LCLS-MFX¶

The Macromolecular Femtosecond Crystallography (MFX) instrument at SLAC's LCLS X-ray free-electron laser. This page walks the beamline as it is being modelled; everything here is reverse-engineered from SLAC's open pcdshub controls stack or inferred, not a commissioned measurement.

Property	Value
Asset	`LCLS-MFX` (root Asset, `tier = Unit`, `parent_id = None`)
Facility	SLAC (bound via `facility_code = "slac"`, `FacilityKind = Site`), CORA's fifth Site and its first XFEL
Status	Off-roadmap modelling exercise (not a CORA pilot)
Technique	serial femtosecond crystallography, fs optical pump-probe, X-ray emission spectroscopy (XES / HERFD)
Beam	SASE hard-X-ray free-electron laser; one shared linac and undulator line feeding many instruments; per-shot photon energy
Control stack	the `pcdshub` EPICS stack, a separate event-driven DAQ (`psdaq`), and the `lightpath` beam-walk engine

Design phase, and a deliberate off-roadmap exercise

MFX is a real, operating instrument, but it is not on the CORA pilot roadmap (APS to MAX IV). It is modelled here, like the Diamond beamlines and FXI, to test that the dry, correct device facts in SLAC's open pcdshub stack seed CORA's intentional model, and to push the model along the one axis no storage-ring pilot reaches: the X-ray-free-electron-laser acquisition paradigm. Every value is reverse-engineered from pcdshub or inferred, carried as confirm until LCLS staff verify it. What CORA still needs the team to confirm is on Open questions.

What MFX adds over the storage-ring exercises¶

Every prior modelling exercise (Diamond I22 SAXS, I03 MX, I11 powder, I15-1 PDF, NSLS-II FXI TXM) is a storage-ring beamline: frame-on-trigger detectors and sub-Hz scalar monitoring. MFX is an XFEL. It collects a per-shot, pulse-ID-tagged event stream at beam rate (120 Hz at LCLS; LCLS-II reaches roughly 1 MHz), times acquisition on beam-synchronous event codes, runs a femtosecond optical pump-probe, and draws from one linac shared across many co-equal instruments. It is the test of whether CORA generalizes beyond the storage-ring acquisition paradigm.

The result is the inverse of I03's. I03 graduated a device Family (Goniometer). MFX graduates none:

The device families fold. Of MFX's full device set, one type had no CORA Family: the von Hamos X-ray emission spectrometer. It introduced the EmissionSpectrometer family, which has since GRADUATED into the catalog once NSLS-II ISS (8-ID) earned the second sighting (SPEC-1). Everything else reuses a catalog Family (Mirror, Filter, Slit, Shutter, Monochromator, Scintillator, Camera, InsertionDevice, TimingController, and the graduated Transfocator, a CRL focusing optic) or an existing loose family (FluxMonitor, Diagnostic, Laser). The offset mirrors fold into Mirror, the solid-Si attenuators into Filter, the pulse picker into Shutter, the intensity-position monitors into FluxMonitor + Diagnostic, all adversarially reviewed.
The genuine gaps are architectural, not taxonomic. What MFX exposes is not missing device kinds but a missing acquisition ontology. These are recorded as deliberate deferrals on Model, each pointing at the existing seam it would extend: per-shot pulse-ID event DAQ (DAQ-1), beam-synchronous event-code timing (TIMING-1), femtosecond pump-probe synchronization (LASER-1), one switched FEL source feeding co-equal instruments (TOPO-1), the attenuator transmission solver (ATT-1), and the computed device-state-to-path-transmission lightpath (LIGHTPATH-1).
The deepest gap gets a design sketch. The per-shot event DAQ (DAQ-1) is the load-bearing mismatch: CORA's acquisition is a single-detector poll loop plus a sub-Hz scalar observation logbook with no pulse-ID key. The shape a fix would take is sketched as a forward-looking design note, gated on a real trigger, not built here.

What MFX keeps the same as the other exercises: the descriptor carries the real pcdshub PV prefixes (as I22 and I03 did), and the model reuses existing Families wherever one fits.

The beamline¶

The systems in three areas the beam passes through, plus the controls that drive them. See the beamline overview for how the areas relate.

Along the beam, in order:

Source: the FEL source and its pulse-energy monitor, then the shared front end and X-ray transport (solid attenuators, offset and transport mirrors, the PPS stopper, and transport-line slits and diagnostics), rendered as the generated source-stage device walk.
Optics and endstation: the MFX-hutch conditioning (pulse picker, attenuator, channel-cut mono, focusing lenses, slits, and per-shot diagnostics), the pump-probe laser, the liquid-jet sample delivery, and the von Hamos emission spectrometer.
Detector: the per-shot area detector and the DAQ data plane it feeds.

Cutting across all three:

Controls: the pcdshub EPICS stack, the EventSequencer beam-synchronous timing, and the event-driven DAQ that CORA references but does not own.

The cross-cutting reference view is the Inventory: the planned Asset tree by parent_id with families, the pcdshub-derived PV handles, and the values still pending confirmation. The Source page is generated from the beamline.yaml descriptor.

Techniques¶

Techniques: what MFX is designed to do, as design intent. Serial femtosecond crystallography, pump-probe, and emission spectroscopy are new Methods over the spine; none of the catalog's tomography Methods fit an XFEL, so all are carried pending (the SLAC Practices).

Governance¶

Governance: who would act at MFX and the trust shape that gates their commands, including the Clearance that would gate the class-4 pump-probe laser. People and agents are facility principals at the SLAC Site.

Model¶

Model: the developer's by-kind index into where each CORA aggregate's MFX content lives, and the architectural gap register, the real product of this exercise.

Not yet documented¶

MFX is a modelling exercise for CORA, so the operations runbook (procedures, recipes, cautions) and the live experiment view are deliberately not written: a runbook for an unmodelled, off-roadmap beamline would be invention, not record. The 2-BM deployment shows the shape they would take.