Blind and Buried Vias: Uses, Limits and Cost

Blind and buried vias let a designer move a signal between layers without drilling through everything above and below it. That is worth real money on a dense board, and it is also one of the few features where a layout decision commits the product to a specific, more expensive fabrication flow. This article explains what the structures are, how they are built, and where their limits are.

The Three Via Types

A through via is drilled from the top of the stack to the bottom and plated through the full thickness. It is the cheapest and the most common, and it blocks the layers it passes through for the whole of its length.

A blind via is drilled from an outer layer but does not reach the opposite side, so it is visible on one face and buried inside the board. A buried via connects two or more inner layers and is not visible from either face.

Both exist for the same reason: a through via consumes real estate on every layer it crosses, because the copper keepout around it has to be maintained on all of them. On a dense board the accumulated keepouts can make the difference between fitting the routing and adding two more layers. Replacing a few through vias with blind or buried structures frees area on the layers the via no longer crosses.

There is a second benefit that has nothing to do with area. The unused part of a through via below the connected layer is a stub, and a stub reflects energy and adds jitter at high frequency. A blind via has no stub by construction, and a buried via has none either, which is why these structures appear on high speed backplanes and on thick boards where a through via stub would be several millimetres long.

How They Are Made

Blind and buried vias are built by sequential lamination. Instead of pressing the whole stack at once, the fabricator builds a subassembly, drills and plates the buried vias in it, then laminates additional layers on top, drills the blind vias, and repeats as needed.

Each additional lamination cycle is a complete pass through press, drilling, plating, imaging and inspection, so the cost rises with the number of cycles rather than with the number of vias. A board with one buried via layer and one blind via layer typically needs two or three press cycles, and that is what the quotation reflects.

Blind vias in a thin outer layer are usually laser drilled, because the depth has to be controlled to a small fraction of the board thickness and mechanical drilling cannot hold that tolerance reliably. Deeper blind vias, or blind vias through several layers, are mechanically drilled with controlled depth, which is a slower and less repeatable operation.

The depth control is the critical process parameter. A blind via that stops a little short leaves insufficient copper at the target layer; one that goes a little deep damages the layer below. Both failures can pass a continuity test and only appear later as an open or a short after thermal cycling, which is why the process window in a first article is examined by cross section rather than by electrical test alone.

Microvias and the HDI Variants

When the blind via is small enough to connect only adjacent layers, it becomes a microvia. These are laser drilled, typically 75 to 150 micrometres in diameter, with a depth of 50 to 100 micrometres, and they are the building block of HDI construction. A microvia to the next layer is the simplest form; a via that spans two layers in one step is a two level or skip via.

Stacked microvias put one microvia directly on top of another, which gives the greatest density and lets the designer route as though the layers were transparent. Staggered microvias place each level offset from the one below, connected by a short copper link on the intermediate layer. Staggered construction is more reliable, because no column of stacked copper has to survive thermal cycling, and it is what most standards prefer for automotive and other high reliability products.

Stack on via is a third variant: a microvia placed on top of a filled and plated through via, which is the structure used to escape a dense ball grid array without consuming routing channels. It is common in the highest density smartphone and module designs and it is also the most demanding, because the underlying via has to be filled flat and plated properly before the next level is built.

Design Rules and Limits

The rules below are the ones that decide whether a design is manufacturable rather than merely attractive on screen.

• Aspect ratio. A laser drilled microvia is normally limited to about 0.75 to 1 in depth to diameter. A blind via that is deeper than its diameter is harder to plate reliably.
• Land size. The capture pad on the target layer has to be large enough to absorb the depth tolerance and the registration error. A microvia needs a land typically 100 to 150 micrometres larger than the hole.
• Depth tolerance. The drilling depth is controlled to within a small fraction of the dielectric thickness, and the layer below the target must not be damaged. Stackups are designed so the target layer has a copper plane to stop on where possible.
• Via fill. Buried vias and any via under a subsequent lamination cycle are usually filled with resin or plated and planarised, because an unfilled via collapses under press pressure.
• Layer pairing. Blind vias are typically limited to a small number of outer layers, and buried via layers must be symmetric in the stackup for the board to stay flat.
• Test access. Each buried structure adds a layer where a probe cannot reach, so test coupons on the production panel carry the structures that cannot be tested on the board itself.

Two limits deserve emphasis because they are where projects get into trouble. The first is the number of lamination cycles: each one adds cost, lead time and yield risk, so a design that uses buried vias on three separate layer pairs may be more expensive than one with more layers and fewer via types. The second is stacking, since a stack of four microvias is a mechanical structure that has to survive reflow and thermal cycling, and not every supplier will build it.

What They Cost

The cost structure is different from most design decisions, because it is driven by cycles rather than by area. A board with one sequential lamination step costs noticeably more than a through via board of the same size and layer count; two steps cost more than twice as much again, because each cycle carries its own tooling, its own inspection and its own yield risk.

Against that, the structures often pay for themselves by removing layers. If blind and buried vias let a design fit on eight layers instead of twelve, the saving in material and in press cycles can exceed the extra cost of the sequential flow. That comparison is the only honest way to evaluate them, and it is worth asking the fabricator to quote both options on the same stackup rather than comparing a via type in isolation.

Yield is the other factor. Buried vias are built before the outer layers exist, so a defect in a buried via cannot be reworked and is often not discovered until final test. That is one reason the reliability of the process matters more than the price difference between two suppliers, particularly for microvia and stacked constructions.

Inspection and Reliability

A blind or buried via cannot be inspected visually, and a simple continuity test proves only that copper exists somewhere in the path. The methods used in practice are cross sectioning of a first article, X-ray inspection where the geometry allows it, resistance measurement on dedicated test structures, and thermal stress or thermal cycling of coupons.

The failure modes are specific and worth knowing. Incomplete plating on the barrel wall is the classic one, and it appears as a high resistance or an open after thermal stress. A blind via that stops short of the target layer produces a connection that tests at the beginning of the barrel rather than through it. Resin voids in a filled via show up later as a blister or a delamination. And a stacked microvia can fail at the interface between two levels, which is why stagger is preferred where reliability is the priority.

Some standards place additional constraints. Automotive and medical customers frequently require staggered microvias, a minimum dielectric thickness between microvia levels, and a specified thermal cycling qualification for the via structure rather than only for the board. Those requirements should be raised before the first build, because they change the stackup rather than only the process.

When a Through Via Is Still Better

Blind and buried vias are not the only way to solve a density or stub problem, and the cheaper options should be considered first.

Back drilling removes the unused stub of a through via by drilling it away after plating, and it is the standard answer on thick backplanes, where it delivers most of the electrical benefit of a blind via at a fraction of the cost. It does not help with routing density, since the via keepouts remain on every layer, but where the problem is signal integrity it is usually the first tool to reach for.

Via in pad, teardrops or a different escape pattern can also recover the space that seemed to require buried vias. A careful review of how much routing is actually blocked, rather than how much looks blocked, often finds that a through via design will fit with a small amount of rework.

FAQ

• What is the difference between a blind via and a buried via? A blind via is visible from one outer surface and connects to an inner layer. A buried via is entirely inside the board and connects only inner layers.
• Do blind vias always require laser drilling? Only when they are small and shallow. Deeper blind vias through several layers are mechanically drilled with depth control.
• How many lamination cycles can a board have? Three is common and up to five is possible, but each cycle adds cost and risk, so most designs use the minimum that fits the routing.
• Can buried vias be repaired? No. Once laminated, a defective buried via cannot be reworked, which is why yield and process control matter more than price on this type of board.

Summary

Blind and buried vias exist to solve two problems: routing channels on a dense board, and via stubs on a fast or thick one. They are built through sequential lamination, so their cost is set by the number of press cycles rather than by the number of holes, and their limits are set by aspect ratio, depth control, via fill and stacking rules.

They are worth the money when they remove layers or when a stub would close the eye on a high speed channel. Where the problem is density alone, a careful look at the escape pattern is cheaper; where the problem is the stub, back drilling is the first option to price. Both should be discussed with the fabricator as options on one stackup, and where the density is extreme, an any layer HDI construction may be the structure that actually meets the requirement.