Operation of a Vacuum Reflow Oven with Preliminary Void Reduction Data

Fred Dimock

Mgr, Process Technology IntInternational

Mike Meilunas Universal Instruments - Advanced Process Lab

Arvind Srinivasan Karthikeyan Auburn University

Vacuum Reflow

The science of using nothing

to remove empty spaces from

something

Outline

Voids

Why inline Vacuum Reflow?

The Inline Vacuum Reflow Oven.

The Vacuum Reflow Cycle?

The Vacuum Trial The test vehicle

Preliminary results from an extended trial at the

Universal Advanced Process Lab

Things to Consider

Voiding is of particular concern forThermal management of QFN and other devices

All about keeping the chip cool by eliminating voids in

Thermal Pads because empty spaces don’t transfer heat

Void issues

• Voids form for many reasons including:▫Paste solvent volatilization▫Paste solids volatilization▫Oxide volatilization▫Outgassing (PCB, Component)

Void formation and growth

• Voids grow by merging

• Then escape the solder if they contact an exposed surface.

Why Vacuum reflow?

“Low void” pastes have shown that void reduction is possible,

but void levels are still higher than desired.

Typical reflow results in ~ 40% voids in thermal pads

Initially there was hope that the solder paste guys could

find a way to reliably lower or eliminate voids with flux

and thermal profile modifications.

Why Vacuum reflow?

“Low void” pastesMichael Meilunas, “Solder Joint Void Analysis: Effects of Paste and Reflow Parameters”

AREA Consortium

Low Void Paste

Hot Profile

Average Cumulative Void % Standard Paste

Cold Profile - Air

Atmosphere Paste Profile MFL100

Standard NC Cold 33%

Air Cold 19%

Low Void Hot 13%

very Hot 13%

Cold 18%

Nitrogen Low Void Hot 13%

very Hot 10%

Vapor phase reflow + vacuum: expensive, slow, messy

Large voids are more likely to contact exposed solder surfaces and escape

Why Vacuum reflow?

Convection reflow + vacuum:

no vapor, faster throughput, minimum mess

Vacuum reflow has been shown to be effective at reducing solder voids

Voids grow as surrounding pressure decreases (Pascal’s Law)

Large voids have an opportunity to combine and become larger voids

Vacuum Chamber is placed

between the last heated zone and cooling section

with automated board transfer in and out of the chamber

Inline vacuum reflow oven

Four Basic Vacuum Process Steps

Not including transport

Pump Down

Vacuum level

Hold time

Equalization

Rate at which vacuum is applied Torr /sec

The amount of vacuum Torr

The time at the chosen vacuum level Seconds

The rate at which the chamber is Torr /sec

returned to room atmosphere

Pump Down Rate at which vacuum is applied

Too fast

Voids explode when leaving the joint

causing solder balls and splatter

We experienced parts moving and flipping

with a 100 Torr/ sec pump down.

0

100

200

300

400

500

600

700

800

30 to 60 Torr / sec

at hard vacuum levels

Vacuum levelThe amount of vacuum

0

100

200

300

400

500

600

700

800

Levels as low as 1 and high as 250 Torr

(and higher) are possible

Experienced significant void reduction

even at 250 Torr

0

100

200

300

400

500

600

700

800

Hold timeThe time at the chosen vacuum level

Hold times as short as 1 sec

and long as a few minutes are possible.

At 1 sec there is little time for voids to

move to the surface

Times greater than 40 seconds have

been proven to not be needed

Equalization Rate at which the chamber is

returned to atmospheric pressure

0

100

200

300

400

500

600

700

800

Currently no known limitations

Experiments done at ~ 85 Torr/sec

Trials at the Universal Instruments Advanced Process Lab

Mike Meilunas Universal Instruments

Arvind Srinivasan Karthikeyan Auburn University

• Two Layer Test Board• 78x104x1mm

• Immersion Sn and ENIG finishes• MLF: 4, 8 and 12mm bodies• WLCSP: 0.4mm pitch• LGA: 0.4mm pitch • SMR: 2512 and 1206• BGA: 1.0mm pitch • D2PAK

Trials at the Universal Instruments Advanced Process Lab

Air & Nitrogen

Numerous pastes

SAC 305

Innolot

Tin Lead

Three paste suppliers

Two board finishes

Two pad designs some with vias

Varied Vacuum level

Hold time

We ran

over 200

boards

Assembly Process

• Screen print solder paste• 100 micron thick stencils

• Paste inspection• Koh-Young 8030-3 & Optical

• Placement & Inspection• Reflow soldering

• BTU Pyramax 125N (control – no vacuum)• BTU Pyramax Vacuum Oven

• KIC SPS Profiler• Post-Reflow Inspection

• Visual• X-ray analysis (Dage XD7600)

Atmosphere Nitrogen

Solder paste SAC 305

Board finish ENIG

Profile RTS - 240 peak

Pad Design Window pane

Pump down rate Fixed

Refill rate Fixed

Multiple

Vacuum levels

Hold Times

Preliminary results from 3 MLFs 100, 52, & 16

MLF 16 no vac vs vac

MLF 100 no vac vs vac

Hold times % Voids @ 60 Torr

Torr Sec MLF100 MLF52 MLF16760 0 38.54 27.81 20.37

60 10 9.3 4.2 3.060 20 8.0 3.4 1.760 30 6.1 2.6 2.560 40 6.2 2.5 2.2

20 5 7.320 10 3.65 2.49 1.1820 20 3.35 1.24 1.1420 30 1.93 0.72 1.0420 40 1.99 1.13 0.66

> 10%< 10%< 3%

Hold times % Voids @ 20 Torr

Conclusions:

Vacuum level % Voids @ 20 Sec hold time

Torr Sec MLF100 MLF52 MLF16760 0 38.54 27.81 20.37

500 20 26.08120 20 6.7 4.1 2.445 20 4.2 2.1 1.320 20 3.35 1.24 1.146 20 1.86 1.86 0.75

> 10%< 10%< 3%

500 30 26.8250 30 9.79 5.62 5.2185 30 7.4 3.1 2.520 30 1.93 0.72 1.046 30 1.5 0.66 0.96

Vacuum level % Voids @ 30 Sec hold time

Conclusions:

Conclusions:

1. Nothing removes most of the empty spaces

from something

3. Vacuum levels as soft as 245 Torr and short as

5 seconds remove some of the empty spaces.

4. Hard vacuum of 20 Torr with a hold time of 30 seconds

results in void levels of about 2% in MFL 100s

2. It is easier to remove voids from small components

than large ones.

Things to consider

Voids located at critical points (“hot spots”)

may be the real issue

Void Size

Large voids are most likely the primary concern

Small voids may not be troublesome

Cumulative voiding may only have a secondary affect

QFN Void Distribution Analysis: QFN100

At 120 Torr and

lower vacuum

levels, the odds of

getting a void larger

than 1% is small

Fred Dimock

Mgr, Process Technology IntInternational

Thank you!

Mike Meilunas Universal Instruments - Advanced Process Lab

Arvind Srinivasan Karthikeyan Auburn University