YB₆₆ – a new soft X-ray monochromator for synchrotron radiation. II. Characterization†

YB₆₆, a complex boron-rich man-made crystal, has been singled out as a potential monochromator material to disperse synchrotron soft X-rays in the 1–2 keV region. Results of a series of systematic property characterizations pertinent for this application are presented in this paper. These include Laue diffraction patterns and high-precision lattice-constant determination, etch rate, stoichiometry, thermal expansion, soft X-ray reflectivity and rocking-curve measurements, thermal load effects on monochromator performance, nature of intrinsic positive glitches and their reduction. The 004 reflection of YB₆₆ has a reflectance of ∼3% in this spectral region. The width of the rocking curve varies from 0.25 eV at 1.1 keV to 1.0 eV at 2 keV, which is a factor of two better than that of beryl(1010) in the same energy range, and enables measurements of high-resolution XANES spectra at the Mg, Al and Si K-edges. The thermal bump on the first crystal arising from the low thermal conductivity of YB₆₆ causes an energy drift of a few eVs with storage-ring current and necessitates periodic energy calibration with metal foils. The positive glitches in the transmission function just above the Mg K-edge have substantially been reduced using an Si or SiC mirror which suppresses the sharp reflectivity increases associated with anomalous scattering for the YB₆₆ 006 reflection at the Y L₃- and L₂-edges. Continual operation over the past five years of a YB₆₆ double-crystal monochromator installed on the JUMBO beamline at Stanford Synchrotron Radiation Laboratory (SSRL) indeed proves the long-term stability of this material in synchrotron radiation under ultrahigh vacuum conditions as indicated by the invariance in rocking-curve characteristics after being exposed to an accumulative power level of ∼3 × 10⁸ J over this period of time.

1. Introduction

YB₆₆, a complex binary compound, has been singled out as a potentially useful soft X-ray monochromator for dispersing synchrotron radiation (Wong et al., 1982, 1990). This material has a cubic crystal structure: space group Fm3c (O⁶_h), and contains over 1600 atoms per unit cell with a lattice parameter of 2.344 nm (Richards & Kasper, 1969). It is refractory and has a melting point of 2373 K. The material is semiconducting like silicon and germanium crystals and is expected to be stable to synchrotron radiation exposure. The strongest (400) reflection, with a 2d value of 1.172 nm, has a calculated Darwin width of <1 eV at 1–2 keV. There are no absorption edges of YB₆₆ in this energy range (the Y L₃-edge being at 2080 eV and the B K-edge at 188 eV). Thus, YB₆₆ satisfies all the X-ray and material property requirements for use as a soft X-ray monochromator in a synchrotron beam, particularly in the 1–2 keV region. The synthesis of high-quality single crystals of this material (Tanaka et al., 1985; Kamimura et al., 1993) opens up a new soft X-ray region (Rowen et al., 1993; Schäfers et al., 1992; Wong et al., 1994; Tanaka et al., 1994) not easily accessible before, for measuring the X-ray absorption spectra of Mg-, Al- and Si-bearing materials. The K-edge energies of these low-Z elements occur in the 1–2 keV region (Mg: 1303 eV; Al: 1559 eV and Si: 1839 eV) (Bearden & Burr, 1967). The K-spectra of these constituent elements in condensed matter, particularly the so-called XANES (X-ray absorption near-edge structure), involve predominently transitions to final p-states which are directly associated with the valence levels and are therefore diagnostic of the chemical bonding in terms of ligand electronegativity, coordination geometry, second or higher neighbour effects as well as the overall crystal structure (Wong et al., 1994; Fröba, Wong, Behrens et al., 1995; Fröba, Wong, Rowen et al., 1995).

In the past few years a multi-laboratory collaborative developmental effort in crystal growth, topography, rocking-curve measurements, soft X-ray reflectivity measurements, systematic homogeneity mapping, micro­probe analysis, etch rates, and long-term stability in a synchrotron beam have been undertaken at NIRIM, BESSY, SSRL and LLNL. Data derived from these property characterizations have been used to understand various parameters controlling the crystal growth process as well as feedback information to improve and attain growth of large and perfect enough crystals for monochromator applications. In this paper the results of various characterization studies leading to the successful development of this novel soft X-ray monochromator material are presented.

2. Experimental

2.1. Crystal growth and sample preparation

Single crystals of YB₆₆ have been grown by an indirect-heating floating zone (IHFZ) method. Since YB₆₆ is a high-resistivity p-type semiconductor (Oliver & Brower, 1971) it is difficult to couple directly with induction heating of the order of a few hundred kHz. In the IHFZ process, which was developed specifically for the growth of YB₆₆ single crystals (Tanaka et al., 1985), the feed rod material is heated by radiation from an inductively heated tungsten ring placed between the feed rod and an RF work coil. Details of this crystal growth method (Tanaka et al., 1985, 1990; Kamimura et al., 1993) and subsequent refined procedures (Tanaka et al., 1998) have been published elsewhere. In order to achieve the growth of high-quality single crystals for monochromator application, a number of growth parameters have been modified systematically based on feedback from characterization results obtained from X-ray topography, double-crystal diffraction analyses (rocking-curves width), and microprobe analysis of composition and impurities. These growth parameters include the number of zone passes, upward and downward drives, growth rate, rotation speed, RF power, and congruent and incongruent compositions. The highest quality currently achieved is represented by a full width at half maximum (FWHM) value of ∼40 arcsec for the (10 0 0) rocking curve using Cu K_α radiation and asymmetrically cut Si(111) as a first crystal. This rocking-curve width is a factor of ∼30 larger than the calculated value. The discrepancy may largely be due to crystal mosaicity, as well as defects and lattice deformations. In the soft X-ray region the measured FWHM values come much closer to the calculated values, within a factor of five or better.

In this work, extensive characterizations have been performed on a number of crystal boules. Both [001]- and [011]-oriented crystals have been grown with typical dimensions of 12 mm in diameter and 60 mm in length. (a) Crystals #6648 and #6652, which have a composition of [B]/[Y] = 62, were grown congruently in an upward drive mode at a growth rate of 12.5 mm h⁻¹ and 25 mm h⁻¹, respectively. (b) Crystal #6716 was used for Laue diffraction using a tungsten source. (c) Crystal #6757, having a composition of [B]/[Y] = 56, was grown under an incongruent melting condition in which the composition of the molten zone is different from, and in equilibrium with, that of the growing crystal. This crystal has been used successfully as a soft X-ray double-crystal monochromator on the JUMBO beamline at SSRL since 1993. (d) Crystal #6825, in which 20 at% of Y was experimentally replaced by Sc in order to test whether such replacement can achieved high crystal quality during growth, was also used for flux measurements in a synchrotron beam.

The boules were sliced with a diamond saw parallel to the growth axis. The crystal quality deteriorated progressively at the zone end due to the large thermal gradient at the termination of the last zone pass. Approximately 10 mm from the zone ends was cut from the slices. Rectangular plates of YB₆₆, 12 mm by 25 mm and 1 mm thick, were used for monochromator crystals. The surfaces, parallel to the {100} planes, were first rough polished with B₄C powder followed successively with 9 µm and 3 µm diamond abrasives and a final polish with 1 µm Al₂O₃ powder on a tin polishing plate.

2.2. Material characterizations

A number of methods have been used to characterize the X-ray properties of the synthesized YB₆₆ materials in terms of monocrystallinity, homogeneity in both chemical composition and diffraction properties, lattice constant, reflectivity in the 1–2 keV region, and materials properties such as thermal expansion, etch rate and stability in a synchrotron beam. These properties are pertinent to the use of this boron-rich material as a soft X-ray monochromator compatible with intense synchrotron radiation under ultrahigh vacuum conditions. The characterization methods may be laboratory-based and synchrotron-facility-based.

2.3. Laboratory-based methods

These include reflection Laue photographs to check for monocrystallinity, high-precision powder diffractometry for cell parameter determination, thermal expansion, micro-probe analysis for composition and impurity distribution and etch rate. Details of the experimental aspects of these laboratory procedures will be described in conjunction with the respective results.

2.4. Synchrotron-based methods

White-light topographic measurements performed at SSRL and the corresponding results are described in detail elsewhere (Tanaka et al., 1998). Reflectivity and energy resolution in the 1–2 keV region were preliminarily obtained using a reflectometer coupled to the double-crystal monochromator KMC beamline (Feldhaus et al., 1986) at BESSY operated with an electron energy of 0.8 GeV and injection current of 540 mA. Monochromatic light from a double crystal of beryl(1010) was fed into the reflectometer chamber in which a YB₆₆ crystal was mounted for reflectivity measurement as a function of the incidence angle (i.e. θ–2θ scans), or as a function of photon energy at fixed Bragg and detector angles.

The rocking curves, transmission function of a double-crystal YB₆₆ monochromator, and K-edge XAFS spectra of a variety of Si, Al and Mg model compounds were measured at the JUMBO beamline (Cerino et al., 1984) at SSRL operated with an electron energy of 3 GeV, injection current of 100 mA and a lifetime of over 20 h. The beam size on the first YB₆₆ crystal was 15 mm in the horizontal and 1.5 mm FWHM in the vertical. The horizontal flux density was essentially uniform. The vertical distribution was asymmetric due to toroidal mirror aberrations, i.e. the smile. The YB₆₆ crystal pair was configured in a non-dispersive mode and with a fixed exit beam height, sending a monochromatic beam from the UHV monochromator chamber to the vacuum sample chamber (10⁻⁵ torr or better) through a 0.4 µm-thick CVD diamond window. The energy position of the monochromator was initially calibrated using its own transmission curve from the position of the L₃- and L₂-edges of Y at 2080 eV and 2156 eV (Bearden & Burr, 1967), respectively, assuming no chemical shift in YB₆₆ from the metal values. X-ray absorption spectra were recorded by monitoring the total electron yield with a channeltron. The spot size on the sample was 2 mm (horizontal) × 1 mm (vertical). The integration time varies from 2 to 8 s point⁻¹ dependent on elemental concentration of Si, Al or Mg in the spectral samples. Heat-load effects on the energy calibration of the double-crystal monochromator were also determined as a function of the SPEAR storage-ring current.

3. Results and discussion

3.1. Laue diffraction patterns and lattice constant

Laue back-reflection diffraction was used to establish the monocrystallinity of the grown YB₆₆ materials. Compared with earlier grown crystals (Oliver & Brower, 1971), which showed double spots (Wong et al., 1990) indicative of twinning and subgrain structure, crystals grown by the IHFZ method are consistently of larger single-crystal domain. A Laue photograph taken with a tungsten tube source operated at 15 kV and 20 mA in a back-reflection mode is given in Fig. 1(a). Using 3000 ASA polaroid film, a sample-to-film separation of 1.5 cm and exposure time of 6 min were found to be optimal. A transmission Laue photograph taken on beamline 2–2 at SSRL is also shown in Fig. 1(b). Both exhibit a single-crystal pattern with the expected fourfold symmetry along the [001] direction.

Figure 1 Laue photographs of YB66(400) (a) taken in a back-reflection mode using a W source operated at 15 kV and 20 mA, and (b) taken with white light at SSRL.

A Guinier powder diffractometer using monochromatic Cu K_α1 radiation and high-purity Ge as an internal standard (a₀ = 0.5657906 nm) was employed to determine the lattice parameter of YB₆₆. Using an XLAT program to refine the 50 reflections observed in the Guinier powder diffraction pattern, values for the cell constant of 2.35052 nm and 2.34996 (21) nm were obtained (Rupp, 1988) for two adjacent slices from the same boule. The numbers in parentheses are the estimated standard deviations, σ, of the last significant digits. More recent work (Higashi et al., 1997) yields the following cell constants of 2.34364 (6) nm and 2.34600 (9) nm for YB₆₂ and YB₅₆ using 551 and 534 independent reflections, respectively.

3.2. Etch rate

The room-temperature etch rate of YB₆₆ in a caustic K₃Fe₃(CN)₆ solution was determined from weight-loss measurements as a function of time. The data are given in Fig. 2. An etch rate of 0.136 µm min⁻¹ was determined from the slope of the weight loss versus time curve and known geometry of the YB₆₆ crystals used. This linear etch rate is useful for controlled chemical removal of damaged surface layer due to cutting and mechanical abrasion.

Figure 2 Weight loss of YB66 in a caustic potassium ferricyanide solution versus etching time.

3.3. Stoichiometry and impurities

Composition mapping on some selected grown YB₆₆ crystals were performed using a Jeol electron microprobe spectrometer, model 733. Concentration profiles of Y, Al and Fe were mapped with a micrometre probe in two step-sizes: 100 µm steps and 2 µm steps over a region of 4–5 mm of the specimens. The latter two elements were suspected impurities introduced during grinding and ball milling of the raw powder mixture in the course of synthesis of the feed rod material for the crystal growth. Two samples were analyzed: (i) sample #6652 with two discs cut one from the seed end and one from the zone end; (ii) sample #6648 with a rectangular slice cut along the growth axis. Two stepping modes were used in the microprobe measurements: (a) 40 points each 100 µm apart and (b) 50 points each 2 µm apart. The results are given in Table 1.

Table 1
Microprobe analysis of Y, Al and Fe content (wt%) in YB₆₆ single crystals

Crystal	Position	Y	Al	Fe	Step size
#6652	Seed end	10.93 ± 0.18	0.11 ± 0.02	0.05 ± 0.01	40 steps, 100 µm apart
Disk A	Zone end	11.11 ± 0.26	0.10 ± 0.01	0.04 ± 0.01	40 steps, 100 µm apart
#6652	Seed end	11.19 ± 0.13	0.12 ± 0.02	0.05 ± 0.01	50 steps, 2 µm apart
Disk B	Zone end	11.21 ± 0.14	0.10 ± 0.01	0.03 ± 0.01	50 steps, 2 µm apart
#6648	Seed end	11.43 ± 0.19	0.40 ± 0.02	0.03 ± 0.01	40 steps, 100 µm apart
Disk A	Zone end	11.11 ± 0.26	0.43 ± 0.01	0.04 ± 0.01	40 steps, 100 µm apart
#6648	Seed end	11.14 ± 0.14	0.40 ± 0.03	0.03 ± 0.01	50 steps, 2 µm apart
Disk B	Zone end	11.18 ± 0.15	0.42 ± 0.02	0.04 ± 0.01	50 steps, 2 µm apart

The standard deviation given in each entry in Table 1 is calculated from the 40 or 50 measurement points. There is no statistically significant variation in the concentration of Y along both crystals (cross sections in the case of #6652 and along the growth axis in the case of #6648). The theoretical wt% of Y in YB₆₆, YB₆₄ and YB₆₂ are 11.079, 11.386 and 11.71%, respectively. The stoichiometry of these two crystals derived from the microprobe analysis is YB_65±1. This value differs somewhat from that of  YB_62±1 obtained by chemical analysis of both Y and B. The higher boron content may be due to surface sensitivity of the microprobe technique. The Al content is three to four times higher in #6648, whereas the Fe content is comparable in both samples. The microprobe data were used to advantage in minimizing both the Al and Fe impurity contents in subsequent powder preparations by (a) eliminating the use of alumina mortar and pestal and (b) removal of Fe using an HCl leach after the reacted feed rod material was pulverized in a stainless-steel ball mill.

3.4. Thermal expansion coefficient

Thermal expansion of YB₆₆ was measured using a conventional dilatometer. A sample of dimensions ∼6 mm × 6 mm × 15 mm was used. The long axis was parallel to the [100] direction. Since YB₆₆ is a cubic crystal, the thermal expansion is isotropic, independent of crystal orientation. The linear expansion coefficients were obtained from the first derivatives of the measured thermal expansion curve plotted as a function of temperature. The results are plotted in Fig. 3. Least-squares analysis of the data yield a linear temperature dependence given by α = 3.521 + 0.01255 T with a correlation factor of 0.998. Together with a shift in Bragg angle, this data will be used to determine the temperature of the first crystal in a double-crystal set-up.

Figure 3 Linear expansion coefficient of YB66 as a function of temperature.

3.5. Rocking curves at Mg K_α energy (1.2536 keV)

Rocking-curve measurement at the Mg K_α (1.2536 keV) is a direct characterization of YB₆₆ for use as a soft X-ray monochromator in the 1–2 keV region. This was performed in a vacuum spectrometer using a θ–2θ geometry with the Mg anode operated at 2.5 kV and 10 mA. Source divergence was 1.3 mrad. A proportional counter, operated at +1410 V and using a P-50 gas (50% Ar + 50% methane) at 200 torr was used as the detector. Beryl(1010) was used as the first crystal and the YB₆₆ (400) samples as the second crystal. In Fig. 4 the rocking curves of sample #6648 cut along the growth axis are plotted as a function of position (in mm) from the seed end to the growth end. Single symmetric rocking curves are obtained to ∼21 mm from the seed end, beyond which the rocking curve broadens with the appearance of a shoulder at 26 mm and a discernible second peak at 29 mm, indicative of a sub-grain boundary at the growth end. It is evident that the (single) crystal quality degrades from the seed end towards the growth end.

Figure 4 One-crystal rocking curves of sample #6648 YB66(400) at the Mg Kα energy as a function of distance from the seed end. The number on each curve denotes the distance in mm from the seed end.

3.6. BESSY data

3.6.1. Absolute reflectivity

The absolute reflectance and energy resolution of YB₆₆ (sample #6757) has been measured as a function of photon energy in the 1–2 keV region using the KMC reflectometer (Kühne & Müller, 1989) at BESSY. The reflectance and energy resolution (FWHM) are shown in Figs. 5(a) and 5(b). The resolution data were not deconvoluted with respect to the performance of the incoming light beam because of insufficient knowledge of the resolution and beam divergence in the reflectometer. However, direct comparison with similar measurements on a beryl(1010) crystal shows that the energy resolution of YB₆₆ is at least a factor of two better. The laboratory measurement with Mg K_α radiation is in good agreement with this synchrotron data as shown by the open triangle points at 1.254 keV plotted also in Figs. 5(a) and 5(b). The result of an energy scan at fixed Bragg angle (open circles at 1.4 keV) yielded an identical result. The peak reflectivity of YB₆₆ is of the order of a few percent with a broad plateau at ∼3.5% between 1.3 and 1.7 keV. It is worth noting that the reflectivity and energy resolution of YB₆₆ is a smooth function of photon energy, while those of beryl (Al₂Si₆Be₃O₁₈) exhibit structures about the Al and Si K-absorption edges at 1.54 keV and 1.84 keV, respectively.

Figure 5 Comparison of (a) reflectance and (b) FWHM of YB66(400) and beryl(1010) in the 1–2 keV region from single-crystal rocking curves measured on the KMC line at BESSY. The open triangles correspond to the Mg Kα data and the open circles were obtained from an energy scan at fixed Bragg angle at 1.4 keV. The data have not been corrected for divergence of the incident light.

3.6.2. Spatial uniformity

To check for spatial uniformity, the reflectivity was also measured at a fixed photon energy of 1.7 keV as a function of position along the growth axis from the seed end to the zone end. The data in Fig. 6 exhibit a constant reflectance value of 3.1 ± 0.1% across a length of ∼17 mm; the corresponding Bragg angle, however, exhibits a monotonic decrease from 51.035° to 50.980° in the same region, indicative of an increase in lattice constant from the seed end to the zone end. The increase in a₀ is attributable to composition variation towards the zone end in the crystal.

Figure 6 (a) Reflectance and (b) Bragg angle of sample #6757 YB66(400) at 1700 eV as a function of distance from seed end.

3.7. JUMBO data

3.7.1. Double-crystal rocking curves

The full width at half maximum (FWHM) of the double-crystal rocking curves plotted for two pairs of YB₆₆ crystals and compared with those for beryl(1010), InSb(111) and quartz(1010) as a function of energy is shown in Fig. 7. Although YB₆₆ crystal #6825 (Sc doped) has significantly higher resolution than crystal #6757, it was unintentionally contaminated with iron from the milling of the feed materials. Iron fluorescence from this crystal pair was unacceptable in an operational monochromator. Crystal #6757 is more typical of the resolution observed, the FWHM of which monotonically increases from ∼0.25 eV at 1100 eV to ∼0.5 eV at 1500 eV and ∼1.0 eV at 2000 eV just below the Y L₃-edge where the reflectivity drops sharply. Both beryl and InSb have lower resolution than YB₆₆, while that of quartz is much higher. The use of quartz as a monochromator for high-intensity synchrotron radiation is limited in two ways. First, it damages very rapidly in the white synchrotron beam. The radiation damaged curve in Fig. 7 is for the case after only several days use with the stored current in SPEAR of <50 mA. Second, the useful energy range of quartz is rather limited from 1500 eV [due to the 2d value of quartz(1010) reflection] to the Si K-edge at 1840 eV.

Figure 7 FWHM of double-crystal rocking curves as a function of photon energy for two YB66 crystals #6767 and #6824 (Sc doped), and comparison with those for quartz (undamaged and damaged), beryl and InSb.

This resolution from the YB₆₆ crystals in the 1–2 keV region is quite suitable for recording relatively high-resolution K-edge XANES spectra for Mg, Al and Si as shown in a later section. Above 2 keV the Y L_3,2,1-edges set in at 2080, 2156 and 2373 eV, respectively, and lower the transmission function of YB₆₆ substantially, making it less attractive for use at higher photon energies when compared with available higher-Z monochromator materials such as Ge and InSb (Fig. 8).

Figure 8 Flux throughput of various double-crystal monochromators in the 1–2.5 keV region.

3.7.2. Thermal load

YB₆₆ is known to have a low thermal conductivity, similar to those of glassy materials (Slack & Oliver, 1971). The performance of the monochromator can be strongly influenced by the thermal load on the first crystal (which `sees' a white beam from the synchrotron) in a double-crystal set-up, particularly with white radiation from a hard X-ray storage ring such as SPEAR. The temperature of the first crystal may be determined by measuring the deviation of its Bragg angle with respect to the second crystal (fixed) as a function of Bragg angle. The measured angular shift combined with the known lattice expansion of YB₆₆ (shown in Fig. 3) have been used to determine the temperature of the first crystal. In Fig. 9 the temperature of the first crystal is plotted as a function of Bragg angle for two incident power levels: ∼22 W and <1 W. The latter power level was facilitated by use of a 25 µm Al filter upstream from the monochromator. At low power a steady temperature of ∼333 K is attained at a Bragg angle of 60°. The fluctuation at low angle is due primarily to low signal level. At higher thermal load, the temperature of the first crystal increases progressively from 473 to 637 K at higher Bragg angle. The temperature increase with Bragg angle is basically due to an increase in thermal load per unit area on the crystal as a result of a reduction of the beam footprint on the first crystal.

Figure 9 Temperature of the first crystal of a double-crystal YB66 monochromator (the C1/C2 pair from sample #6757) on the JUMBO beamline at SSRL with SPEAR operating at 3 GeV electron energy plotted as a function of Bragg angle at two estimated thermal power levels.

The lattice expansion of the first crystal under the thermal load induces an energy shift which varies monotonically with the storage-ring current. In Fig. 10 a calibration curve using the first inflexion point (given by the first maximum in the first derivative spectrum) in the Al metal K-edge spectrum defined at 1559 eV (Bearden & Burr, 1967) can be determined as a function of SPEAR current. Data above 50 mA correspond to the normal mode for multibunch operation and those below 40 mA correspond to a timing mode for single-bunch operation. Thus, at the Al K-edge the energy calibration varies by 2–3 eV between fills and must be corrected for if accurate edge-shift data are desirable. Energy shifts at the Mg K-edge and Si K-edge as a function of SPEAR current are also comparable.

Figure 10 Calibration energy of the C1/C2 double-crystal YB66 monochromator at the Al K-edge (1559 eV) as a function of SPEAR current.

Compared with highly perfect Si crystals and other semiconducting monochromator materials such as Ge and InSb, the spatial uniformity of individual YB₆₆ crystals as seen in the data given in Figs. 5 and 6 is thus far limited to, at best, 2–3 cm in length. Therefore, in a double-crystal monochromator set-up, the relative locations in each of the crystal pair used for monochromatization is very important and have to be mapped out carefully for optimal performance in terms of both flux throughput and energy resolution. Such mapping using hard X-ray topography has indeed been performed for the C1/C2 YB₆₆ pair (crystal #6757) before installing on the JUMBO beamline at SSRL. Furthermore, the FWHM of the rocking curve at 1500 eV has also been mapped as a function of beam position on the second crystal at various beam positions on the first crystal (Fig. 11). Thus, in routine XAFS scans, both the `angular' phase (Bragg angle) and `translational' phase between the two crystals must be optimized as a function of photon energy range intended for the scan.

Figure 11 Rocking-curve width of the C1/C2 monochromator as a function of beam position on the second crystal at various positions of the beam on the first crystal as labelled on each curve.

3.7.3. Positive glitches

Careful examination of the flux data presented in Fig. 8 reveals that there are two positive glitches in the transmission function of YB₆₆ at 1385.6 and 1438 eV in the double-crystal configuration. These glitches limit the usefulness of YB₆₆ for spectroscopy in the region just above the Mg K-edge at 1303 eV (Fig. 12). For the case of dilute specimens where the ratio of sample signal to incident signal is small, these glitches do not normalize out and interfere strongly with analysis of both the XANES and EXAFS data.

Figure 12 Glitches in the transmission function of a double-crystal YB66(400) monochromator above the Mg K-edge. The inset shows the corresponding transmission function with a 45° azimuthal rotation of the YB66 crystals, showing the disappearance of the negative glitch at 1439 eV and the persistence of two positive ones.

The nature of these glitches has been elucidated using a combination of photoemission experiments, reflectivity measurements and anomalous scattering calculations (Tanaka et al., 1997). They are now understood to be due to the transmission at an energy 1.5 times higher than the 004 reflection and correspond to the Y L₃- and L₂-absorption edges at 2080 and 2156 eV, respectively. Reflectivity measurements and structure-factor calculations for the 006 reflection confirmed that these glitches are caused by the sharp reflectivity increases associated with anomalous scattering for the 006 reflection at the Y L₃- and L₂-edges. To minimize the intensity of the anomalous high harmonics at the glitches, a cut-off mirror is used. Reflections from an Si or SiC surface will have a sharp cut-off for the Si K-absorption edge at 1839 eV. For example, the calculated reflectivity for an SiC mirror at a glancing angle of 1.25° is over 80% below 1480 eV, but only 5% at 2080 eV. This was realized by the installation of an SiC mirror downstream from the YB₆₆ monochromator on JUMBO. Fig. 13 shows the relative intensity of the flux in the region of the glitches and their suppression as a function of glancing angle of the beam on the mirror. The data show that at 1.38° most of the higher-order flux due to the glitches is eliminated without substantial reduction of the transmission at the energy of interest. Fig. 14 shows the Mg K-edge spectra of a series of Mg-bearing materials ranging from yoderite [Mg₂Al_5.3Fe_0.5Ca_0.2Si₄O_17.6(OH)_2.4], diopside (MgCaSi₂O₆), MgSiN₂ to MgTiO₃ recorded with (right-hand panels in Fig. 14) and without (left-hand panels in Fig. 14) the SiC cut-off mirror at a glancing angle of 1.25°. This series of Mg compounds is chosen to illustrate the effectiveness of the cut-off mirror in suppressing the two positive glitches as a function of increasing Mg content from yoderite to MgTiO_3.

Figure 13 Reduction of the 006 glitches with increasing glancing angle of an SiC cut-off mirror located downstream from the YB66 double-crystal monochromator.

Figure 14 Mg K-edge spectra of yoderite, diopside, MgSiN2 and MgTiO3 recorded without (left-hand panels) and with (right-hand panels) the SiC cut-off mirror to suppress the two positive glitches at 1385.6 and 1438 eV.

4. Concluding remarks

The collaborative efforts in crystal growth, topographic evaluation and material characterization described in this work, together with an interactive feedback between these research areas, have led to a successful development of YB₆₆ as a soft X-ray monochromator for synchrotron radiation. Since the first installation of a double-crystal YB₆₆ monochromator (Rowen et al., 1993), continual operation on the JUMBO beamline at SSRL over the past five years has proved that this boride material is synchrotron-radiation stable under UHV conditions as shown by the invariance in rocking-curve characteristics after being exposed to an accumulative radiation power of ∼3 × 10⁸ J over this period of time. This capability opens a new spectroscopic window to investigate low-Z materials containing Si, Al and Mg with XAFS. In Fig. 15 the K-edge of some selected Si, Al and Mg model compounds are plotted and show systematic edge shift and increase in white-line intensity of the 1s → 3p transition as a function of ligand type, electronegativity and coordination number. For the Si compounds, except for stishovite, which is a high-pressure form of silicon dioxide, all have Si in fourfold coordination by the respective ligands in the crystal structure. In Al₄C₃ and AlN, the Al is in fourfold coordination, whereas in Al₂O₃ and AlF₃ it is sixfold coordinated. In MgSiN_2, MgO and pyrope, a Mg-bearing garnet, the Mg coordination is 4, 6 and 8, respectively. Again, the XANES spectra exhibit systematic positive edge shift together with a progressive white-line intensity increase. Furthermore, the transmission function of YB₆₆ permits a large energy scan of the XAFS spectrum for both Mg and Al, thus permitting larger k-range data for Fourier transform analysis (Wong et al., 1994).

Figure 15 Normalized K-edge XANES spectra of selected Si, Al and Mg model compounds showing systematic edge shift and increase of white-line intensity of the 1s → 3p transition as a function of ligand type (electronegativity) and coordination number.

L- and M-edge spectroscopy of many of the 4p elements and rare-earths are now possible with the use of the YB₆₆ monochromator since their respective edge energies are in the 1–2 keV region, as illustrated in Fig. 16. An example is given for the case of the Br L_3,2 spectra of an ionic compound KBr and KBrO₃ in which the formal valence of Br is +5 in the bromate ion. The L_3,2 spectra of Br in the two compounds are quite distinct and characteristic of their chemical and structural environment.

Figure 16 (a) List of elements with L and M absorption edges in the 1–­2 keV region. (b) Normalized L3,2-edge XANES of Br in KBr and KBrO3 showing the richness of the spectral features with bonding and structure of Br in each of the compounds.

The availability of intense monochromated photons in the 1–2 keV region of the electromagnetic spectrum will no doubt impact a variety of disciplines such as materials science, mineralogy, catalysis, metallurgy, environmental and biosciences. Analysis and understanding of the model Si, Al and Mg compound systems have begun to shed light on more complex materials containing these constituent elements in a variety of aluminosilicate materials. These include zeolites, which are open three-dimensional aluminosilicates important as catalysts in petroleum cracking and other chemical syntheses; clay minerals important in geoscience, soil chemistry and environmental clean-up; aluminium alloys important in hi-tech structural components; Mg centres important in novel catalysts and in biomolecules; and, of course, silicon-based materials relevant to the semiconductor industry. Results obtained thus far on Si, Al and Mg model compounds (Wong et al., 1994; Fröba, Wong, Behrens et al., 1995; Fröba, Wong, Rowen et al., 1995; Wong et al., 1997) indeed reveal a wealth of fine-structure features in the XANES spectra that may be systematically correlated with the bonding, local coordination, type of neighbouring atoms as well as bond angle about the central atom. Successful installations of the YB₆₆ monochromator have also been reported recently at the SRS facility in Daresbury, UK (Smith et al., 1998), and the UVSOR facility in Okazaki, Japan (Kinoshita et al., 1998).

Footnotes

¹Part I. Wong et al. (1990).