manuals

MIDAS Manuals — High-Energy Diffraction Microscopy

Version: 11.0 Contact: hsharma@anl.gov

---

Manual Index

Manual	Topic	When to Use
GPU_Acceleration.md	GPU-accelerated computation	Building with CUDA, GPU indexing, fitting, reconstruction, and integration
FF_Calibration.md	FF-HEDM geometry calibration	Setting up a new experiment; determining detector parameters
FF_Analysis.md	FF-HEDM grain indexing and fitting	Extracting grain orientations, positions, and strain tensors
FF_Match_Stack_Reconstructions.md	Grain matching and layer stitching	Tracking grains across load states; combining multi-layer scans
FF_Radial_Integration.md	Radial integration / caking	Converting 2D detector images to 1D intensity vs. 2θ profiles
FF_Interactive_Plotting.md	Interactive FF-HEDM visualization	Exploring grains, spots, and raw data interactively in a browser
FF_Visualization.md	FF-HEDM result visualization	Static plotting and export of FF-HEDM grain maps
FF_Dual_Datasets.md	Dual-dataset FF-HEDM analysis	Combining two FF datasets (e.g., different load states)
PF_Analysis.md	Point-Focus HEDM analysis	Analyzing data from a focused beam geometry
PF_Interactive_Plotting.md	PF-HEDM sinogram, intensity & tomo viewer	Exploring PF-HEDM sinograms, intensity patches, and tomo reconstructions interactively
NF_Calibration.md	NF-HEDM detector calibration	Calibrating NF detector distances and beam center
NF_Analysis.md	NF-HEDM reconstruction workflow	Reconstructing spatially resolved orientation maps
NF_MultiResolution_Analysis.md	Multi-resolution NF-HEDM	Iterative NF reconstruction at increasing grid resolution
NF_GUI.md	NF-HEDM interactive GUI	Calibrating detectors, visualizing NF data, and inspecting results
GUIs_and_Visualization.md	All GUIs and visualization tools	Master guide to all desktop, web, and utility visualization tools
Forward_Simulation.md	Forward simulation (simulateNF)	Validating reconstructions by simulating expected diffraction patterns
GSAS-II_Integration.md	Importing MIDAS output into GSAS-II	Rietveld refinement of caked powder data
Tomography_Reconstruction.md	Absorption-contrast CT reconstruction	Reconstructing tomographic slices from projection data
FF_Benchmark.md	FF-HEDM benchmark testing	Validating the FF-HEDM pipeline with simulated data
FF_Integrator_Benchmark.md	Integrator peak fitting benchmark	Validating calibration → integration → peak fitting with CeO₂ data
FF_Phase_Identification.md	Multi-phase identification	Identifying crystallographic phases from diffraction images

---

Getting Started Checklist

For a first-time user performing a combined FF + NF experiment:

1. Calibrate the FF-HEDM detector using a powder calibrant (CeO₂ or LaB₆) → FF_Calibration.md
2. Integrate / cake the powder data (if WAXS analysis is needed) → FF_Radial_Integration.md
3. (Optional) Import caked data into GSAS-II for Rietveld refinement → GSAS-II_Integration.md
4. Run FF-HEDM analysis on the specimen → FF_Analysis.md
5. Visualize FF results interactively → FF_Interactive_Plotting.md
6. Calibrate the NF-HEDM detectors → NF_Calibration.md
7. Run NF-HEDM reconstruction using FF orientation seeds → NF_Analysis.md or NF_MultiResolution_Analysis.md
8. Validate results with forward simulation → Forward_Simulation.md
9. Match/stitch grains across load states or layers → FF_Match_Stack_Reconstructions.md
10. (Optional) Reconstruct tomographic slices for absorption-contrast imaging → Tomography_Reconstruction.md

---

Introduction

MIDAS (Microstructure Diffraction Analysis Software) is a suite of tools for reconstructing three-dimensional microstructures from High-Energy Diffraction Microscopy (HEDM) data. Developed at the Advanced Photon Source (APS) at Argonne National Laboratory, MIDAS supports the complete data-reduction pipeline — from raw detector frames to grain maps, strain tensors, spatially resolved orientation fields, and tomographic reconstructions.

HEDM exploits the high penetration depth of synchrotron X-rays to non-destructively characterize the internal grain structure of polycrystalline materials. Unlike surface-sensitive techniques such as EBSD, HEDM probes the full bulk of a specimen while it remains intact, making it ideal for in situ mechanical loading, thermal cycling, and other experiments where preserving the sample is essential.

---

HEDM Techniques Supported by MIDAS

MIDAS supports three complementary diffraction geometries and an absorption-contrast tomography module. Each probes a different length scale and provides different information about the microstructure.

graph LR
    subgraph "HEDM Geometries"
        FF["Far-Field (FF-HEDM)<br/>Detector ≈ 1 m"]
        NF["Near-Field (NF-HEDM)<br/>Detector ≈ 5–10 mm"]
        PF["Point-Focus (PF-HEDM)<br/>Focused beam"]
    end

    FF --> G1["Grain centroids<br/>Average orientations<br/>Elastic strain tensors"]
    NF --> G2["Spatially resolved<br/>orientation maps<br/>Grain morphology"]
    PF --> G3["High-resolution<br/>grain orientations<br/>from focused beam"]

Loading

Far-Field HEDM (FF-HEDM)

A large-area detector is placed approximately 1 meter downstream of the sample. As the sample rotates, each grain produces a set of discrete diffraction spots that trace out arcs on the detector. By associating spots across many rotation angles, MIDAS determines:

• The center-of-mass position of each grain.
• The average crystallographic orientation (as an orientation matrix).
• The full elastic strain tensor of each grain (6 independent components).

FF-HEDM is well-suited for specimens containing tens to thousands of grains. The diffraction spots are analyzed using a peak-search → indexing → fitting pipeline that groups spots by grain, determines the orientation matrix, and refines the full strain tensor. See FF_Analysis.md for the complete workflow.

Near-Field HEDM (NF-HEDM)

The detector sits only a few millimeters from the sample. At this distance, individual diffraction spots from neighboring grains are spatially separated and can be mapped back to specific voxels in the sample. MIDAS reconstructs a 3D grid where every voxel is assigned a crystallographic orientation and a confidence metric.

What NF-HEDM produces:

• 3D grain-shape maps with sub-micrometer spatial resolution — the morphology and topology of every grain in the illuminated volume.
• Intragranular orientation gradients — misorientation spreads within individual grains, revealing sub-grain boundaries, deformation substructure, and dislocation density variations.
• Grain boundary networks — the full 3D interface between all grains, including boundary normals, inclination angles, and misorientation across each boundary.

How it works:

NF-HEDM reconstruction is a voxel-by-voxel forward-modeling process. For each candidate voxel position and orientation, MIDAS simulates the expected diffraction pattern and compares it against the measured data. The orientation that produces the best agreement (highest confidence) is assigned to that voxel. The process typically proceeds as follows:

1. Orientation seeding — Candidate orientations come from a prior FF-HEDM analysis. Each FF grain provides a seed orientation that is tested at every voxel within the grain's expected spatial extent.
2. Multi-distance detector setup — NF-HEDM at APS 1-ID typically uses multiple high-resolution detectors stacked at different distances from the sample (e.g., layers at 5 mm, 6 mm, 7 mm). Each distance provides an independent constraint, improving the spatial fidelity of the reconstruction.
3. Grid-based reconstruction — The sample volume is discretized into a regular grid (typically 1–5 µm spacing). Each voxel is tested against all candidate orientations, and the best match is stored in a .mic file.
4. Multi-resolution refinement — For large volumes, reconstruction can proceed iteratively: a coarse grid identifies the approximate grain structure, then finer grids refine boundaries and intragranular features. See NF_MultiResolution_Analysis.md.

NF-HEDM detector calibration is separate from FF calibration because the detector geometry must be known more precisely at the shorter working distances. The NF calibration procedure determines the sample-to-detector distance and beam center for each layer. See NF_Calibration.md.

Interactive analysis is supported through a GUI that provides real-time visualization of raw NF diffraction images, reconstructed microstructure layers, and confidence maps. See NF_GUI.md.

Tip

For best NF results, always run FF-HEDM first to generate high-quality orientation seeds. The quality of the NF reconstruction depends heavily on the completeness and accuracy of the seed orientations.

Point-Focus HEDM (PF-HEDM)

A variant of FF-HEDM that uses a focused (pencil) beam instead of a box beam. Because the illuminated volume is much smaller, the number of overlapping spots is reduced, which improves indexing for fine-grained samples or high-symmetry materials. The sample is scanned spatially through the focused beam, providing layer-by-layer grain mapping. See PF_Analysis.md.

Tomography (Absorption-Contrast CT)

MIDAS includes a fast tomographic reconstruction engine based on the gridrec algorithm — a Fourier-domain filtered back-projection method using prolate spheroidal wave function (PSWF) interpolation. It reconstructs absorption-contrast cross-sections from rotation series of projection images.

Key capabilities:

• Dark-field and white-field normalization of raw projection images (dark + 2 white frames).
• Multiple reconstruction filters — Shepp-Logan, Hann (default), Hamming, and Ramp.
• Rotation-axis shift search — reconstructs at multiple candidate center positions to find the optimal alignment.
• Ring artifact removal — sinogram-space filtering to suppress ring artifacts from detector pixel defects.
• OpenMP parallelism — multi-threaded slice reconstruction with automatic memory management.

Two entry points are provided:

Entry Point	Input	Use Case
process_hdf.py	HDF5 file with /exchange/data layout	Standard APS data exchange format
midas_tomo_python.py	NumPy arrays (dark, whites, projections)	Programmatic / custom data formats

Both generate a parameter file and call the MIDAS_TOMO C binary, which uses FFTW for FFT acceleration. The output is a set of reconstructed slices as float32 binary data, with dimensions rounded up to the next power of 2.

For full details — parameter reference, data formats, rotation-center alignment, and troubleshooting — see Tomography_Reconstruction.md.

---

End-to-End Workflow

A typical MIDAS analysis proceeds through the stages shown below. Not every stage is required for every experiment — for example, WAXS caking and tomography are independent of the grain-reconstruction pipeline.

flowchart TD
    RAW["Raw Detector Data<br/>(HDF5 / GE / TIFF)"] --> CAL
    RAW --> TOMO

    subgraph "Calibration"
        CAL["FF Calibration<br/>(powder reference)"]
        CAL --> NFCAL["NF Calibration<br/>(if doing NF-HEDM)"]
    end

    CAL --> INT["Radial Integration<br/>(caking)"]
    INT --> GSAS["GSAS-II / Rietveld<br/>Refinement"]

    CAL --> FF["FF-HEDM Analysis<br/>(indexing + fitting)"]
    FF --> VIZ["Interactive<br/>Visualization"]
    FF --> SEED["Orientation Seeds"]
    SEED --> NF

    NFCAL --> NF["NF-HEDM Analysis"]
    NF --> MIC["Microstructure<br/>(.mic file)"]

    CAL --> PF["PF-HEDM Analysis"]
    PF --> VIZ

    FF --> FWD["Forward Simulation<br/>(validation)"]
    NF --> FWD

    TOMO["Tomography<br/>(gridrec CT)"] --> SLICES["Reconstructed<br/>Slices"]

    style RAW fill:#1a1a2e,stroke:#e94560,color:#fff
    style CAL fill:#16213e,stroke:#0f3460,color:#fff
    style NFCAL fill:#16213e,stroke:#0f3460,color:#fff
    style FF fill:#0f3460,stroke:#e94560,color:#fff
    style NF fill:#0f3460,stroke:#e94560,color:#fff
    style PF fill:#0f3460,stroke:#e94560,color:#fff
    style INT fill:#533483,stroke:#e94560,color:#fff
    style GSAS fill:#533483,stroke:#e94560,color:#fff
    style VIZ fill:#2b2d42,stroke:#8d99ae,color:#fff
    style MIC fill:#2b2d42,stroke:#8d99ae,color:#fff
    style FWD fill:#2b2d42,stroke:#8d99ae,color:#fff
    style SEED fill:#2b2d42,stroke:#8d99ae,color:#fff
    style TOMO fill:#533483,stroke:#e94560,color:#fff
    style SLICES fill:#2b2d42,stroke:#8d99ae,color:#fff

Loading

---

Coordinate Systems

Understanding the coordinate system is critical for correctly loading data into MIDAS and interpreting results.

Laboratory Frame

MIDAS uses the ESRF convention for the laboratory coordinate system and here is the relation APS coordinates:

MIDAS_Coordinates	Direction	APS_Coordinates
X_L	X-ray beam propagation direction (downstream)	Z_L
Z'_L	Tomographic rotation axis	Y'_L
Y_L	Plane normal of the (X_L, Z'_L) plane	X_L
Z_L	X_L × Y_L	Y_L

The angle between Z_L and Z'_L is the wedge angle (Ω). For standard HEDM setups, the wedge angle is zero — the rotation axis is perpendicular to the beam.

Note

The right-handed rotation about Z'_L is denoted ω (omega). A full ω-scan typically covers 360° (or ±180°) to capture all accessible diffraction conditions.

Detector Frame

The detector plane is described by three tilt angles relative to the laboratory frame and a beam-center position (BC). The tilt about X_L (called tx) requires Friedel pairs from a single crystal and cannot be determined from powder rings alone.

Azimuth η	Detector Axis
0°	+Z_D
+90°	−Y_D
−90°	+Y_D
±180°	−Z_D

Image Transformation

When loading raw detector images, MIDAS may need to apply a transformation to make the image consistent with the laboratory coordinate system. Transformations are specified via the ImTransOpt parameter:

ImTransOpt	Transformation
0	No change
1	Flip left/right
2	Flip top/bottom
3	Transpose

Important

Getting ImTransOpt right is essential. If the image orientation does not match the coordinate system, calibration will fail silently and all downstream results will be incorrect. Verify using a physical feature on the detector (e.g., a beam-stop wire, a fiducial marker) whose position you can track.

Sample Frame

The sample coordinate system is an intermediate frame unknown to the diffraction geometry. MIDAS treats crystals in the sample purely as scatterers that produce spots as the sample rotates through diffraction conditions. It is the user's responsibility to:

1. Track the sample orientation relative to the beam during mounting.
2. Use fiducial markers (e.g., Au dots, notches, tomography) to relate the lab frame to the sample frame for post-analysis interpretation.

---

If you encounter any issues or have questions, please open an issue on this repository.