BGA Package Warpage Through the Reflow Profile

Package warpage is the change in shape of a ball grid array package as it passes through the reflow profile, and it is the reason a joint can fail on a package that was measured as flat before assembly. The package is a composite of materials with different expansion rates, so it bends as the temperature rises, and the direction of that bend changes across the melting range of the alloy.

Why Packages Warp

The substrate, the mould compound and the die each expand at a different rate, and the copper inside the substrate is distributed unevenly. Temperature therefore produces a bending moment across the package, and the direction of the bend depends on which layer dominates at that moment.

The behaviour is not linear. A package may be convex at room temperature, flat through the soak zone and concave at the peak, so a coplanarity figure taken cold describes only the state in which the part is shipped and not the state in which it is soldered.

The Reflow Temperature Range

Warpage reaches its maximum near the melting range of the alloy, between roughly 200 and 240 degrees Celsius for a lead free process, which is exactly the interval in which the balls must contact the pads and form joints. The height of the warpage at that moment, rather than at any other point in the profile, is the figure that matters.

The peak temperature and the time spent above liquidus both influence the result. A higher peak increases the warpage but shortens the time available for it to relax, while a longer liquidus time gives the package more opportunity to distort while the alloy is fluid. Warpage is therefore quoted at the temperature where the alloy is molten, and not at the temperature where the measurement happens to be convenient.

Coplanarity and Ball Contact

Coplanarity describes the deviation of the ball array from a plane, and it combines the package warpage at temperature with the ball height variation of the part as received. A package that is flat at 25 degrees Celsius can still have a ball at the corner that is 80 micrometres below its neighbours at 230 degrees Celsius.

Contact is lost when the gap between a ball and its pad exceeds the collapsed height of the joint that can still be formed. In practice a few tens of micrometres of extra separation at the corner is enough to produce an open joint once the paste volume and the pad tolerances are taken into account.

Shadowing and Open Joints

When the package bows away from the board, the joints at the edge of the array must be formed across a larger gap, and the balls near the centre of the bow lose contact entirely. The result is a ring of open or partially formed joints that follows the direction of the bend rather than the thermal pattern of the oven.

The same distortion changes the local solder volume: joints that remain in contact may receive more alloy than intended as the package settles, which shows up as a wide, flattened joint beside an open one. Shadowing of this kind is a package and profile problem, and it is rarely corrected by changing the paste. Where the printed paste is thicker than intended the same warpage produces more opens, because the deposit holds the package further from its neighbours.

Measuring Warpage

Warpage is measured either by a shadow moire system that maps the surface as the part is heated, or by a digital image correlation method that tracks the displacement of a pattern on the package. Both produce a plot of deflection against temperature, and it is the value near the melting range that is quoted.

Measurements taken at room temperature are not a substitute. A cold coplanarity report describes handling and shipping, while the figure that predicts assembly yield is the deflection measured between 200 and 240 degrees Celsius, at the heating rate used in production.

Carrier and Support Effects

How the board is supported during reflow changes the measured behaviour of the assembly even when the package is unchanged. A thin board that sags in a centre supported carrier lifts the package away from the paste just as effectively as package warpage does, which is why board warpage has to be measured alongside the package.

Support pins placed under the package area reduce board sag, and a carrier with a flat, well supported base reduces it further. Where an open joint appears only on thin boards, the board rather than the package is usually responsible, and the two effects have to be separated with a measurement rather than with an assumption. A support placed in the wrong position can also lift a thin board into the package instead of holding it away.

Profile Adjustments That Help

The most effective single change is often a longer soak at a temperature below the solidus, which lets the package and the board reach a similar temperature before the alloy melts. The joint then forms on a package that is already close to its final shape.

A slower ramp from the soak zone to the peak also helps, because it gives the package time to distort while the alloy is still solid. Raising the peak temperature to compensate for warpage is the least effective option, since it increases the distortion it is meant to overcome.

Defects and Diagnosis

The signature is a ring of open or bridged joints at one edge or one corner of the array, with good joints elsewhere, and it repeats on every board. Joints that are open in the middle of a package and sound at the edges point in the other direction, to a paste or pad problem.

X ray inspection shows the open joints and the wide joints together, and the pattern is what identifies the cause. A solder joint that is open at the package edge, wide near the centre and accompanied by no paste defect is a warpage signature rather than a printing one.

Process Window and Records

The window is described by peak temperature, time above liquidus, soak length, ramp rates, board support and the paste volume. It is established for each package and board combination, because the same package behaves differently on a thick board and on a thin one.

Records should carry the package lot, the measured warpage at temperature where it is available, the support arrangement and the profile used. The BGA profile that works with one package should not be assumed to work with the next revision of the same device, because the internal construction is what sets the warpage.

FAQ

What causes package warpage in reflow? The different expansion rates of the substrate, mould compound and die produce a bending moment. The package bends differently at soak and at peak, so the deflection near the melting range is the figure that matters.

Can a reflow profile fix warped packages? It can reduce the effect. A longer soak below the solidus and a slower ramp to the peak let the package distort before the alloy melts, which is more effective than raising the peak temperature.

How is warpage measured? Normally by shadow moire or digital image correlation through a thermal cycle, reporting deflection against temperature rather than a single room temperature coplanarity figure.