Table of Contents Table of Contents
Previous Page  8 / 58 Next Page
Information
Show Menu
Previous Page 8 / 58 Next Page
Page Background edfas.org

ELECTRONIC DEVICE FAILURE ANALYSIS | VOLUME 18 NO. 2

8

unnecessary and only increases the cost of the system.

The

Z

axis uncertainty must be equal to or better than

the required profile integrity. If it is desired to reproduce

a surface profile to ±1 µm, the

Z

axis repeatability need

not be less than 0.5 µm. If

X-Y

-positional repeatabil-

ity is not constrained directly, it can be defined as the

maximum profile slope divided by the

Z

axis maximum

uncertainty. Therefore, if a 1 µm

Z

axis limit is required,

and the maximum profile slope is 20 µm/mm, the

X-Y

uncertainty needs only to be less than 50 µm. Nowhere

is there a requirement for 50 nm resolution in any axis or

“submicron” accuracy.

If a machine is required to thin plastic-packaged

devices and produce a thinned die with less than ±5 µm

in thickness variation, the following specifications are

all that are required. Tighter specifications only result in

increased purchase and maintenance costs:

• X

and

Y

axis resolution: 1.0 µm

• Z

axis resolution: 0.5 µm

• X, Y

axis independent repeatability: 2.0 µm

• Z

axis independent repeatability: 1.0 µm

• Spindle runout: 10 µm

• Axis orthogonality: 0.05 mrad, maximum

• Variation fromstraight line travel: 0.003mmper 25mm

of travel, including runout

• Axis pitch and yaw: 0.05 mrad, maximum

• Stage deflection: 50 Newtons/µm, maximum

The axis resolution defines the minimum required

scale resolution. It should be 50% or less of the repeat-

ability. The spindle runout limits the effective increase in

tool diameter caused by “wobble.” The axis orthogonality

and axis pitch and yaw each limit axis-to-axis interaction

to 0.05 µm per millimeter of travel. Because all materials

and machines are elastic, the stage deflection specifica-

tion is required to limit the

Z

axis positional change as a

result of the tool forces.

In total, over a 25 mm

2

area, the

X-Y

-positional uncer-

tainty is approximately 15 µm with 2 µm repeatability.

The

Z

-positional uncertainty is approximately 11 µm

with repeatability of less than 2 µm. Any more than this is

unnecessary and costly, and any less does not guarantee

performance.

The axis positional uncertainty comes into play if

the surface profile is measured by different equipment.

If this is done, all of the positional uncertainties of both

the measuring system and the processing machine add.

Because the positional uncertainty is much greater than

repeatability, the integrity of profile reproduction comes

into question.

FRONTSIDE DELAYERING

REQUIREMENTS

Delayering requirements are much more complex.

Delayering on a flat lap requires that the device rotational

axis and the scan axis be parallel to the platen rotational

axis to a degree of precision that is not normally encoun-

tered. Maintaining 10 nmplanarity of a 10mm

2

die sample

requires that all axes be within 0.002 mrad. In addition,

the vertical runout of the platenmust be less than 0.2 µm.

Measuring the runout is problematic, as are the align-

ment measurements, but all are possible with the right

measurement equipment and personnel. Delayering on

a backside thinning machine is even more problematic

because axis alignment is either not available or difficult

to adjust to the accuracy required. Delayering requires

that the spindle be orthogonal to the

X-Y

plane within

0.0067 mrad for a 3-mm-diameter tool. A larger tool

diameter tightens the requirements. Normally, the spindle

orthogonality of a backside system is more than 10 times

that required to do die delayering. Additionally, 10 nm

Z

axis repeatability is not truly available. This indicates that

using a flat lap is difficult and using a backside thinning

system is not realistically possible. If one is delayering

a 5 µm design-rules die, almost anything can be used.

Currently for the latest design rules, only a very carefully

set up flat lap can meet the requirements.

CONCLUSIONS

Claiming or stating resolution as accuracy is disin-

genuous because accuracy is a function of many different

parameters. Accuracy, repeatability, and resolution must

be matched to the process requirements. Increasing

resolution, repeatability, and accuracy beyond what is

required will not increase sample quality. The desired

process results should determine the equipment specifi-

cations. There are some critical parameters that currently

are not specified by some equipment suppliers, such as

the geometric relationships and straight line movement

variations of each of the axes. To ensure that the desired

results are obtained, these parameters must be defined.

For some processes, there are no readily available, simple,

and easy equipment solutions…yet.

REFERENCES

1. “Accuracy,”

Dictionary.com

,

dictionary.reference.com/browse/ accuracy?s=t,

definition 2.

2. T. Doiron and J. Beers:

The Gauge Block Handbook,

NISTMonograph

180, 2005.

(continued on page 10)