1) The document summarizes a project to identify the root cause of crystallization defects appearing in the blackout frit coating on inner oven door glass for a premium household appliance manufacturer. 2) Statistical analysis including a full factorial experiment confirmed with 95% confidence that frit thickness was the prime cause of crystallization, while frit mix was a secondary contributing factor. 3) Recommended corrective actions focused on reducing frit thickness, including modifying the screen used for coating and monitoring thickness during the coating process.

1. Page 1
Inner Oven Door Glass
Project Summary Report
Statistical Engineering
Document By: Dave Johnson and Chris Rosko Revision
Date: 10/07/10
Team Leader: David C. Johnson, Consultant, Technical Problem Solutions
Assistant Leader Christopher Rosko, Consultant, Technical Problem Solutions
Project Start Date: September 13, 2010
Project Completion Date: October 8, 2010
Problem Definition
‘A’ Corporation, the manufacturer of premium household appliances ,out of San Luis Potosi,
Mexico has just launched a new oven with a ‘B’ Corporation supplied inner door glass. The ‘A’
Corporation leadership has expressed dissatisfaction with the appearance quality of the inner oven
door glass.
Focused Defined Project
Find and eliminate the Prime X causing crystallization in the blackout frit coating on the inner oven
door glass.
Strategy
The strategy choice used on this problem was a layer to layer based on full contrast and leverage
(See page 8).
Measurement System
A sensory scoring based on visual level of crystallization was used to evaluate the samples (See
page 13).
Clue Generation
Refer to Convergence Diagrams (See pages 4-7).
Statistical Confirmation
Full Factorial using convergence split variables frit thickness and frit mix statistically confirmed
with 95% confidence that thickness is the Prime X and mix is the Secondary X . (EC = 16; see
pages 10-12). Frit thickness is therefore the root cause with greatest impact on generating
crystallization, and frit mix is a significant secondary contributor.
Corrective Action
1) Continue use of multi-vane mix auger.
2) Apply balanced, even squeegee pressure through entire print cycle.
3) Schedule more frequent screen wipeouts with reduced run size & frit screen load (target 25 and
adjust based on performance data).
4) Monitor wet film thickness at beginning & end of batch run (Suggest using a wet film comb gage
with graduations starting at 10 microns). Set in process control limits based on final inspection
data.
Executive Summary

2. Page 2
Inner Oven Door Glass
Project Summary Report
Statistical Engineering
Document By: Dave Johnson and Chris Rosko Revision
Date: 10/07/10
Executive Summary (Cont.)
Corrective Action Cont’d.
5) Stop any future frit reclamation use efforts.
6) Continue with planned experiment & implementation of modified screen mesh. Either the 230x48
or 255x40 screen provides a reduced theoretical wet film thickness from the current 196x55
screen which is consistent with SE project findings.
7) Continue use of the wet sponge inspection technique at final inspection to help identify glass
panes with crystallization.
8) Incorporate the crystallization boundary samples augmented with inspector daily refresh
acknowledgement at final inspection.
Explanation of the Physics
Enamel frit properly applied is formulated to allow the heating process to burn off (out gas) the
organic medium (solvents) early in the oven bake process before the sintering of the frit solids to the
glass substrate. If sintering begins prior to complete solvent burn-off; the latent, explosive, out gas
of entrapped solvent in a softened, molten, frit – glass flux results in porosity & micro fractures along
the glass interface surface. Excessive frit thickness and poor mixture exacerbate the solvent
entrapment. Optical analysis of the samples containing crystallization confirm both the presence of
porosity through the frit enamel coating and glass substrate micro-fractures at the frit to glass
interface. Humidity and moisture accentuate this condition due to light refraction. The difference in
the Indices of Refraction of the ceramic glass (1.50+) verses water (1.33) causes a prism effect
(crystallization). It was demonstrated that this appearance condition could be accelerated using an
environment test chamber. Suspect crystallization samples placed in an environmental chamber
and held for 2 hours at 200° F & 98% humidity revealed the crystallization defect.

3. Page 3
Inner Oven Door Glass
Project Summary Report
Statistical Engineering
Document By: Dave Johnson and Chris Rosko Revision
Date: 10/07/10
Problem Convergence Diagram
Dimensional Appearance Other
Decision Split: Management
direction
Improve Customer Satisfaction
with Mabe Glass
Decision Split: Pareto – Highest
frequency of complaints/returns
Decision Split: Pareto – Highest
frequency of returns / in-house scrap
(50%)
Decision Split: Pareto – Highest
frequency of complaints
Customer Returns Other
Find & eliminate the X Prime causing crystallization in
blackout frit coating on the Mabe inner oven door
panel glass.
Management Metric
Tin Oxide Blackout Frit Other
Scratches Crystallization Other
Focused Defined Project

4. Page 4
Inner Oven Door Glass
Project Summary Report
Statistical Engineering
Document By: Dave Johnson and Chris Rosko Revision
Date: 10/07/10
Project Convergence Diagram
Sensory Crystallization Score
Random
Event Defect Feature
Decision Split: Problem is a defect
Find and eliminate the X Prime causing crystallization in
blackout frit coating on the Mabe inner oven door panel
glass.
Elimination Split: Problem occurs
Randomly in 2D, Non-Randomly in
3D. SEM analysis shows
crystallization occurs at the glass
surface along the frit interface.
Layer to
Layer
Region to
Region
Strategy Split: Layer to Layer was
chosen based on Contrast &
Leverage.
Decision Split: Problem occurs on
both sizes. Focus will be on 17”
glass panel & the solution will be read
across to the 20” glass
17” glass 20” glass
Non Random
(Layer to Layer)
Variable Attribute
Elimination Split: Assigned sensory
scoring to frit appearance to allow for
variable measurements.
Other
CEL = 2 Sensory Crystallization Score
BOB WOW
0 1 4 7
Green Y is the severity rating
of the visual crystallization
appearance.
Customer Enthusiasm Limit
(CEL) requires the Green Y
be targeted at 1.
Y Response Distribution
2 10
Focused Defined Project
Contamination No Contamination Elimination Split: SEM element
mapping indicates no contamination.

5. Page 5
Inner Oven Door Glass
Project Summary Report
Statistical Engineering
Document By: Dave Johnson and Chris Rosko Revision
Date: 10/07/10
Project Convergence Diagram Cont’d
CEL = 2 Sensory Crystallization Score
BOB WOW
0 1 4 7
Y Response Distribution
2 10
Screen Print
Parts Process
Elimination Split: Same material
can generate both BOB and WOW
conditions.
Elimination Split: Same oven
conditions produced both BOB and
WOW conditions.
Elimination Split: Same screen
frame mesh produced both BOB and
WOW conditions.
Elimination Split: Operation Search –
The Y Response is exposed after
enamel apply (oven cure).
Glass Preparation Enamel (Frit) Print
Oven
Mix Thickness
Elimination Split: Wipeout solvent
was organic-based & consistent with
frit material. Samples with excess
wipeout solvent did not create
crystallization. Both mix and
thickness are possible X Prime
candidates.
Frame MeshEnamel Apply
Wipeout
X Prime
Candidate
X Prime
Candidate
Confirmation: Full Factorial
statistically confirms with 95%
confidence that Thickness is the X
Prime & Mix is the Secondary X
(EC = 16).
Failed to ProveProved

6. Page 6
Inner Oven Door Glass
Project Summary Report
Statistical Engineering
Document By: Dave Johnson and Chris Rosko Revision
Date: 10/07/10
Project Convergence Diagram Cont’d
1) Continue use of multi-vane mix auger.
2) Apply balanced, even squeegee pressure through
entire print cycle.
3) Schedule more frequent screen wipeouts with
reduced run size & frit screen load (target 25 and
adjust based on performance data).
4) Monitor wet film thickness at beginning & end of
batch run (Suggest using a wet film comb gage
with graduations starting at 10 microns). Set in
process control limits based on final inspection
data.
5) Stop any future frit reclamation use efforts.
6) Continue with planned experiment &
implementation of modified screen mesh. Either
the 230x48 or 255x40 screen provides a reduced
theoretical wet film thickness from the current
196x55 screen which is consistent with SE project
findings.
7) Continue use of the wet sponge inspection
technique at final inspection to help identify glass
panes with crystallization.
8) Incorporate the crystallization boundary samples
augmented with inspector daily refresh
acknowledgement at final inspection.
Corrective Action

7. Page 7
Inner Oven Door Glass
Project Summary Report
Statistical Engineering
Defect Contrast Diagram
Contrast Rationale
Defect to Defect
Same Point
Measurement. Variable using sensory rating
of crystallization severity.
Some contrast. No leverage. Concentration
Diagram show crystallization at some points and
not at others.
Good contrast. Some leverage. Some regions
had more crystallization than others in 2D view.
Good contrast. Little leverage. Glass panels
exhibited both BOB & WOW conditions, but this
choice does not provide clues to physics of the
failure mode.
Some contrast. No leverage.
Both glass panel sizes had BOB & WOW
conditions present.
Some contrast. No leverage.
Both shifts produced panels with BOB & WOW
conditions present.
No data – Choice unnecessary at this time
Point to Point
Same Layer
Region to Region
Same Part
Part to Part
Same Size
Size to Size
Same Shift
Shift to Shift
Same Day
Day to Day
Complete contrast. Good leverage.
Optical & SEM analysis shows crystallization
resides at the glass interface in the form of
micro fractures of the glass surface.
Layer to Layer
Same Region
Document By: Dave Johnson and Chris Rosko Revision
Date: 10/07/10

8. Page 8
Inner Oven Door Glass
Project Summary Report
Statistical Engineering
Operations Search
Conclusion: X Prime resides in the Screen Print Operation because it is the first operation that
exhibits both BOB / WOW conditions.
Input
Robax Glass Blank
Finished Pane
WOW condition not
present
Cut Glass
Rationale
WOW condition not
present
WOW condition not
present
Both BOB / WOW
conditions exist.
Contrast present.
Glass Pane
Prepared Glass / Frit
Coated Glass Pane
Cured Glass Pane
Grind / Edge
Wash
Screen Print
Oven Bake
Final Inspection / Pack
Operation
Glass Pane
Output
Finished
Pane
Prepared
Glass Pane
Coated Glass
Pane
Finished Part
Cured Glass
Pane
Document By: Dave Johnson and Chris Rosko Revision
Date: 10/07/10
Both BOB / WOW
conditions exist.
Contrast present
Both BOB / WOW
conditions exist.
Contrast present

9. Page 9
Inner Oven Door Glass
Project Summary Report
Statistical Engineering
Document By: Dave Johnson and Chris Rosko Revision
Date: 10/07/10
Contrast B is the X Prime Candidate
Contrast A is the Secondary X Candidate

10. Page 10
Inner Oven Door Glass
Project Summary Report
Statistical Engineering
Document By: Dave Johnson and Chris Rosko Revision
Date: 10/07/10
Contrast B is the Prime X!
Contrast A is the significant Secondary X

11. Page 11
Inner Oven Door Glass
Project Summary Report
Statistical Engineering
Document By: Dave Johnson and Chris Rosko Revision
Date: 10/07/10

12. Page 12
Inner Oven Door Glass
Project Summary Report
Statistical Engineering
Document By: Dave Johnson and Chris Rosko Revision
Date: 10/07/10
10 Heavy Crystallization
Moderate Crystallization
Light Crystallization
No Crystallization
7
4
1
Y Response = Crystallization as viewed from ‘A’ Side of part
Q1Q4
Q3
Q2
Sensory Scoring Legend
Observations were recorded by quadrant per the above scoring
legend to understand if the defect was random or non random in
2D occurrence.

13. Page 13
Inner Oven Door Glass
Project Summary Report
Statistical Engineering
Document By: Dave Johnson and Chris Rosko Revision
Date: 10/07/10
Frit
Micro Fracture /
Crystallization
Defect
Scanning Electronic Microscope – View @2300X
Glass Surface
Fracture
* When viewed as
shown above, the
fracture has a prism
effect with light, and
has thus been called
crystallization.
Severe Crystallization
B Side Surface View
Hand Painted Frit Sample
Crystallization
Edge Profile View
Crystallization Defect
Glass Surface