Scan region in Omicron / RHK .sm4

How an .sm4 file from an Omicron microscope with an RHK controller states the scanned area, and how the reader turns it into NeXus under the scan region convention.

Raw keys

Raw key	Meaning	How the reader uses it
RHK_Xoffset, RHK_Yoffset	Centre of the scan area	scan_offset_value_x, _y
RHK_Xscale, RHK_Yscale	Signed distance between adjacent pixels	\|scale\| is the pixel pitch; the sign of RHK_Yscale is the slow direction
RHK_Xsize, RHK_Ysize	Pixels per line, number of lines	scan_points_x, _y
RHK_X/@unit, RHK_Y/@unit	Unit of the lateral axes	Unit of offset, range, start, end and step
RHK_ScanType	Names the Forward and Backward pass	Channel naming; one scan control group per page
RHK_Angle	Rotation of the frame	scan_angle_x, _y, not applied

No range is stored: it is range = N × |scale|, the edge-to-edge width, counting N pitches and not N - 1.

How the scan direction was decided. The slow direction comes from the sign of RHK_Yscale, positive for an upward scan and negative for a downward one. The fast axis stays unsigned: RHK_ScanType names the Forward and the Backward pass, but it does not say which way either ran, and RHK_Xscale carries the same sign on both pages, so it describes the coordinate mapping of the stored array rather than the travel of the tip. An RHK document defining RHK_ScanType, or two scans of one area with opposite RHK_ScanType and a feature that fixes the direction, would settle it.

Evidence

The centre reading is graded vendor manual and empirical test on vendor data. The RHK R9 User Manual states that "Move to Center: moves the Scan Area to the center of the Scan Range by setting the XY Offsets to 0" (p. 195), so the XY offset is the position of the scan-area centre within the scanner range and may take any value in it. The page header stores the same quantity: in Figure_6c.SM4 the embedded R9 parameter block gives X offset ::1.5773e-007 m against a header RHK_Xoffset of 1.6301e-07 m, within 5 to 9 nm, where a first-pixel reading would put them about 50 nm apart.

Scan pairs confirm it. The method and the data are described centrally.

Pair, identical offsets	Correlation peak / next	Measured	Centre predicts	First pixel predicts
VT231211_A1_0064 (10 nm) in _0065 (30 nm)	0.71 / 0.54	(0.53, 0.41) nm	(0.00, 0.00) nm	(9.96, 9.96) nm
VT231205_A1_0064 (up) and _0063 (down), 20 nm	0.44 / 0.38	(0.94, 4.26) nm	(0.00, 0.00) nm	(0.00, 19.96) nm

The 10 nm scan sits in the middle of the 30 nm one. The up and the down scan overlap, which the first-pixel reading forbids: with RHK_Yscale of opposite sign it would put one scan in the 20 nm below RHK_Yoffset and the other in the 20 nm above it, two strips that never meet. Three further pairs with different offsets agree as well.

Gwyddion's rhk-sm4.c reads the offsets and never applies them, and spym drops them for image pages; RHK's own "SM4 Data File Format" document is not publicly available.