2.1. X-ray source and optics hutches

Before the extremely brilliant source upgrade (EBS; https://www.esrf.eu/about/upgrade) of the ESRF storage ring (see below), the X-ray source for the ID30A branch comprised two 1.4 m undulators with a periodicity of 21.2 mm (U21.2) operated in air with a minimum gap of 11 mm and optimized to emit their highest brilliance at an energy of 12.9 keV. A schematic overview of MASSIF-3 is shown in Fig. 1. There are two independent optical hutches (OH), OH1 that is shared with ID30B, and OH2 that contains most of the optical elements for the MASSIF beamlines. The canting angle between the sources for ID30A and ID30B is ±2.2 mrad, providing sufficient space between the two beams to install all optical equipment. Table 1 summarizes the current (and future, see below) beamline characteristics.

Figure 1 Schematic layout of the MASSIF-3 beamline. Major components are shown with their distance relative to the source. The monochromator/beamsplitter for MASSIF-2 is shown only for consistency but is currently not in place.

2.2. EH2 configuration

The experimental hutch of MASSIF-3 contains both the beamline's horizontal focusing element and its sample position and sample environment. Horizontal focusing of the X-ray beam is achieved using a bent multilayer mirror (B₄C/[Ru/B₄C]₇₀/Cr, manufactured in-house at ESRF) located 1.0 m upstream of the sample position. Combined with the vertical focusing provided by the white-beam CRLs (see above), this results in a focal spot size of 18 µm × 14 µm [FWHM as measured by scanning a tungsten wire across the beam; Fig. 2(c)] at the sample position. Potential contamination of diffraction intensities by X-ray photons of higher harmonic reflections from the Si(111) monochromator (mostly the third-harmonic reflections with photon energy ∼38.4 keV) are intrinsically suppressed or rejected due to the combination of three factors: (i) the X-ray detector's quantum efficiency drastically decreases above 20 keV, (ii) the focusing distance of the CRLs depends on the photon energy such that higher energy photons are focused far behind the sample position, and (iii) the reflectivity of the multi-layer mirror drastically decreases at higher energies (Bigault et al., 2003).

Figure 2 Experimental setup of MASSIF-3. (a) Photograph of the sample environment and (b) setup of the micro-spectrophotometer with the X-ray beam indicated as a red arrow and the optical pathway in transparent orange. (c) 3D plot of the X-ray beam profile measured with a 5 µm pinhole (abscissae in micrometres and ordinate in arbitrary units). (d) Image of the X-ray beam at the sample position as viewed through the on-axis viewer using the built-in scintillator (at 0.06% transmission and a storage ring current of 180 mA).

Upstream of the horizontally focusing mirror, a high-vacuum attenuator box containing a total of eight absorbing foils [two pyrocarbon (1 and 2 mm in thickness) and six Al (0.1, 0.2, 0.35, 0.5, 1, and 1.5 mm thickness)] is installed. This allows to accurately adjust the photon flux at the sample in small steps. Downstream of the mirror, a pair of beam-defining secondary slits (ss2) (JJ-XRAY, Lyngby, Denmark) has been installed to allow for the optimal alignment of the mirror and also to remove parasitic X-rays scattered from the mirror. This pair of slits is mounted in a `slitbox' which also contains a scattering foil diode (I₀), read by a Keithley picoammeter which allows constant monitoring of the X-ray beam intensity, a rotating millisecond fast shutter controlled by ESRF-developed IcePAP electronics (Janvier et al., 2013), and a second scattering foil diode (I₁) also read through a Keithley picoammeter which allows monitoring of the X-ray beam intensity after the fast shutter. As at the other ESRF Structural Biology Group MX beamlines, the I₀ reading is correlated with photon flux using a calibrated diode periodically placed at the sample position, thus allowing the constant measurement of photon flux impinging of the sample.

2.3. Sample environment

The MASSIF-3 sample environment [Fig. 2(a)] comprises a MD2 micro-diffractometer (Arinax, Voreppe, France; permanent loan from EMBL Grenoble Outstation), an Eiger X 4M detector (Dectris, Baden-Daettwil, Switzerland; Casanas et al., 2016; Förster et al., 2019) mounted on a fast translation table, and a FlexHCD robotic sample changer (McCarthy et al., 2018). Samples on the rotation axis of the MD2 are kept at cryogenic temperature using a 700 series cryo-stream (Oxford Cryostreams, Oxford, UK) equipped with an annealing system (Giraud et al., 2009). The entire experimental setup [Fig. 2(a)], including the slitbox described above, is installed on a large granite table to allow for the alignment of all elements relative to the X-ray beam using three vertical (tz1, tz2 and tz3) and two horizontal (tyf and tyb) motors. These motors are operated in a closed-loop feedback mode and can be moved independently, or conjointly as pseudo motors for controlling the table height (thgt), translation (ttrans), roll (txrot), pitch (tyrot), and yaw (tzrot). The MD2 micro-diffractometer is, in contrast to all other MX beamlines, not permanently equipped with a Mini-Kappa goniometer head (MK3, Brockhauser et al., 2013) to keep the sphere of confusion to less than 1 µm. However, the mounting of an MK3 device is available upon request.

The maximum speed of the MD2 rotation axis is currently 720° s⁻¹, which allows, in combination with the fast Eiger X 4M detector, the recording of complete datasets in less than a second (Table 2). Additional built-in accessories of the MD2 include beam-cleaning apertures and a YAG screen for automatic alignment of the rotation axis and the X-ray beam [Fig. 2(d)]. The MD2 is equipped with a standard G-shaped beamstop/capillary setup, but optionally a motorized beamstop with a maximum distance of 120 mm from the sample can be mounted, leaving more space for experiments that require installation of specific setups that are significantly bulkier than standard SPINE pins. The FlexHCD robotic sample changer allows to use SPINE sample holders stored either in SC3 pucks (Cipriani et al., 2006) or in UniPuck (https://smb.slac.stanford.edu/robosync/Universal_Puck) containers. Each system needs its own specific FlexHCD gripper to transfer sample holders, and the exchange of grippers can be performed automatically within less than a minute via the beamline control GUI MXCuBE (https://MXCuBE.github.io/MXCuBE; Gabadinho et al., 2010; Oscarsson et al., 2019).

The Eiger X 4M detector provides diffraction images in HDF5 format (https://portal.hdfgroup.org/display/support), which allows the storage of many (typically 100) images in a single file together with extensive metadata, thus reducing file system clutter. Most crystallographic software programs [e.g. XDS (Kabsch, 2010), DIALS (Winter et al., 2018); CrystFEL (White et al., 2012)] are able to handle this data format, but for certain steps of automatic data processing [namely strategy calculations using EDNA/BEST (Bourenkov & Popov, 2006; Incardona et al., 2009)] diffraction images are automatically converted to miniCBF format (Bernstein & Hammersley, 2006). The maximum resolution that can be achieved with the Eiger X 4M at the shortest sample-detector distance available on MASSIF-3 is 1.52 Å at the edge of the detector, 1.19 Å at its corners.

MASSIF-3 is currently recommended for diffraction data collection from small crystals (<15 µm) or from larger crystals where use of a relatively small beam will allow the use of helical data collections (Flot et al., 2010) or the identification of the best diffracting segment of the crystal for a diffraction experiment (i.e. diffraction cartography; Bowler et al., 2010]). As on all other ESRF MX beamlines, a HC1 humidity device (Arinax, Moirans, France) that allows the recording of diffraction data under controlled relative humidity at room temperature is available upon user request (Sanchez-Weatherby et al., 2009; Russi et al., 2011). The beamline is also well adapted for experiments requiring the on-line collection of in crystallo fluorescence and/or UV/vis absorption spectra exploiting the use of a microspectrophotometer [McGeehan et al., 2009; von Stetten et al., 2015; Fig. 2(b)]. Additionally, MASSIF-3 is an excellent vehicle for serial synchrotron crystallography (SSX) experiments using a high-viscosity extruder (HVE; Botha et al., 2015, results to be published elsewhere) in close collaboration with the group of Ilme Schlichting from the MPI for Medical Research in Heidelberg, Germany.

2.4. Beamline control and data collection

Low-level control of all beamline components and communication with the PMAC controller of the MD2 is provided by the newly developed ESRF standard software BLISS (Beamline Instrumentation Support Software; https://gitlab.esrf.fr/bliss/bliss; Guijarro et al., 2017). BLISS commands are mostly accessed transparently from the beamline control GUI MXCuBE2 (Gabadinho et al., 2010; Oscarsson et al., 2019), but, for troubleshooting, a terminal interface is also available. All motors except those of the MD2 and that of the millisecond fast shutter are controlled by ESRF-developed IcePAP electronics (Janvier et al., 2013) and pneumatically driven devices are controlled by WAGO control modules (https://www.wago.com).

Users collect diffraction data via the MXCuBE2 GUI (Gabadinho et al., 2010; Oscarsson et al., 2019) which is the standard interface on all MX beamlines at the ESRF. The metadata of each data collection are stored in the ISPyB database (Delagenière et al., 2011) and can be recalled through the web-service EXI (https://exi.esrf.fr). Specific automated workflows are available and executed from within MXCuBE for complex tasks such as fast mesh scans, X-ray centring, helical data collections (Flot et al., 2010), MeshAndCollect (Zander et al., 2015), as well as MXpress workflows for fully automatic data collection (Svensson et al., 2015; McCarthy et al., 2018). MASSIF-3 also features a more advanced workflow for helical data collection (Fig. 3). This workflow is especially useful for cases where crystals are thin elongated needles or plates and where start and end points for the helical trajectory are difficult to assign due to the opaqueness and/or refraction (i.e. lensing effects) of frozen mother liquor. In this workflow, initial start/stop points for the trajectory are refined and precisely determined by X-ray diffraction using two orthogonal line scans at each point. A data collection strategy taking into account that the radiation damage will be distributed over a larger crystal volume (Zeldin et al., 2013) is then calculated using characterization images collected at the midpoint of the trajectory and is subsequently executed (Fig. 3).

Figure 3 Schematic representation of the procedure for optimized helical data collection. Two orthogonal line scans (green arrows) at each end of the needle-shaped crystal (blue) precisely locate the start and end positions. Then a characterization is performed in the middle of the crystal (red dot), and the data set is collected along the crystal (orange line).

As on the other MX beamlines at the ESRF, all collected datasets are automatically processed using a series of software pipelines [EDNAproc, GrenADeS (Monaco et al., 2013), XDSAPP (Sparta et al., 2016), DIALS (Winter et al., 2018), autoPROC (Vonrhein et al., 2011)], the results of which are stored in the ISPyB database and are accessible, along with files containing the reduced data, through the web-service EXI (https://exi.esrf.fr; Delagenière et al., 2011).

Data backup is easily handled, using an in-house-developed GUI that allows the users to select a specific beamline, experiment number, date, and USB drive, and to automatically copy their data in parallel to the experiment.