supporting a nantokite-coated grid acquired through a red fluorescence line filter. (b) Image at 9 keV from a cuprite-coated coupon supporting a nantokite-coated grid acquired through a green fluorescence line filter. (c) Normalized XEOL−XANES spectra extracted from ROI 1 and ROI 2 in the red and green filtered image stacks. Inset: XEOL− XANES reference spectra of cuprite and nantokite powder.

using the image stack acquired through the red fluorescence filter. The position of the edge was determined for each spectrum (column of pixels along the energy axis) using a modified Savitsky−Golay differentiation46 in esaProject, converted to gray scale, and plotted in a 2D map. The difference in binding energy of the two products is observable in the shift of the absorption edge position, as demonstrated in Figure 5c (inset). This corresponds with the observations in Figure 6. The darker gray areas clearly correspond to regions on the grid (i.e., a greater edge shift from nantokite), and lighter areas correspond to edge positions extracted from areas where cuprite is present.

emission they observed. Red emission from cuprite is suppressed at room temperature. However, the use of a red filter resulted in a significantly different image (Figure 5a) favoring regions of the grid rather than the coupon. The red emission clearly carries the nantokite XANES information (Figure 5c) and is presumably due to a midband gap impurity since the normally observed room temperature photoemission from CuCl is in the far violet (383 nm).45 Figure 5c shows the XEOL spectra extracted from ROI 1 and ROI 2 indicated in Figure 5a,b. Using the green filter, both ROIs show the presence of cuprite; i.e., they indicate that cuprite is present on both mesh bars and on the coupon. This is expected due to the rapid hydrolyzation of nantokite to cuprite during the rinsing stage. When red filtered, the ROIs give different XEOL−XANES spectra: From ROI 1 (mesh bar) we see nantokite (note the edge shift compared to the spectrum acquired through the green filter from the same region). From ROI 2 the spectrum is (unexpectedly) a mixture of those for

CONCLUSIONS A novel XEOL-based microscopy system, XEOM 1, that can produce chemical maps from the surfaces of metals and other materials in air and controlled environments has been described. XEOM 1 adds lateral resolution to the information previously obtained by ODXAS for crystalline and amorphous materials. XEOM 1 acquires images from low-flux-density X-ray beams of millimeter dimensions. The minimum image exposure time for a 2048 × 2048 image is around 10 s, giving an acquisition time for a stack of images across the absorption edge of 20 min to a few hours. This results in statistical precision similar to that obtained with high-flux microbeams, but with a power density of several orders of magnitude lower, which reduces radiation damage to the sample. It is also likely to reduce X-ray-stimulated reactions with the surroundings an important consideration when attempting to use controlled environments. Image stacks were acquired from samples consisting of copper coupons and grids each provided with

■

11794



Article

(11) Rogalev, A.; Goulon In Chemical Applications of Synchrotron Radiation, Part II: X-ray Applications; Sham, T. K., Ed.; World Scientific: Singapore, 2002; pp 707−760. (12) Sham, T. K.; Jiang, D. T.; Coulthard, I.; Lorimer, J. W.; Feng, X. H.; Tan, K. H.; Frigo, S. P.; Rosenberg, R. A.; Houghton, D. C.; Bryskiewlcz, B. Nature 1993, 363, 331−333. (13) Soderholm, L.; Liu, G. K.; Antonio, M. R. J. Chem. Phys. 1998, 109, 6745−6752. (14) Dowsett, M. G.; Adriaens, A.; Jones, G. K. C.; Poolton, N.; Fiddy, S. Anal. Chem. 2008, 80, 8717−8724. (15) Taylor, R. P.; Finch, a. a.; Mosselmans, J. F. W.; Quinn, P. D. J. Lumin. 2013, 134, 49−58. (16) Adriaens, A.; Quinn, P.; Nikitenko, S.; Dowsett, M. G. Anal. Chem. 2013, 85, 9556−9563. (17) Ahmad, Z. Principles of Corrosion Engineering and Corrosion Control; 1st ed.; Butterworth-Heinemann: Oxford, U.K., 2006; pp 120−121. (18) Feldman, Y.; Lyahovitskaya, V.; Leitus, G.; Lubomirsky, I.; Wachtel, E.; Bushuev, V. a.; Vaughan, G.; Barkay, Z.; Rosenberg, Y. Appl. Phys. Lett. 2009, 95, 051919. (19) Martis, V.; Nikitenko, S.; Sen, S.; Sankar, G.; van Beek, W.; Filinchuk, Y.; Snigireva, I.; Bras, W. Cryst. Growth Des. 2011, 11, 2858−2865. (20) Stanley, H.; Banjeree, D.; van Breemen, L.; Ciston, J.; Liebscher, C.; Martis, V.; Hermida Merino, D.; Longo, A.; Pattison, P.; Peters, G.; Portale, G.; Sen, S.; Bras, W. CrystEngComm 2014, 16, 9331−9339. (21) Hand, M. Development of an X-ray Excited Optical Luminescence Microscope (XEOM). Ph.D. Dissertation, University of Warwick, June 2014. (22) Martínez-Criado, G.; Alen, B.; Homs, A.; Somogyi, A.; Miskys, C.; Susini, J.; Pereira-Lachataignerais, J.; Martínez-Pastor, J. Appl. Phys. Lett. 2006, 89, 221913. (23) Martínez-Criado, G.; Alén, B.; Sans, J. a.; Homs, a.; Kieffer, I.; Tucoulou, R.; Cloetens, P.; Segura-Ruiz, J.; Susini, J.; Yoo, J.; Yi, G. Nucl. Instrum. Methods Phys. Res., Sect. B 2012, 284, 36−39. (24) Poolton, N.; Towlson, B.; Hamilton, B.; Evans, D. Nucl. Instrum. Methods Phys. Res., Sect. B 2006, 246, 445−451. (25) Nasir, M. E.; Hamilton, B.; Allsop, D. W. E. Nucl. Instrum. Methods Phys. Res., Sect. B 2010, 268, 251−253. (26) Ishii, M.; Uchihashi, T. Phys. B (Amsterdam, Neth.) 2003, 340− 342, 1142−1146. (27) Larcheri, S.; Rocca, F.; Pailharey, D.; Jandard, F.; Graziola, R.; Kuzmin, a; Kalendarev, R.; Purans, J. Micron 2009, 40, 61−65. (28) Larcheri, S.; Rocca, F.; Jandard, F.; Pailharey, D.; Graziola, R.; Kuzmin, A.; Purans, J. Rev. Sci. Instrum. 2008, 79, 013702. (29) Rosenberg, R. A.; Zohar, S.; Keavney, D.; Divan, R.; Rosenmann, D.; Mascarenhas, A.; Steiner, M. A. Rev. Sci. Instrum. 2012, 83, 073701. (30) Sham, T.; Gordon, R. AIP Conf. Proc. 2010, 1234, 133−136. (31) Stambolov, T. The Corrosion and Conservation of Metallic Antiquities and Works of Art; Central Laboratory for Objects of Art and Science: Amsterdam, 1985. (32) Scott, D. J. Am. Inst. Conserv. 1990, 29, 193−206. (33) Wang, J.; Xu, C.; Lv, G. Appl. Surf. Sci. 2006, 252, 6294−6303. (34) Beldjoudi, T.; Bardet, F.; Lacoudre, N.; Andrieu, S.; Adriaens, A.; Constantinides, I.; Brunella, P. Rev. Metall. (Les Ulis, Fr.) 2001, 803−808. (35) Sabbe, P.-J.; Dowsett, M. G.; De Keersmaecker, M.; Hand, M.; Thompson, P.; Adriaens, A. Submitted for publication to Appl. Surf. Sci., 2014. (36) Dowsett, M.; Adriaens, A.; Martin, C.; Bouchenoire, L. Anal. Chem. 2012, 84, 4866−4872. (37) Dowsett, M.; Hand, M.; Sabbe, P.-J.; Adriaens, A. Submitted for publication to Heritage Sci., 2014. (38) Dowsett, M. G.; Adriaens, A. Anal. Chem. 2006, 78, 3360−3365. (39) Brown, S. D.; Bouchenoire, L.; Bowyer, D.; Kervin, J.; Laundy, D.; Longfield, M. J.; Mannix, D.; Paul, D. F.; Stunault, a.; Thompson, P.; Cooper, M. J.; Lucas, C. a.; Stirling, W. G. J. Synchrotron Radiat. 2001, 8, 1172−1181.

different artificial corrosion layers, which facilitated the straightforward evaluation of the imaging characteristics of the microscope. The first XEOM images show clear chemical contrast from copper corrosion products, and upon optical filtering, the XEOL emission from different corrosion products could be imaged separately. Bespoke image software can process the XEOM image stacks to reduced the data sets, which allows XEOL−XAS spectra to be extracted from the entire image, from a user-defined ROI, or from individual pixel columns; these can subsequently be processed, for example, into 2D edge-shift (oxidation state) images. The results demonstrate that chemical imaging on the micrometer scale is feasible by this relatively nondestructive and potentially noninvasive technique.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Phone: +32 9 264 4826. Fax: +32 9 264 4960. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS We are greatly indebted to the Research FoundationFlanders (FWO) and the Paul Instrument Fund of the Royal Society for financial support and to FWO for access to the DUBBLE beamlines. P.-J.S. and M.H. respectively acknowledge Ghent University and the U.K. Engineering and Physical Sciences Research Council (EPSRC) for funding. BM28 (XMaS) is a beamline funded as a midrange facility by the EPSRC. We acknowledge the Computing Group at ESRF for providing assistance with interfacing and also Adrian Lovejoy, David Greenshields, and Bob Day of the Electronics Workshop at The University of Warwick for their help. The XEOM hardware was designed and built by EVA Surface Analysis, United Kingdom, and the proof of concept device was funded by the Paul Instrument Fund of the Royal Society.

■

REFERENCES

(1) Cotte, M.; Susini, J.; Dik, J.; Janssens, K. Acc. Chem. Res. 2010, 43, 705−714. (2) Yilmazbayhan, A.; Motta, A. T.; Comstock, R. J.; Sabol, G. P.; Lai, B.; Cai, Z. J. Nucl. Mater. 2004, 324, 6−22. (3) Eckermann, F.; Suter, T.; Uggowitzer, P. J.; Afseth, A.; Davenport, A. J.; Connolly, B. J.; Larsen, M. H.; Carlo, F. De; Schmutz, P. Corros. Sci. 2008, 50, 3455−3466. (4) Palmer, T. .; Elmer, J. .; Babu, S. . Mater. Sci. Eng., A 2004, 374, 307−321. (5) Vasquez, M. J.; Halada, G. P.; Clayton, C. R. Electrochim. Acta 2002, 47, 3105−3115. (6) Janssens, K.; Vittiglio, G.; Deraedt, I.; Aerts, A.; Vekemans, B.; Vincze, L.; Wei, F.; Deryck, I.; Schalm, O.; Adams, F.; Rindby, A.; Kn, A.; Simionovici, A.; Snigirev, A. X-Ray Spectrom. 2000, 91, 73−91. (7) Janssens, K.; De Nolf, W.; Van Der Snickt, G.; Vincze, L.; Vekemans, B.; Terzano, R.; Brenker, F. E. TrAC, Trends Anal. Chem. 2010, 29, 464−478. (8) Leyssens, K. Monitoring the Conservation Treatment of Corroded Cupreous Artefacts: The Use of Electrochemistry and Synchrotron Radiation Based Spectroelectrochemistry. Ph.D. Dissertation, Ghent University, June 2006. (9) Young, M. L.; Casadio, F.; Schnepp, S.; Pearlstein, E.; Almer, J. D.; Haeffner, D. R. Appl. Phys. A: Mater. Sci. Process. 2010, 100, 635− 646. (10) Reguer, S.; Mirambet, F.; Dooryhee, E.; Hodeau, J.-L.; Dillmann, P.; Lagarde, P. Corros. Sci. 2009, 51, 2795−2802. 11795



Article

(40) Borsboom, M.; Bras, W.; Cerjak, I.; Detollenaere, D.; Glastra Van Loon, D.; Goedtkindt, P.; Konijnenburg, M.; Lassing, P.; Levine, Y. K.; Munneke, B.; Oversluizen, M.; Van Tol, R.; Vlieg, E. J. Synchrotron Radiat. 1998, 5, 518−520. (41) Ravel, B.; Newville, M. J. Synchrotron Radiat. 2005, 12, 537− 541. (42) Hamza, S.; Khan, M. A.; Lewonczuk, S.; Ringeissen, J. Solid State Commun. 1990, 75, 29−33. (43) Grayburn, R.; Dowsett, M. G.; Hand, M.; Sabbe, P.-J.; Thompson, P.; Adriaens, A. Accepted for publication to Corros. Sci., 2014. (44) Park, J.-W.; Jang, H.; Kim, S.; Choi, S.-H.; Lee, H.; Kang, J.; Wei, S.-H. J. Appl. Phys. 2011, 110, 103503. (45) Rajani, K. V.; Olabanji Lucas, F.; Daniels, S.; Danieluk, D.; Bradley, a. L.; Cowley, a.; Alam, M. M.; McNally, P. J. Thin Solid Films 2011, 519, 6064−6068. (46) Press, W. H.; Teukolsky, S. A.; Vetterling, W. T.; Flannery, B. P. Numerical Recipes in C: The Art of Scientific Computing; 2nd ed.; Cambridge University Press: New York, 1992; p 189.

11796
