You are on page 1of 21

EVALUATION OF POT METAL SAMPLERS FOR RELIABLE ANALYSIS OF

ALUMINUM AND IRON IN MOLTEN ZINC

Larry F. Crawford
International Steel Group - Research
Bethlehem, PA

C. Ramadeva Shastry
International Steel Group - Research
Bethlehem, PA
2

EVALUATION OF POT METAL SAMPLERS FOR RELIABLE ANALYSIS OF


ALUMINUM AND IRON IN MOLTEN ZINC

Larry F. Crawford and C. Ramadeva Shastry


International Steel Group - Research
Bethlehem, PA

Abstract

An evaluation of four commonly used zinc pot metal samplers for the ability to take
reproducible, high quality samples for the determination of Al and Fe in zinc is reported in this
paper. The samplers evaluated were a poured vertical mold, a poured horizontal mold, a vacuum
assisted mold, and an immersion sand mold. The samples taken by each sampler were tested for
sampler equivalency and between-sample and within-sample homogeneity by optical emission
spectrometry. Metallographic examination was conducted to correlate sample heterogeneity with
the amount and distribution of dross entrapped by the different sampler types. The results
indicate that dross contributes to increased standard deviations in aluminum and iron contents in
the galvanizing melt. In the galvannealing melt, dross affects mainly the iron averages and
standard deviations. In summary, the results show that all sampler types other than the poured
vertical mold provide repeatable aluminum analysis on the initial sample surface. However, if
multiple analyses are desired on the same sample by successive resurfacing, the horizontal mold
sampler or the vacuum assisted sampler is the best option for repeatable, consistent % aluminum
measurement.
3

EVALUATION OF POT METAL SAMPLERS FOR RELIABLE ANALYSIS OF


ALUMINUM AND IRON IN MOLTEN ZINC

Larry F. Crawford and C. Ramadeva Shastry

Introduction

Columbus Coatings, Inc. (CCI) is the newest continuous hot dip galvanizing line owned by
International Steel Group (ISG). The line produces high quality coated sheet for the automotive
market. In order to ensure consistent product quality, melt aluminum is closely monitored and
controlled. A well-controlled pot metal % Al is critical to producing high quality coated sheet
and is used by operations to determine which ingots to add to maintain acceptable % aluminum
levels. For accurate control of pot chemistry, pot metal samples must show good repeatability
for aluminum. The sampler used to obtain pot metal samples must take representative samples
that yield repeatable analyses by Optical Emission Spectrometry (OES), the technique commonly
used for melt analysis in galvanizing lines such as CCI.

Currently, there are four major commonly used pot metal samplers that yield samples suitable for
the OES. These are: a poured horizontal mold sampler, a poured vertical mold sampler, an
immersion sand mold sampler and a vacuum assisted mold sampler. Figure 1 shows photographs
of the four samplers. Figure 2 shows the corresponding pot metal samples generated from the
samplers before they were trimmed and surfaced for OES analysis. Only one of these sample
types has been investigated in the past for homogeneity. Gagne et. al. [1] conducted a round-
robin study of the poured vertical disc sampler. These authors noted significant differences in the
analysis results between laboratories, techniques, and even between samples in the same test.
They attributed these differences to a nonuniform distribution of Fe-Al-Zn intermetallics within
the sample. They also showed that the accuracy of the analysis could be improved by removing
the intermetallics from the melt through filtration. They did not investigate other sampler types
for homogeneity.

From a practical standpoint and for the ease of operation, there is a need for a simple and direct
sampling device that provides samples for a highly repeatable aluminum analysis. A study was
therefore initiated at CCI on the four commonly used pot samplers to determine which type
provides a representative and repeatable sample for analysis of aluminum. The samplers were
tested for between-sample and within-sample homogeneity as well as sampler equivalency.

Experimental Work and Results

The galvanized samples were taken on January 23, 2003 and the galvanneal samples were taken
on February 25, 2003. Two separate zinc pots are used for the two melts at CCI. Experiments
were designed to determine sampler equivalency, between-sample homogeneity, and within-
sample homogeneity. Each experiment is discussed in detail below. In all cases, the samples
were obtained at a depth of 12” – 18” below the melt surface. For the poured mold samples, the
4

melt was taken from the pot with a long-handled ladle. The immersion sand mold samples and
the vacuum assisted samples were obtained by immersing the appropriate sampler in the melt.
The ladle and the other samplers were preheated above the melt surface before immersion.
Samples were taken in the front end of the pot at a spot facing the strip as it exits from the melt
to provide a well-mixed, homogeneous area for sampling.

The samples were analyzed on the OES at CCI. Control samples (NF33 for galvanneal and
BS139 for galvanized) were analyzed periodically to confirm that the OES did not drift
significantly during an experiment. These standards are produced in-house by ISG in compliance
with ISO Guide 34 on reference material production to help ensure that high quality reference
materials are produced and used [2]. All samples were prepared for analysis on the manual lathe
at CCI. The average values of aluminum and iron reported and the corresponding standard
deviations are based on four burns on each sample.

Sampler Equivalency - The experiment was designed to test the equivalency of sampler types.
Each sampler type was used to obtain pot metal samples sequentially, as close in time as
possible. Samples were taken sequentially twice without interruption between the sequences (Set
1 and Set 2). Samples were prepared by removing the top surface only and running the samples
on the OES consecutively. Results are shown in Table 1. The results for aluminum are also
presented graphically in Figure 3. Figure 3 includes the 95% confidence interval for the
aluminum averages. Together, the data in Table 1 and Figure 3 show each sampler to be
equivalent in taking representative pot metal samples and not subject to contamination. The
repeatability of analysis is excellent for % aluminum in both galvanize and galvanneal melts as
shown in Figure 3. The repeatability for iron is excellent in the galvanize melt, but poor in the
galvanneal melt due to the occurrence of zinc-iron bottom dross particles (Table 1). This topic is
discussed in greater detail in the section below.

Between-Sample Homogeneity - This experiment is designed to confirm that multiple samples


taken at the same time with each sampler type were homogenous, i.e. no significant difference
exists between the samples taken with same sampler type at the same time. Seven to nine pot
metal samples were taken from the galvanizing and galvannealing pots using each sampler type.
Samples were prepared for OES by removing the top surface only and analyzing the samples
from each sampler type consecutively. The results are given in Table 2. Taking <2% to be
acceptable, the results show that the relative standard deviation of the % aluminum average is
acceptable for the horizontal mold, the vacuum assisted mold and the immersion sand mold
samplers. The relative standard deviation, which is based on pooled one sigma (RSDp), is less
than 1.8% for these samplers. The vertical mold sampler standard deviation, however, is not
acceptable, with an RSDp of 4.4% for the galvanize samples and 4.2% for the galvanneal
samples. In addition, the averages for each sampler type did not show any consistent trend that
would indicate instrument drift during the experiments. Considering that the different sampler
types were not analyzed on the same day, they cannot be used for direct comparison of the
averages due to instrument drift. Direct average comparison was addressed in the Sampler
Equivalency section.
5

Between-sample homogeneity in the galvanizing melt is depicted for the poured vertical mold
and the vacuum assisted mold samplers in Figure 4. Notice in this figure that the % aluminum
averages are plotted on the primary y-axis on the left and the % iron averages and standard
deviations are plotted on plotted the secondary y-axis on the right. Notice also that the scale on
secondary y- axis is different from the scale on the primary. The plot for the vacuum assisted
mold sampler shows excellent between-sample homogeneity by displaying nearly constant
average aluminum and low standard deviations. The poured vertical mold sampler, on the other
hand, shows a greater variability in average % aluminum values and standard deviations. This
variability is attributed to the presence of aluminum and iron rich top dross particles in the OES
analysis area. The variability in OES analysis due to dross was established first in separate tests
where locations of dross particles were identified on the sample face before the OES burns. An
example of top dross particles encountered in the vertical mold sampler is shown in Figure 5.
The poured horizontal mold sampler and the immersion sand mold sampler show between-
sample homogeneity similar to the vacuum assisted mold sampler.

For both sampler types shown in Figure 4, the % iron averages and the corresponding standard
deviations trend together, i.e., when one is up so is the other. This trend is also roughly similar to
that of % aluminum. The % iron variability for the vacuum assisted mold sampler is less than it
is for the poured vertical mold sampler. For aluminum, part of the % iron variability is attributed
to the presence of top dross particles in the OES analysis area. Additional factors may also
contribute to the variability, in particular for the vertical mold sampler. These include poor
sample quality due to porosity and limited width area available for the OES burns in the vertical
mold sampler. The reason for top dross particles to be more prevalent in the poured vertical
sampler has not been investigated. It is believed related to the difference in the pattern of
solidification in the vertical mold as compared to the other samplers.

Plots of between-sample homogeneity in the galvanneal melt are shown in Figure 6. Here too the
poured vertical mold sampler shows greater variability in % aluminum averages and standard
deviations as compared the other samplers (also see Table 2). Results for the horizontal mold
sampler shown in Figure 6 are very similar to the results for the vacuum assisted mold and the
immersion sand mold samplers as well. For any of these non-vertical mold sampler types,
between sample variability in the galvanneal melt may be rated as excellent for aluminum.
However, all four samplers show much higher variability in iron than aluminum. The variability
in % Fe correlates directly with the variability in the corresponding % Fe standard deviations.
The reason for the high iron variability is attributed to the presence of bottom dross particles in
the galvannealing melt. An example of bottom dross in the vertical mold sampler is shown in
Figure 7. The bottom dross particles contain, on average, about 7.6% Fe and only about 2.2%
Al, the balance being zinc. As a result, they contribute more significantly to the variability of %
iron but leave % aluminum relatively unaffected. Because the type and amount of dross particles
are different in the two melts, the % Fe variability is different in the galvannealing melt as
compared to the galvanizing.

Within-Sample Homogeneity - This experiment was designed to confirm that heterogeneity


does not exist within a sample. There are two potential homogeneity problems that may exist;
these are around the circumference of the sample face and through the depth of the sample, i.e.
6

significant chemistry differences occur as the sample is resurfaced multiple times. To determine
if systematic chemistry differences exist around the face of a sample, three samples from each
sampler type were surfaced and burned at 6-8 known locations around the circumference of the
sample, shown in Figure 8. This was repeated at three different depths, minimal, half, and
maximum removal. The OES results did not show any systematic, significant chemistry
variation by location around the circumference of the sample face for any sampler type.
However, the results did suggest variability with depth, which is consistent with the results
discussed below, particularly with the immersion sand mold and the poured vertical mold
samplers. Galvanize results are shown in Figure 9 and galvanneal results are shown in Figure 10.

In order to further examine the depth-related heterogeneity, three samples from each sampler
were resurfaced and analyzed multiple times throughout the depth of the samples. The average
and standard deviation from each sample depth were compared to determine if significant
variations existed as a function of depth. The galvanneal pot metal samples taken by the
immersion sand mold sampler showed that the average aluminum concentration and standard
deviation increased after about three faces were removed and then decreased toward maximum
sample removal. This is displayed graphically in Figure 11, which shows plots of the average %
aluminum and pooled standard deviation of the three samples analyzed as a function of sample
depth. The vertical mold sampler showed higher than expected standard deviations for a large
number of the analysis points taken. All other sampler types did not display any significant
heterogeneity throughout the depth of the sample, as demonstrated, for example, in Figure 12 for
the horizontal mold sampler. The through-thickness heterogeneity of the immersion sand mold
sampler and the vertical mold sampler is consistent with the through-thickness heterogeneity
reported by Gagne et. al [1]. The similarity between these two sampler types in this regard may
be related to the similarity in mold filling characteristics.

Conclusions
1. A methodology was established to evaluate zinc pot metal samplers on the basis of
sampler equivalency, between-sample homogeneity and within-sample homogeneity. The
methodology permits the selection of samplers for repeatable analysis of galvanize and
galvanneal melts for aluminum by optical emission spectroscopy.
2. Based on analysis on the initial face, the four commonly used samplers evaluated in this
study are equivalent in taking representative pot metal samples and not subject to
contamination. The four samplers are the poured vertical mold sampler, the poured
horizontal mold sampler, the immersion sand mold sampler and the vacuum assisted
mold sampler.
3. Relative % aluminum standard deviation based on pooled one sigma (RSDp) was used to
compare between-sample homogeneity. Based on this metric, between-sample
homogeneity is considered unacceptable for the poured vertical mold sampler in both the
galvanize and galvanneal melts. Between sample homogeneity for the other three
sampler types is considered excellent.
4. In the galvanize melt, the high relative standard deviation in aluminum in the vertical
mold samples is due to the presence of dross particles. The % iron averages and the
7

corresponding standard deviations in this case follow a trend roughly similar to that of
aluminum.
5. In the galvanneal melt, all four samplers show much higher variability in iron compared
to aluminum. The variability in iron correlates directly with the variability in the
corresponding % iron standard deviations. The latter is attributed to the presence of
bottom dross particles in the samples. Replicate samples and a greater number of burns
per sample are necessary to obtain more repeatable % iron averages in galvanneal melts.
6. There is no systematic, significant chemistry variation by location around the
circumference on the analysis surface for any sampler type. This around-the-
circumference homogeneity prevails through the sample thickness for all samplers.
7. The vertical mold sampler and the immersion sand mold sampler display significant
heterogeneity through the sample depth. For the immersion sand mold sampler, the
average Al concentration and standard deviation are much higher in the middle of the
sample than on locations near the external surfaces. The vertical mold sampler shows
high aluminum standard deviations for a large number of analyses at different depths.
8. All samplers other than the vertical mold sampler provide repeatable analysis on the
initial surface. If multiple analyses are desired on the same sample after successive
resurfacing, the horizontal mold sampler or the vacuum assisted mold sampler is most
likely to provide the most repeatable % aluminum results.

References

1. M. Gagne, H. Guttman, J. L’Ecuyer, G. G. Brummitt, G. L. Adams and D. Kleimeyer, “The


Analysis and Control of Aluminum in Galvanizing Baths”, Galvanizers Association Meeting,
October 22-25, 1990, Niagara Falls, New York, pp. 126-145.

2. Anon: ISO Guide 34:2000 – General requirements for the Competence of Reference Materials
Procedures, ISO, International Organization for Standardization, Geneva, Switzerland.

Acknowledgements

We gratefully acknowledge the assistance of Scott Wilson, Tim O’Toole, Gary Bennett, Mark
DeBruin and Audie Seevers, all of CCI, during the experimental phase of this work. Assistance
in OES by Beth Crawford and Russ Hartranft, and expert metallographic support by Phil Fekula
are appreciated. Sam Lawrence is acknowledged for electron probe microanalysis of dross
particles. We appreciate the support and encouragement in this work by Wayne Bontempo of
CCI and Theresa Simpson at ISG Research. We are grateful to the management at ISG-
Columbus Coatings Inc. and ISG Research for permission to prepare and publish this paper.
8

Table 1. Results of Sampler Equivalency Experiments

Galvanize Melt Galvanneal Melt


Sampler Type % Al % Fe % Al % Fe
Set 1 Set 2 Set 1 Set 2 Set 1 Set 2 Set 1 Set 2
Poured Vertical 0.202 0.204 0.022 0.023 0.116 0.118 0.063 0.069
Poured Horizontal 0.209 0.207 0.021 0.019 0.122 0.122 0.052 0.085
Immersion 0.202 0.205 0.018 0.020 0.119 0.120 0.059 0.053
Vacuum Assisted 0.202 0.204 0.019 0.018 0.119 0.121 0.077 0.082

Table 2. Between-Sample Homogeneity for Aluminum

Galvanize Melt Galvanneal Melt


Sampler Type No. of Avg. RSD No. of Avg. RSD
Samples %Al % Samples %Al %
Poured Vertical 9 0.209 4.4 9 0.118 4.2
Poured Horizontal 7 0.226 1.3 8 0.118 1.4
Immersion 9 0.202 1.8 9 0.114 1.2
Vacuum Assisted 9 0.202 1.3 9 0.120 1.2

Notes:
1. Direct comparison of aluminum between samplers is not valid because analyses were done on
different days.

2. RSD % = [Average (pooled) standard deviation/average aluminum] x 100


9

Poured Vertical Mold

4 cm

Poured Horizontal
Mold

4 cm

Immersion Sand Mold

4 cm

Vacuum Assisted
Mold

4 cm

Figure 1. Photographs of the four commonly used pot metal samplers.


10

Poured Vertical

2 cm

Poured Horizontal

2 cm

Immersion

2 cm

Vacuum Assisted

2 cm

Figure 2. Pot metal samples obtained from commonly used samplers.


11

Sampler Equivalency
Galvanize Pot Metal

0.26
0.24
Average Al (wt.%)

0.22
0.2
Set 1
0.18
Set 2
0.16
0.14
0.12
0.1
Vacuum Immersion Horizontal Vertical
Assisted
Sampler Type

Sampler Equivalency
Galvanneal Pot Metal

0.15
0.14
Average Al (wt%)

0.13
0.12 Set 1
0.11 Set 2
0.1
0.09
0.08
Vacuum Immersion Horizontal Vertical
Assisted
Sampler Type

Figure 3. Results of aluminum measurements on sequential samples taken with four different
12

samplers.

Vertical - Galvanize Melt


Al
Averag
0.3 0.04 e
Al Std
0.25 Dev
0.03
0.2
Al (wt.%)

Fe (wt%)
Fe
0.15 0.02
Averag
0.1 e
0.01 Fe Std
0.05
Dev
0 0
0 2 4 6 8 10
Sample ID

Vacuum Assisted - Galvanize Melt

0.25 0.035
0.03 Al
0.2 Averag
Average Al (wt.%)

0.025 e
Al Std
Fe (wt%)

0.15 0.02 Dev

0.1 0.015 Fe
Averag
0.01 e
0.05 Fe Std
0.005 Dev
0 0
0 2 4 6 8 10
Sample ID

Figure 4. Galvanize melt between-sample homogeneity for poured vertical mold


and vacuum assisted mold samplers
13

Figure 5. Typical top dross particles observed in vertical mold galvanize melt samples.
14

Vertical Mold - Galvanneal Melt

Concentration (wt% m/m)


0.14
0.12
0.1 Al Avg
Al Std Dev
0.08
Fe Avg
0.06
Fe Std Dev
0.04
0.02
0
0 2 4 6 8 10
Sample ID

Horizontal Mold - Galvanneal Melt

0.14

0.12
Concentration (wt% m/m)

0.1
Al Avg
0.08 Fe Avg
0.06 Al Std Dev
Fe Std Dev
0.04

0.02

0
0 2 4 6 8 10
Sample ID

Figure 6. Galvanneal melt between-sample homogeneity for poured vertical mold


and poured horizontal mold samplers.
15

Figure 7. Typical bottom dross particles observed in vertical mold galvanneal melt samples.

12:00

10:30 1:30

9:00 3:00

7:30
4:30

6:00

Figure 8. OES burn locations for evaluation of within sample homogeneity.


16

Vertical Mold Sample 10V

0.3
0.28
0.26
0.24
0.22
% Al

0.2
0.18
0.16
0.14
0.12
0.1

X
1: 0-X
3: 0-X
4: 0-X
6: -X
7: 0-X
9: -X
:3 X
3: 0-I
4: -I
6: 0-I
7: -I
9: 0-I
:3 I
1: 0-I

I
10 0-
0-
00

00

0-
10 00-
30

30
3

3
0
:0

:0
3
0

0
12

12
Location around Sample Face

Horizontal Mold Sample 29

0.3
0.28
0.26
0.24
0.22
% Al

0.2
0.18
0.16
0.14
0.12
0.1
3:30-X

X
4:00-X
6:30-X
7:00-X
93 X
10:000-X
:3 -X
4:00-I
6:30-I
7:00-I
3 I
109:000-I
12 30-I

3:30-M

12 -M
3:30-I

: I

4:00-M
6:30-M
7:00-M
93 M
10:000-M
:3 -M
: -
1:00-

1: 0-

0-
1:00-

0
:0
:
12

Location around Sample Face

Figure 9. Within-sample homogeneity of galvanize melt samples. I = initial face,


M = midway through thickness and X = maximum removal. The 95% probability interval based
on 2 x standard deviation is shown by the dashed horizontal lines.
17

Immersion Sand Mold Sample 12E

0.3
0.28
0.26
0.24
0.22
% Al

0.2
0.18
0.16
0.14
0.12
0.1
0 M
0 M
0 M
10 00 M
:0 -M
12 M
0 I
4: 0-I
0 I
10 00 I
:0 -I
12 0-I

0 X
0 X
6: 0-X
10 00 X
:0 -X
X
2: 0-

6: 0-
8: 0-

0-
2: 0-
4: 0-

8: 0-
2: 0-
4: 0-
6: 0-
8: 0-

0-
:0

:0

0
:0
12

Location around Sample Face

Vacum Assisted Mold Sample 10M

0.3
0.28
0.26
0.24
0.22
% Al

0.2
0.18
0.16
0.14
0.12
0.1
4: 0:-M

12 -M
4: 0:-I

12 0-I

M
10:00-M
:0 -M
8:00-I
10 00 I
:0 -I

4: 0:-X

X
M
6:00-I

8 00 X
10:00-X
:0 -X
6:00-X
0 -
2::00

0 -

0-
0 -
6:00-

2::00
2::00

8 0

0
0
12

Location around Sample Face

Figure 9 (contd.). Within-sample homogeneity of galvanize melt samples. I = initial face,


M = midway through thickness and X = maximum removal. The 95% probability interval based
on 2 x standard deviation is shown by the dashed horizontal lines.
18

Vertical Mold Sample 12V

0.19
0.17
0.15
0.13
% Al

0.11
0.09
0.07
0.05
3:30-M
4:00-M
6:30-M
7:00-M
9 30 M
10:00-M
:3 -M
12 - M
3:30-I
4:00-I
6:30-I
7:00-I
9 30 I
10:00-I
12 0-II

3:30-X
4:00-X
6:30-X
7:00-X
9 30 X
10:00-X
:3 -X
X
:3 -
1:00-

1: 0-

0-
1:00-

0
:0
:
12

Location around Sample Face

Horizontal Mold Sample 6H

0.19
0.17
0.15
0.13
% Al

0.11
0.09
0.07
0.05
1: 0-X
3: 0-X
4: 0-X
6: 0-X
7: 0-X
9 -X
:3 X
-X
3: -I
4: 0-I
6: 0-I
7: 0-I
9: -I
:3 I
1: -I

I
10 00-
0-
30

30
0

10 :00-
30

0:
0
3
0
:0

:0
3
0
3
0
12

12

Location around Sample Face

Figure 10. Within-sample homogeneity of galvanneal melt samples. I = initial face,


M = midway through thickness and X = maximum removal. The 95% probability interval based
on 2 x standard deviation is shown by the dashed horizontal lines.
19

Immersion Sand Mold Sample 4E

0.26
0.24
0.22
0.2
0.18
% Al

0.16
0.14
0.12
0.1
0.08
0.06
0 M

12 M
0 I

12 0-I

0 M
0 M
10 00 M
:0 -M
0 I
0 I
10 00 I
:0 -I

0 X

X
0 X
0 X
10 00 X
:0 -X
2:00-
4: 0-
6: 0-
8: 0-

2: 0-

0-
4: 0-
6: 0-
8: 0-
2: 0-

0-
4: 0-
6: 0-
8: 0-

:0
:0
:
12

Location around Sample Face

Vacum Assisted Mold Sample 2M

0.19
0.17
0.15
0.13
% Al

0.11
0.09
0.07
0.05
4: 0:-M

12 -M
4: 0:-I

12 0-I

6:00-M
8:00-M
10 00 M
:0 -M
6:00-I
8:00-I
10 00 I
:0 -I

4: 0:-X

X
6:00-X
8 00 X
10:00-X
:0 -X
0 -
2::00

0 -

0-
0 -

2::00
2::00

0
12

Location around Sample Face

Figure 10. (Contd.). Within-sample homogeneity of galvanneal melt samples. I = initial face,
M = midway through thickness and X = maximum removal. The 95% probability interval based
on 2 x standard deviation is shown by the dashed horizontal lines.
Pooled Standard Deviation Average Al (wt %)
(wt %) In
iti

0.1
0.11
0.12
0.13
0.14
0.15
0.16

In Se al F
iti co ace

0
0.005
0.01
0.015
0.02
0.025
0.03
0.035
0.04
0.045
0.05
Se al F nd
co ac Th Fac
nd e ird e
Th Fac Fo Fa
ir e ur ce
th
Fo d F
ur ace F
th Fi ace
fth
Fi Fac Si Fac
fth e x e
Si Fac Se th F
x ve ac
20

nt e
Se th F e h
ve ac F

function of depth below the surface. Galvanneal Melt.


nt Ei ac
h e gh e
t
Depth
Ei Fac N Fac
gh e in
t th e

Depth
Immersion Sampler
Immersion Sampler

N Fa
in ce Te Fac
th
Te Fa El nth e
c ev F a
en ce
El nth e
ev Fa th
en ce F
th M ace
ax
M Fac Fa
ax e ce
Fa
ce

Figure 11. Average % Al and pooled standard deviation for immersion sand mold sampler as a
21

Horizontal Mold Sampler

Average Al (wt %) 0.16


0.15
0.14
0.13
0.12
0.11
0.1
Initial Face Second Third Face Fourth Face Fifth Face
Face

Depth

Horizontal Mold Sampler


Pooled Standard Deviation

0.05
0.045
0.04
0.035
(wt %)

0.03
0.025
0.02
0.015
0.01
0.005
0
Initial Face Second Third Face Fourth Fifth Face
Face Face

Depth

Figure 12. Average % Al and pooled standard deviation for poured horizontal mold sampler as a
function of depth below the surface. Galvanneal melt.

You might also like