From your data to a decision you can explain.

01 · Start with the spectrum

1. Create an account, verify your email and wait for approval when required. The web account is separate from Windows application registration.
2. Explore the synthetic example to learn the controls. Its signals and results are demonstrations, not experimental measurements.
3. Import processed files, select a ramp and review gradients. Click the peaks of interest in the spectrum.
4. In Diffusion, apply reviewed parameters before calculating. Inspect attenuation, excluded points and warnings.
5. Compare geometries, estimate mass only with an applicable calibration, and save the project with its decisions.

02 · Import processed Bruker or CSV

Bruker · 1r / 2rr

Select the complete folder with its subfolders. The importer detects both 1r spectra and 2rr matrices, including different pdata numbers. Acquisitions shows each series or processing separately, with source, pulse program, nucleus, temperature and solvent. Alternative processings are not mixed within a fit. If 1r and 2rr exist in the same pdata, 2rr takes priority and a warning is recorded.

For 1r retain procs and acqus. For 2rr also retain proc2s, acqu2s and associated lists: dimensions, QF layout, counts and metadata determine whether rows can be associated with gradients. A candidate series requires reviewing its conditions. A map with a D axis and a spectroscopic 2D dataset can be explored but are not fitted as attenuation. An unknown axis can only be confirmed with evidence when row correspondence is compatible and processing is not contradictory. Padding rows are separated from fitting.

Gradients may come from GPZ[6] or an associated list. gradient_percent.csv declares percentages; difflist and gradlist retain their original values and their units are not guessed. Open Review rows & gradients and choose to declare original-list units or edit documented percentages. For G/cm or T/m, provide the calibrated maximum gradient in T/m and its source; calculation uses that same maximum. Reconvert the list to change it. Raw fid/ser files require prior processing and are not transformed here.

A folder may contain several 1r ramps, multiple 2rr DOSY datasets with different parameters, and original ser/fid data. Available processed data are inspected and their groups remain separate. A ser file without its processed counterpart does not provide a spectrum or enable fitting here; process it in TopSpin first, then select the folder again.

Single-spectrum CSV

Use the literal header ppm,intensity,gradient_percent. The third column is optional; when present, it must be constant throughout the file. The ppm axis must be strictly increasing or decreasing. Select multiple CSV files for a series.

ppm,intensity,gradient_percent
7.250,120,8
7.249,450,8
7.248,180,8

Attenuation CSV

The first column is gradient_percent, followed by one intensity column per signal. Optional name__included columns preserve inclusion with true/false or 1/0. This table does not invent a spectrum or ppm position.

gradient_percent,signal_A,signal_A__included,signal_B
8,1000,true,820
20,760,true,690
40,210,false,390

These fragments only illustrate the format. Use all relevant data, finite intensities and gradients within ±100%. The exported results table is not an importable intensity table.

03 · Associate ramps and review conditions

Detection uses metadata, acquisition order and gradient changes. Folder numbers do not define percentages or require starting at 10 and ending at 32. Select a series in Acquisitions and open Review rows & gradients. Changes in pulse program, nucleus, D/P, NS, solvent, gradient shape, axis or processing are separated. The full temperature span of a group cannot exceed 0.1 K. When gradients are declared, resets automatically separate new blocks; manual decisions are retained by spectrum identity and affected fits are invalidated. Do not mix incompatible pulse programs or timings in a fit.

D20/P30 and initial control values are suggestions to check. Apply & recalculate records your review; it does not certify an experimental calibration.

04 · Track a maximum, not the baseline

Click near a resonance or type its ppm position. A positive maximum is sought within a limited local window and distance; the marker shows an actual point of the visible spectrum. Zoom with the wheel, pan with Ctrl + drag or Pan, and reset with a right click.

For overlapping peaks, zoom and check that the same resonance is tracked throughout the ramp. Molecular regions and peak references are indicative: a single chemical shift does not uniquely identify a molecule.

Candidate molecules when you click a peak

The peak panel and Candidate molecules automatically show catalogue names compatible with the original monitored crest in the active ramp. Changing the peak or series refreshes proposals; identification from an earlier result is not reused. Each candidate includes its published window, multiplicity when recorded, companion positions and source or DOI. Companions are references to review, not automatically detected signals.

NUC1 and SOLVENT are checked across all active spectra. If information is missing, Review nucleus and solvent lets you declare it and Confirm context saves it only for that ramp, without changing data or calculations. No ¹H/CDCl3 default is assumed. Mixed, contradictory or unsupported context requires acquisition review. No compatible reference molecule does not mean there is no substance: Possible standards and Indicative regions and types are presented separately.

Assign this label saves an explicitly chosen hypothesis with crest, context, references and provenance. It does not confirm identity or apply an internal standard, shape or D/Rh correction. Remove label clears that choice. Labels and context are saved with the project; review a previous label after changing acquisition or context.

Rectangle zoom and axis fitting

Enable Box zoom and drag a rectangle at least 6 pixels wide and tall. Either direction and a reversed ppm axis are supported. Escape cancels; Home or a right click restores the complete extent. A click without dragging on an existing marker selects that resonance and opens its panel, even while Box zoom or Pan is active. Turn those modes off to add new peaks; in Diffusion, turn off Box zoom to include or exclude points.

The wheel zooms XY around the pointer; Shift + wheel changes X only, Alt + wheel Y only. Fit restores the complete data. Fit Y uses original samples inside the visible X interval, so off-screen peaks do not dominate its scale. Resizing preserves zoom. On touch screens, tap selects, drag pans and two fingers zoom; in rectangle mode, one finger draws the window.

Peak panel and appearance

1. Select a marker or a signal in Resonances. The panel shows the tracked maximum and, after calculation, its D and Rh values and applied models. Selecting another signal updates its content.
2. Drag the header or the Move panel button to move it away from the spectrum. You can also focus that button with Tab and use the arrow keys; Shift increases the step. The panel stays inside the visible area.
3. Press × or Escape within the panel to close it. Click the marker or its list entry again to reopen it; Return to the selected peak restores its position beside the maximum.

Switch appearance in the sidebar alternates light and dark themes. Light mode uses darker, opaque traces, with coordinated colours for spectra, markers and results. Changing the theme preserves intensities and calculations; if a peak is outside the scale, use Fit Y or Fit.

05 · Review attenuation and exclusions

In Diffusion, inspect ln(I/I₀) against squared normalized gradient. Click a point to include or exclude it; the decision is retained per signal and ramp. A non-positive value cannot enter a logarithm directly.

Detect noise tail proposes exclusions when the end no longer supports useful attenuation. It does not impose a universal cutoff at 0.5. Review proposals against the original signal and table; Restore points lets you reconsider. A high R² does not establish identity, absence of overlap or model validity.

06 · Separate measurement and corrections

External-reference calibration uses its slope and conditions on a compatible gradient scale. It differs from correction using an internal signal in the same ramp. For temperature, specify the target and decide whether to include the viscosity change.

For an internal reference, select the signal, confirm identity and provenance, and enter its expected D at measurement temperature. Suggested solvent, TMS, TSP or TTMS references require experimental confirmation. Reference and analyte must share the ramp, nucleus and solvent.

Reference and temperature factors are applied once from measured D. Review their values and do not reuse an already corrected D as input.

SMSQ10 and duration per lobe

The Sinnaeve analytic model represents a 10% sin² rise, 80% plateau and 10% cos² fall. Enter δ = P30/1000 ms per lobe. STE-BPP uses δeff = 2δ internally; PGSE/STE uses δ. Do not also double the input. Check Δ and τ against the acquired pulse program.

K(PGSE/STE) = γ² δ² × 0.81 × (Δ − 0.3010342623428058 δ)
K(STE-BPP) = γ² δeff² × 0.81 × (Δ − 0.32525856558570143 δeff − τ/2)
D = −slope / (Gmax² K)

Times in these equations are in seconds and γ is angular. The 0.81 area factor alone does not replace the timing term. Double sequences need their own reviewed model.

Probe management and absolute calibration

Open Probes to use the TBI, TBO, TBI LF and BBFO recipes supplied for Legacy. All retain Dref = 1.902 × 10⁻⁹ m²/s, Δref = 50 ms and the literal values in the table. Temperature correction is separate. The δref field is the total effective duration; the experiment input remains P30/1000 per lobe.

δeff = 2 P30 / 1000; Δ = D20 × 1000 [ms]
Fexp = 0.81 (Δ − 0.32525 δeff − 1/2)
coef = (1.902 / |slope_ref|) (δref/δeff)² Fref/Fexp
D / (10⁻⁹ m²/s) = |slope_sample| × coef
Absolute: D = |slope_sample| / (Gmax² K)
Derived: Gmax = sqrt(|slope_ref| / (Dref Kref)) [SI]

Duplicate a recipe to edit it or add an absolute probe with Gmax in T/m and evidence. Absolute mode retains the experiment waveform, sequence and timing. Export/import JSON profiles between Windows and web. The web library is stored per account and browser; custom profiles also travel with the project. Each series retains its applied profile, revision, method, conditions and SHA256. Editing the library does not change earlier results. Opening a saved project does not recalibrate it.

1. First select the ramp you want to calibrate and open Probes. The four original recipes are protected; Duplicate creates an editable custom profile. Add probe starts an absolute profile.
2. For a custom profile, enter name, family, nucleus and provenance. Add serial numbers or instruments when documented. Allow compatible automatic detection controls automatic selection; disabling it still permits manual application.
3. Choose Relative recipe to use a standard and slope, or Absolute to enter Gmax and its provenance. Derive Gmax from standard computes an equivalent from a valid reference; it does not create a new calibration measurement. Save profile updates the library; Apply to analysis records the choice for the active ramp. Review its conditions and recalculate.
4. Export JSON includes custom profiles for transfer to another browser or to Windows. Import JSON validates their fields; an incompatible identifier requires resolving the conflict and does not silently replace another calibration. Delete profile removes it from the library while retaining snapshots already applied in projects.

Detection uses probe metadata, nucleus, pulse program, waveform and timing. Serial or instrument matches take precedence over family; ambiguous matches require a choice. A family-level supplied recipe does not validate every physical probe of that type. The previous experimental TBO profile below remains separate and requires its own conditions.

Detected probe ≠ calibrated gradient

Profile TBO_HDO_20260904_temperature_adjusted_v1 belongs to probe Z148626_0002, instrument spect, ¹H, BF1 500.13 MHz, stebpgp1s1d and GPNAM[6] SMSQ10.100. Automatic application requires compatible evidence and matching SHA256 hashes of waveform and pulse-program files. Missing evidence requires documented review; contradictions block transfer. A TBO/TBI/BBFO family name or folder title is insufficient.

The profile retains its literal 0.32525 timing coefficient and effective Gmax ≈0.4807882547 T/m. Nominal 0.535 T/m does not replace it. Analytic 0.32525856558570143 is a separate model; equation and calibration are not silently interchanged. Profile ID, hash, source and model accompany the saved decision.

Estimated viscosity and internal reference

Curves use log10(η / mPa·s) = A + B/T + C·T + D·T² for pure liquids within their published interval. CDCl3 uses the CHCl3 curve with an isotope-approximation notice. There is no extrapolation or claim to measure your sample viscosity. D2O, mixtures and matrices without a curve require documented η.

Reference suggestions examine maxima in the active ramp and rank candidates by nucleus, solvent and catalogue compatibility. Confirm identity, bottle and overlap before applying. TTMS is an ambiguous laboratory label; it does not automatically mean TTMS-Si, TTMS-O, TMS or HMDSO. A ppm-only reference needs calibrated D at a known temperature or applicable Rh for the matrix; quantitative values are not invented.

07 · Interpret Rh, shape and apparent mass

Rh is an equivalent hydrodynamic radius derived from D, temperature and viscosity. Compare sphere, cylinder and ellipsoid in 3D; drag to rotate and use the wheel to zoom. Adjust Q = b/a or P = a/b. The ellipsoid coincides with a sphere when Q = 1.

Geometry and Wirtz/Chen microfriction are assumptions, not a structure resolved from diffusion alone. Declare solvent radius when applying microfriction.

Molecular mass offers infinite dilution and finite concentration; in the latter C is solute mass / solvent mass. Review calibration family, solvent, temperature and applicability. Apparent mass from a scalar D is not Mn, Mw or dispersity. Outside numerical support or at a discontinuity, the application does not invent a mass. Linear-polymer calibrations do not automatically validate stars, brushes or branched resins.

Read the molecular-mass card

The main card presents apparent mass for the selected resonance, its ppm position and the applied model, without an equation in the result area. Below it are the effective D, temperature and viscosity; the concentration model adds the solute/solvent ratio. These are the conditions used by that estimate, not another correction applied to the result.

The comparison lists resonances with a calculated mass. Select a row to change the selected signal; its bar represents mass relative to the largest value in the list, not a molecular distribution. Open Conditions and scope of the estimate to review warnings and limits. If Outside model support or No solution in this model segment appears, review the data and applicability rather than interpreting a missing value as zero.

Catalogue provenance and corrections

Windows are queried by nucleus and solvent. The Babij2016 extension adds ¹H patterns for 2-MeTHF, CPME, isopropyl acetate and L-ethyl lactate in CDCl3/DMSO-d6. Candidates retain source, conditions, multiplicity and companion signals. Specific windows are not drawn as general bands and do not assign shape.

Two published corrections replace historical matches: tert-butanol CH3/CD3OD 1.40→1.22 ppm and pyridine ¹³C CH(3,5)/CD3CN 127.76→124.77 ppm. Historical rows remain documentary records. Chemical-shift references do not certify D, Rh or identity.

08 · Preserve the project and export

• Download project. Saves a local .diffatonce file with data, parameters and decisions. Do this before closing or reloading: there is no results autosave. Reopening recalculates from saved information.
• Save to my account. Explicitly sends processed spectra and project state. Limits: 5 MiB per project, 20 projects and 50 MiB per account. Keep a local copy if larger.
• My projects. Open or delete copies in your account. A revision conflict requires opening the current version before overwriting. Download a copy before deletion.
• HTML / PDF / PostScript / CSV. Export HTML and use Print → Save as PDF. PostScript generates .ps; CSV exports the results table. The report records available conditions, decisions, models and warnings.

Original Bruker files remain on your computer. Cloud saving does send processed project intensities: only store data you are authorized to upload.

09 · Registration, access and privacy

An account asks for a name or alias, email and a 14–72 byte password. It grants no administrator permissions. The initial policy requires email verification and approval. When email is configured, request an expiring, single-use link to verify or recover access. If none arrives or delivery is not configured, use Support: the administrator can check email ownership and record manual approval.

The administrator can pause the service, close registration or block an account. Blocking does not delete projects. With the tab open, download a local copy if a notice appears; server requests recheck permission.

Email, alias and operational account activity are retained. Private cookies are required for the session; the laboratory includes no advertising analytics. Consult the general notice to contact us about retention and rights.

Online account and Windows installation

These are separate access systems. An online-laboratory account uses email, password, email verification and approval when required by policy. A Windows installation uses a copy identifier and a signed authorization. Approving an installation ID under Copies and installations does not activate a web account; verifying a Legacy Online email does not authorize a Windows copy.

In Windows, Request sent · awaiting approval confirms receipt of registration and awaits the administrator; Check approval queries its state. Request not confirmed means a valid response is missing and connectivity should be reviewed. For a web-account problem, provide your account email to support; for Windows, provide product, version, state and visible ID. Do not send passwords or identity files.

Windows registration and authorization help →

10 · Web application FAQ