International
Tables for
Crystallography
Volume I
X-ray absorption spectroscopy and related techniques
Edited by C. T. Chantler, F. Boscherini and B. Bunker

International Tables for Crystallography (2022). Vol. I. Early view chapter
https://doi.org/10.1107/S1574870720007569

X-ray excited optical luminescence

Tsun Kong Shama*

aDepartment of Chemistry, Western University, 1151 Richmond Street, London, Ontario N6A 5B7, Canada
Correspondence e-mail: tsham@uwo.ca

The phenomenon and the technique often referred to as X-ray excited optical luminescence (XEOL) are described in both the energy and time domains. Emphasis is placed on excitation with X-rays from a tuneable synchrotron light source and the tracking of XEOL upon preferential excitation of a given core level of an atom in a chemical environment. The combined measurements of XEOL and X-ray absorption at the near edge, X-ray absorption near-edge structure (XANES), are illustrated with examples and the prospect of XEOL imaging is also noted.

Keywords: X-ray excited optical luminescence; energy and time domains; synchrotron radiation; X-ray absorption near-edge structure; optical XANES; site specificity; energy-transfer dynamics; XEOL imaging.

1. Introduction

Photoluminescence in the UV, visible and near-infrared region of the electromagnetic spectrum from a material (phosphor) upon X-ray irradiation, a phenomenon contemporarily known as XEOL (X-ray excited optical luminescence), is as old as the discovery of X-rays. XEOL is now used daily in research laboratories worldwide to `see' X-rays. The study of the conversion of X-ray energy into optical photon emission by matter has gained noticeability in the last couple of decades following the advancement of synchrotron light sources, which now provide unprecedentedly bright and tuneable X-rays. These magnificent light sources for the first time allow investigation of the optical response upon preferential excitation of a specific core level across an absorption edge of an atom in a chemical environment, providing element, site and chemical specificity for materials analysis in tracking the origin of the luminescence. In addition, the time structure of a synchrotron light source also provides the capability to study the post-excitation energy transfer and radiative decay dynamic processes. This chapter discusses the physical origin of these processes in both the energy and time domains and how they are intimately related to the morphology, crystallinity and sizes of the light-emitting material, which in turn provides a unique tool for materials analysis. The emphasis here is placed on soft (30–1500 eV) and tender (1.5–5 keV) X-ray excited XEOL and its interplay with materials properties via combined studies of XEOL and XANES. An overview of XEOL in the pre-synchrotron era and from hard X-ray excitations can be found in a review by Rogalev & Goulon (2002[link]).

2. XEOL: the energy domain

2.1. XEOL, photoluminescence and cathodoluminescence

XEOL as a phenomenon and a technique can be summarized as `X-ray photons in, optical photons out, the optical photons are analysed in conjunction with element-specific excitation' (Sham et al., 1993[link]). It is useful to begin with a comparison of XEOL with the related but more conventional techniques of optical emission studies excited by UV–visible sources and energetic electrons, i.e. photoluminescence (PL) and cathodoluminescence (CL), respectively (Lakowicz, 2006[link]; Ronda, 2008[link]; Rodnyi, 1997[link]). PL involves valence excitations (for example HOMO to LUMO in molecules and valence band to conduction band in semiconductors) and their corresponding radiative decay channels (fluorescence/phosphorescence). CL involves the inelastic scattering of energetic electrons which transfer their energy to excite core electrons to the continuum; the subsequent decay of the core hole produces Auger electrons and fluorescent X-rays (Auger is the dominant process for low-Z elements); both the primary electrons and Auger electrons will thermalize in the solid, producing a thermalization track (Erbil et al., 1988[link]; Hu et al., 2002[link]; Sham & Rosenberg, 2007[link]; Sham, 2014[link], 2015[link]). The thermalization track is determined by the kinetic energy-dependent escape depth/inelastic mean free path1 (IMFP) of the electrons, often referred to as the universal curve in X-ray photoelectron spectroscopy (XPS; Fadley, 2015[link]; Emura et al., 1993[link]). This consideration is especially important for nanomaterials with a size that is smaller than the path for complete thermalization (see below). Thermalization is a cascade process in which energetic electrons lose their energy with each collision, producing shallower holes and secondary electrons as they travel in the solid (for example in a semiconductor) until they slow down and ultimately settle in the bottom of the conduction band and the corresponding holes in the top of the valence band. These electrons and holes can form excitons (a hydrogen-like atom of an electron and a hole with a binding energy of the order of meV to tens of meV below the valence band; VB), which can recombine radiatively, emitting a photon with nearly the energy of the band gap (minus the binding energy of the exciton), often termed the optical band gap or simply the band gap (BG) to the first approximation. Energy transfer to defect states in the band gap will occur at the expense of BG emission, leading to defect emission (DE) at longer wavelengths. BG emission is often sharp and fast, whereas DE is often broad and slow. The quality of the crystallinity of the specimen can often be determined by the branching ratio of BG emission and DE.

Although XEOL does not require synchrotron radiation, a tuneable synchrotron light source with tuneability varying from soft to hard X-rays can access nearly all of the important core levels of all of the elements in the periodic table. Thus, XEOL combined with X-ray absorption near-edge structure (XANES), a technique that probes the local structure of the absorbing atom, can provide element, site and chemical specificity, which are not available with laboratory X-ray line sources (Bremsstrahlung is too weak), PL and CL studies (Hu et al., 2002[link]; Sham & Rosenberg, 2007[link]; Sham, 2014[link], 2015[link]).

2.2. XEOL, X-ray absorption length and electron-escape depth

The availability of tuneable soft and tender X-rays (for example ∼30 eV–5 keV) from synchrotrons has made the XEOL technique unique and very attractive. Hard X-rays usually penetrate significantly into the bulk, while soft X-rays have very short penetration depths, resulting in a situation where all X-rays are absorbed by the specimen (Sham et al., 1993[link]). This total absorption situation has some interesting bearings for photon-in photon-out spectroscopy such as XEOL studies of thick film/bulk structure versus thin films/nanostructures, sometimes leading to the inversion of XANES recorded in the optical yield.

Fig. 1[link](a) shows the attenuation length (1/e attenuation, sometimes known as the one absorption length) of X-rays at a wide range of energies in ZnO, a common phosphor. This is an important parameter in the XEOL technique. One can see that the attenuation length in the hard X-ray region, the Zn K edge (∼10 keV), is more than 10 µm, whereas at the Zn L edge (∼1000 eV) and the O K edge (∼540 eV) the attenuation length is submicrometre. For instance, the attenuation length just below the Zn L3 edge is ∼900 nm and that above it is 200 nm. This difference has some interesting implications for the optical yield across the O K edge and Zn L edge. Fig. 1[link](b) shows the morphology and crystallinity dependence of XEOL from a ZnO nanoneedle (NN), a nearly perfect nanocrystal and a ZnO nanowire (NW), with noticeable defects under TEM inspection (Sun et al., 2005[link]).

[Figure 1]

Figure 1

(a) The attenuation length of X-ray absorption in ZnO at the O K edge and the Zn K and L edges. (b) XEOL from a highly crystalline ZnO nanoneedle (NN) and a ZnO nanowire (NW) with noticeable defects. The insets show the corresponding TEM images.

Although XEOL begins with the absorption of X-rays, the light-emission process is largely the work of secondary electrons, the production of which is dependent on the absorption process. As noted above, energy transfer from X-rays absorbed by the specimen to the optical channel is carried out via the thermalization of energetic electrons, and the thermalization track can be understood on the basis of the `universal curve' of the IMFP (escape depth) of electrons (Fadley, 2015[link]; Emura et al., 1993[link]). In thin films and nano­structures, where the thermalization process is often truncated, incomplete thermalization can lead to a trackable variation in optical yields, providing valuable information on the depth profile of the specimen. The XEOL process is illustrated in Fig. 2[link], where the left panel shows the excitation and the thermalization process leading to the competing band gap and defect luminescence and the right panel shows the thermalization of an Auger electron and the universal curve of electron-escape depths.

[Figure 2]

Figure 2

Left panel: X-ray absorption at and above the edge (vertical arrows). The core hole primarily decays via an Auger process (dotted arrows). The photoelectron and Auger electron thermalize (dashed arrows), producing secondary electrons (e) and holes (h) in their path which settle in the bottom of the CB and the top of the VB, respectively. Radiative e + h recombination yields near-band-gap (*, exciton) and defect emission (downward arrows). Right panel: excitation of atom A in a solid with photon energy Ei ≥ the threshold energy of an absorption edge of A and the emission of corresponding Auger electrons with energy EAuger at a given angle of incidence α and an angle of emission β. z is the depth axis. lα = z/sinα and lβ = z/sinβ are the attenuation lengths of the incident photon and the Auger electrons, respectively. The circled region magnified on the right shows energy transfer from the Auger electron to the optical channel via inelastic scattering as the electron is thermalized in the solid. The length of the thermalization track can be determined by the universal curve/escape depth (bottom).

The universal curve can be used to determine the length of the thermalization track of electrons with a given energy. For example, while a 5000 eV electron will thermalize nearly completely at three attenuation lengths (∼15 nm based on the escape depth), a 1000 eV electron will thermalize after ∼4 nm and the thermalization track for a 100 eV electron is only ∼2 nm. Thus, XEOL is particularly useful for the study of the optical properties of nanomaterials and thin films with a dimension comparable to or smaller than the thermalization path.

2.3. XEOL and X-ray absorption fine-structure (XAFS) spectroscopy

XEOL is a powerful tool when used together with X-ray absorption fine structure (XAFS). XAFS is an element-specific technique that tracks the modulation of the absorption coefficient (relative to a free atom) of an element in a chemical environment and is often divided into XANES, the near-edge region where multiple scattering dominates, and extended X-ray absorption fine structure (EXAFS), the extended region where single scattering dominates. In the soft X-region, XANES is the region of interest in connection with XEOL.

2.3.1. XEOL from optically thin and thick samples and XEOL yield across an edge

The XEOL process is generally complex but trackable (Sun et al., 2005[link]). To the first approximation, one assumes that the optical photon intensity is proportional to the absorption,[I_{\rm XEOL} \propto I_{\rm abs} = I_{\rm o}[1 - \exp(-\mu t)], \eqno(1)]where Iabs is the photon flux absorbed, Io is the incoming photon flux, μ is the absorption coefficient and t is the sample thickness. Thus, for a thin sample, with μt ≪ 1,[I_{\rm XEOL} \propto \mu t \eqno(2)]and the optical yield generally tracks the absorption behaviour of the light-emitting materials. For a thick sample, all of the X-rays are absorbed, and the optical response is entirely dependent on the proportional parameter, which is often non­linear to the absorption coefficient. The XEOL yield can be expressed as[I_{\rm XEOL} \simeq \textstyle \int \limits_0^t f(n)n(x)\,{\rm d}x, \eqno(3)]where f(n) is the quantum yield of the luminescence in the bulk (energetic electrons are completely thermalized) and n(x) is the spatial distribution of the secondary electron excitations (electrons in the CB and holes in the VB) created along the thermalization track of the energetic electrons (photoelectron and Auger; see Fig. 2[link]) at a distance x from the surface. Assuming that f(n) is a constant, which is a reasonable assumption since this is a bulk parameter, the XEOL yield depends entirely on n(x). It is known that in semiconductors the electrons and holes can diffuse long distances, over many lattice constants, in the solid to the surface and recombine non­radiatively instead of forming excitons or being trapped by defects; energetic electrons can also escape the solid without being fully thermalized. For example, in surface atoms an energetic electron can escape the surface without participating in thermalization; in nanostructures, the thermalization path can be truncated if the size of the nano­structure is smaller than the thermalization path. These events will significantly hamper the XEOL yield intensity. It has been shown that for total absorption (VUV and soft X-ray; Rogalev & Goulon, 1997[link])[n(x) = {{\tau} \over {\mu ^2 L^2 - 1}} \left\{\exp \left [-\left ({x \over {L}} \right)\right] - \exp(-\mu t)\right \}, \eqno (4)]where L = (Dτ)1/2 is the diffusion length with diffusion rate D and τ is the total lifetime. Thus, for a thick sample, it has been shown that for a semi-infinite sample[I_{\rm XEOL}\simeq f{1 \over {1 + \mu L}}. \eqno (5)]

Thus, for a thick sample the XEOL yield will decrease with increasing absorption; for an abrupt change in absorption coefficient across an absorption edge, the optical yield will be inverted. Fig. 3[link] illustrates the normal and inverted XAFS at the Zn K edge recorded with XEOL or photoluminescence yield (PLY) for a thin ZnO NN film and a thick ZnO (0001) single crystal (Heigl et al., 2007[link]).

[Figure 3]

Figure 3

(a) Zn K-edge XANES recorded in PLY from the BG emission of a ZnO nanoneedle (NN) and (b) time-gated zero-order emission from a ZnO (0001) single crystal (defect emission dominates) with a 20–150 ns time window in a 153 ns dark gap.

For samples of intermediate thickness, multicomponent systems and organic light-emitting materials, and very localized de-excitations, the diffusion model is not always valid and n(x) is often not a linear function of μ, sometimes leading to the partial inversion of XANES recorded in XEOL (see Fig. 4[link]a), but the data can usually be interpreted at least semi-quantitatively in most cases (Sham & Rosenberg, 2007[link]; Sham, 2014[link], 2015[link]; Heigl et al., 2007[link]).

[Figure 4]

Figure 4

The left panel shows (a) the PLY and total electron yield (TEY) XANES at the Si K edge of a porous silicon (PS) sample deposited reductively with rhodium on the surface (PS is a reducing agent) and (b) the Rh L3,2 edge with PLY and TEY (Kim et al., 2004[link]). The PLY XANES diminishes at the SiO2 resonance (∼1848 eV) since the luminescence is largely of an elemental silicon origin (quantum confinement; see inset). The Rh L3,2-edge XANES in PLY is inverted since rhodium is not a light emitter. The right panel shows the luminescence from a thin film of GeO2 nanowires (NW) on a SiO2 substrate excited at the Ge L edge and O K edge, which appear totally different (peak A is from SiO2 and peaks B and C are from GeO2); this is due to the different penetration depths of the soft X-rays (see bottom illustration; Armelao et al., 2012[link]). At the Ge L edge the photon can penetrate the thin film and excite the luminescence from the SiO2 substrate. Insets (a) and (b) show the Ge L3,2-edge and O K-edge XANES, respectively; inset (c) shows the angle-resolved XEOL confirming that peaks B and C are from the surface region (GeO2). Inset (d) shows the variation of the XEOL across the Ge L3,2 edge.

2.3.2. XEOL and chemical specificity

XEOL is unique in probing specific sites of a multi-element composite that is responsible for luminescence. There are several cases. (i) When a composite containing more than one site (absorption edge) is probed by XEOL, sites responsible for the luminescence will yield the right-side-up XANES in PLY; sites that are dark will exhibit an inverted edge due to the drop in quantum efficiency, since at the edge of the dark site an abrupt increase in the fraction of the total photon flux is now absorbed by the dark site with a lower quantum yield. (ii) Similarly, XEOL using tuneable X-rays can also be used to track the inhomogeneity of a specimen with unevenly distributed optical sites. (iii) XEOL can also be used to track the energy transfer across the interface of hetero nanostructures. Figs. 4[link] and 5[link] illustrate some of these cases.

[Figure 5]

Figure 5

Left, top panel: XEOL excited across the Si K edge of a silicon nanowire core, silicon oxide shell specimen showing three emission bands and the photon energy-dependent branching ratio (right inset) across the resonance of elemental silicon and silicon dioxide (left inset) (Sham et al., 2004[link]). Left, bottom panel: partial luminescence yield XANES from the three emission bands showing the origin of the luminescence from the silicon oxide shell (460 nm), elemental silicon (530 nm) and the Si–SiO2 interface (630 nm). Right, top panel: XEOL from a heterostructure of a CdSe–Si nanowire (Rosenberg, Shenoy, Sun et al., 2006[link]). Each component exhibits its individual emission characteristics (Si, ∼520 nm; CdSe, ∼640 nm) which can be revealed with time-resolved XEOL (see below). Right, bottom panel: XANES of the heterostructure at the Se L3,2 edge. The inset shows the morphology of the specimen.

2.4. 2D XANES–XEOL

Using a CCD detector, one can detect XEOL in a wide range of wavelengths from UV to visible to near-IR (200–900 nm) simultaneously (Wang et al., 2014[link]; Ward et al., 2011[link]; Li et al., 2015[link]). If a scan is performed across an absorption edge, this will produce a 2D map with the excitation energy and emission energy as the x and y axes, respectively, and with the emission intensity colour-coded, resulting in a set of data that can be displayed as a 2D XANES–XEOL map. This technique allows the study of XEOL from a light emitter from an elemental and excitation-channel specific perspective. Fig. 6[link] illustrates how a 2D XANES–XEOL map is obtained at the O K edge of a ZnO nanowire specimen which emits via the BG channel in the UV and DE channel in the visible (Sun et al., 2005[link]; Wang et al., 2014[link]). This technique is especially powerful for the study of multichannel emission and composites. This situation is illustrated in Fig. 7[link] using the XEOL from a solid solution of GaN and ZnO (Ga1−xZnxN1−xOx). Although both GaN and ZnO have band-gap energies of >3 eV, it has been found that GaN–ZnO solid solutions can have band-gap energies of <3 eV, i.e. that the band gap narrows with an increase in x (ZnO concentration) and falls within the range ∼2.4–2.8 eV (516–442 nm) for x = 0.05–0.42, for example (Ward et al., 2011[link]). A full range of XEOL can be collected across all edges of interest in a 2D map.

[Figure 6]

Figure 6

Snapshot of the sequence of XEOL obtained when the excitation photon energy is scanned from below to about the O K edge of a ZnO nanowire specimen. (a) From left to right, the top panel represents XANES recorded in DE yield at ∼515 nm and the bottom panel shows the XEOL at the selected excitation photon energy (marked by an arrow) and the accumulated 2D map up to that energy, where the horizontal cut is the emission wavelength, the vertical cut is the XEOL spectrum and colour represents the intensity (partial optical yield). (b) Similar display with PLY from the BG channel at ∼375 nm.

[Figure 7]

Figure 7

(a) 2D XANES–XEOL map across the N and O K edges with TEY overlaid (blue); the black line denotes the window where optical XANES were collected. (b) XEOL cuts taken across the N and O K edges at selected excitations. (c) Wavelength-selected PLY (red) versus TEY across the N and O K edges. (d) 2D XANES–XEOL across the Zn and Ga L3,2 edges with TEY overlaid (blue). (e) XEOL cuts taken across the Zn and Ga L3,2 edges. (f) Wavelength-selected PLY (red) versus TEY across the Zn and Ga L3,2 edges (Ward et al., 2011[link]).

2.4.1. 2D XANES–XEOL map of ZnO nanostructure

Fig. 6[link] shows how a 2D XEOL–XANES map is generated upon an energy scan across the O K edge of a ZnO nanowire specimen which exhibits XEOL in both the UV (band-gap emission) and the visible (defect emission) (Wang et al., 2014[link]).

In Fig. 6[link](a), the top panel shows the normalized PLY at ∼515 nm, with excitation energies of 532 eV (just below the edge), 536 eV (at the white-line maximum), 552.5 eV (near the end of the XANES region) and 574 eV (the beginning of the EXAFS region); the bottom panel shows the corresponding 2D XANES–XEOL map up to the selected excitation energy (marked by an arrow). It is apparent from Fig. 6[link](a) that the defect emission exhibits a normal XANES (upper panel). It is interesting to note that the yield of the BG emission relative to that of the DE channel increases with increasing photon energy (increasing probing depth; see Fig. 1[link]a). The branching ratio of the two competing emission bands across the edge is associated with the origin of the luminescence (for example surface versus bulk) and the effectiveness of energy transfer to these channels. In this case, the results indicate that the DE is from the near-surface region and the BG emission is from the bulk. That is, as the photon energy increases from below the edge to the edge threshold, the penetration depth reaches a minimum at the white-line maximum, as seen in Fig. 1[link](a), and the energy transfer to these two optical channels varies depending on the distribution of the defects. At excitation energies beyond the white line, the photon probes deeper and energy transfer to the BG emission increases if the defect density decreases. This is what is observed. Similar behaviour can be seen in Fig. 6[link](b), where the XANES was recorded with the BG PLY at ∼375 nm. The edge jump is nearly flat, indicating that at the edge energy transfer to the DE channel dominates, suggesting that there is a higher defect density in the near-surface region.

2.4.2. 2D XANES–XEOL map for GaN–ZnO solid solution

Fig. 7[link] displays the 2D XANES–XEOL map and the spectroscopic results of Ga(1−x)ZnxO (GZNO) solid-solution nanoparticles (x = 0.25, size distribution 40–120 nm; Ward et al., 2011[link]).

Fig. 7[link](a) is the 2D XANES–XEOL map, in which the emission wavelength and intensity are displayed as the photon energy is scanned across the N and O K edges. Figs. 7[link](b) and 7[link](c) display the XEOL (one single peak at 630 nm) and the corresponding optical yield XANES across both the N and O K edges, respectively.

From Figs. 7[link](a)–7[link](c), we see that XEOL exhibits a single peak with a maximum at ∼630 nm (∼2 eV). This energy is too small and its width too broad to be attributed to BG emission. It is from defects. The luminescence intensity decreases at the N and O K edge, producing an inverted PLY spectrum. This observation indicates that energy transfer to the defect optical channel becomes less efficient when the sampling depth deceases (increasing absorption) abruptly above the edge, reducing the thermalization path: i.e. that a significant fraction of the energetic electrons created at or near the surface escape the surface without contributing to thermalization, hence reducing the quantum yield.

Across the Zn and Ga L3,2 edges (Figs. 7[link]d–7[link]f), the X-ray penetration depth is greater than the size (diameter) of the GZNO nanoparticles and XEOL tracks the optical photons due to energy transfer to the defect optical channel, where the Zn L edge is more effective (shorter thermalization path) than the Ga L edge, as revealed from Fig. 7[link](f): the zinc edge jump relative to that of gallium is significantly higher in the PLY. Similar measurements have also been carried out for a TiO2 nanotube system (Li et al., 2015[link]).

3. Time-resolved X-ray excited optical luminescence (TRXEOL)

3.1. Time structure of synchrotron light sources and TRXEOL

Fig. 8[link] shows the time structure of a typical synchrotron (left panel), where the bunch length determines the pulse width and the separation between electron bunches determines the duration of the dark window. The corresponding counting scheme is shown in the right panel.

[Figure 8]

Figure 8

Schematic for time-resolved XEOL; the left panel shows the typical characteristics of a storage ring operating with multi-bunches. The right panel shows how the decay curve, TR-XEOL and TR optical XAFS can be collected. The red and blue horizontal bars depict fast and slow time windows, respectively.

The typical bunch length for a third-generation storage ring is ∼30 ps and the dark window (gap between bunches) varies from picoseconds to hundreds of nanoseconds to microseconds depending on the number of bunches filled. Using the SR pulse, one can trigger a time event in which the optical photons emitted after excitation at time 0 are tracked. This provides the decay curve of the optical emission. If the optical monochromator is set at zero order, one collects all wavelengths. One can now apply time-gating techniques in which one collects all optical photons emitted at a selected time interval. For example, the APS operating in the 24-bunch top-up mode provides a dark gap of 153 ns; one may call 0–10 ns a fast window (red bar in Fig. 8[link]) and 10–153 ns a slow window (blue bar). One then obtains time-resolved XEOL which only tracks optical emission within the time window. Time-gating can also be used to track the optical XAFS, revealing the origin of luminescence (Sham & Rosenberg, 2007[link]; Sham, 2014[link]; Rosenberg, Shenoy, Tien et al., 2006[link]).

3.2. TRXEOL using an optical streak camera

The lifetime of the optical decay of XEOL from GZNO has been investigated using an optical streak camera (OSC) with a fast sweep (Regier et al., 2010[link]). The results are summarized in Fig. 9[link]. The OSC operates by converting a time event (triggered by a 35 ps synchrotron pulse) into a spatial profile, and thus is able to achieve picosecond time resolution. The benefit of OSC-based TRXEOL is the ability to simultaneously collect both lifetime-decay and emission-wavelength data, thus producing a temporally resolved 2D map of the XEOL. The resulting 2D map contains XEOL on the horizontal axis and temporal information on the vertical axis, with intensity colour-coded. The ability to differentiate between multiple luminescence decay processes that are occurring with different lifetimes by applying appropriate integration windows (both temporal and spectral) to the resultant streak image is the greatest advantage that this technique provides (Ward et al., 2021[link]) over other TRXEOL methods (Sham, 2014[link]; Heigl et al., 2007[link]; Regier et al., 2010[link]; Ward et al., 2013[link]).

[Figure 9]

Figure 9

(a) 5 ns streak image of XEOL (340–640 nm) from GZNO excited at 550 eV. (b) XEOL decay lifetime taken from blue (400 nm) and red (550 nm) windows of the streak image. (c) Fast and slow XEOL taken from 0.5–0.6 ns, 1.45–1.55 ns and 0.5–5.0 ns time windows with maximum intensity normalized to unity.

Fig. 9[link](a) shows the 2D map of the OSC for GZNO and Figs. 9[link](b)–9[link](d) display selected cuts in both the energy (wavelength) and time domains (Li et al., 2015[link]). Fig. 9[link](b) displays the decay curve associated with the vertical cut around 400 and 500 nm as shown in Fig. 9[link](a). A slower decay is observed for the longer wavelength emission, as expected on the basis of Coulomb attraction. Fig. 9[link](c) shows the emission wavelength at two given time windows (horizontal cut) as marked in Fig. 9[link](a), showing emission at much shorter wavelengths that is not apparent in the ungated XEOL; for example, Fig. 9[link](d) clearly shows that emission from the fast time window, which is not noticeable in the ungated XEOL, is revealed at much shorter wavelengths (471–512 nm), while the intense defect emission observed in the ungated spectrum is now absent. This observation is consistent with previous observations that the defect emission has a much slower timescale (Regier et al., 2010[link]; Ward et al., 2013[link]). The band gap for the GZNO sample has previously been estimated to be ∼2.53 eV (490 nm) from UV–Vis diffuse reflectance measurements (Mosselmans et al., 2013[link]). The TRXEOL results are in good agreement with the wavelength of maximum emission (integrated over the entire window: 0.5–5 ns). For nanostructures with surface and bulk defects, the optical band gap is often obscured in XEOL by the intense defect emission but can easily be retrieved with fast time gating, since, as noted above, the decay of defect emission is slow and will not be detected in the fast time-gated TRXEOL. It has generally been recognized that the decay behaviour depends on the size, morphology and crystallinity of the nanostructure in that (i) the decay lifetime is very fast for BG emission and slow for defect emission, (ii) for the same optical channel, the smaller the crystal, the faster the decay and the better the crystallinity, the faster the decay and (iii) the branching ratio of the BG versus DE emission intensity is also dependent on the crystallinity; the more perfect the crystallite, the more intense the BG emission. Hence, the possibility of using TRXEOL to reveal the optical band-gap emission is further confirmed and can readily be recorded using the OSC (Ward et al., 2021[link]). In addition, the OSC offers the opportunity to reveal more than one optical band-gap emission bands with different decay lifetimes arising from a distribution of nanostructures of slightly different sizes and morphologies.

4. XEOL imaging

Like its PL counterparts, XEOL is another tool for imaging (Sham et al., 2010[link]; Pailharey et al., 2007[link]; Rosenberg et al., 2012[link]; Martínez-Criado et al., 2012[link]; Hand, 2014[link]; Cao et al., 2020[link]). Many nanostructured semiconductors and inorganic and organic light-emitting materials can convert the X-rays that they absorb into optical photons; the conversion process is directly related to the morphology and crystallinity of the materials such as crystallite morphology and dimensions, as well as defects and impurities (Sham, 2018[link]). Since the dynamics for converting X-ray into optical photons depends on the energy-transfer process via inelastic scattering of energetic electrons in a thermalization process, for which the timescale can span many orders of magnitude from pico­seconds to microseconds and beyond. Fig. 10[link] shows a couple of examples of XEOL imaging of rare-earth phosphors excited in the tender (Ce and Tb L3 edge) and soft X-ray (O K edge) region using a microprobe and a nanoprobe, respectively (Heigl et al., 2013[link]; Wang et al., 2018[link]). It can be seen from the left panel of Fig. 10[link] that XEOL from a given emission channel, such as the UV emission from Ce3+ or green emission from Tb3+ in an inorganic phosphor, has a different depth and site selectivity to images recorded from X-ray fluorescence (XRF), hence it produces slightly different maps. The right panel of Fig. 10[link] shows the optical image and corresponding nano O K-edge XANES of individual europium-doped Y2O3 nanoparticles recorded with an optical yield using scanning transmission X-ray microscopy (STXM; Wang et al., 2018[link]).

[Figure 10]

Figure 10

Imaging with partial X-ray excited optical luminescence yield; the left panel shows the L3 edge XANES of a rare-earth phosphor with X-ray and optical photon detections (top) as well as the XEOL spectrum (bottom); the middle panel shows the 2D image with XRF (left) and optical photons (right) in the UV and blue originating from Ce3+ and Tb3+, respectively; the right panel shows the XEOL image from doped Y2O3 nanoparticles (top) together with the nano-XANES in the region of interest (ROI), which is colour-coded.

5. Prospects

Recent developments in XEOL, in particular two 2D soft X-ray absorption–optical emission techniques, as well as time-resolved XEOL (TRXEOL), are reported here. These techniques are powerful tools for the characterization of materials as the spatial resolution of the photon beam and the energy resolution and sensitivity of detectors and detection schemes continue to advance. It should be noted that resonant inelastic X-ray scattering (RIXS), where one tracks the inelastically scattered X-rays with enhanced cross-section at the absorption edge, and resonant X-ray emission (RXE) can now be tracked together with XEOL simultaneously, making multi-dimensional spectroscopy a possibility. Many beamlines can now provide capabilities which will allow imaging and microanalysis using fluorescence X-rays as well as XEOL in both the energy and time domains. There is also increasing interest in tracking persistent photoluminescence using XEOL. These very slow decays from PPL nanomaterials are ideal for bioimaging triggered by an external X-ray source (Mian et al., 2022[link]). Other XEOL imaging capabilities have been implemented elsewhere (Mosselmans et al., 2013[link]; Sham et al., 2010[link]; Pailharey et al., 2007[link]; Rosenberg et al., 2012[link]). With more advanced third-generation synchrotron light sources now coming into service worldwide (for example NSLS II and the Taiwan Photon Source) and many upgrades to existing third-generation light sources, as well as the forthcoming fourth generation, diffraction-limited storage rings and X-ray free-electron lasers (XFELs), these 2D or even multidimensional photon-in photon-out techniques will undoubtedly help to advance our capability in materials analysis. It should also be noted that while the synchrotron pulse is typically ∼30 ps, that of XFELs is ∼100 fs with tuneable repetition rates, and time-resolved imaging via energy-transfer channel in the ultrafast time domain is a possibility.

Acknowledgements

The author acknowledges the financial support of NSERC, CFI, CRC of Canada, OIT of Ontario and the University of Western Ontario. The author is also grateful to many former graduate students, postdoctoral fellows, visiting scientists and staff at the synchrotrons (CSRF/SRC, CLS and APS), as well as many collaborators, who have contributed to many of the results presented here.

References

First citationArmelao, L., Heigl, F., Kim, P. G., Rosenberg, R. A., Regier, T. Z. & Sham, T. K. (2012). J. Phys. Chem. C, 116, 14163–14169. Google Scholar
First citationCao, C., Toney, M. F., Sham, T., Harder, R., Shearing, P. R., Xiao, X. & Wang, J. (2020). Mater. Today, 34, 132–147.Google Scholar
First citationEmura, S., Moriga, M., Takizawa, J., Nomura, M., Bauchspiess, K. R., Murata, T., Harada, K. & Maeda, H. (1993). Phys. Rev. B, 47, 6918–6930.Google Scholar
First citationErbil, A., Cargill, G. S. III, Frahm, R. & Boehme, R. F. (1988). Phys. Rev. B, 37, 2450–2464.Google Scholar
First citationFadley, C. S. (2015). Hard X-ray Photoelectron Spectroscopy (HAXPES), edited by J. Woicik, pp. 1–34. Cham: Springer.Google Scholar
First citationHand, M. J. (2014). PhD thesis. University of Warwick, United Kingdom. http://wrap.warwick.ac.uk/61954/ . Google Scholar
First citationHeigl, F., Jürgensen, A., Zhou, X. T., Hu, Y. F., Zuin, L. & Sham, T. K. (2013). J. Chem. Phys. 138, 061104.Google Scholar
First citationHeigl, F., Jürgensen, A., Zhou, X. T., Lam, S., Murphy, M., Ko, J. Y. P., Sham, T. K., Rosenberg, R. A., Gordon, R., Brewe, D., Regier, T. & Armelao, L. (2007). AIP Conf. Proc. 879, 1202–1205.Google Scholar
First citationHu, Y. F., Tan, K. H., Kim, P. S., Zhang, P., Naftel, S. J., Sham, T. K., Coulthard, I. & Yates, B. W. (2002). Rev. Sci. Instrum. 73, 1379–1381.Google Scholar
First citationKim, P.-S. G., Zhang, P. & Sham, T. K. (2004). Langmuir, 20, 4690–4695.Google Scholar
First citationLakowicz, J. R. (2006). Principles of Fluorescence Spectroscopy, 3rd ed. New York: Springer.Google Scholar
First citationLi, J., Liu, L. & Sham, T. K. (2015). Chem. Mater. 27, 3021–3029.Google Scholar
First citationMartínez-Criado, G., Alén, B., Sans, J. A., Homs, A., Kieffer, I., Tucoulou, R., Cloetens, P., Segura-Ruiz, J., Susini, J., Yoo, J. & Yi, G. (2012). Nucl. Instrum. Methods Phys. Res. B, 284, 36–39.Google Scholar
First citationMian, F., Bottaro, G., Wang, Z., Yiu, Y. M., Rancan, M., Sham, T. K. & Armelao, L. (2022). Appl. Surf. Sci. 577, 151896.Google Scholar
First citationMosselmans, J. F. W., Taylor, R. P., Quinn, P. D., Finch, A. A., Cibin, G., Gianolio, D. & Sapelkin, A. V. (2013). J. Phys. Conf. Ser. 425, 182009.Google Scholar
First citationPailharey, D., Mathey, Y., Jandard, F., Larcheri, S., Rocca, F., Kuzmin, A., Kalendarev, R., Purans, J., Dalba, G., Graziola, R. & Dhez, O. (2007). J. Phys. Conf. Ser. 93, 012038.Google Scholar
First citationRegier, T. Z., Vogt, J. M., Sammynaiken, R., Sham, T. K., Garrett, R., Gentle, I., Nugent, K. & Wilkins, S. (2010). AIP Conf. Proc. 1234, 838–841.Google Scholar
First citationRodnyi, P. A. (1997). Physical Processes in Inorganic Scintillators. Boca Raton: CRC Press.Google Scholar
First citationRogalev, A. & Goulon, J. (1997). J. Phys. IV, 7, C2-565–C2-568.Google Scholar
First citationRogalev, A. & Goulon, J. (2002). Chemical Applications of Synchrotron Radiation, Part II, X-ray Applications, edited by T. K. Sham, pp. 707–760. Singapore: World Scientific.Google Scholar
First citationRonda, C. (2008). Editor. Luminescence: From Theory to Applications. Weinheim: Wiley-VCH.Google Scholar
First citationRosenberg, R. A., Shenoy, G. K., Sun, X. H. & Sham, T. K. (2006). Appl. Phys. Lett. 89, 243102.Google Scholar
First citationRosenberg, R. A., Shenoy, G. K., Tien, L. C., Norton, D., Pearton, S., Sun, X. H. & Sham, T. K. (2006). Appl. Phys. Lett. 89, 093118.Google Scholar
First citationRosenberg, R. A., Zohar, S., Keavney, D., Divan, R., Rosenmann, D., Mascarenhas, A. & Steiner, M. A. (2012). Rev. Sci. Instrum. 83, 073701.Google Scholar
First citationSham, T. K. (2014). Adv. Mater. 26, 7896–7901.Google Scholar
First citationSham, T. K. (2015). J. Electron Spectrosc. Relat. Phenom. 204, 196–207.Google Scholar
First citationSham, T. K. (2018). Synchrotron Radiation in Materials Science: Light Sources, Techniques, and Applications, Vol. 1, edited by C. Fan & Z. Zhao, pp. 123–136. Weinheim: Wiley-VCH.Google Scholar
First citationSham, T. K., Gordon, R. A., Garrett, R., Gentle, I., Nugent, K. & Wilkins, S. (2010). AIP Conf. Proc. 1234, 133–136.Google Scholar
First citationSham, 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. & Bryskiewicz, B. (1993). Nature, 363, 331–334.Google Scholar
First citationSham, T. K., Naftel, S. J., Kim, P. G., Sammynaiken, R., Tang, Y. H., Coulthard, I., Moewes, A., Freeland, J. W., Hu, Y. F. & Lee, S. T. (2004). Phys. Rev. B, 70, 045313.Google Scholar
First citationSham, T. K. & Rosenberg, R. A. (2007). ChemPhysChem, 8, 2557–2567.Google Scholar
First citationSun, X. H., Lam, S., Sham, T. K., Heigl, F., Jürgensen, A. & Wong, N. B. (2005). J. Phys. Chem. B, 109, 3120–3125.Google Scholar
First citationWang, Z., Guo, X. & Sham, T. (2014). Nanoscale, 6, 6531–6536.Google Scholar
First citationWang, Z., Wang, J., Hou, D. & Sham, T. K. (2018). Microsc. Microanal. 24, 480–481.Google Scholar
First citationWard, M., Regier, T., Vogt, J., Han, W.-Q. & Sham, T. K. (2021). J. Phys. Chem. C, 125, 19438–19444.Google Scholar
First citationWard, M. J., Han, W. Q. & Sham, T. K. (2011). J. Phys. Chem. C, 115, 20507–20514.Google Scholar
First citationWard, M. J., Smith, J. G., Regier, T. Z. & Sham, T. K. (2013). J. Phys. Conf. Ser. 425, 132009.Google Scholar








































to end of page
to top of page