SEPTEMBER 1976
LEVELS OF TRACE ELEMENTS
IN THE AMBIENT AIR AT
SELECTED LOCATIONS IN
THE NORTHERN GREAT PLAINS
r.
ENVIRONMENTAL PROTECTION AGENCY it KB
ROCKY MOUNTAIN-PRAIRIE REGION I ^10^ f
REGION VIII % -^AlAZ t
^4 PROl^
-------
n
lie,
pf/M
LEVELS OF TRACE ELEMENTS
IN THE AMBIENT AIR AT
SELECTED LOCATIONS IN THE
NORTHERN GREAT PLAINS
SOC-L
£PA
®°gion
999
Der)v,
8 Ubn
18th St
or, CO
dry
Suite
8Q202-
500
2466
EPA Contract No. 68-02-1383, Task 7
Prepared for:
EPA Project Officer: Terry L. Thoem
Environmental Protection Agency
1860 Lincoln Street
Denver, Colorado 80202
Prepared by:
F. G. Mesich, Radian Corporation
H. L. Taylor, Acculabs Research
TS-4a
-------
ABSTRACT
Trace element levels were determined by spark-source
mass spectometry for five (5) locations in the Northern Great
Plains area. Samples were collected using low volume membrane
samplers and analyzed for some fifty (50) elements.
Quarterly composite samples were analyzed. Comparison
of composite results was made with individual filter results
during one quarter.
A discussion of recommended methodology for ambient
trace element sampling is presented.
i
-------
TABLE OF CONTENTS
Page
1.0 INTRODUCTION AND PROGRAM RATIONALE ........ 1
2.0 EXPERIMENTAL METHODOLOGY 6
2.1 Filter Collection 6
2.2 Sample Preparation 6
2.3 Analysis 7
3.0 EXPERIMENTAL RESULTS 9
3.1 Major Constituents in Composites 9
3.2 Minor or Trace Elements in Composites ... 9
4.0 DISCUSSION OF RESULTS 30
5.0 REFERENCES 35
APPENDIX
ii
-------
LIST OF TABLES
Page
I Mean Values of Total Air Volume For Each
Composited Sample 10
II Fourth Quarter Individual Total Air Volume. . . 11
III Major Species in Newcastle Quarterly
Composites ^
IV Major Species in Glendive Quarterly
Composites 15
V Major Species in Garrison Quarterly
Composites 16
VI Major Species in Ft. Peck Quarterly
Composites 17
VII Major Species in Belle Fourche Quarterly
Composites 18
VIII Detailed Analyses For Newcastle Composites. . . 19
IX Detailed Analyses For Glendive Composites ... 20
X Detailed Analyses For Garrison Composites ... 21
XI Detailed Analyses For Ft. Peck Composites ... 22
XII Detailed Analyses For Belle Fourdie
Composites 23
iii
-------
LIST OF TABLES (Cont'd)
Page
XIII Individual Fourth Quarter Filters From
Newcastle 25
XIV Individual Fourth Quarter Filters "From
Glendive.............. 26
XV Individual Fourth Quarter Filters From
Garrison 27
XVI Individual Fourth Quarter Filters From
Belle Fourehe 28
XVII Individual Fourth Quarter Filters From
Ft. Peck 29
XVIII Comparison of Fourth Quarter Composites
With Range and Average of Fourth Quarter
Individual Filter Trace Element Analyses.... 33
XIX Mean Grain Loadings From High-Volume
Samplers Near the Membrane Sampler Sites.... 34
iv
-------
1.0 INTRODUCTION AND PROGRAM RATIONALE
The development of the energy industry In the Northern
Great Plains area will result in an increase in the atmospheric
level of particulates. Of particular interest is the possible
increase in the levels of certain potentially hazardous trace
elements as a result of coal conversion facilities. However,
the data base concerning existing concentrations of trace elements
in the ambient air in undeveloped areas is virtually nonexistent.
Therefore, this program was designed as a supplement to an ambient
air monitoring project performed by PEDCo Environmental under
EPA Region VIII sponsorship. The PEDCo program was to collect
ambient air quality data including sulfur oxides, nitrogen oxides
and particulates at representative locations in the Northern
Great Plains area. The filters collected by PEDCo were to be
shipped to Radian's analytical subcontractor, Accu-Labs Research,
Inc. for trace element analysis by spark source mass spectrometry
(SSMS). The samples were collected at the following locations
shown in Figure 1-1: Ft. Peck and Glendive, Montana; Garrison,
North Dakota; Belle Fourche, South Dakota; and Newcastle, Wyoming.
As initially planned, samples were to be collected at
twenty-six locations by Hi-Volume (Hi-Vol) sampling techniques
with the 8 x 10" filters submitted for trace element analysis
by SSMS. Samples were to be taken at each site at six-day inter-
vals for one year. These samples were to be composited and
analyzed for their trace element content on a quarterly basis
to determine long-term variation in composition.
During the planning for this program, it was recommended
by Radian that cellulosic filter material be used for the-collection
of trace element samples. This would insure a low ash and low
-------
Figure 1-1 Saspling Locations
-2-
-------
background residual of trace elements in the filter media. Also,
this type of filter is highly uniform which permits suitable
blank corrections to be performed. During early meetings with
EPA officials, it was established that much of the sampling had
already been performed utilizing fiberglass Hi-Vol filters, and
that it would not be possible to change the Hi-Vol media for this
project. To determine the feasibility of using the fiberglass
media, a study was undertaken to establish the level of trace
element impurities in selected fiberglass blank samples that had
been used in the program as well as some typically loaded samples.
Also the statistical variation of these ii^purities was established
to determine how accurately blank corrections could be performed.
The results of this study are outlined in Reports 1 and 2 attached
as Appendix I to this report. In summary, these reports indicated
that data obtained from these samples would be of little use
because of the poor blank characteristics of the fiberglass filter
pads used to obtain the samples.
After consultation with the project officer, it was
decided to modify the scope of work to switch from the fiber-
glass Hi-Vol filters to the membrane back-up filters taken at
each site. It was agreed that this would necessarily compromise
the data to some degree but would allow the data that was
acquired to be meaningful. The compromises were as follows:
(1) the air volumes of the membrane filters were lower than the
Hi-Vols, hence, less sample was collected; (2> no weights were
obtained by the field contractor on the actual loading of the
filters, thus, even though blank levels are low, an unfavorable
ratio of particulate to.blank can still exist; C3} due Co Che
limited amount of sample, it would not be possible to obtain
data on the element mercury, because it is determined by an
alternate acomic fluorescence procedure.
-3
-------
To perform Che analysis on the membrane filters, it
was absolutely necessary to have blank membrane filters from
the same lot as those used to obtain the samples. These were
requested from the field contractor and received. After
commencement of the analysis, it was soon realized that not
only did the blank membranes not represent those used in collec-
tion of the samples, but that, definitely, several types of.
membranes were employed in collection of samples.
This was evidenced by attempting dissolution of the
filter media. Some filters dissolved readily in acetone while
others left a web of insoluble supporting material. On ashing,
the materials gave ashes of varying colors, again raising
suspicions that the filter substrates were of different
materials. However, these problems not withstanding, the
analysis of the membrane filters provides a usable data base
for a number of elements. These are documented in the results
section.
The analysis of filter-collected atmospheric particulate
matter can be accomplished with relative ease by following a few
basic procedures. It is recommended that in any ambient trace
element sampling and analysis program the following key items
should be considered:
• selection of a low ash/low trace element
background filter material,
• careful equilibration and weighing of
filters before and after particulate
collection,
-4-
-------
collection of a sufficient amount of
particulates for analysis, i.e. a 24-hr
Hi-Vol sanrple or a 6-7 day membrane
filter sample, and
collection of meteorological data at
the sample site to permit data analysis
-------
2.0
EXPERIMENTAL METHODOLOGY
2.1 Filter Collection
The samples were collected at the following sites:
(1) Newcastle, Wyoming;
(2) Glendive, Montana;
(3) Garrison, North Dakota;
(4) Ft. Peck, Montana; and
(5) Belle Fourche, South Dakota,
Figure 1-1 is a map showing the collection points. The sitea
were selected to give a representative cross-section of back-
ground air quality conditions in the Northern Great Plains area
2-2 Sample Preparation
The circular, 12.5 cm diameter membrane filters were
prepared for spark source mass spectrographic analysis in the
following manner.
(1) The sample was divided into quadrants,
in such a fashion that the ink-stamped
identification number was located in
only one quadrant. This quadrant was
not used for any of the analyses. When
analyzing individual filters, one-half
of the filter was carried through the
remainder of the preparation steps as
listed below. When compositing was
performed, one-fourth of each sample
to be composited was combined for sub-
sequent preparation.
-6-
-------
(2) 0.100 gram of ultra-high purity graphite
was weighed into an acid-washed (Vycor)
crucible which had been tared to constant
weight by successive heatings in a muffle
furnace. The portion or portions of
filter(s) to be analyzed were then placed
into the crucible, which was heated in
a muffle furnace at 450°C for ashing.
(3) After ashing, 10 micrograms of indium
were added to the graphite-ash mixture
to act as an internal standard. This
mixture was then thoroughly homogenized.
(4) The mixture was compressed into elec-
trode pins, which were then ready for
analysis.
2.3 Analysis
The mechanism and principles of operation of the AEI
MS 702R high resolution spark source mass spectrometer have been
described in many publications CI. 2, 3) and consequently will
not be included in this report.
A pulsed radio frequency potential is employed to
vaporize and ionize the sample. The positive ions which are
formed by this process are separated by electrostatic and
magnetic fields and impinged on a photographic plate. This
permits not only high sensitivity to be obtained (by the inte-
gration principle of ion beams) but also high resolution.
The results appear as a series of lines with differing
densities on the photoplate. The lines are measured optically
with the darker lines indicating higher concentrations.
-7-
-------
Selected isotropic spectral lines are recorded. These are then
compared to values obtained for reference standards, and
concentrations are calculated.
In this particular case, since the basic matrix of the
samples is graphite, synthetic standards made in graphite are
used for reference materials.
Figure 2-1 below shows an actual photoplate from an
SSMS scan. The lines of different densities represent the
various mass-to-charge ratios measured. Each line indicates
a particular mass-to-charge. By selection of the appropriate
isotopes,the concentrations of each element can be measured.
I
11
¦ i
i I '
» I |i I
• III*
i 910
m
Figure 2-1 SSMS Photoplate
-8-
-------
3.0 EXPERIMENTAL RESULTS
3.1 Maior Constituents in Composites
Table I lists the mean value of the total volume of
air passed through each of the composited samples from each
sampling site. Table II lists the total air volume passed
through each of the individual fourth-quarter samples from
each site.
Tables III through VII list the major constituents in
the quarterly composite samples from each of the sampling sites
in micrograms per cubic meter. These values may be compared to
the total particulate loading recorded using a Hi-Vol with a
fiberglass filter at the same site.
The samples were generally lightly loaded. These
elements represent the major constituents. The major elements
varied slightly from location to location probably reflecting
the mineralogy of the background dust.
3.2 Minor or Trace Elements in Composites
All samples were subjected to a semi-quantitative scan
for over 50 elements. In a typical case, some 40+ were detected.
The actual detection limits vary slightly from day to day and
from sample to sample depending upon internal standard adjust-
ments, sample size differences, isotope abundance and spectral
background. Therefore, no uniform detection llosit can be set
for each element. The detection limit for elements in each
sample is reflected by a less than (<) value.
Tables VIII- XII give a tabulation of the elements studied.
Because of the very low levels of most elements, an exponential
format is used. For example 2E-5 is equivalent to-2 x 10
-9-
-------
TABLE I
MEAN mUEB OF TOTAL AIR VOTIJME
FOR EACH COMPOSITED SAMPLE
Site Quarter Con^oaite Mean Volume (a3)
Newcastle 1 282
Newcastle 2 263
Newcastle 3 2§l
Newcastle ^ 252
Glsndlr* 1 120
Glendiva 2 120
Glandive 3 123
Glendive ** 123
Garrison 1 82
Garrison 2 209
Garrison 3 198
Garrison 189
Ft. Peck 1 ^
Ft- Feck
Ft. Peck
Ft. Peck
Belle Fourche
2 177
3 175
tt 158
Belle Fourche 1 153
Belle Fourche 2 ish
Belle Fourche 3 ijj*
H 122
-10-
-------
TABLE II
FOURTH QUARTER INDIVIDUAL TOTAL AIR VOLUMES
Data
6-5-75
6-11-7S
6-17-75
6-23-75
6-29-75
7-5-75
7-11-75
7-17-75
7-23-75
7-29-75
8-4-75
8-10-75
8-16-75
8-22-75
8-28-75
Gl«ndive
6-5-75
6-11-75
6-17-75
6-23-75
6-29-75
7-5-75
7-11-75
7-17-75
7-23-75
7-29-75
Volume
254
265
257
209
265
269
261
261
265
228
254
246
259
254
239
108
152
130
152
147
124
113
119
H3
124
-11-
-------
TABLE II
(Continued)
Site
Garrison
Date
6-5-75
6-11-75
6-17-75
6-23-75
6-29-75
7-5-75
7-11-75
7-17-75
7-23-75
7-29-75
8-4-75
8-10-75
8-16-75
8-22-75
8-28-75
9-3-75
Belle Fourche 6-5-75
6-11-75
6-17-75
6-23-75
6-29-75
7-5-75
7-11-75
7-17-75
7-23-75
7-29-75
8-4-75
8-10-75
8-16-75
Volume (m^)
159
165
159
159
159
159
159
171
184
159
223
184
171
165
159
165
122
122
122
122
122
123
123
123
123
123
123
123
123
-12-
-------
TABLE XI
(Continued)
Site
Ft. Peck
Date
6-5-75
6-17-75
6-23-75
6-27-75
7-5-75
7-7-75
7-17-75
7-23-75
7-29-75
8-4-75
8-10-75
8-16-75
8-22-75
Volume (a3)
157
159
157
159
160
157
157
147
153
157
169
170
170
-13-
-------
TABLE III
MAJOR SPECIES IN NEWCASTLE QUARTERLY COMPOSITES
Oig/m3)
Newcastle Newcastle Newcastle Newcastle
1st Quarter 2nd Quarter 3rd Quarter 4th Quarter
Comp. Comp. Coop. Comp.
Aluminum
Calcium
Chlorine
Fluorine
Iron
Lead
1.1
0.1
0.07
0.01
0.4
0.05
0.8
0.3
0.02
0.003
0.1
0.01
0.3
0.2
0.02
0.004
0.09
0.02
0.8
0.4
0.008
<0.02
0.4
0.01
Magnesium
Phosphorus
Potassium
Silicon
Sodium
0.2
0.2
0.1
Major
0.3
0.3
0.1
0.04
Major
0.2
0.1
0.09
0.1
Major
0.2
0.2
0.06
0.3
Major
0.4
Sulfur
Titanium
0.1
1.9
0.02
1.5
0.07
1.5
0.005
0.8
Mean High Volume
Sample Loadings (ref 4)
24.9
8.3
11.3
40.3
-------
TABLE IV
MAJOR SPECIES IN GLENDIVE QUARTERLY COMPOSITES
(pg/m3)
G1endive Glendtve Glendive Glendive
1st Quarter 2nd Quarter 3rd Quarter 4th Quarter
Comp. Comp. Comp.
Aluminum
2.5
0.4
0.7
1.3
Calcium
0.8
0.3
0.3
0.9
Chlorine
0.2
0.04
0.05
0.02
Fluorine
0.4
0.05
0.1
0.03
Iron
0.08
0.003
0.03
0.01
Lead
0.03
0.002
0.002
0.002
Magnesium
0.1
0.009
0.09
0.4
Phosphorus
0.02
0.07
0.02
0.02
Potassium
0.08
0.009
0.07
0.05
Silicon
Major
Major
Major
Major
Sodium
0.1
0.3
0.03
2.0
Sulfur
0.3
0.03
0.2
0.008
Titanium
1.7
3.3
2.4
1.6
Mean High Volume
Sample Loadings (ref 4)
18.7
11.4
13.5
23.7
-------
TABLE V
MAJOR SPECIES IN GARRISON QUARTERLY COMPOSITES
(yg/m3)
Garrison
1st Quarter
Comp.
Garrison
2nd Quarter
Comp-
Garrison
3rd Quarter
Comp.
Garrison
4th Quarter
Comp.
Aluminum
Calcium
Chlorine
Fluorine
Iron
Lead
Magnesium
Phosphorus
Potassium
Silicon
Sodium
2.4
1.0
0.2
0.2
0.05
0.03
0.1
0.1
0.02
Major
.06
0.5
0.2
0.1
0.08
0.03
0.007
0.1
0.04
0.1
Major
0.4
1.0
0.2
0.01
0.01
0.02
0.02
0.2
0.1
0.09
Major
0.5
3.6
0.4
0.01
0.03
0.1
0.008
0.4
0.05
0.07
Major
2.0
Sulfur
Titanium
0.06
4.9
0.1
1.9
0.1
1.0
0.03
1.4
Mean High Volmae
Sample Loadings (ref 4)
31.5
40.0
18.6
34.1
-------
TABLE VI
MAJOR SPECIES IN FT. PECK QUARTERLY COMPOSITES
(yg/m3)
Ft. Peck Ft. Peck Ft. Peck Ft. Peck
1st Quarter 2nd Quarter 3rd Quarter 4th Quarter
Comp. Comp. Comp. Comp.
Aluminum
0.5
0.1
0.6
1.4
Calcium
0.1
0.4
0.3
0.1
Chlorine
0.3
0.05
0.07
0.008
Fluorine
0.06
0.03
0.09
0.03
Iron,
0.01
0.006
0.06
0.06
Lead
0.003
0.009
0.01
0.003
Magnesium
0.04
0.09
0.07
0.2
Phosphorus
0.03
0.07
0.2
0.05
Potassium
0.04
0.07
0.2
0.05
Silicon
Major
Major
Major
Major
Sodium
0.6
0.2
0.5
0.9
Sulfur
0.06
0.2
0.03
0.1
Titanium
1.8
1.7
1.1
1.6
Mean High Valwm
Sample Loadiag (ref 4) 18.8
7.1
13.0
6.8
-------
Aluminum
Calcium
Chlorine
Iron
Fluorine
Magnesium
Lead
Phosphorus
Potassium
Silicon
Sodium
Sulfur
Titanium
Mean High VoliMfc
Sample Loading (ref 4)
TABLE VII
MAJOR SPECIES IN BELLA FOURCHE QUARTERLY COMPOSITES
(Vig/m3)
Belle Fourche
1st Quarter
Conp.
Belle Fourche
2nd Quarter
Contp.
Belle Fourche
3rd Quarter
Comp.
Belle Fourche
4th Quarter
Comp.
2.6
1.3
0.008
0.4
0.02
0.5
0.03
0.2
1.3
2.6
0.02
0.3
0.05
0.6
0.03
0.3
0.7
0.6
0.04
0.07
0.01
0.08
0.009
0.06
1.8
0.6
<0.02
0.06
<0.02
0.2
0.02
0.04
0.2
Major
0.5
0.2
3.3
0.1
Major
0.09
0.3
0.2
0.06
Major
0.6
0.1
2.8
0.08
Major
1.3
0.06
1.2
19.1
8.9
14.9
39.2
-------
TABLE VIII
DETAILED ANALYSES FOR NEWCASTLE COMPOSITES
(yg/m3)
IfwrnwtT*
^oScter 2ad Quarter
Aaciaoay
AtmoIc
S«rii»
fcttrUlua
BlaHich
Calcium
Ctriua
OiXorln*
Chroalua
Cobalt
Coppor
Fluosiao
gftif'f
Amusini
leoa
Ljachutua
Ua
-------
TABLE XX
DETAILED ANALYSES FOR GLENDIVE COMPOSITES
(yg/m3)
IKSV §Sl£tor Sorter
SmauM Zmaatim gffiffSPtt***
fho«phoru«
3*l*alua
Silicon
lilvar
fettia
fetoaeltaa
Sulfur
»4
4i-i n-i m
aw iw «fw
feMtMBllM ?TTi 4tM»S
Aatinoor ;ri «-4 M-* <4g-5
fTj 18-3 UM «-3
ta*lw <38-3 <38-3 <«S4
taxruiuai *2*; r.g-4 48-4 ;.s-4
8lM,,th «2-s «-s «~s
C«lai.ua «•£' )g.t 48-1 «-l
Cilciua »-4 a-*
C«c4« •* , 41-2 38-2 28-2
Qtlorloa Js-2 12-2 «-3
ChcMtw ** <88-« t« ® £3 2^
_ U-4 STf 28-3 38-4
Unaluat S8-4 f* : <3e.<
JESS. <11 s:l I; 2:;
SS1" Si J«-J SM
Zirconium
-20-
-------
TABLE X
DETAILED ANALYSES FOR GARRISON COMPOSITES
(Ug/ffi3)
CmtUob C«tt1#
gitalin J®"* lf.|
T«mdtisa 21-3 <3^.5
TMtrtiua *^*1 18-4
Tttrlw «-* 9S-3
8iae »; 28-3
Zireeaiua *"
Garrison
Gtrtaam
3rd Quarter
4th Quarter
ISO
m
is-}
m*
i*-4
21-4
5*-l
<«M
18-4
<48-5
ilff
28-1
4fcl
i§-4
48-3
ut-t
»*3
2*4
91-5
JJM
48-3
18-2
«3W
5«-5
%•)
3t-5
<1*4
-------
mi
nETAlLED ANALYSES FOR FT. PECK COMPOSITES
Cyg/ffl3)
Ft. fwif Ffi» ^aflUe
<2i-5
Bhodlwa u«4
<«-J
•wehaalua <31-5
Tla
<«*-l
<«-3
«*»«- 33 ^
Sa4ftl* <1I-J **0 £S
»«U*« <»j 38-5 f|* JrJ
S3 .
J2rt in •»-* H"0
**¦*¦¦¦ JsL; a-* J*-*
Anel«oay 2S-4 »-4 2*-*
4*»aia S*i 31.3 «-3 9*-*
<53
-------
TABLE XII
DETAILED ANALYSES FOR BELLE FOURCHE COMPOSITES
(yg/m8)
lii'oSSSf
eomaMttg Com watte Cg2 11-2 <2*-2
FluortM »-* 2S-* 71-3 *»-5
eautai **"* 2s-3 <«-«
CwuiUta , U-t 71-2
ta» £3 31-4 a-4
Uaetuiuaa »"! 3«.2 «-3 2t-2
UU »* Sl-3 <*«-*
Uthiua »-4 6£.1 8X-2 21-1
MiSBMln 3£J SE-3 <31-3 JM
(ftagUMM 5g-4 4S-5 <21-4
MolytxUtnat 3g_3 3E-3 «-4
Sicfc«l **"' <32-4 CUS-4
thodlia 1 <31+1 *****
UO«®» <™, 3E-s «-3 2»-S
Sllw: 21-3 fT, <6E-i W»
«**"* xi-2 «"3 <«-5
Stmeiua *•"* ±£7 u-1 Sl-2
Sulfite Jt-1 -tf-4
-------
All detection limits reported are actual calculated values.
Therefore, depending upon internal standard adjustments, sample
size differences, choice of analysis, isotope, and spectral
background, these values may differ somewhat from s&t^le to
sample. Tables XIII-XVII list the results for the individual
samples collected at Newcastle, Glendive, Garrison, Ft. Peck,
and Belle Fourche sites, respectively.
-24-
-------
TABLE XIII
INDIVIDUAL FOURTH QUARTER FILTERS FROM NEWCASTLE
(Alt fc>S«
*/>m *mm *mn% 4/23;?^ umn mm n}t(f}
U lltHficMK ri|»n)
tMiaoo)
toHU
•ul«
•ocftltwk
fttWMCk
Mttw
Cnlw
OnrtM
QkwIn
Cobalt
Ctillua
Caraaalwa
llM
LttUw
Hajnf f tum
IIU|»«W
tfelytotaflUNfc
Sidcl
Sntat
t«l Udfa*
ruttMM
Pot •*«*<«!
Iktaiu*
Nuadtua
feMdiw
ff iKiliM
—-——— "r*.vr i
S*l*«ltt|i
snicM
sa««r
Sodlua
MMUlw
Sulfur
TtMlliMW
Tborlua
tin
Uranlu*
V«na4tt»
KUdte
tttili*
Zinc
Itnmta*
<78-2
K-S
<78-5
2E-1
se-i
»-}
-4
3E-2
<18-4
3E-1
2E-2
<7B-5
•8—1
«M0
SI-S
<38-5
58-5
18-3
48-4
58-4
<*8-S
<•8-5
4M-5
58-5
98-5
48-5
U4
11-1
Ht-4
•t-5
18-4
<38-5
18-4
28-3
<48-4
28-4
<18-5
in
28-4
<88-3
41-1
18-4
48-1
18-2
a-s
18-4
tt-4
e
m
48-2
<«-J
it;
**4
se
48-1
<48-5
28-3
18-1
<48-*
78-4
28-4
48-1
38-4
<78-3
38-3
18-4
38-2
18-*
18-4
<48-*
101
28-4
38-1
58-3
48-1
38-2
<18-4
18-4
<58-5
<28-5
18-g
<48-5
48-1
<18-5
<48-A
38-3
<28-5
18-4
28-4
-------
AMma
K-2
<2E-i
4cs«alc
«•!
•arte*
K-S
Iwillta
» mm* ifnfjf
<48-3
ie-4
J«+i
M-4
It-J
<»-»
*-#
«»*
imm ">y»f
<18-1
-------
TABLE XV
INDIVIDUAL FOURTH QUARTER FILTERS FROM GARRISON
tnAi1)
I
K»
I
Silicon
Sitwr
Sxttw,
Mfat
JMUa
%r*
Star
VfcuMaa
tiuritlap
ntriun
Ziupaiu*
-------
TABLE XVI
INDIVIDUAL FOURTH QUARTER FILTERS FROM BELLE FOURCHF.
mm,
*-»
m\m smm
mm
<21-1
mm vnm
j/uol.
<2*-l
~r «
«-4
s-t
(
Hwtlx
IMmIv
***** '
Mwiat
latlwaiuB
Sc
It-i
<51-5
}£•
5«-5
W-l
'K-*
'M-4
21-1
11-1
«l*-2
U-4
lt-4
U-4
R-t
H-2
U-l
<*£-4
11-4
<11-4
51-2
K-l
U-2
<*-1
2E-2
U-l
'2£-4
H-4
«U-4
U-4
H-4
<21-4
<11-4
4t-4
n*-4
«4t-»
41-1
•U-i
41-1
-------
TABLE XVII
IIB>IVIDUAL FOURTH QUARTER FILTERS FROM FT. PECK
Bmrtaa
¦MylltM
lloatk
Cikliia
CHtw
ckixtw
OtroBlna
Cobalt
tlwilai
Imi
ItiMhiiw
S3l
IMklM#
(uMmiUmi
kutlw
ttitlim
Ilw
tlcMia
ICO
tt-4
*-*
1E-J
<»-l
lt-4
IE-1
tt-l
a-t
*• Am ll|m)
441/24 imm mim "22^1 1Z3GS mmi imrn imm t/t«m mim .mm
® ,W IW »•» »-l U-l II* Ha u-l .. .
B>1
U-3
ME-I
18-3
Ml
28-1
tt-i
<«-a
tt-3
M-J
tt-2
«tt-J
<4t-t
<41-4
48-2
»-4
38-2
»-3
tt-2
tt-2
<18-4
48-1
<48-1
2E-1
«M-S
tt-I
«a-i
-------
4.0 DISCUSSION OF RESULTS
The results of each set of filters demonstrate a
fair consistency. Four factors must be kept in mind when
using or analyzing thes^ data:
(1) Spark source mass spec create tery used
In Che survey made is at best ±30%,
with the true accuracy dependent on
background, standards and the amount
and nature of sample. For these
samples, the results are felt to
give a reasonable order of magnitude
level of the various elements.
(2) The nature of the filter material is
in question. Although the sampling
contractor supplied blanks and sup-
posedly used the same filters in all
cases, the behavior of the individual
filters under analysis raises some
doubts as to the uniformity of filter
materials.
(3) The sample loading was very low, too
low, in fact for the sampling con-
tractor to be able to weigh the filters.
/
All results are baaed, therefore, on
flow measurements.
(4) The filter loadings available were from
separate fti-V©l samples tak#n coEucuv&ptly.. .
A direct comparison indicate# that cor-
respondence with the membrane samples
is poor. This is probably due to the
difference in sampling rates and location
of the two samples.
-30-
-------
Irrespective of these problems, it is felt that the
results presented in this report give a usable semi-quantitative
indication of the relative trace element backgrounds in the
Northern Great Plains area.
To show some measurement of the reliability of the
data, the individual results of the fourth quarter samples
were averaged and compared with the results of the fourth quarter
composites. These data, along with the ranges for the individual
analyses are given in Table XIII.
With few exceptions, even though the individual
samples show a wide range, the composites and average* are generally
within a good enough agreement to lend credence to the data base.
Comparison of individual filters is complicated by
the absence of total loading measurements. The data shown in
Table XIX are from the High-Volume samplers located near the
membrane samplers used for trace element analysis. Unfortunately
the strong differences in total grain loadings are not reflected
directly in the trace element analyses. Two factors may be
responsible for this finding: CD the membrane samples were
operated at much lower flows than the Hi-Vol samplers, perhaps
affecting total grain loading though differences in turbulence;
and (2) the composition of the particulates is variable as a
function of location and meteorological conditions. The very low
levels of the trace elements, frequently below the detection
limits of the instrument, also make variations difficult to
detect.
Ambient standards for the trace elements are virtually
nonexistent. Cadmium and mercury emissions are regulated in some
smelter applications, the maximum permissible level of beryllium
averaged over a 30-day period is 0.01 yg/m3 (CO-182). This
-31-
-------
standard is applied to the following industries.
(a) extraction plants, ceramic plants,
foundries, incinerators, and
propellant plants which process
beryllium oxide, beryllium alloys,
or beryllium-containing waste; and
(b) machine shops which process
beryllium, beryllium oxides, or
any alloy when such alloy contains
more than 5 percent beryllium by
weight.
This is significantly higher than the ambient levels found
of ^10"6 yg/ms in the Northern Great Plains.
The data acquired during this program represent
a usable data base for the ambient conditions near each sampling
point. The levels of the trace elements are consistently
low indicating that any gross increases in individual elements,
for example 50-100%, caused by emissions from a point or mobile
source should be detectable. It is strongly recommended,
however, that any future work in this area pay strict attention
to sound sampling procedures. Trace element sampling requires
the use of special low ash filters, not fiberglass, to provide
a low background. Also sampling should be designed such that
weighable quantities of dust are collected. Trace element
sampling should.be specially designed into an ambient duet
monitoring program. In this way, reliable, credible results
are insured. The very low concentrations of ambient trace
elements and the potential significance of small changes in
individual elements makes this of vital importance.
-32-
-------
TABLE XVIII
i
COMPARISON OF FOURTH QUARTER COMPOSITES WITH RANGE AND AVERAGE OF FOURTH QUARTER INDIVIDUAL FILTER
TRACE ELEMENT ANALYSES (ttg/tt3)
IwmlU Glw»JI»« C»rrl»o«
Awhm »mm 6mwlt« ItanKf twwulW town* Iwn"
lellr FimrcM
ft. frck
frwmt »an^« twnnItT
tlMlM
in
1C-2/K0
«E-t
it-1
3E-2/2EB
tea
<2K-t
<28-2/88-1
48*
'K-l <
tot l««H
<48-4
K-i
w8-6
<48-6/38-4
*4
W-5/JE-1
•8-4
4E-4
<6E-S/9E-4
3E-3
3E-4
38-4/48-4
•8-4
»E-1
AmIm
8-2
<48-3/88-2
48-2
18-2
nHian
<«•»
«48-4/<48-4
<2E-4
<98-4
8-4
<48-4/<28-4
<28-4
<*-4
r*uutM
4£-l
K-i/ica
38-1
BE-2
5E-3/2E-1
<58-2
28-1
<28-2/»»
18-2
M-l
IkWtMl
<18«*4
«»e-4/
28-2
Zlrcoolua
2E-1
IE-4/IE-I
IE-3
»2E-4/«-4
SE-4
28-1
18-4/48-1
W-3
18-1
<4E-2/2E-l
1E-6/4E-4
8-4
28-1
38-2
6E-3/SE-2
IE-2
18-4
18-3
18-4/28-3
48-4
-------
TABLE XIX
MEAN GRAIN LOADINGS FROM HIGH VOLUME SAMPLERS LOCATED NEAR THE MEMBRANE SAMPLER SITES
Newcastle
Glendive
Garrison
Ft. Peck
Belle Fourche
1st Quarter
24.9
18.7
31.5
18.8
19.1
(jig/ra3)
2nd Quarter
8.3
11.4
40.0
7.1
8.9
3rd Quarter
11.3
13.5
18.6
13.0
14.9
4th Quarter
40.3
23.7
34.1
6.8
39.2
-------
5.0 REFERENCES
1) Brown, R. , Jacobs, M. L. and Taylor, H.E., American
Laboratory, 4, p. 72, (1972).
2) Taylor, H. E., and Brown, R., Inst, in Mining and
Met. Ind., 2, p. 21, C1974).
3) Trace Analysis by Mass Spectrometry. Ed. Ahearn,
A. J., Acad. Press, New York, p. 297, (1972).
4) "The Environmental Protection Agency Northern
Great Plains Ambient Air Monitoring Network, Vol. I",
Environmental Protection Agency, Rocky Mountain
Region, Region VIII, (November, 1975).
5) Code of Federal Regulations. 40 Protection of
Environment. Revised edition. Washington, D.C.,
General Services Admin., Office of the Federal
Register, (1973).
-35-
-------
APPENDIX
-------
Report Ho. 1
Statistical Evaluation of Blank Fiberglass
Filter Pads Used in Northern Great Plains Resource Project
February 18, 1975
by:
H.E. Taylor
R. Brown
Contract No. 68-02—13S3
A-l
-------
I. PURPOSE
Hie purpose of this study was to determine the r.ian levels and
statistical variation of trace eler.ants in blank fiberglass filter pads
prior to the analysis of filters loaded with air particulates. This
data was compared with typical results acquired or. two loaded samp lets
provided by the Environmental Protection Agency. All data was obtained
by spark source mass spectrometry.
Multi-element analytical data was acquired on each of eiaht s
8H" x 11" blank fiberglass filter pads. Each filter was statistically
sampled by removing eight one inch square segments from one inch wide
diagonals cut from comer to comer of Che filter pad*. That# eight
square inches of filter were ground into a fine powder in an agate
mortar and pestle. A separate composite of each of the eight: blank
filter pads was prepared in this fashion, such that a statistically
accurate composite was available for each pad.
Table I tabulates the printed number on each blank pad that
was sampled. In addition, the numbers of the two typical loaded
samples are also listed.
II. SCOPE
TABLE I
Identification
Filter No.
Weight (g)
Blank
Blank
Blank
Sample
Sample
Blank
Blank
Blank
Blank
Blank
002060
004030
009030
013060
016060
021039
0 2 3060
026030
800036
800038
4.2643
4.3124
4.3975
4.3088
4.2712
£.3161
3.8444
4.4354
4.9575
.8053
A-2
-------
0.1000 gram of each powdered composite was blended with 0.1000
gram of ultra high purity graphite, and internal standards of indium
and rhenium were added. Each of these mixtures was compacted into
electrodes and spectra from the nass spectrometer was obtained for each
sample under identical instrumental parameters.
The two typical loaded filters were prepared in the following
fashion. The unloaded margin of sample Mo. 1 was cut and removed
from the loaded area. This margin was ground into a powder, and
0.1000 gram was prepared as described above. The loaded area of
the filter was sampled as described above into eight randomly taken
one inch squares from the diagonals of the filter. The ground composite
of each sample was thoroughly mixed, and 0.1000 gran was subs amp led
and blended with 0.1000 gram of high purity graphite. By back
calculating from the weight of one square inch of filter material,
the 0.1000 gram sample was shown to be equivalent to approximately
1.7 square inches of surface area.
III. RESULTS
Table II lists the data from the determination of trace elements
in the blank fiberglass pads. This data is presented both in the
units of ppm wt. in the bulk filter material, and also in terms of
Ug/sq. inch based upon the mean weight/unit area of the blank pads.
The data is presented in this form so as to be more directly
comparable to the typical loaded filter data.
Error terms are also presented. These were determined based
upon accepted statistical data treatment techniques. The 90%
confidence interval was used to represent.the error data.
TABLET II
' Sample Data
Blank Data Bisalc #1 Saaple 1 tap!* 2
wean Cppm *t.) t error* mean (Ug/in.2) t error* (Ug/ia.2) Cttg/la-2) Cy g/in. 2)
u
1*2
1.2
0.065
0.065
0.034
0.07L
o.m
Th
6.0
6.9
0.32
0.37
0.21
0.21
0.36
71
0.42
0.17
0.023
0.009
0.034
0.029
o.ac
Fb
6.4
5.1
0.35
0.2ft
1.2
0.31
0.42
V
0.46
0.41
0.025
0.022
0.048
0.048
0.056
Hf
1.7
1.5
0.092
0.091
0.040
0.0 S3
0.15
A-3
-------
Clement
mean Cooca wt.)
TABLE- IT (con'-t)
Blank Data
i error8 mean (ust/in.2)
± error*
Loaded Sample
Blank #1 Sample
(us/in.2) (UK/in.
Data
1 Sample
2)
Lu
0.07
0.02
0.004
0.0001
0.004
-0.004
O.OOS
Yb
0.36
0.30
0.030
0.016
0.037
0.028
0.039
Is
0.03
0.02
0.002
0.001
0.002
0.002
0.002
Er
0.14
0.08
0.008
0.004
0.008
0.017
0.013
Ho
0.06
0.05
0.003
0.003
0.004
0.007
0.004
Dy
0.90
0.83
0.049
0.045
0.036
0.071
0.072
Tb
0.15
0.13
0.009
0.007
0.007
0.0 LI
0.011
Gd
0.61
0.4&
0.033
0.026
0.036
0.029
0.043
Eu
0.55
0.43
0.030
0.023
0.014
0.029
0.054
So
0.90
0.79
0.049
0.043
0.021
0.05Q
0.084
Nd
16
9.9-
0.86
0.53
0.77
0.56
0.78
Pr
2.6
1-.3
.0.14
0.070
0.12
0.51
a. 12
Cs
37
20
2.0
1.1
1.7
0.83
1.3
La
9.0
5.3
0.49
0.29
0.25
0.29
0.47
Ba
130
68
7.0
3.7
5.1
3.0
7.8
Cs
0.029
0.048
0.002
0.003
0.001
0.004
0.004
I
0.23
0.18
0.012
a.oio
0.018
0.021
0,021
Te
0.06
0.05
0.003
0.003
< 0.004
< 0.004
<0.004
Sb
0.10
a.os
0.005
0.007
0.006
0.016
0.006
Sa
2.4
3.0
0.13
0.16
0.55
0.10
0.14
Cd
ff.32
0.15
0.017
0.008
0.043
0.037
0.019
Ag
0.04
a.02
0.002
0.001
0.004
0.002
0.005
Mb
1.0
0.66.
0.053
0.036
0.05?
0.059
0.18
Nb
8.0
4.5
0.43
0.24
0.59
0.36
0.36
Fr
150
63
8.1
3.4
5.9
7.1
9.a
Y
11
7.4
0.59
0.40
0.43
0.39
0.44
Sr
67
1.3
3.6
0.07a
4.0
4.0
4.0
Rb
1.7
0.26
0.092
a.oi4
0.15
0.065
0.066
Er
1.1
1.1
0.059
0.059
0.015
0.21
0.073
Se
0.12
0.12
0.006--
a.oa&
0.010
0.014
O.014
A-4
-------
TABLE II Cecwirt)
Element
sean Cdpib wt.)
Blank Daca
± error* mean Cujr/in.2)
t error'
As
1.6
0.97
0.086
0.052
Ge
0.32.
0.33
0.017
0.018
Ca
4.2
2.1
0.23
0.11
Zrt
12
9.9
0.65
0.53
Cu
10
6.4
0.54
0.35
SI
14
6.1
0.75
0.33
Co
0.52
0.6 J
0.028
0.034
Fe
1900
1040
100
56
Ma
17
9.4
0.92
0.51
Cr
39
26
2.1
1.4
V
75 •
61
4.0
3.3
Ti
1600
1800
86
97
Sc
20
33
1.1
0.033
Ca
Maj
Maj
Mai
Mai
K
340
*480
IS
26
CI
140
120
7.5
6.5
s
110
96
5.9
5.2
p
360
260
19
14
Al
Maj
Mai
Mai
Mai
Mfc
sz
2.12
2700
1100
Re
6900
8700
370
470
F
2250
2750
12Q
150
B
Ma J
Mai
Mai
Mai
Be
0-53
0.50
0.029
0.027
LI
27
31
1.5
1.7
Loaded Sample Data.
Blank. #1 Sample 1 Sample 2
(ug/lp-) (iig/ig.2) (ug/in.2)
0.13
0.20
o.ia
0.015
0.012
0.016
0.37
0.25
0.25
1.2
1.3
0.96
1.1
1.2
1.4
1.1
1.1
t.L
0.045
0.034
1.0 U
100
170
150
1.2
1.2
0.84
2.7
1.4
1.9
5.9
5.9
4.S
104
100-
84
1.2
0.89
2.4
Mai
Mai
Mai
21
14
14
14
14
14
17
100
28
17
14
6.1
Mai
Mai
Mai
3400
3400
3500
530
590
360
71
71
160
Mai
Mai
O.O 33
0.071
0.072
3.9
3.9
3.9
a 90% confidence interval
-------
IV. CONCLUSION
As can be seen fror. the data presented in Table II, most all elerasnts
are present as impurities in the fiberglass materials, and as was predicted,
they are not uniformly distributed throughout all filter pads.
The degree of elemental enhancement of loaded samples over blank
levels is not sufficiently high to allow accurate blank corrections to
be performed. This is illustrated by noting that in Table II, the loaded
values are often lower than the value of the upper 90% confidence limit for
the blank neasurenents. Perhaps with core heavily loaded samples, the
background levels would be less significant.
In summary, it is apparent that this fiberglass material is not
suitable for the sampling of particulate matter when trace element analysis
is to be performed. The combination of high background levels of trace
elements in the fiberglass material and the inability to separate the
particulate loading from the fiberglass material prior to analysis precludes
the ability to achieve accurate analyses.
A-6
-------
REPORT NO. 2
Statistical Evaluation of Blank Fiberglass
Filter Pads in Northern Great Plains Resource Project
March 17, 1975
by:
H.E. Taylor
R. Brown
Contract Uo. 6S-02-1383
A-7
-------
I. PURPOSE
1'lic purpose of this study was to deterni-.a the mean levels and
statistical variation of mercury in blank fiberglass filter pads prior
to the analysis of filters loaded with air particulates. This data
was compared with typical results acquired on two loaded samples provided
by the Environmental Protection Agency. All data was obtained by combustion/
amalgamation atomic fluorescence spectrophotometry.
Analytical data was acquired on each of eight separate 8 V by 11"
blank fiberglass filter pads. Each filter was statistically sampled
by removing eight one inch square segments from one inch wide diagonals
cut from corner to corner of the filter pads. These eight square inches
of filter were ground into a fine powder in an agate mortar and pestle.
A separate composite of each of the eight blank filter pads was prepared
in this fashion, such that a statistically accurat* composite was avail-
able for each pad.
Table I tabulates the printed number on each blank pad that was
sampled. In addition, the numbers of the two typical loaded samples are
also listed.
II. SCOPE
TABLE I
Identification Number and Weight of Eight Blank
Fiberglass Filter Pads and Two Typical Loaded Sar.ples
Identification
Filter Ko
Ueight (g)
Blank
Blank
Blank
Blank
Blank
B±ank
Blank
Blank
Sample
Sample
026030
800038
023060
800030
016060
021030
009030
013060
002060
004030
4.2643
4.3124
4.39 75
4.30SS
4.2712
4.3161
3.8444
4.4354
4.9575
4.8053
A-8
-------
Each of the composited samples vas analyzed by heating 0.1000
gram to 850° C. in a dynanic oxygen atirosphere. The vapor is passed
through a quartz combustion tube at 1000° C. and amalgamated on gold
wool• The nercury is thermally deanalgamated in an argon stream, and
p.•.ssed. through an atomic fluorescence cell where it is excited by 253.6
nanometer radiation. The fluorescence signal is measured by a solar
blind photomultiplier tube, and the resultant signal is amplified and
displayed on a strip chart recorder. The sample signals are compared
to the signal observed from NBS standard reference material 1633-fly ash.
Table II lists the data from the deternination of mercury in the
blank fiberglass pads. This data is presented both in the units of ppm
vt. in the bulk filter material, and also in terms ofyg/sq. in. based
upon the mean weight/unit area of the blank pads. The data is presented
in this form so as to be more directly comparable to the typical loaded
filter data.
Error terms are also presented. These were determined based upon
accepted statistical data treatment techniques. The 90% confidence
interval was used to represent the error data.
III. RESULTS
TABLE II
Blank Data
Mean (ppm wt.) 0.050
Error (ppm wt.) ±0.020
Mean (pg/in.^) 0.002
Loaded Parole Data
2
Sample l(yg/in. ) 0.016
Sample 2(pg/in.^) 0.010
Error (yg/in.*") ±0.0008
A-9
------- |