Critical Dimensions in Microfluidic Drawings
A microfluidic drawing does two jobs. It describes the geometry, and it communicates which parts of that geometry are functionally important.
The first job is generally handled well by a solid CAD model. The second depends on annotation choices — and where those choices are unclear, the consequence tends to surface late, in the form of a part that matches the model but does not perform as the experiment requires.
Why Identifying Everything as Critical Reduces Clarity
A drawing where every dimension carries a tight tolerance conveys a similar amount of actionable information as one where none does: no priority ordering.
Manufacturing involves trade-offs between setup, sequence, tooling selection and inspection effort. Where a manufacturer can see which characteristics govern function, those trade-offs can be made deliberately. Where all dimensions are presented as equally important, two outcomes are possible: the priorities are inferred incorrectly, or all features are treated as tight — which can raise cost and lead time for characteristics that had substantial functional margin.
There is also an inspection consequence. Inspection scope follows critical characteristics. A long list of critical dimensions implies either a correspondingly long measurement scope, or a scope conversation that would have been more efficiently held at the drawing stage.
A working heuristic: if you cannot state what changes functionally when a dimension drifts, it may not need to be designated critical.
Characteristics That Are Often Functionally Important
Which characteristics matter is application-specific. The categories below recur frequently enough to be worth checking against your own design.
Channel cross-section
Where flow rate, residence time, shear condition or droplet formation are governed by channel geometry, cross-sectional dimensions are directly implicated. In that case, specify them — and specify where they are measured. A channel with any taper along its length has different dimensions at different positions, so “channel width” without a stated location is ambiguous.
If your application is demonstrably insensitive to moderate variation in channel cross-section, state that explicitly. Otherwise, treat channel width and depth according to their functional importance.
Feature positions relative to external interfaces
Where a chip mounts to a stage, mates with a manifold, or aligns with an optical path, the positional relationship between those features and their datum is functionally important. Depending on the application, the absolute position of a channel within the substrate may be less important than its position relative to a locating feature.
Ports and external interfaces
Ports and external interfaces deserve particular attention because sealing and alignment requirements are often different from the fluidic geometry itself.
Worth specifying:
- Port position relative to the declared datum
- Port diameter and depth, and whether threaded or straight
- Which fitting or tubing standard the design targets, if any
- Whether the port face requires a particular finish for sealing
An under-specified interface can produce a part whose internal geometry is acceptable but whose connection to tubing or a manifold is unreliable.
Substrate thickness where optics or fit depend on it
If you are imaging through the chip with a specific objective and working distance, the material thickness between the channel and the outer surface may be functionally important. If the chip must fit a slot or holder, overall thickness is.
If substrate thickness is not function-critical, identify it as flexible or provide an acceptable range rather than imposing an unnecessarily narrow value. This may provide more manufacturing options.
Interfaces with existing hardware
Dimensions that interface with equipment you already own are worth stating explicitly, including which face serves as the reference.
Characteristics That May Not Require Tight Control
- Cosmetic finish on surfaces not used for observation
- Edge treatment and chamfers, where handling and sealing do not depend on them
- Outer profile dimensions where the chip sits in free space
- Internal corner radii in regions where flow behaviour is not sensitive
- Feature positions in areas with substantial clearance
Leaving these open is a deliberate choice, not an omission. It allows the manufacturability review to propose a practical route, which can reduce cost and lead time.
Declare a Datum
Dimensioning features from one another in a chain, without a declared reference, creates two difficulties: tolerances accumulate along the chain, and there is no agreed reference against which to inspect.
Establishing a datum — typically a corner, edge or locating feature — and dimensioning functionally important features from it addresses both. Where a chip mounts to external hardware, the mounting interface is often a suitable datum, since it governs alignment in use.
Many early-stage research prototypes do not require a fully developed GD&T scheme, but they do benefit from a clear datum strategy and directly dimensioned critical features.
If you do want to apply formal geometric tolerancing, the governing standards are ASME Y14.5-2018 (R2024) and, in the ISO GPS system, ISO 1101:2017 together with ISO 5459:2024 for datums and datum systems. To reduce ambiguity, use one governing dimensioning and tolerancing convention consistently and identify it in the drawing notes.
State the Tolerance the Application Requires
A tolerance states the acceptable range for a characteristic. Two patterns worth avoiding:
Inheriting a default tolerance block. A template’s general tolerance was written for a different class of part. Applied to a microfluidic channel, it may be either uninformative or unachievable.
Specifying tighter than required as a precaution. Tightening beyond functional need does not add margin to the experiment. It can add cost and lead time, and may convert a straightforward feature into a demanding one.
The productive question is: within what range does the application still work? That range is the tolerance. If it is not yet known, saying so is a legitimate answer and is better resolved in review than estimated on a drawing.
Achievable tolerance is also not a fixed constant across a part. It depends on the feature, the material, the surrounding geometry and the manufacturing route. A tolerance that is straightforward on one feature may be impractical on another feature of the same component. This is one reason a preliminary manufacturability review before quotation is worth the time.
Account for the Effect of Bonding
The channel in your CAD model is the channel geometry before the cover layer is joined.
Bonding processes can alter channel geometry. This is documented in the literature across several bonding routes: thermal fusion bonding is performed with the substrates heated near or above the glass transition temperature, and channel deformation is a recognised consequence [1]; in solvent bonding, overexposure to solvent can deform the microchannels, so solvent concentration and exposure time are controlling factors [2]. Work comparing sealing methods has quantified channel deformation by cross-section imaging before and after bonding [3].
Whether this change matters depends on the function of the device and the magnitude of the geometry change. If final channel cross-section is function-critical, identify it before the bonding route and inspection plan are agreed.
This also affects what “critical dimension” means in practice. A dimension measured on an open channel before bonding is not necessarily the dimension the fluid encounters in the finished device — which is worth stating explicitly if it matters to your result.
Check That Critical Characteristics Can Be Verified
A tolerance on a characteristic that cannot practically be measured functions as an expectation rather than a specification.
Before finalising a drawing, ask of each critical characteristic: how would this be verified? Where the answer is unclear — an internal feature with no line of sight, a dimension inside a closed channel, a feature near the resolution limit of available measurement methods — either the specification needs reconsideration, or verification will be indirect and that should be agreed in advance.
Either outcome is workable. The situation to avoid is discovering the ambiguity after parts exist, when the drawing states one thing, the measurement reports another, and no agreed verification method exists to resolve it.
A Practical Drawing Package
For a prototype microfluidic chip:
1. A 3D file (STEP or STP) — the complete geometry.
2. A 2D drawing (PDF or DXF) — showing the declared datum, critical characteristics with tolerances, port specifications and fitting standards, any optical or surface requirements, and a revision identifier.
3. A short statement of intent — a few sentences on what the chip does and which features are functionally critical. A short statement of intent gives the reviewer information that the geometry alone does not contain: which features carry the function and which requirements have flexibility.
Revision Identifiers
Place a revision identifier on the drawing and update it when geometry changes. Microfluidic designs often iterate, and a project may pass through several revisions during development.
Without a revision identifier, the question of which version produced a given batch can become difficult to answer — frequently at the point where results are being written up.
Summary
- Designate few characteristics as critical, on functional grounds
- Declare a datum
- Specify the tolerance the application requires
- Fully specify ports and external interfaces
- Account for the effect of bonding on final channel geometry
- Confirm that each critical characteristic can be verified
- Send a 3D file, a 2D drawing, and a statement of intent
- Include a revision identifier
References & Further Reading
- Giri, K.; Tsao, C.-W. Recent Advances in Thermoplastic Microfluidic Bonding. Micromachines 2022, 13(3), 486. DOI: 10.3390/mi13030486 — supports: thermal fusion bonding is performed near or above Tg, with channel deformation a recognised consequence.
- Shakeri, A.; Khan, S.; Abu Jarad, N.; Didar, T.F. The Fabrication and Bonding of Thermoplastic Microfluidics: A Review. Materials 2022, 15(18), 6478. DOI: 10.3390/ma15186478 — supports: solvent overexposure can deform microchannels; exposure time and solvent concentration are controlling factors.
- Mahmoodi, S.R.; Sun, P.K.; Mayer, M.; et al. Gas-assisted thermal bonding of thermoplastics for the fabrication of microfluidic devices. Microsystem Technologies 2019, 25, 3923–3932. DOI: 10.1007/s00542-019-04380-9 — supports: channel deformation quantified by cross-section imaging before and after sealing.
- ASME Y14.5-2018 (R2024), Dimensioning and Tolerancing. American Society of Mechanical Engineers.
- ISO 1101:2017, Geometrical product specifications (GPS) — Geometrical tolerancing — Tolerances of form, orientation, location and run-out. Fourth edition, 2017-02. ISO/TC 213. Reviewed and confirmed 2022; current.
- ISO 5459:2024, Geometrical product specifications (GPS) — Geometrical tolerancing — Datums and datum systems. Third edition, 2024-10. ISO/TC 213. Cancels and replaces ISO 5459:2011, which is withdrawn.
These references support general statements about thermoplastic microfluidic fabrication and engineering drawing practice. They describe published work by third parties and standards published by third-party bodies. They do not verify, endorse or characterise Yeda Precision’s manufacturing capability.
Ready to have a design reviewed?
Send your STEP, STP, DXF or PDF for a preliminary manufacturability review.
Related: How inspection scope is agreed · PMMA microfluidic chips