Data outputs

Every measurement writes a single file (or pair, for legacy mode) into <data_directory>/<sanitized-username>/. The base filename is built as:

{unix_timestamp}_{sanitized_sample_name}_{mode_tag}_{source_value_str}

where mode_tag is R, VSRC, ISRC, 4PP, vdP, or DATA, and source_value_str summarises the source level (e.g. 1.00mA, 0.500V, 2.00mA_delta, sweep_0.0to1.0). Username and sample name are sanitized to strip path separators and .. traversal before they touch the filesystem.

The file extension is chosen by the active exporter (selected in Settings → Output).

Format choices

Backend	Setting value	Files produced	Best for
CSV (default)	csv	<sample>_<timestamp>.csv	Excel / Origin / pandas. Human-readable. Optional gzip on finalize.
HDF5	hdf5	<sample>_<timestamp>.h5	Compact archival, mixed-type columns, attribute-encoded metadata. Always internally gzip-compressed. Requires optional h5py.
CSV + Legacy JSON	csv+legacy_json	<sample>_<timestamp>.csv + .json	Back-compat with pipelines written against pre-2.0 ResistaMet. The CSV has only a header row (no # metadata block); the JSON contains the metadata + a full data copy.

Format version is 2.0 for the CSV and HDF5 backends, 1.0 for the legacy dual emit. Recorded in the metadata so consumers can branch on it.

CSV layout

# resistamet_format_version: 2.0
# user: alice
# sample: cu-foil-spot-1
# mode: four_point
# started_at: 2026-05-25T14:33:21.451
# software_version: 1.12.0
# instrument: KEITHLEY INSTRUMENTS INC.,MODEL 2420,1230523,C30
# gpib_address: GPIB0::24::INSTR
# sampling_rate_hz: 10.0
# nplc: 1.0
# settling_time_s: 0.2
# params.source_current_A: 0.0001
# params.voltage_compliance_V: 5.0
# params.probe_spacing_cm: 0.1016
# params.thickness_um: 0.5
# params.k_factor: 4.532
# params.alpha: 1.0
# params.model: thin_film
# params.target_samples: 20
# params.auto_zero: on
# units: s,V,A,Ω,Ω/□,Ω·cm,S/cm,V,A,,
elapsed_s,V,I,V_over_I,Rs_ohm_sq,rho_ohm_cm,sigma_S_cm,V_unc_V,I_unc_A,compliance,event
0.1,0.001045,0.0001,10.45,47.36,2.37e-3,422.1,3.0e-4,3.1e-8,OK,
0.2,0.001047,0.0001,10.47,47.45,2.37e-3,421.3,3.0e-4,3.1e-8,OK,
...
# --- run completed ---
# ended_at: 2026-05-25T14:33:23.512
# total_samples: 20
# duration_s: 2.061

Two metadata blocks: one at the top (everything known at run start), one at the bottom (filled in at finalize). The trailing block is delimited by # --- run completed ---. Comments use # prefix; the column header row is the first non-comment line. The units: line is parallel-indexed to the column header so downstream code can look up units by column name without hard-coding.

Crash safety: rows are flushed and fsync'd as they arrive, so the partial CSV is the recovery artifact — there's no separate checkpoint sidecar for the v2.0 backend (the legacy dual-emit backend writes .json.tmp checkpoints).

parse_metadata(path) in resistamet_gui/data_export.py reads both blocks and returns a flat dict; handles .csv and .csv.gz transparently.

HDF5 layout

Single dataset named data of compound dtype (every column as a variable-length UTF-8 string for type-mixing safety). Metadata lives in attrs keyed by the same dotted-path names the CSV uses (params.source_current_A, params.thickness_um, etc.), plus columns and units attribute arrays. End-metadata is appended to attrs at finalize. Chunked (1024 rows per chunk), gzip-compressed at level 6.

Per-mode columns

Resistance

Column	Unit	What it is
elapsed_s	s	Time since Start
V_meas	V	Voltage measured at the DUT (across the probes if 4-wire)
I_meas	A	Current sourced (instrument-reported, matches the test-current setpoint)
R_ohm	Ω	Instrument-reported R — preserves Enhanced R / offset-comp / source-readback features (NOT V_meas/I_meas recomputed here). When Cable Null is active the res_cable_null offset is subtracted from this column only; V_meas and I_meas are written as the instrument measured them.
R_unc_ohm	Ω	Per-reading σ_R. When Enhanced R is on, this comes from the datasheet's Enhanced column; otherwise from V/I RSS propagation.
compliance		OK on a normal reading, V_COMP when the voltage side hit compliance
event		Empty or the label of the most recent M keypress event marker

Metadata params: test_current_A, voltage_compliance_V, measurement_type, auto_range, auto_zero, offset_compensated_ohms.

Voltage Source

Column	Unit	What it is
elapsed_s	s	Time since Start
V_set	V	Sourced voltage (the setpoint — actual output is V_set ± σ_V_source)
I_meas	A	Measured current
R_calc	Ω	V_set / I_meas
I_unc_A	A	σ_I from the per-range current-measurement spec
R_calc_unc_ohm	Ω	σ_R via RSS through R = V_set/I_meas (combining V-source spec + I-measure spec)
compliance		OK on a normal reading, I_COMP when the current side hit compliance
event		Event-marker label written by the M keyboard shortcut

Metadata params: source_voltage_V, current_compliance_A, current_auto_range, duration_hours, auto_zero.

Current Source

Mirror of Voltage Source: V_meas/I_set instead of V_set/I_meas, V_unc_V instead of I_unc_A.

Metadata params: source_current_A, voltage_compliance_V, voltage_auto_range, duration_hours, auto_zero.

Four-Point Probe

Column	Unit	What it is
elapsed_s	s	Time since Start
V	V	Voltage measured between inner two probes
I	A	Current sourced through outer two probes
V_over_I	Ω	V/I — the raw ratio before geometric corrections
Rs_ohm_sq	Ω/□	Sheet resistance = K · α · V/I (K modified by ASTM F84 corrections when applicable)
rho_ohm_cm	Ω·cm	Resistivity = Rs · thickness
sigma_S_cm	S/cm	Conductivity = 1 / rho
V_unc_V	V	σ_V from per-range voltage-measurement spec
I_unc_A	A	σ_I from per-range current-measurement spec
compliance		OK on a normal reading, V_COMP when the voltage side hit compliance
event		Event-marker label written by the M keyboard shortcut

Per-reading uncertainty on Rs / ρ / σ is recoverable downstream: σ_X/X = √((V_unc/V)² + (I_unc/I)²). The aggregated per-spot stats use a different formula combining statistical and instrument uncertainty — see Concepts → Uncertainty.

Delta mode splices four more columns before compliance:

Column	Unit	What it is
V_plus	V	Voltage at +I polarity
V_minus	V	Voltage at −I polarity
R_f	Ω	Forward V_plus / +I
R_r	Ω	Reverse V_minus / −I

Per the F84 §11.2.2.2 forward/reverse diagnostic — R_f / R_r should approach 1; large deviation flags thermoelectric or rectification effects.

Metadata params: source_current_A, voltage_compliance_V, probe_spacing_cm, thickness_um, k_factor, alpha, model, target_samples, plus the F84 inputs (fpp_diameter_cm, fpp_geometry, fpp_temperature_c, fpp_dopant_type) and delta-mode settings when active.

Van der Pauw

One row per geometry (4 rows total per spot), both polarities captured in the same row:

Column	Unit	What it is
elapsed_s	s	Time the row was captured
geometry		One of 4 F76 geometry labels
group		Geometry pair (A/B) for the §11.1 homogeneity check
source_high / source_low / sense_high / sense_low		Contact numbers (1–4) for this geometry — derived from F76 Fig. 2
label_pos		Operator-facing label for the positive-polarity measurement
V_pos	V	Voltage at +I
label_neg		Operator-facing label for the negative-polarity measurement
V_neg	V	Voltage at −I
current_A	A	Sourced current magnitude (same |I| at both polarities)

The final sheet resistance, resistivity, f-factors, Q ratios, and the §11.1 homogeneity result land in the metadata at finalize, not as data rows. They appear in the trailing # --- run completed --- block with dotted keys:

• vdp_result.sheet_resistance (Ω/□), vdp_result.sheet_resistance_uncertainty
• vdp_result.rho_avg (Ω·cm), vdp_result.rho_avg_uncertainty
• vdp_result.rho_a, vdp_result.rho_b (the two F76 group resistivities)
• vdp_result.q_a, vdp_result.q_b, vdp_result.f_a, vdp_result.f_b
• vdp_result.homogeneous (boolean — true when |ρ_A − ρ_B|/ρ_avg ≤ F76 §11.1 threshold)
• vdp_result.asymmetry_pct (the |ρ_A − ρ_B|/ρ_avg × 100 number)
• vdp_result.current_a, vdp_result.thickness_cm
• vdp_result.voltages.<label> — every raw V measurement, keyed by F76 geometry label

Metadata params: source_current_A, voltage_compliance_V, thickness_cm, settling_s, readings_per_polarity, standard: ASTM F76-08 Method A.

I-V Sweep

Column	Unit	What it is
point		Sweep index (0…N)
V_source	V	Sourced voltage at this step (for V-source sweeps) — for I-source sweeps this column is named I_source and the next is V_meas
I_meas	A	Measured current (for V-source sweeps)
compliance		OK on a normal point, COMP when that point hit compliance

Metadata params: source_function, start, stop, step, compliance, delay_s, direction.

Reading the data back

A small example to load a v2.0 CSV with metadata into pandas:

import pandas as pd
from resistamet_gui.data_export import parse_metadata

path = "measurement_data/alice/cu-foil-spot-1_20260525_143321.csv"

# Metadata as a flat dict
meta = parse_metadata(path)
print(meta["mode"], meta["nplc"], meta["params.source_current_A"])

# Data rows — pandas skips the `#`-prefixed header automatically
df = pd.read_csv(path, comment="#")
print(df.head())

For HDF5:

import h5py

with h5py.File(path.replace(".csv", ".h5"), "r") as f:
    print(dict(f.attrs))                # metadata
    data = f["data"][:]                 # structured numpy array
    print(data.dtype.names)             # column names

For per-spot 4PP summaries, the Export Summary… button on the 4PP tab writes a separate (non-v2.0) CSV with a header block and two sections:

4-Point Probe Summary
Sample,<sample-name>
User,<username>
Model,<thin_film | semi_infinite | finite_thin | finite_alpha>
Spacing s (cm),<value>
Thickness t (cm),<value>
Alpha,<value>

Metric,Mean,StdDev
Sheet Resistance (Ω/□),<mean>,<std>
Resistivity (Ω·cm),<mean>,<std>
Conductivity (S/cm),<mean>,<std>

Per-Spot Results
Spot,N,Rs Mean (Ω/□),Rs Std,Rs RSD%,ρ Mean (Ω·cm),ρ Std,σ Mean (S/cm),σ Std
<spot-name>,<n>,...

The "Per-Spot Results" section is only written when ≥1 spot has been saved with the Save Spot button. When ≥2 spots are saved, an Inter-spot Uniformity block is appended:

Inter-spot Uniformity
Rs Mean-of-Means (Ω/□),<value>
Rs Std-of-Means (Ω/□),<value>
Inter-spot RSD%,<value>

Mean/Std use np.nanmean / np.nanstd(ddof=1); NaN and infinite values are written as the literal string N/A.