semi合集-English.pdf - 第6485页

SEMI G78-0699 © SE MI 1999 18 A1-1 .1 5 Step 13 — Calcu late Normaliz e d (b y lot) Die Offset (11) = Average D ie Offset (3) - Av erage Lot Offset (9) [Accu racy] loti setj w afk diel loti setj wafk diel loti setj wafk …

SEMI G78-0699 © SEMI 199917

A1-1.11 Step 9 — Calculate

W-W-X or W-W-Y = 3σ of Normalized (by setup) Wafer Offset (8).

A1-1.11.1 A 3σ variation will be calculat ed based on the normalized wafer offset and will be cited as the Wafer-to-

Wafer variation (W-W-X, W-W-Y).

A1-1.11.2 Discusssion of W-W-X or W-W -Y (Wafer-to-Wafer variation in the X-direction, Wafer-to-Wafer

variation in the Y-direction)

A1-1.11.2.1 This equation will produce a v alue for the 3σ variation from Wafer-to-Wafer in the X and Y

directions. If this number were large, it would tend to indicate that the prober does not behave repeatably from

wafer-to-wafer. Possible causes are:

• loading problems

• temperature stability issues, and

• wafer rotation issues

A1-1.12 Step 10 — Calculate

Average Lot Offset (9) = average of Average Setup Offset (7).

LOTX xoff

LOTY yoff

loti setj

(9)

A1-1.12.1 Discussion of Average Lot Offset (LOTX or LOTY)

A1-1.12.1.1 This will be one of the values added to the graph. This is simply the average of all of the average die

offsets in the X-direction and all of the average die offsets in the Y-direction. If one needed to pick a single number

in X and Y to describe the accuracy of the prober – this is the one. This can be very misleading though – what this

number really tells you is what value the distribution is centered around. Refer to the graph to get a visual sense of

this data.

A1-1.13

Step 11 — Calculate

Normaliz ed (by lot) Setup Offset (10) = Average Setup Offset (7) - Average Lot

Offset (9)

loti setj

loti setj loti setj

xnorm

xoff

ynorm yoff

LOTX

LOTY

=−

(10)

A1-1.13.1 Discussion of Normalized (by lot) Setup Offset

A1-1.13.1.1 This is another normalization step used to calculate the true setup-to-setup error in X and Y.

A1-1.14 Step 12 — Calculate

S-S-X or S-S-Y = 3σ of Normalized (by lot) Setup Offset (10).

A1-1.14.1 A 3σ variation will be calculat ed based on the normalized setup offset and will be cited as the Setup-to-

Setup variation (S-S-X, S-S-Y).

A1-1.14.2 Discussion of S-S-X or S-S-Y (Setup-to-Setup variability in the X-direction, Setup-to-Setup variability in

the Y-direction)

A1-1.14.2.1 These two values describe how much variability there is between setups.

SEMI G78-0699 © SEMI 1999 18

A1-1.15 Step 13 — Calculate

Normaliz ed (by lot) Die Offset (11) = Average Die Offset (3) - Average Lot Offset

(9) [Accuracy]

loti setj wafk diel

loti setj wafk diel loti setj wafk diel

xoff

yn yoff

LOTX

LOTY

,, ,

,, , ,, ,

=−

(11)

A1-1.15.1 Discussion of Normalized (by lot) Die Offset

A1-1.15.1.1 This step is useful if you need to compare the variation between several different lots. It is not

completely necessary for the next step, since subtracting a constant from a string of numbers does not change the 3σ

variation, only the center point.

A1-1.16 Step 14 — Calculate

TTLX or TTLY = 3σ of Normalized (by lot) Die Offset (11) [Repeatability]

A1-1.16.1 Discussion fo TTLX or TTLY

A1-1.16.1.1 This is the value for determini ng (with 3σ confidence) whether your probe marks will always fall

within your desired spec. They are the hatch-marked lines.

A1-1.17 Step 15 — This normalized die offset includes Die-to-Die, Wafer-to-Wafer, and Setup-to-Setup offsets

thus a 3σ variation of this offset will be called Total Prober

Variation (TTLX and TTLY).

A1-1.17.1 Calculate Total Prober Error Range (3

LOTX + TTLX, LOTX - TTLX = Total Prober X-Error Range (3σ)

LOTY + TTLY, LOTY - TTLY = Total Prober Y-Error Range (3σ)

A1-1.17.2 Suggested Graphs:

X-Offset Y-Offset

LOTX + TTLX LOTY + TTLY

LOTX LOTY

LOTX - TTLX LOTY - TTLY

Average Die X-Offset (3) Average Die Y-Offset (3)

A1-1.17.2.1

It is recommended to create on e graph illustrating the prober accuracy in X and one graph for Y.

A1-1.18 Discussion of Graphs (one exa mple for X [Figure X] and one for Y [Figure Y])

A1-1.18.1 Average Lot X-Offset is the va lue from Step 10. This number shows where the probe mark distribution

is centered (the accuracy of the prober).

A1-1.18.2 Average Lot X-Offset +3σ is t he result from step 10 + the result from step 14

A1-1.18.3 Average Lot X-Offset -3σ is th e result from step 10 – the result from step 14

A1-1.18.4 These two lines show the ± 3σ range of your distribution. This describes the repeatability of the prober.

NOTE: There will always be explainable and unexplainable accuracy and repeatability errors on a prober. If these two lines fall

within your desired specification, you will probably not have any problems. The example graph for X-offset illustrates a well-

behaved, or “passing” prober. The example graph for Y-offset, on the other hand, is representative of a prober which failed to

meet the desired accuracy specification.

A1-1.18.5 The upper and lower desired specification limits have been added to the graphs.

A1-1.18.6 The final piece of data added t o the graph is the Average Die Offset. This, as stated in Step 1, is very

good at illustrating any trends throughout the lot.

A1-1.18.7 Tips for graph making:

• Add gridlines across the X-axis to separate wafers.

• Print all graphs (X and Y, different lots, different probers) with the same scale in X and Y.

• Add a legend to your graphs, so that they will be readable by your “audience” as well as yourself.

Avg Lot X-Offset

Avg Lot X-Offset +3s

Avg Lot X-Offset -3s

Lower Spec Limit

Upper Spec Limit

Avg Die X-Offset

Figure 7

X-Offset