'AJ~k9"V""

mllllljflllllflllllllllll‘lSl
LIBRARY38 1

Michigan State
Unlvordtyfi

 

 

 

 

 

 

 

This is to (:81 tify thal‘the

thesis entitled

Tritium Transport Through a Fractured Till in Michigan

presented by

John Maris Gobins

has been accepted towards fulfillment
of the requirements for

 

M.S. dpmvein Geological Sciences

MAX;

/ Majo rofessox

Date 12 Ma! 1989

0.7639 MS U is an Affirmative Action/Equal Opportunity Institution

 

 

 

RETURNING MATERIALS:
MSU Place in book drop to remove this

‘ checkout from your record. FINES
LIBRARlF-b will be charged if book is returned
after the date stamped below.

 

 

 

 

 

 

TRITIUM TRANSPORT THROUGH A FRACI‘URED TILL IN MICHIGAN
By

John Maris Gobins

A THESIS
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

MASTER OF SCIENCE

Department of Geological Sciences

1989

:w.L

St. "I

ABSTRACT
TRITIUM TRANSPORT THROUGH A FRACTURED TILL IN MICHIGAN
By

John Maris Gobins

The concentration of tritium in 40 ground-water samples collected from a fractured
clay-rich till in southeastern Michigan was found tO range from 0 to 44 TU. Bomb
tritium was identified in samples at depths of 1.5 to 42 m below the water table while
pre-bomb tritium was identified in samples from depths of 5.0 to 54 In. The distribution
of tritium in the ground-water reservoir was simulated using a one-dimensional solute
transport model. Based on the shallowest occurrence of non-tritiated water, the
dispersivity value of the matrix flow system in the upper till unit was found to be less
than 0.025 m. Bomb tritium penetration tO depths beyond those predicted by the model

can be explained by the presence of a fracture flow system.

ACKNOWLEDGEMENTS

Dr. Grahame Larson, thesis chairman, is thanked for suggesting the project and
being actively involved until its completion. Drs. David Long and Roger Wallace are
thanked for serving on the thesis committee and providing valuable suggestions during
the course Of the study.

Two adjunct committee members are thanked for their significant contributions.
Robert Hayes at the Michigan Department of Natural Resources made available the
hydrogeologic data and provided numerous insights. Richard Mandle at the US.
Geological Survey provided most of the guidance during the computer simulations Of
ground-water flow.

Mitchell Adelman collected and delivered the water samples.

Valuable insights were also provided by William Iversen and Roger Noyce at the
Michigan Department of Natural Resources, Norman Grannemann, Michael McDonald,
and David Westjohn and at the US. Geological Survey, and colleagues William
Monaghan, Michael Miller, John Gillespie, Manrico Delcore, David Regalbuto, Gregory
FOOte, and Pierre Bruno.

Dr. John Wilband and Karlis Kaugars provided computer assistance.

The Environmental Isotope Laboratory at the University of Waterloo is thanked for
its prompt and courteous service. Dr. James Hancock Of the Horticulture Department

and Dr. James Tiedje Of the Crop and Soil Science Department at Michigan State

iii

University are thanked for permission to use the liquid scintillation counters.
The Water Research Institute at Michigan State University is thanked for funding
this study.

MIlS paldies maniem vecakiem!

iv

TABLE OF CONTENTS

LIST OF TABLES

LIST OF FIGURES

INTRODUCTION

STUDY AREA

GEOLOGY

AGE AND ORIGIN OF THE DRIFT

HYDROLOGY

NUMERICAL SIMULATION OF GROUND-WATER FLOW

SAMPLING METHODS

STABLE ISOTOPE RESULTS

TRITIUM RESULTS

DISCUSSION

CONCLUSION

APPENDICES
APPENDIX A - Regional Geologic and Hydrologic Analysis
APPENDIX B - Cross Sections Through the Study Area
APPENDIX C - Simulated Hydraulic Head Distributions
APPENDIX D - Lateral Tritium Distribution

APPENDIX E - Site Soil Types

vii

viii

9

9

15

17

20

22

26

27

35

39

41

43

APPENDIX F - Tritium Systematics
APPENDIX G - Sample Preparation for Tritium Analysis
Pre-distillation
Electrolytic Enrichment
Post-distillation
APPENDIX H - Liquid Scintillation Analysis
Introduction
Theory Of Liquid Scintillation Counting
Factors Affecting Counting Efficiency
Quench Monitoring and Curve Generation
Counting Window Generation
APPENDIX I - Tritium Activity Calculations
APPENDIX J - Two Sigma Error Determination
APPENDIX K - Tritium Input Function

BIBLIOGRAPHY

46

49

49

50

51

53

53

53‘

55

58

60

62

65

67

69

Table 1. Stable isotope data.

Table 2. Tritium data.

LIST OF TABLES

vii

18

21

Figure 1.

Figure 2.

Figure 3.

Figure 4.

Figure 5.

Figure 6.

Figure 7.

Figure 8.

Figure 9.

LIST OF FIGURES

Geologic map showing location of study area and
surficial deposits Of Genesee County (after
Wiitala et al., 1963).

Geologic cross section showing simulated head
distribution along line X-X’ Of Figure 4 (contour
interval = 0.5 m). All elevations in m.a.s.l.

Elevation versus percentage Of soil borings
indicating coarse material (asterisks- all
borings, circles- western half Of site data).

Observed potentiometric surface contour maps and
location Of cross section X-X’ (all elevations in
m.a.s.l.).

A Upper Till (contour interval = 0.5 m)

‘B Middle Sand (contour interval = 0.25 m)

C Bedrock (contour interval = 0.5 m)

Elevation of screened interval tOp versus
elevation Of static water level and geologic
units sampled. Open circles- upper till, Open
squares- middle sand, filled circles- lower till,
filled squares- bedrock.

Stable isotope composition of water sampled from
drift wells in Michigan. Solid line- Lake Simcoe
meteoric line (Desaulniers et al., 1981).

Simulated and Observed tritium values in the
upper till unit at the facility.

Regional bedrock surface contour map (contour
interval = 2 m).

Regional bedrock surface lithology map (filled
circles- shale, Open circles- sandstone).

viii

10

16

19

24

30

31

Figure 10.

Figure 11.

Figure 12.

Figure 13.

Figure 14.

Figure 15.

Figure 16.

Figure 17.

Figure 18.

Regional drift isopach map (contour interval =
2 m).

Regional map showing height Of bedrock
potentiometric surface above bedrock surface
(contour interval = 2 m).

Regional bedrock potentiometric surface map
(contour interval = 2 m).

Map showing location of wells sampled and

location of cross sections in relation tO the
Berlin & Farro site boundary.

Cross sections A-A’ and B-B’.

Cross sections CC and D—D’.

Figure showing the effect Of varying lithologies
on simulated head distribution along cross
section X-X’ in Figure 4 (contour interval =

0.5 m).

Lateral distribution Of tritium within the upper
till unit (all values in TU).

Tritium input function.

32

33

34

36

37

38

4O

42

68

INTRODUCTION

The clay-rich tills which characterize much Of the Great Lakes basin have
traditionally been viewed as low permeability deposits that have limited potential for
water supply (Norris, 1963). For this reason, they have been commonly used for
containment Of waste material and it has even been proposed that, in places, the tills
may provide a suitable media for the disposal of hazardous waste (Desaulniers et al.,
1981)

Within the last several decades, however, researchers have increasingly noted the
occurrence of fractures within till deposits Of the British Isles and Great Plains region of
North America (Horberg, 1952; Fookes, 1965; Kazi and Knill, 1973; Grisak et al., 1976:
Eyles and Sladen, 1981; Hendry, 1982; Keller et al., 1986; Cravens and Ruedisili. 1987).
These fractures may have Openings as large as several centimeters and have been
Observed to depths Of over 10 meters (Eyles and Sladen, 1981; Keller et al., 1986). The
presence Of such fractures in tills Of the Great Lakes basin would be of considerable
interest since a regional low-level radioactive waste disposal facility could be associated
with till deposits within Michigan (MILLRWC, 1987).

The purpose Of this study is to examine the ground-water flow system in a fractured
clay-rich till in southeastern Michigan and to determine the effect Of the fractures on
the movement Of "bomb" tritium through the till. After a ground-water flow model is
developed for the till, a one-dimensional solute transport model is employed to simulate

the distribution of bomb tritium with depth.

STUDY AREA

The study area for this investigation is the Berlin and Farro Liquid Incineration
Facility (BFLIF) which is located approximately five and one-half kilometers south of
the town Of Swartz Creek, Gaines Township, Genesee County, Michigan ('1‘. 6 N., R. 5
13., Sec. 23). The 0.14 kmz facility lies exclusively on till plain (Figure 1) and at one
time included an incinerator, a sludge lagoon, and a landfill. As a result Of repeated
improper handling Of toxic and hazardous wastes by the facility Operators during the
1970’s, the facility is currently included in the CERCLA (Superfund) National Priority
List for remedial action and is presently under the control of the Michigan Department
of Natural Resources (D’ Appolonia, 1983).

Since 1982, several hydrogeologic studies have been initiated at the BFLIF, mainly
at the request of the Michigan Department of Natural Resources, to provide the state h
with information necessary to meet Environmental Protection Agency requirements for
federal funding Of cleanup Operations. These studies include magnetometer, resistivity.
and seismic surveys, fracture trace analysis, test drilling, and ground-water and soil
sampling (Woodward and Clyde, 1989). TO date, 56 Observation wells and more than
100 test borings have been drilled on or near the facility. The majority of Observation
wells are screened in glaCial drift, but some are completed in bedrock. Remedial action

has included removal Of landfill and lagoonal contents and excavation Of buried drums.

GEOLOGY
The BFLIF is located on a gently rolling till plain, deposited by the Saginaw glacial

lobe, which is bordered on the north by the Flint moraine and on the south by the

 

XATIN

End moraine

 

Till plain

E

Outwash plain

 

 

Lake plain

Bar/ beach

10 km

 

Figure 1. Geologic map showing location of study area and surficial deposits of Genesee
County (after Wiitala et al., 1963).

4
Gaines moraine (Levrett and Taylor, 1915; Figure 1). Surface elevation ranges between
250 In in the northeast and central parts of the facility to 247.5 In in the western part.

The Pennsylvanian Saginaw Formation is the uppermost bedrock formation beneath
the facility. The elevation Of the bedrock surface ranges between 220 In along the
eastern boundary Of the facility to less than 205 m in isolated low spots in the central
and southwestern parts Of the facility. Boring logs show that it consists Of light gray to
grayish-brown and brownish-gray very fine- to coarse-grained sandstone. Interbedded
with the sandstone are thin layers Of shale and laminae of dark gray clays. Crossbeddin g
in the sandstone is common and consists Of alternating bands of dark and light, fine- and
coarse-grained quartz. Cements are usually non-calcareous. The interbedded shale is
typically gray tO dark gray, firm to very hard, fissile, micaceous, and thinly bedded.
Although the rock quality designation (RQD) for the Saginaw ranges from 0% to 98%.
the average RQD value is 75% to 90% (Woodward and Clyde, 1989).

The unconsolidated sediments which overlie the bedrock are glacial in origin and
were deposited during the WOOdfordian stage Of the Wisconsinan glaciation (Farrand
and Eschman, 1974). Based on monitor well and boring logs, two relatively distinct
layers of till, separated by a 0-5 m thick, discontinuous and variable, water-bearing sand
and gravel layer, underlie the facility (Figures 2, 14, and 15).

The lower till is an 18 to 27 m thick, very stiff to hard, moist to dry, gray clayey silt
tO silty clay with occasional sand lenses and trace amounts of gravel. Characteristic Of
this till is its resistance tO penetration with blows per foot typically exceeding 50. The
sand and gravel lenses within this till vary in composition from silty sand to fine-grained

sand tO coarse-grained sand and gravel and are typically dense and wet. In areas

4.33: 3 80396—0 =< .AE nd n 3235 3988 v 253m
.0 .x-x on: «no? caustic emu: econ—25a ”arson“ .5303 «.38 ammo—000 .N Ouswfi

 

 

 

 

 

 

 

 

 

 

 

 

 

 

ICON
2003mm .3
I 03 M
V
Iowa u
j.» 526.. O
N
loam \w)
e
oz<m made-3 III..- ttttttttt I ttttttttttttt IIILUI.“ .s
I II I I III III II III II I...IIIIIII..HIIH nuuuHIIIhnnuIIuflIhnnWIIWwa Hunnmmtnmlwm IOVN NU
IlvTI/II lIIlIlllIL tomN
op 33 832 ha 3! mm mm MN 33 . ha 32 Np mm
x x

6
where the bedrock surface consists Of shale, shale fragments have been incorporated into
the base Of the lower till, and in areas where the bedrock surface consists Of sandstone,
the lower till is sandier at its base.

The upper till is a 12 to 18 m thick, soft to stiff, moist to dry, brown to gray, silty
to sandy to gravelly clay with occasional sand and gravel lenses and stringers. The
penetration resistance Of this till varies between 10 and 40 blows per foot but may be
less in some Of the sand lenses. The sand and gravel lenses vary in composition from
silty sand and clayey gravel to fine-grained sand to coarse-grained sand and gravel and
are typically loose and wet.

The large number Of soil borings makes it possible tO analyze in some detail the
distribution Of sand lenses within the upper till unit. Equal elevation sections of soil
texture were generated at 0.3 m vertical intervals, starting at the surface and continuing
to a depth of 12 In. These were subdivided into two textural classes. The coarse
textural class includes gravel and sand and is described in the Unified Soil Classification
System by the symbols GW, GP, GM, GC, SW, SP, SM, and SC. The fine textural class
includes silt, clay, and organic deposits and is described in the same system by the
symbols ML, CL, CL, MH, CH, and OH. Intervals not recovered were recorded as part
of the coarse textural class. The percent Of wells and borings indicating coarse material
at each vertical interval was calculated and plotted as a function Of depth (Figure 3).
One curve was generated based on all Of the study area logs and another was based on
logs from the western half Of the site where sand lenses appeared to be more common.

The findings indicate the presence Of a 5 m thick zone in the upper till unit,

characterized by intervals Of coarse sediment, approximately 4 to 5 m below the surface

 

 

250
T /J. m COMPLETE SITE DATA
_ -.. '- ooeooWESTERN HALF OF SITE DATA
/'.\ - i i ,
(I; ..
0 245A
E T
\.../
CZ) A
E _
> 240-
L'LJ .J
__t
UJ _
235 I I r l r I l I 1

 

O 20 40 5'0 80 100
% COARSE

Figure 3. Elevation versus percentage of soil borings indicating coarse material
(asterisks- all borings, circles- western half Of site data).

8
(Figure 3). Individual borings typically encounter 1 to 3 m Of coarse sediment in the
upper 12 m of the drift. The cumulative thickness of the coarse sediment in each
boring may be represented by a single sand layer or by several layers separated by clay
stringers. Since the thickness of the coarse zone and the thickness of the lenses is
different, it can be inferred that some vertical ofi’set exists between the lenses.
Laterally, almost 50% Of all wells and borings indicate coarse material at an elevation of
244 m and, when logs from only the western half Of the site are used, this value
increases to nearly 65%. This indicates that, at this elevation, the majority of coarse-
grained lenses occur in the western half of the study area. Lateral extent of individual
lenses is more difficult to determine but in some instances, a lens encountered in one
boring has not been encountered in another boring as little as 5 m away.

Fractures within the upper till unit have been intersected by approximately one-
third Of the soil borings at depths of l to 5 In. These fractures are typically vertical to
sub-vertical and discontinuous and may be empty or filled with trace amounts of sand
and/or root material. The fracture aperture is generally less than several millimeters and
fracture spacing near the surface is approximately 5-10 cm based on Observations at a
excavation 1 km north Of the facility. Walls of the fractures are typically smooth and
contain a thin brown oxidation rind developed on the gray matrix material. Fractures
with this type Of appearance in tills have also been described by other workers in North
America (Williams and Farvolden, 1969; Grisak et al., 1976; Hendry, 1983a). Brown and
gray mottles are often associated with the fractures within and just beneath the soil

ZOI'IC.

AGE AND ORIGIN OF THE DRIFT

While an exact age has not been assigned to the drift beneath the facility, it is
believed to be Late Wisconsinan. A similar drift stratigraphy has been described for
Saginaw Lobe drift approximately 100 to 150 km southwest of the study area by
Monaghan and Larson (1986) and has been correlated with drift sequences of the Lake
Michigan Lobe in southwestern Michigan and the Huron-Erie Lobe in central Ohio.
They describe two tills separated by an intervening variable sand and gravel unit. The
lower and upper tills have been assigned ages of approximately 16,500 yr BF. and
14,780 yr B.P., respectively. They associated the sand and gravel unit which separates
the two tills with the Erie Interstade (15,500 to 15,000 yr B.P.) which represents a major
retreat of the ice from the region (Dreimanis and Goldthwait, 1973).

As a result of its hardness and scarcity of sand and gravel lenses, the lower till at
the facility is believed to represent lodgement till. The propensity Of sand lenses in the
upper till and its lesser consolidation, on the other hand, would suggest that it formed as '

ablation till.

HYDROLOGY

The ground surface at the facility generally slopes westward and, accordingly,
surface water is removed in that direction. Approximately 150 In west of the facility,
Slocum Drain collects the runoff.

Three "water-bearing" zones have been identified at the facility (Figure 4). The
uppermost zone includes the sand lenses in the upper till unit between an elevation of

approximately 240 m to 245 In. The middle zone includes the sand layer which

10

246.76
j

 

 

 

 

 

 

 

 

 

I 247.93 0
\
\ \
B - /
244V
q,“ 245.25 . 245.37
//////’ .24542
45.25 0//
245.33
. ”5.5
. 245.41 / . 245.67
'245.29 o 245.55
245.93 .
\
. 244.98
0
244.52 /
245-0° . 245.40

/ 245.50 .2‘5.‘ 7
/ .245.“

 

 

/
/246-00
0 100 200!!!
1 J

r

° 246.10

Figure 4. Observed potentiometric surface contour maps and location of cross section
X-X’ (all elevations in m.a.s.l.).
A Upper Till (contour interval = 0.5 m)
B Middle Sand (contour interval = 0.25 m)
C Bedrock (contour interval = 0.5 m)

11
separates the upper and lower tills and whose elevation ranges from 230 m to 240 m.
The lowermost water-bearing zone is represented by the bedrock. The potentiometric
surfaces of all three water-bearing zones are shown in Figure 4.

The potentiometric surface of the upper water-bearing zone slopes southwestward
and northeastward from a northwest-southeast-oriented divide and has a horizontal
hydraulic gradient of essentially zero in the central part of the facility to 0.01 in the
southwestern and southeastern parts. The potentiometric surface of the middle water-
bearing zone, on the other hand, slopes toward the northwest and has a horizontal
hydraulic gradient of approximately 0.00125. The potentiometric surface of the lower
water-bearing zone slopes to the north and has a horizontal hydraulic gradient of
approximately 0.005.

Vertical hydraulic gradients between the various water-bearing zones were calculated
using water levels recorded in nested wells two weeks before sampling. The average
vertical hydraulic gradient between the upper till unit and the middle sand unit is 0.22
and downward. Between the middle sand unit and the bedrock, the vertical hydraulic
gradient in the northwest part of the facility is 0.013 and downward. However, in the
eastern part and south Of the facility, the average vertical hydraulic gradient between the
bedrock and middle sand unit is 0.01 and upward. Also, the similarity of water levels
recorded in the upper till unit near the center of the facility suggests that some of the
sand lenses may be hydraulically interconnected. Although some shallow wells did show
annual water level fluctuations as great as 2 m, annual water level fluctuations for most
of the monitor wells were generally less than 1 m.

Slug and hail tests were performed in the bedrock and drift to Obtain hydraulic

12

conductivity values. In the bedrock wells, inflatable packers were used to isolate several
zones for testing. In the drift wells, however, the interval tested was represented by the
length of the screened interval. The Hvorslev (1951) method was used to calculate
hydraulic conductivity. The average for 20 bedrock hydraulic conductivity measurements
was found to be 9.9 x 10‘7 m/s and to range from 4.6 x 10'8 to 1.0 x 10'5 m/s. The
average hydraulic conductivity of the lower till unit, based on one measurement, was
found to be 6.7 x 10‘8 m/s. The average hydraulic conductivity for the middle sand unit,
based on 11 measurements, was found to be 4.4 x 10'5 m/s and to range from 1.9 x 10‘6
to 2.0 x 10‘4 m/s. The average hydraulic conductivity of the sand lenses in the upper till
unit, based on nineteen measurements, was found to be 5.1 x 10'5 m/s and to range

from 7.1 x 10'7 to 2.1 x 10'4 m/s.

NUMERICAL SIMUALTTON OF GROUND-WATER FLOW

A three-dimensional finite-difference flow model (McDonald and Harbaugh, 1988)
was applied in two dimensions along a northwest-southeast-Onented cross-section X-X’
through the central part of the facility area (Figures 2 and 4). The modeled section was
divided into 336 rectilinear nodes, 21 in the horizontal direction (X) and 16 in the
vertical direction (Z). The upper till unit was divided into eight layers, the middle sand
unit into two layers, the lower till unit into five layers, and the bedka into one layer.
The top layer was modeled as unconfined while the remaining layers were modeled as
confined. A recharge well was used for the bedrock layer along the southeastern
boundary of the cross section to account for flow across that boundary. No recharge

wells, however, were used for the drift along the southeastern boundary of the cross

13
section because flow across that boundary was assumed to be negligible. Discharge wells
were used for the bedrock and the drift along the northwestern boundary of the cross
section to account for flow across that boundary. Recharge was applied evenly across
the entire cross section at a rate of 0.10 m/yr. To Obtain this value, peak discharge
rates for Swartz Creek, which drains the till plain containing the facility and has a
gaging station approximately 12 km northeast of the facility, were subjected to baseflow
separation (U.S.G.S, 1980).

The horizontal hydraulic conductivity (Kb) values used in the simulation were those
obtained from the slug and bail tests. For the bedrock and middle sand units, the
highest field-measured Kb values were applied because most of the water in these
areally relatively extensive geologic units would move through the zones of highest
permeability. For the lower till unit, the Kt: value Obtained from the one field
measurement was applied. For the clay-rich zones of the upper till unit, the Kh value
applied was ten times greater than the value used for the lower till to account for the
effects of weathering (Keller et al., 1986; Cravens and Ruedisili, 1987). For the sand
lenses in the upper till unit, the average of their field-measured Kb values was applied.
Vertical hydraulic conductivity (K,) values were estimated by assuming certain
anisotropies: 1:1 for sand and gravel and 1:10 for clay, silt, and bedrock.

The simulated head distribution beneath the study area is shown in Figure 2 and is
in general agreement with observed heads. Slight adjustments were made to the
recharge rate but the hydraulic conductivities, anisotropies, and recharge/discharge well
flow rates were not adjusted.

Equipotential data generated from the model indicate that between the land surface

14
and the middle sand unit, ground-water flow direction is predominantly near-vertical and
downward. Along the modeled cross section, the sand lenses in the upper till unit have
little effect on the ground-water flow direction. The simulated horizontal gradient in
the middle sand unit is very small and is in agreement with gradients Obtained from
measured water levels in the same unit.

The use of a recharge well in the bedrock along the southeastern boundary of the
model simulates the observed upward hydraulic gradients between the bedrock and
middle sand unit along the southern part of the facility. However, further to the north,
equipotential lines indicate mainly downward hydraulic gradients between the middle
sand unit and the bedrock and are in agreement with observed heads. The northward
decreasing head values simulated by the model are consistent with measured head
values. I

Since there was some uncertainty as to the composition of the upper till unit in the
northern third of the simulated cross section, a sensitivity analysis was performed by
treating the composition of that part of the cross section as either all clay, all sand, or a
clay/sand mixture with a distribution similar to that beneath the facility. The results
indicate that the composition of the upper till unit in the northern part of the cross
section has little influence on the hydraulic head distribution beneath the facility (Figure
16).

Ground-water velocity (V) can be calculated from the flow model using

15

where i is the simulated hydraulic gradient, K is the hydraulic conductivity, and n is the
porosity. The vertical ground-water velocity for the upper till unit is 4.8 x 10'9 m/s
when applying a K' of 6.7 x 10'9 m/s, an effective porosity of 0.35 (Hendry, 1988), and

an average vertical hydraulic gradient Of 0.25 which was derived from the flow model.

SAMPLING METHODS

Ground-water samples from 40 monitoring wells were collected by the Michigan
Department of Natural Resources field personnel during the second week of October,
1987. These wells, terminating in glacial deposits and bedrock, have, at their bases.
screened intervals of 1.5 and 3 m, respectively, and are completely cased down to the
tops of these intervals. The elevations and lengths of screened intervals, as well as the
geologic units that they draw from, are shown in Figure 5.

Before collecting water samples, each well was purged to insure that the sample
obtained is representative of water from the screened interval. Samples for tritium
analysis were delivered to 1 liter Nalgene plastic bottles, tightly capped, and labeled. In
addition, 125 ml plastic bottles were filled with water from six selected wells and shipped
to the University of Waterloo Environmental Isotope Laboratory (UWEIL) for 18O and
2H analysis by mass-spectrometry. The accuracy of the analyses is +/— 0.2 o/oo for 18O
and +/- 2 o/OO for 2H.

The tritium concentration of each sample was first analyzed by direct liquid
scintillation counting in the tritium laboratory at Michigan State University (detection
limit of approximately 12 TU and counting error of +/-10 TU). Samples which
contained more than 12 TU were also analyzed by direct liquid scintillation counting at

UWEIL (detection limit of 6 TU and counting error of +/- 8 TU). Samples which

16

2863 -8339. 35. .=: 530. $286 3::
.23 062E -8513. some .=: coma: 5.286 EEO .uoEEmm 2E: u_wo_oom
see. .95. 5:; 3:5. .3 83:56 war“; no. EEBE 8528 Lo co_E>o_m .m oSmE

A._.w.m.Ev hmmw mmmEMHamw mm Jm>m._ mm.—.<>> 0_._.<._.w no ZO_.F<>m._m

 

 

ooeg oosvm ooevm oo.mv~ 8.4%
g — h _ — — _ _ _ _ F — _ _ ~ _ 00F
I

. I

rooN
xooeuom -

I I I
88

o I
.5 $33 roam

. .l
romw

0 0

Bow 23:2 o e no no r
o 0 room

:2. Ban: 00 o W 00 m o o I
romm

('I'S'e'w) dO.L 'IVAHBLNI GBNBEHOS :IO NOLLVAHTB

17
contained less than 12 TU were electrolytically enriched in the tritium laboratory at
Michigan State University and counted at UWEIL (average detection limit of 1.5 TU

and counting error of +/- 1.4 TU).

STABLE ISOTOPE RESULTS

The stable isotopes 18O and 2H (deuterium) are part of the water molecule and can
therefore be used to trace the origin and history of a water mass. The mass differences
of the various isotopic forms of water lead to slight differences in their vapor pressures.
which in turn are responsible for the isotopic fractionation of water during evaporation
and condensation stages of precipitation formation. The degree of fractionation depends
on the rate of the reaction and the temperature, and increases with increasing rate of
reaction and with decreasing temperature (Dansgaard, 1964). Unless high temperatures
are encountered in the subsurface, exchange of 18O with oxygen-bearing minerals can be
considered negligible. The isotopic content of precipitation is also influenced by
geographic factors in that it becomes increasingly lighter as its parent air mass moves
over continental areas and preferentially releases the heavier isotopes.

It has been demonstrated by Craig (1961) that most precipitation, excluding that in

tropical areas, falls on a meteoric water line which on a global basis is described by
dc] 2H = 8 del 18o + 10.

Thus, a ground-water sample whose isotopic composition falls on the meteoric line can
be used as an indicator of the average isotopic content of precipitation in its respective

recharge area. Water samples whose isotopic composition falls off of the meteoric line

18

can be said to have undergone evaporation, reaction with minerals, or mixing with water
of non-meteoric origin.

Del values, in parts per thousand or mil (o/oo), are used to report the deviation of
the isotopic ratios of the sample from those in an international standard. The deviations
are calculated using the equation

Rsample

del sample = ------
Rstandard

-1 X1000

 

where R is the abundance ratio (180/160 and 2H/lH). The standard for reporting del
18O and del 2H is standard mean ocean water (SMOW). The results of the stable

isotope analyses from the facility are shown in Table 1.

Table 1. Stable isotope data.

Sample # Sampled Interval del 18O del 2H Geologic Unit

Elevation (masl) Sampled
BF-14 244.48 - 242.96 —9.01 -67.2 Upper Till
BF-15 238.81 - 237.29 -8.82 -67.0 Middle Sand
BF—5 223.94 - 222.41 -8.73 -66.5 Lower Till
BF—2 215.95 - 214.43 -8.90 -67.3 Lower Till
BF-ll 205.86 - 202.81 -9.14 —67.2 Bedrock
BF—10 195.19 - 192.15 -9.25 -67.8 Bedrock

The isotopic composition of the four samples taken from drift wells is plotted in Figure
6 and is consistent with a trend observed in other samples taken from drift wells in

Michigan (Regalbuto, 1987; Long et al., 1988; Long and Larson, 1988). These samples
have been plotted against a meteoric line (del 2H = 7.5 del 18O + 12.6) for the Lake

Simcoe, Ontario area which lies approximately 300 km east of the study area

19

—30~

AAAAAGenesee County (this study)

....- tnghom County (Long and Larson, 1988)
*H'H' Leelonou County (Regalbuto, 1987)

00000 Bay County (Long et al., 1988)

 

 

I T I

 

I I I I I 1
-20 —18 —16 -14 —12 —1O —8 -6

del 18 O (o/oo)

Figure 6. Stable isotope composition of water sampled from drift wells in Michigan.
Solid line- Lake Simcoe meteoric line (Desaulniers et al., 1981).

20
(Desaulniers et al., 1981).

The results indicate no significant isotopic trends with depth. Unlike 60 km to the
north in Bay County (Long et al., 1988) and 150 km to the east in southwestern
Ontario (Desaulniers et al., 1981) where very negative values in thick clay sequences
have been used to suggest the presence of water recharged when the climate was cooler
during the late Pleistocene and early Holocene, no such extremely negative values have
been measured in ground-water samples from the facility. Therefore, the stable isotopic
contents of the samples suggest that the ground water beneath the facility was recharged

during a climate similar to that of present and during the relatively recent geologic past.

TRITIUM RESULTS

Large amounts of tritium (H), a radionuclide of hydrogen with a half-life of 12.43
years (Mann et al., 1982), were introduced into the hydrologic cycle from 1953 to 1962
as a result of atmospheric nuclear weapons testing. Prior to 1953, the natural tritium
concentration Of precipitation was 3 TU, where 1 TU is equal to 1 tritium atom per
1018 hydrogen atoms (Robertson and Cherry, in press). Post-1953 precipitation is
characterized by tritium concentrations which are several orders Of magnitude higher
than in pre-1953 precipitation. Therefore, any water which entered the ground prior to
1953 cannot have tritium contents in excess of 0.45 TU. For a review of tritium
systematics, refer to Appendix F.

The results of the tritium analyses from the facility are shown in Table 2. Any
sample whose tritium concentration is 0.45 TU above its counting error, expressed in

TU, contains "bomb" tritium and can be considered to have recharged after 1953.

21

016606 60061166 6680115

566616 6 10 II- 10

510

66-12 6.66 6.60
66-13 16.61 6.00
66-15 (6 6.66
66-22 6.65 6.00
66-25 (6 6.06
66-21 16.60 6.00
66-26 (6 6.00
66-36 16.61 6.60
66-35 66.16 6.06
66-36 22.31 6.00
66-31 21.61 6.00
66-36 11.63 6.00

66610660 00661166 6656116

66068 666
866666 6 CPI 066 06 09 Is 191 668 066: 10 +/- 10
61-11 116.21 116.56 1.22 1.22 13.15 .66 3.66 551.35 5.65
66-6 3.10 .01 .26 .22 13.66 .66 .60 .35 1.26
66-16 3.65 .62 .26 .21 13.52 .66 .60 .10 1.26
66-16 3.16 .13 .26 .22 12.63 .66 .00 .62 1.22
66-20 6.21 .66 .21 .23 13.16 .66 .02 3.12 1.32
66-21 3.16 .11 .26 .22 13.62 .66 .00 .56 1.32
66-26 3.66 .06 .26 .21 13.65 .66 .60 .31 1.31
66-31 6.63 1.66 .26 .26 12.62 .66 .63 6.56 1.26
66-33 3.60 .21 .26 .22 13.26 .66 .01 1.33 1.26
66-66 6.21 .66 .21 .23 13.56 .66 .62 3.21 1.35
61-12 66.61 61.11 1.10 1.66 16.52 .66 6.60 562.62 6.10
66-3 3.65 .15 .26 .21 12.65 .62 .61 .16 1.26
66-6 3.66 .36 .26 .22 16.66 .63 .01 1.56 1.36
66-6 3.11 .21 .26 .22 13.60 .63 .01 1.16 1.33
66-11 3.56 .06 .25 .21 13.65 .63 .60 .66 1.26
66-16 6.51 1.61 .26 .26 15.16 .63 .06 5.61 1.51
66-23 3.10 .20 .26 .22 15.62 .66 .61 1.16 1.51
66-26 3.11 .21 .26 .22 16.35 .63 .01 1.61 1.36
66-32 3.11 .21 .26 .22 15.20 .66 .01 1.53 1.65
66-36 3.11 .21 .26 .22 16.26 '.63 .61 1.16 1.31
66-61 6.11 1.21 .26 .26 15.36 .66 .65 6.62 1.56
61-13 66.13 65.23 1.12 1.11 15.36 .62 3.66 556.22 6.66
66-1 3.11 .21 .26 .22 15.62 .62 .01 1.66 1.55
66-2 3.11 .21 .26 .22 16.56 .62 .61 1.51 1.63
66-5 6.65 1.15 .26 .26 15.16 .62 .05 6.65 1.61
66-1 3.16 .20 .26 .22 13.16 .62 .01 1.65 1.36
66-16 3.11 .21 .26 .22 15.11 .62 .01 1.22 1.61
66-11 3.12 .22 .26 .22 13.36 .61 .61 1.16 1.32
66-26 3.63 .13 .25 .21 16.61 .63 .61 .61 1.51
66-36 6.21 .11 .21 .23 13.65 .61 .03 6.02 1.36
66-62 6.36 .66 .21 .23 12.11 .61 .03 6.36 1.36

Table 2. Tritium data.

22

A plot of tritium concentration versus depth indicates that, with the exception of
three samples from depths of 15, 26, and 45 m, bomb tritium has not penetrated deeper
than 4 to 8 m below the ground surface and is therefore confined to the upper till unit
(Figure 7). However, within the upper till unit, the concentration of tritium at any
particular depth is not the Same and indicates that tritium is not moving downward along
a front. For example, a plot of the lateral distribution Of tritium within the sand lenses
of the upper till unit shows that in the central part Of the study area, where water level
readings suggest that the sand lenses may be interconnected, tritium is not evenly
distributed (Figure 17). Pre-bomb or bomb tritium is also probably present in the
middle sand unit at a depth of 17.5 In, in the lower till unit at a depth of 33 In, and in
the bedrock at a depth of 55 at. These seemingly anomalous values could be attributed

to fracture flow, leakage down the annular space of the casing, or analytical error.

DISCUSSION

In order to evaluate the effect of hydrodynamic dispersion (D) on the movement Of
bomb tritium through the upper till unit at the BFLIF, mathematical simulations were
generated which take into account advection, dispersion, and diffusion. Implicit in these
simulations is the assumption that ground-water flow through the upper till unit is
vertical and downward. This is supported by the flow model already discussed. The
middle sand, lower till, and bedrock were not included in the simulations because flow
velocities and directions in those units are different from those in the upper till. The
simulations are based on a computer code (Javandel et al., 1984) that generates

analytical solutions to the following one-dimensional solute transport equation:

dZC de dC
D ........ - ---..-- - LC =
dxz dx dt
D = av + D.

where x is the vertical and downward distance of travel over time t, v is the average
linear ground-water velocity in the x direction, C is the concentration of tritium (solute),
L is the radioactive decay constant for tritium, and a and D. are the coefficients of
dispersivity of the aquifer material and effective molecular diffusion for tritium in the
saturated porous medium, respectively. The method of superposition (Egboka et al.,
1983) was applied to the simulations to account for the temporal variability of bomb
tritium entering the till over a period of 34 years. The tritium input function used was
generated from the tritium concentration in recharging precipitation for Chicago, Illinois
(IAEA, 1969, 1970, 1971, 1973, 1975, 1979, 1983, 1986).

Figure 7 shows the simulated and actual tritium concentrations for the upper till
unit at the facility. The simulations were obtained by applying dispersivity values of
0.025, 0.25, and 2.5 m while maintaining a constant value of 2.1 x 10'9 m/s for velocity
as calculated from laboratory permeability values for the till and vertical hydraulic
gradients from the flow model. The radioactive decay constant value is 0.000153. A
value of 5 x 10'11 mzls was used for the difl‘usion coefficient and is based on reported
values for diffusion of conservative solutes in clayey materials (Freeze and Cherry, 1979;
Grisak et al., 1980; Gillham et al., 1984).

The results Of the simulations indicate that no single dispersivity value for the upper

24

 

 

 

0 30 60 90 120
O I 1 “1 I I 1 I I l I I l
1 \
\
\
2 ‘~
1 "l 1 \
t \
D ‘1 I
2 D .‘ ,’
—I '6 I '
A I, l”’
E '. o ’7'"
3“ I,"
v ./BB ”l
I ,’
1'— ], ,' a
0.. 4 I ,1
LIJ , r’
C) ,I o
c f,
5 : . l,’
‘6
'. ,I' ..... ENRICHED 9.5 +/- 1.4 TU)
6 '6 count: UNENRICHE (6 +/- 8 TD)
, . ’6 O a 0.025 m
I -- - a a 0.25 In
'3. ..... (I E 265 m
7 -" I

Figure 7. Simulated and observed tritium values in the upper till unit at the facility.

25
till unit can adequately account for the distribution of tritium. For example, when
applying a dispersivity of 2.5 m, few measured concentrations correspond to predicted
concentrations and most are significantly greater or less than predicted. Likewise, when
a dispersivity of 0.25 In is applied, most measured concentrations are significantly greater
or less than predicted.

The simulation generated using a dispersivity value of 0.025 III also does not explain
all the tritium data in Figure 7, but it does explain the presence of non-tritiated water at
a depth of 3.8 m, especially since the solute transport model assumes matrix flow. The
presence of tritiated water below depths predicted can likewise be explained by fracture
flow. Such a system has been presented by Williams and Farvolden (1969), Grisak and
Cherry (1975) and Hendry (1983a). Also, Grisak and Pickens (1980) have
mathematically simulated solute transport in such a system and have demonstrated that
diffusive transport is usually dominant in the unfractured matrix of till and advective-
dispersive transport is dominant in the fractures.

Although a one-dimensional solute transport model like the one applied in this
study does not directly account for fracture flow, its application is justified by its
usefulness as a preliminary indicator of the presence of a non-matrix flow system and by
the minimum amount of data required. However, to accurately simulate the movement
of solutes through deposits such as fractured tills, especially if tills are to be considered
for containment of hazardous or low-level radioactive waste, solute transport models like
the one developed by Grisak and Pickens (1980) should be applied. Such models
normally require information about fracture aperture, spacing, and flow velocity, as well

as diffusion and distribution coefficients, and porosity for the matrix material. It is

26
probable, however, that the acquisition of such information would be considerably less

difficult than the callhration Of such models against field data.

CONCLUSION

The concentration of tritium in 40 ground-water samples collected from two till
units and underlying bedrock in southeastern Michigan was found to range from 0 to 44
TU. Bomb tritium was identified in samples at depths of 1.5 to 42 m below the water
table while pre-bomb tritium was identified in samples from depths of 5.0 to 54 m. The
distribution of tritium in the ground-water reservoir of the upper till unit was simulated
using a one-dimensional solute transport model which assumes matrix flow. Based on
the shallowest occurrence of non-tritiated water, the dispersivity value of the matrix flow
system in the upper till unit was found be less than 0.025 In. Bomb tritium penetration
to depths beyond those predicted by the model is explained by the presence of a
fracture flow system in the tills.

The results of this study demonstrate that the tritium distribution observed in
Michigan tills is similar to that observed in tills in the Great Plains region and that
further study Of these deposits is required if they are to be used for containment of low-

level radioactive or hazardous waste.

APPENDICES

APPENDIX A

REGIONAL GEOLOGIC ANALYSIS

More than 70 bedrock water well records within a 3.5 km radius of the study area
were examined to define regional trends in the bedrock surface, bedrock potentiometric
surface, drift thickness, bedrock surface composition, and height of potentiometric
surface above bedrock surface.

Local bedrock was found to lie between 200 and 220 m in elevation and contain .
slopes of 12 to 20 m per km. However, no regional trends in bedrock surface are
observed (Figure 8).

Approximately 60% of the wells indicate shale as the uppermost bedrock unit while
the remaining 40% of the wells encountered sandstone as the uppermost bedrock
formation (Figure 9). Although sandstone is encountered more toward the eastern and
southeastern parts of the study area, it would be speculative to draw a contact based on
this data because the Paleozoic formations are rather variable in their composition and
irregular bedrock topography could explain some of the apparent lateral variability. The
possibility Of a contact in this area, however, should not be ruled out completely because
regional maps indicate that a Saginaw-Marshall contact runs beneath southeastern

Gaines Township. Apparently, the Michigan Formation is not represented in this area.

27

28

Local drift thickness ranges from more than 50 m in the southwest to less than 30
m in the northeast (Figure 10). Although drift thickness is largely a function of bedrock
topography when dealing with a planar ground surface, a gradual increase in ground
surface elevation to the south appears to account for the increasing drift thickness in
that direction.

The local drift consists predominantly of clay with subequal amounts Of silt, sand,
and gravel. Most well records indicate the presence Of 3 to 5 m of brown clay at the
surface underlain by thick gray clay with occasional sand and gravel lenses. The brown
and gray clay probably represent zones of oxidized and unoxidized material, repsectively
(Williams and Farvolden, 1969; Grisak et al., 1976; Desaulniers et al., 1981; Hendry,
1982; Keller et al., 1986; Cravens and Ruedisili, 1987).

A north-south oriented surface water divide, located approximately 2 km southwest
of the study area, separates the Flint River drainage basin to the east from the
Shiawassee River drainage basin to the west. Locally, Kimball Drain removes any water
east of the divide and Cargill Creek drains areas west of the divide. Drainage in the
area is improved through extensive use of tile fields and drains.

Local ground water supplies include thin, shallow (10-20 In deep) sand and gravel
layers within the thick clay sequences. These sand and gravel bodies may be related to
the middle sand unit in the study area. These aquifers are typically confined and
produce less than 10'3 m3/s. The Saginaw Formation is the principal aquifer in the
bedrock, producing water from fractured interbedded shale-sandstone sequences at the
bedrock surface. The confined conditions of the Saginaw are attested to by the level Of

the water above bedrock surface which ranges from 50 m in the south to less than 30 m

29
in the north (Figure 11). The bedrock potentiometric surface slopes to the north with a

horizontal hydraulic gradient of 0.005 (Figure 12).

 

 

Figure 8. Regional bedrock surface contour map (contour interval =====

 

 

 

 

 

 

 

 

31

 

 

 

. 3
o 0.0 0 o
c
O
. o
o
o
co 0 o 0 c 0. 0.51m
41‘ 9 0° 0
o
o ’ e
O
l- . ’
o
o o
o
. o o
O O
00
. O
o
. O
O O
'0
O
4 4 e e 0 c

 

 

 

Figure 9. Regional bedrock surface lithology map (filled Circles- shale, open circles-

sandstone).

 

 

 

 

 

 

 

 

 

 

 

Study area

 

 

 

 

 

 

 

 

 

Figure 12. Regional bedrock potentiometric surface map (contour interval = 2 m).

APPENDIX B

CROSS SECTIONS THROUGH THE STUDY AREA

35

36

53.53 can chew a. 5.5m on. o. cone—2
E 8289. 3.08 Lo .5330. can BEES £03 mo c0330. 9:30;» 32 .2 oSwE

 

 

 

 

 

 

 

_
mam
— q a
SOON cow 0
e at;
m
mun \
.mo
o8
_
o a.
z“
.< 1111 9...
:o
On
a. one A... o 7
one \ v

ELEVATION (m.a.s.l.)

1>

-SO27

250*

SB 26

245‘

240*

 

 

235T

230q

 

225J

WW

 

 

$826

\

 

Figure 14. Cross sections A-A’ and B-B’.

 

~8855

 

 

  
  

 

37

 

B

Upper Till

5 Middle Sand

Lower Till

 

 

  

 

 

 

 

 

 

 

 

E] Till

sand/gravei

O 100 m
L._.___.J

 

 

 

Sand/gravel

 

252 -
250-!

3 2484

g

242-

ELEVATION (ma
to
£1

 

250-

2457

240—

.s.l.)

a
?

ELEVATION (m.
N N
<?" ‘53

g

215‘

 

210-

38

-SB 29
8 67
\MW 22

 

 

 

 

..............................

5255525231551" """ ‘ Upper Ti"

 

O SJOm

 

    

 

 

I: Till

Sand/gravel

 

 

 

 

Upper Till

   

 

 

 

 

 

 

Lower Till

 

 

0 100m
L___.l

Figure 15. Cross sections CC and D-D’.

   

 

[:1 run

 

Sand/grave'

APPENDIX C

SIMULATED HYDRAULIC HEAD DISTRIBUTIONS

39

 

Elevatlon (m.a.s.l.)

 

 

 

 

 

 

\..--

\

Elevatlon (m.a.s.l.)

c........~

 

 

 

 

 

 

 
    

 

 

 

 

250'
4.V..,.‘;;;v...;:,._;;;_. ._ -__._' “egg..- _ ~_ _ _ _ __ .—
2 240‘ “_—-—2‘6 _—.-‘~-¢-¢w-uw~‘

O _ ‘ fl ‘ p I .\
Q .i .. . 7 ”i“
q ~ \ z " " " — l
E 2% ‘I_ ‘ I”

v ’
‘. ,b’
§ ‘~ 9
., 220- -.
, -
g E III I”
l
h I
a 210- : .. .......... 5 ......
\ »l‘ A.’
200-
~

Figure 16. Figure showing the effect of varying lithologies on simulated head
distribution along cross section X-X’ in Figure 4 (contour interval = 0.5 m).

APPENDIX D

LATERAL TRITIUM DISTRIBUTION

41

42

ON 0

ADP E 82$, :3 an: :3 Ban: 9: £53» 635.: we :ozzntfin 38:5 .5 22mm

 

 

 

T _
EOON OO_.
O.
Q; to. Qme 2.0 «N.
O to O.
. e
m F. 0. mp. O
«.0 e O 0 me.
06. méo «an

 

fie

APPENDIX E

SITE SOIL TYPES

The soils at the BFLIF are representative of the Del Rey-Lenawee association and
are level to gently sloping, somewhat poorly drained and poorly drained silt loams and
silty clay loams having a silty clay loam substrate (Holcomb, 1972). Soils of the
Metamora series and the Selfridge series are also represented at the site. The
Metamora series consists of nearly level to gently sloping soils formed in sandy and
loamy deposits underlain by light sandy clay loam. The Selfridge series consists of
somewhat poorly drained, nearly level to gently sloping loamy sands having a silty clay
loam substrate. The characteristics of each soil type, as described in the Genesee
County soil survey of 1972, are summarized below (Holcomb, 1972).

The Del Rey silt loam is the dominant soil in the undulating areas of the heavier
textured till plains and borders of lake plains. Typical areas have short, complex slopes,
wet depressions, and random drainage ways that empty into well-defined drainage
channels. These soils have a moderate shrink-swell potential, and grain size analysis
indicates that 55-90 percent pass sieve No. 200. As these soils are nearly level to gently
sloping, runoff is slow or very slow and some depressions are ponded. Permeability is

moderately slow to slow. Del Rey soils are found in the eastern half of and along the

43

44

southern border of the site, as well as in the extreme northwest corner of the site.

Lenawee silty clay loam is extensive on lake plains but may be found on till plains.
It is developed in drainage channels within areas of undulating soils and is generally
associated with the higher, somewhat better drained Del Rey soils. The Lenawee soils
have moderate shrink-swell potential, and 85-90 percent pass sieve No. 200. As these
soils are level or nearly level, runoff is very slow or ponded, and the high content of
fine silt and clay results in moderately slow or slow permeability rates. These soils occur
in isolated bands in the western, northern, and southeastern parts of the site.

The Metamora sandy loam occurs on till plains and outwash plains and occupies
low, smooth ridges and knolls and short slopes of drainageways. The shrink-swell
potential of these soils is low to moderate, and, depending upon the horizon analyzed,
20-90 percent pass sieve No. 200. Permeability is moderately rapid in the upper layers
but moderate to moderately slow in the deeper, finer textured material. Metamora soils
occupy a small area in the northwest quarter of the site.

Selfridge sandy loam is developed in outwash and may be transitional between the
till plains and lake plains. The soil occurs on hills, knolls, and short slopes bordering
intermittent upland drainageways. The shrink-swell potential of these soils is low to
moderate, and, depending upon the horizon analyzed, 15-90 percent pass sieve No. 200.
These sandy, nearly level or gently sloping soils have slow or very slow runoff and
readily absorb most of the rainfall. The upper sandy layers have moderately rapid
permeability, but the finer textured underlying material has moderately slow permeability.
The Selfridge soils are found in the northeast corner of the site.

A zone of brown and gray mottles at various depths in the soil profile records the

45
fluctuation in water level. This mottled zone is typically underlain by a brown zone
which marks the upper and lower limits of oxidizing conditions below the water table.
Beneath the brown zone, reducing conditions below the water table are indicated by a
gray zone. Although the mottled zone is usually within a meter or two of the ground
surface, the contact between the base of the brown zone and the underlying gray zone

has been recorded as deep as 5 m in some soil borings.

APPENDIX F

TRITIUM SYSTEMATICS

A variety of tracers can be used to determine rates of water movement in geologic
systems. Traditionally, tracers have been categorized according to detection means.
Dyes are detected by color, electrolytes are detected by conductivity, and radioisotopes
are detected by nuclear radiations. Ideally, a tracer should accurately describe water
flow for the system under study. To this end, the tracer should not appreciably change
the hydraulic or chemical properties of the water, and it should not alter the
transmission characteristics of the medium through which the water is flowing.

Hydrogen isotopes meet these requirements because they are part of the water
molecule, and therefore, do not appreciably change the properties of water or the
porous medium. The additional advantage of tritium is its radioactivity which can, with
proper instrumentation, be detected in minute quantities. Tritium occurs naturally in the
water cycle and is continuously, in time and space, added to the ground-water system.

The natural production of tritium takes place primarily in the atmosphere, but small
amounts are produced in the lithosphere and hydrosphere. Solar radiation or reaction
of cosmic rays with other gasses in the upper atmosphere produces neutrons which

cosmically radiate 14N to produce 12C and 3H (tritium). The rate of production of

46

47

tritium is greatest at a height of approximately 10 km and for neutrons of 15-20 MeV,
and its mean lifetime in the stratosphere is about 2 years (Vinogradov et al., 1968).

The tritium so formed combines readily with oxygen atoms in the stratosphere and mixes
with molecular water in the atmosphere and on the surface of the earth. Analysis of
natural water for tritium is based on detection of the beta radiation which accompanies
the radioactive decay of 3H (tritium) to 3He (helium). The utilization of tritium as a
tracer assumes that it is being produced at a constant rate in the upper atmosphere and
that it has reached a state of equilibrium between production and decay.

Inception of atmospheric nuclear weapons testing in the early 1950’s served to
introduce large quantities of artifical tritium into the hydrologic cycle. It is estimated
that prior to 1953, cosmic-ray produced tritium concentrations were in the 5-10 TU
range (Payne, 1972), although more recent studies suggest that this value may have been
as low as 3 TU (Robertson and Cherry, in press). A water of concentration 1 TU has 1
tritium atom for every 1018 hydrogen atoms. The tritium unit is useful because it
represents the order of magnitude of tritium in natural waters prior to 1953, and it is
the approximate limit of activity detectable using present day counting systems.

Artificial tritium input into the hydrologic cycle continued for about 10 years during
which time tritium concentrations in precipitation increased to levels which were as
much as four orders of magnitude greater than pre-1953 levels. It is estimated that
during the 10-year period prior to the test ban treaty of 1962, approximately 200 kg of
artificial tritium were added to the planet. Calculations based upon formation rates of
natural tritium suggest that the planet contained less than 2 kg of tritium prior to 1953

(Vinogradov et al., 1968).

48

The use of tritium in hydrologic investigations was first proposed in 1953, and
shortly thereafter came the realization that tritium could be used to determine the age
of ground water relative to 1953 (Libby, 1953). Waters which enetered the ground after
1953 have tritium concentrations which may be several orders of magnitude higher than
waters which entered the ground prior to 1953. Since the half-life of tritium is 12.43
years (Mann et al., 1982), and almost three half-lives have elapsed since 1953, the
waters which entered the ground before 1953 should presently have tritium
concentrations of less than 0.4 TU. This pre-1953 water is commonly referred to as
"dead" or "pre-bomb" water whereas post-1953 water is referred to as "tritiated” or
"bomb" water.

Tritium has been applied as a tracer for calculating circulation and residence times
(Begemann and Libby, 1957; Brown, 1961; Dincer and Payne, 1971; Hendry et al.,
1983b), ground-water age (von Buttlar and Wendt, 1958; von Buttlar, 1959; Nir, 1964),
storage (Eriksson, 1958), recharge rates and flow velocities in coarse-grained materials
(Allison and Hughes, 1972; Allison and Holmes, 1973; Atakan et al., 1974; Dincer et al.,
1974; Vogel et al., 1974; Allison and Hughes, 1975; Rehm et al., 1982; Andres and
Egger, 1985; Knott and Olimpio, 1986; Delcore and Larson, 1987), recharge rates and
flow velocitites in fine-grained materials (Hendry, 1982; Hendry, 1983a; Bradbury, 1984;
Keller et al., 1986; Cravens and Ruedisili, 1987; Hendry, 1988), diffusion rates in
fractured materials (Foster, 1975), and dispersion in coarse-grained materials (Rabinowitz

et al., 1977; Egboka, 1983; Larson et al., 1987; Robertson and Cherry, in press).

APPENDIX G

SAMPLE PREPARATION FOR TRITIUM ANALYSIS

PRE-DISTILLATION

The intial step in the analysis of 3H consists of the complete distillation of the
sample to remove all dissolved solids and suspended particles. The system used consists
of a boiling flask, Kjeidahl bulb, Allihn condenser, and receiving flask (refer to Delcore,
1985 or Regalbuto, 1987 for an illustration of the apparatus). The system is first dried
under vacuum, and once drying is complete, dry air is admitted to the system.
Approximately 70 g of glass beads are placed in the boiling flask to insure more even
boiling of the sample, and afterwards, 225 ml of undistilled sample are poured into the
flask. Then, 0.25 g of copper powder are added to the sample to aid in retention of
sulfur, the radioisotope of which is a beta-emitter and whose energy emission spectrum
overlaps that of tritium. Vacuum is reapplied to the system. and removed when bubbling
of the sample, due to degassing, stops. Dry air is readmitted to the system, the line is
connected to a pressure compensator, and heat is applied to begin distillation. If, during
distillation, the pressure compensator indicates excess pressure, vacuum is applied
momentarily. When the level of the sample in the flask reaches the top of the glass

beads, the heat is removed to allow any water in the Kjeidahl bulb to drain. The

49

50
receiving flask, with the distilled sample, is removed from the apparatus and sealed from
the atmosphere. To prevent any possibility of sample cross-contamination, a heat source
is lightly applied to the Kjeidahl bulb and connecting glass, and air is pumped through
the condensing tube until the system is completely dry.

For direct counting, 10 ml of this pre-distilled sample are pipetted into a plastic
liquid scintillation vial to which 13 ml of scintillation cocktail are added, and the vial is
placed in the dark for one to two hours before counting. Direct counting is done in
groups of twelve samples and consists of one standard (40 DPM/ml), two dead water

samples to monitor background, and nine unknown samples.

ELECTROLYTIC ENRICHMENT

Enrichment of the samples is necessary because specific tritium activities in natural
waters are very low. Electrolytic enrichment is used because it has been proven to yield
reproducible results and is a simple process in which large numbers of samples can be
enriched at the same time (Florkowski, 1981).

The electrolysis system consists of: gas exhaust lines, a low-temperature bath,
Ostlund electrolysis cells, electrodes, and a power supply. Electrolysis is simultaneously
performed on twelve samples of which ten are unknown samples, one is a standard (4
DPM/ml), and one is a sample of dead water. The dead water is obtained from a 600
m deep well at a fish hatchery in Thompson, Michigan, and a supply of the water is
distilled in the tritium laboratory prior to use with electrolysis. A 160 ml quantity of
distilled sample water is transferred to an Ostlund cell, and to this, a 5 ml aliquot of

electrolyte is added. The electrolyte used is 4 molar NaOH solution with dead, distilled

51
water as the solvent. The cells are set in the water and glycol bath, and the electrodes
are put into the cells and connected to the electrical wires. The iron electrode is the
cathode and the nickel electrode is the anode. The cells are capped, connected to the
exhaust fan by way of gas exhaust lines, and allowed to equilibrate with the temperature
of the bath for about 15 minutes. The electrolysis process is most efficient at low
temperatures, and therefore, the water bath and cells are kept at 1 degree Centigrade
(Hoffman and Stewart, 1966). To insure that an equal amount of current will pass
through each cell, the electrodes are connected in series to the power supply unit.

The current used is calculated based on a total electrode area of 34 square
centimeters and a maximum current density of 0.09 amps per square centimeter (Ostlund
and Werner, 1962). Since electrolysis efficiency decreases as the electrolysis temperature
increases, and since heat dissipation is a function of the water level inside the cell, it is
necessary, as the volume of sample in the cell decreases, to limit the amount of heat
produced by electrolysis. Therefore, the current supplied to the electrodes is decreased
as the volume of sample in the cells decreases, and a circulation pump is used in the
bath to provide circulation of the coolant. When the volume of the samples in the cells
has been reduced to between 11 and 13 ml, the power unit is shut off, the electrical and
exhaust connections are removed, and, to minimize atmospheric contamination, capped

distillation heads are substituted for the electrolysis heads.

POST-DISTILLATION
Upon completion of electrolysis, 11 to 13 ml of enriched sample usually remain in

the cell. This water must be quantitatively removed from the cell and separated from

52
the NaOH. These two steps are accomplished by freeze-up distillation. The post-
distillation line consists of Ostlund cells, modified distillation heads, weighing bulbs,
liquid nitrogen, pressure compensator (balloon), vacuum, and dry air (Delcore,1985;
Regalbuto, 1987). The weighing bottle, modified distillation head, and balloon are
weighed before and after each post-distillation.

The Ostlund cells are removed from the cooling bath and connected to the
weighing bulbs via the modified distillation heads. The balloon is mounted onto the
stem of the distillation head and secured with a rubber band. Vacuum is then applied.
and if the system appears vacuum-tight, the vacuum is shut off at the stopcock. The
weighing bottle is submerged in the liquid nitrogen to a level approximately 2 cm below
the point at which the bottle neck begins to widen. One flexible heating coil is wrapped
around the lower part of the cell and heated to 60 degrees Centigrade while the other
heating coil is wrapped around the upper part of the cell and the distillation head and
heated to 100 degrees Centigrade. During heating, the point at which the bottle neck
begins to widen should be the transition zone between the hot bottle neck above and
the cold bulb below; some adjustment of the liquid nitrogen level will probably be
necessary. Heating is continued for an additional 30 minutes beyond apparent dryness,
during which period it will probably become necessary to submerge the weighing bottle
another 1-2 cm into the liquid nitrogen.

When heating has been completed, the weighing bottle is weighed to determine the
exact amount of sample in it. Lastly, 10 ml of the enriched, post-distilled sample are
pipetted into a plastic liquid scintillation vial in preparation for liquid scintillation

counting.

APPENDIX H

LIQUID SCINTILLATION ANALYSIS

INTRODUCTION

The concentration of natural tritium in ground water samples is difficult to
determine because it is often low (less than 1 TU) and because the energy of the beta
emission spectrum of tritium is low (less than 18.6 keV). Traditionally, internal gas or
liquid scintillation counters have been used to measure the low concentrations.
Although internal gas counting yields low background counting rates and is very sensitive
while not requiring enrichment, conversion of the sample to hydrogen gas and
destruction of the sample itself are just two of the disadvantages with this generally time
consuming and complicated procedure. Liquid scintillation counting requires that the
water sample be mixed with a liquid scintillation solution (cocktail), and, because this
procedure is known to yield reproducible results while being far simpler, it was chosen

as the counting technique for this study.

THEORY OF LIQUID SCINTILLATION COUNTING
Liquid scintillation counting is an analytical technique which is defined by the

incorporation of a radionuclide into uniform distribution with a liquid chemical medium

53

54
capable of converting the kinetic energy of nuclear emissions into emitted photons,
(Packard, 1987). The rate that these light scintillations are produced within a finite time
period is proportional to the rate of radioactive decay, and the number of photons or
intensity of the light scintillations is proportional to the energy of the decaying beta
particle (Beckman, 1976).

Radioactive decay of an element is the initial step in the scintillation process and is
the result of an imbalance between the nuclear components of the atom, the proton and
neutron. This imbalance, which results in an unstable elemental condition, eventually
yields to rearrangement and subsequent energy release in the form of radiation. For
tritium, the radiation emitted consists of a beta particle, which is an electron originating
in the nucleus, and a neutrino. The energy of the radiation is characteristic of the
radionuclide and, in the case of tritium, it is 18.6 keV. This energy is shared between
the beta particle and the neutrino and, in any particular radiation, the beta particle can
contain all, some, or none of this energy (Packard, 1987).

The second step in the scintillation process entails efficient transfer of the beta
particle energy to the cocktail. This requires a cocktail that consists of a solvent and
solute and acts as a solvent for the sample material. While solvent molecules, in an
excited state, are not readily recognizable, they are, nontheless, capable of transferring
energy from one another and to the solute molecules. This transfer of kinetic energy to
the solute results in flourescence of the solute and emission of light energy as photons
(Packard, 1987).

The final step in liquid scintillation analysis involves the detection of the emitted

photons and the conversion of light energy into an electrical signal. This conversion can

55
be achieved with a device known as a photomultiplier tube (PMT), and a pair of these,
one on either side of the sample, is usually employed in a scintillation counter. If both
PMTs record an event over a very short time interval, the machine circuitry is instructed
to record that event as a true count and to convert the energy of the event to a digital
value. As a result of each PMT producing a signal from the same event, the
geometrical location of the event within the vial becomes less significant (Packard,

1987)

FACTORS AFFECTING COUNTING EFFICIENCY

Liquid scintillation counting of tritium decay events is an inefficient process and the
result of various inherent interferences. These interferences include other solutes within
the sample or cocktail, photoluminescence, vial material, reaction of the cocktail with
the vial material (wall effect), type and volume of cocktail, and background radiation.

The most important type of interference internal to the sample is quenching, and it
can result in a loss of counting efficiency exceeding 50 percent and a shift of the energy
spectrum toward lower energies. The three types of quenching, caused predominantly
by the cocktail, include photon quenching, in which there is incomplete transfer of beta
particle energy to solvent molecules, chemical quenching, in which energy is lost as it is
transferred from solvent to solute, and color quenching, in which light photons are
attenuated in the solution (Packard, 1987). The energy loss in chemical quenching takes
place when impurities quench the sample by either competing with the solute for energy
transfer or by chemically making the solute less reactive to energy transfer (Beckman,

1976). Most sample material causes little or no quenching as any quenching agents in

56
the sample are removed by distillation prior to counting; however, in solute-rich samples
(e.g. brines), some quenching agents may pass through the distillation process. These
effects are generally insignificant when compared with quenching caused by the cocktail.

Photoluminescence occurs when the cocktail is activated by sunlight or flourescent
lights in the laboratory. The decay of photoluminescence appears to be a function of
the severity of exposure but is usually quite rapid (less than 0.1 hr) (Packard, 1987).
This type of interference can best be avoided by storing the samples in dark areas,
loading them into the machine while the room is dark, and keeping the smoked glass
cover on the machine closed at all times. This phenomennon was recorded on several
occasions with blank samples which, having been exposed to sunlight during transport,
yielded abnormally high initial readings followed by gradually decreasing readings to a
stable level.

The influence of the vial composition on the overall background counting rate was
determined by measuring background with polyethylene and borosilicate vials that
contained dead water. The findings indicated that background readings were higher with
the borosilicate vials, and therefore, polyethylene vials were chosen for this study.

When polyethylene scintillation vials are used, the counting accuracy of the sample
is affected by the reaction of the cocktail with the vial material. Long-term background
measurements using blank water tend to increase with time as the cocktail reacts with
the vial material, but the same trend is not observed with borosilicate vials. The smell
of cocktail emmanating from the plastic vials after one or two days suggests that the vial
walls are not impermeable to fluid and/or vapor movement. Corroborating these

findings is the observation that sample vials, which have been sitting on the shelf for

57
several months, have lost about one-half of their contents, which, incidently, is
approximately equal to the amount of cocktail added to each vial. Borosilicate vials
which, in some cases, were more than one year old, did not show any appreciable loss of
contents. The counting efficiency of a sample is also a function of the cocktail type
because all cocktails do not use the same solvent and/or solute. Currently, Insta-Gel and
Pica-Flour LLT are the two most commonly used cocktails for counting water of low
tritium content. Experimentation with the two cocktails indicated that Pica-Flour LLT
yields higher efficiencies and lower background values and was, therefore, chosen for
this study.

Significant quenching of the sample can also be brought about by varying the
sample-to-cocktail ratio in the vial. Excess cocktail serves to reduce efficiency by
providing excess solvent molecules which may absorb beta energy but fail to transfer it
to the solute molecules. Insufficient volumes of cocktail do not provide enough solvent
molecules to transfer the beta energy to the solute molecules. There should exist,
therefore, an optimum sample-to-cocktail ratio that maximizes counting efficiency and
minimizes lower limit of detection. Experimentation with sample-to-cocktail ratios
ranging from 0.3 to 3.0 indicated that a ratio of 0.75 yields excellent efficiency while
preserving a low limit of detection. Since the scintillation vials used for this study had a
capacity of 23 ml, addition of 10 ml sample and 13 ml cocktail would result in a ratio of
0.77; therefore, this ratio was selected.

External to the sample, fluctuating background radiation is the most important
interference because measurement of environmental samples for tritium typically yields

values which are slightly above background radiation levels. In this study, background

58
radiation generally comprised more than 75 percent of the total sample counts amassed.
For this reason, it is very important to minimize all of the possible interferences internal
to the sample.

In order to gain a better understanding of background radiation fluctuations, several
long-term measurements of background radiation were made. These indicate that
background may vary by as much as 25 percent. Daily cyclical patterns were observed in
background levels, but, contrary to conventional thinking, the highest background levels.
were not always recorded during the day. These variations could also be a function of
the daytime activity in the laboratory where the scintillation counter is located. It has
been reported that longer term fluctuations in background are indeed real and related
to seasonal variation and sunspot activity (Hut, 1986). The intercomparison study also
indicates that at most of the laboratories, scintillation counters are placed in the

basement so as to minimize background radiation even more.

QUENCH MONITORING AND CURVE GENERATION

Although all samples were distilled prior to counting, quench was initially monitored
so as to confirm the efficiency of the distillation process. The Packard Tri-Carb 1500
liquid scintillation analyzer used for this study monitors quench using the transformed
spectral index of the external standard (tSIE). After the sample vial has been lowered
into the counting chamber, an external gamma source (Barium-133) is momentarily
positioned next to the vial and an artificial excitation of the solvent molecules takes
place. The resultant Compton spectrum shows a shift toward lower energies when

compared with the spectrum of the unquenched standard, and the tSIE value is

59

calculated from this difference. Quench monitoring in this study indicated that the
distillation process is indeed efiective for reducing the amount of quenching in all
samples to a similar level.

A quench curve shows the relationship between counting efficiency and the quench
indicating parameter (tSIE) and provides information necessary to calculate the true
sample activity of the particular radionuclide. To generate a quench curve, a series of
hot standards of various quench are counted and quench is monitored. These standards
may be purchased, made in the laboratory, or quench curves may already be stored in
the memory of the counter. For this study, a set of quenched standards was made in
the laboratory because greater curve resolution was desired in the region encompassing
the quench values of the unknown samples. These quenched standards were generated
by mixing 4 drops of carbon tetrachloride from a Pasteur pipette and 52 ml of cocktail
in a beaker. After stirring, 13 ml of this quenched cocktail were pipetted into a
scintillation vial which contained 10 ml of 40 dpm/ml standard. The second standard
was made by using 11 ml of quenched cocktail arid 2 ml of unquenched cocktail. The
third, fourth, fifth, and sixth standards were mixed with 9, 7, 5, and 3 ml of quenched
cocktail, respectively and 4, 6, 8, and 10 ml of unquenched cocktail, respectively. These
six standards, a standard containing no carbon tetrachloride, and one blank sample were
then counted and provide the data for the quench curve. The data was plotted and the
equation for the quench curve was defined as the line of best fit.

The study indicated that the pre-distillation process was effective in reducing quench
in all samples to a similar level. Therefore, the counting efficiency for a known

standard was used to calculate the tritium activity of its respective batch of unknown

samples.

COUNTING WINDOW GENERATION

Although the energy distribution of tritium ranges from 0—18.6 keV, the distribution
of decay events across this range is not even. Therefore, a series of experiments was
conducted to determine the optimum channel setting that would best reproduce
standards of known concentration. A 56.6 deml standard and a blank were counted
in 0.5 keV increments, beginning at 0 keV and ending at 9.0 keV. The efficiency of
each 0.5 keV increment was calculated and the efficiency of a sum of increments could
be obtained by summing the efficiencies for those increments.

The data indicate that, although background intensity is fairly even from 0-9.0 keV,
tritium energy distribution is actually centered in the 0-6 keV range. The incremental
counts per minute (cpm) of the standard and background across this range were
summed and then each of the incremental values was normalized. The 0-0.5 keV part
of the spectrum was not used because the amount of efficiency gained by including it
would be minimal and because that part of the spectrum could be susceptible to
distortion caused by wall efiect (Packard, 1987). Three channel settings were tested:
0.6-5.4 keV, 0.8-3.2 keV, and 0.74.3 keV. The first window was selected simply because
it represents most of the usable part of the spectrum. However, the second and
narrowest window was obtained by plotting the normalized cpm for the standard and the
blank on a graph of normalized cpm versus keV. Since distribution of the background
intensity is fairly even across the 06 keV range, a line of best fit was calculated to

define the background distribution. The intersection of this line and the normalized

61
cpm curve of the standard define the upper and lower limits of the window. The third
window is simply an arithmetic intermediate between the first and second windows.

To test the performance of the various windows, the 100 T.U. standard from the
1986 International Atomic Energy Agency intercomparison study was counted. In
addition, several dilutions of the standard were made to determine the performance of
the windows at lower tritium concentrations. The results indicate that the best
approximation of the true tritium content could be obtained by averaging the values
obtained by the three windows. Rarely did all three windows yield the same results;
more typically, one window indicated a value 5-10 % too high, another indicated a value
5-10 % too low, and the remaining window indicated the true value. It is anticipated
that background fluctuations which occur in the higher energy regions of some windows

and not others, result in the inconsistency of window performance.

APPENDIX I

TRITIUM ACTIVITY CALCULATIONS

The formulae used to calculate tritium activities from counting data are described by
Wyerrnan (1978). Counting parameters which are required for calculation of

electrolytically enriched tritium standard activities include:

 

 

 

 

 

 

Initial standard volume (ml) V0
Final standard volume (ml) Vf
Initial standard activity (DPM/ml) DPM0
Gross standard counting rate (cpm) CPMsg
Background wanting rate (cpm ) CPMb
Standard counting efficiency (%) EFFs

The enrichment factor obtained for the standard is used to calculate the activities of the

unknown samples in the particular electrolysis batch.

The background counting rate was determined by counting the sample of blank
water which had been electrolyzed with each particular batch. The net standard
counting rate (CPMsn) is obtained by subtracting the background counting rate (CPMb)

from the gross standard counting rate (CPMsg),

62

63

CPMsn = CPMSB - CPMb.

The electrolytic efficiency of the standard (EE”) is determined by multiplying the
ratio of the final to the initial volumes of the standard (VWO) and the ratio of the final

to the initial activities of the standard (DPMtlDPMo),

EB” = (vgvo) * (DPMIIDPMO).

During electrolysis, gases, which bubble through the liquid and eventually escape
from the system, may carry away both water Vapor and spray. Variable vapor loss may
be the result of uneven cooling of cells in the water bath, electron density differences
due to unequal electrode areas, and electrode spacing variations (Hufen et al., 1969).

The variable vapor loss is compensated for by the electrolysis fractionation factor (B),

B = ln (Va/Vt)” / -ln E5”.

The electrolysis fractionation factor of the standard is then used to calculate the

electrolysis efficiency of the unknown samples (BEN),

EEus = (Volvf)s.(l/B)'

At this point, the gross counting rate of the unknown sample (CPMusg) can be used

to calculate its tritium content (TUus),

64

TUus = (CPMusg - CPMb) / (EBB) * (Va/v0s * v * EFFs * K,

where K = 0.0071 DPM/ml, a TU conversion factor.
Direct count sample tritium contents are calculated with the following equation:

Tu“s = (CPMusg - CPMb) / v * EFFs * K.

APPENDIX J

TWO SIGMA ERROR DETERMINATION

All samples, enriched and unenriched, were counted to a two-sigma error
(confidence level of 95.5 %). Since the gross count rate (CPMg) is a combination of
the net sample count rate (CPMn) and the background count rate (CPMb), the
uncertainty in the net sample count rate (Un) can be determined from the uncertainty
in the gross count rate (Us) and the uncertainty in the background count rate (Uh).
The following equations are adapted from Packard (1987), and uncertainties are
indicated as counts per minute.

The uncertainty in the background count rate (Uh) is defined by:

 

where 2 = sigma error desired and tb = total counting time of background, in minutes.

The uncertainty in the gross counting rate (Us) is defined by:

 

Us = 21(CPM2 * t8) / t 2,

where 2 = sigma error desired and t' = total counting time of sample, in minutes.

65

66

Therefore, the uncertainty in the net count rate (Un) is defined by:

 

Un =q(Ub)z + (Ug)2‘

Since the uncertainty in background count rate is an important component of the
net count rate uncertainty when dealing with low tritium concentrations, it is necessary
to count samples for longer periods of time to insure that background fluctuations are
smoothed. Therefore, samples were counted until the returns in certainty from
additional counting became minimal. This usually corresponded to a counting time of
480 minutes for the standard and unknown samples and 960 minutes for the blank
samples. The twelve-position sample chamber allowed one standard, two blanks, and
nine unknown samples to be counted over a four-day period. The lowest limit of
detection (LLD) was defined as the minimum tritium concentration at which the net

sample count rate (CPMn) exceeded the uncertainty in the net count rate (Un).

APPENDIX K

TRITIUM INPUT FUNCTION

67

TU

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

10000 a
i AAAAA UNCORRECTED
- ' ° ° ' ' CORRECTED
' a ‘1
10003
2 Al 0 A
: To ‘r
d ‘1 A 0 o " 4’ ‘i 1‘ A}
4
100? A A A o 0 it) a A
: 0 ‘L 0 0 A A I} 0 A A
j of 0 1’ 0 o 0 A 0 ‘f 0 0 fl
.1 0 0 0 ‘ if 79 A) A
ll
10 g
3 t
1 E
i.
0.1 j
1953 1958 1963 1968 1973 1978 1983 1988

Figure 18. Tritium input function.

BIBLIOGRAPHY

BIBLIOGRAPHY

Allison, GB, and M.W. Hughes, 1972. Comparison of recharge to groundwater under
pasture and forest using environmental tritium. Journal of Hydrology, v. 17, p.
81-95.

Allison, GB, and J.W. Holmes, 1973. The environmental tritium concentration of
underground water and its hydrological interpretation. Journal of Hydrology, v.
19, p. 131-143.

Allison, GB, and M.W. Hughes, 1975. The use of environmental tritium to estimate
recharge to a south-Australian aquifer. Journal of Hydrology, v. 26, p. 245-254.

Andres, G., and R. Egger, 1985. A new tritium interface method for determining the
recharge rate of deep groundwater in the Bavarian Molasse basin. Journal of
Hydrology, v. 82, p. 27-38.

Atakan, Y., W. Roether, K.-O. Munnich, and G. Matthess, 1974. The Sandhausen
shallow-groundwater tritium experiment. Isotope Techniques in Groundwater
Hydrology, International Atomic Energy Agency, Vienna, p. 21-43.

Begemann, F., and W.F. Libby, 1957. Continental water balance, ground water
inventory and storage times, surface ocean mixing rates and world-wide water
circulation patterns from cosmic-ray and bomb tritium. Geochemica et '
Cosmochimica Acta, v. 12, p. 277-296.

Bradbury, KR., 1984. Major ion and isotope geochemistry of ground water in clayey
till, northwestern Wisconsin, USA. Practical Applications of Ground Water
Geochemistry, edited by B. Hitchon and El. Wallick, National Water Well
Association, Dublin, Ohio, p. 284-289.

Brown, R.M., 1961. Hydrology of tritium in the Ottawa valley. Geochemica et
Cosmochimica Acta, V. 21, p. 199-216.

Craig, H., 1961. Isotopic variations in meteoric waters. Science, v. 133, p. 1702-1703.

Cravens, S.J., and LC. Ruedisili, 1987. Water movement in till of east-central South
Dakota. Ground Water, v. 25, p. 555-561.

69

7O
Dansgaard, W., 1964. Stable isotopes in precipitation. Tellus, v. 16, p. 436-468.

D’Appolonia, 1983. Hydrogeologic study of the Berlin & Farro site, Swartz Creek,
Michigan. Draft report.

Delcore, MR, 1985. Rate of recharge to a heterogeneous aquifer: An investigation
using bomb tritium. M.S. Thesis, Michigan State University.

Desaulniers, D.E., J.A. Cherry, and P.Fritz, 1981. Origin, age and movement of pore
water in argillaceous Quaternary deposits at four sites in southwestern Ontario.
Journal of Hydrology, v. 50, p. 231-257. '

Dincer, T., and ER. Payne, 1971. An environmental isotope study of the south-western
karst region of Turkey. Journal of Hydrology, v. 14, p. 233-258.

Dincer, T., A. Al-Mugrin, and U. Zimmermann, 1974. Study of the infiltration and
recharge through the sand dunes in arid zones with special reference to the stable
isotopes and thermonuclear tritium. Journal of Hydrology, v. 23, p. 79-109.

Dreimanis, A., and RP. Goldthwait, 1973. Wisconsinan glaciations in the Huron, Erie,
and Eric lobes. The Wisconsinan Stage, edited by RF. Black, R.P. Goldthwait,
and HE. Williams, Geological Society of America Memoir 136, p. 71-106.

Egboka, B.C.E., J.A. Cherry, R.N. Farvolden, and E0. Frind, 1983. Migration of
contaminants in groundwater at a landfill: A case study, 3. Tritium as an
indicator of dispersion and recharge. Journal of Hydrology, v. 63, p. 51-80.

Eriksson, E., 1958. The possible use of tritium for estimating groundwater storage.
Tellus, v. 10, p. 472-478.

' Eyles, N., and J.A. Sladen, 1981. Stratigraphy and geotechnical properties of weathered
lodgement till in Northumberland, England. Quarterly Journal of Engineering
Geology, v. 14, p. 129-141.

Farrand, W.R., and DP. Eschman, 1974. Glaciation of tthe southern peninsula of
Michigan: A review. Michigan Academician, v. 7, p. 31-56.

Florkowski, T., 1981. Low level tritium assay in water samples by electrolytic
enrichment and liquid scintillation counting in the International Atomic Energy
Agency laboratory. Methods of Low Level Counting and Spectrometry, IAEA,
Vienna, p. 335-351.

Fookes, P., 1965. Orientation of fissures in stiff overconsolidated clay of the Siwalik
System. Geotechnique, v. 15, p. 195-206.

Foster, S.S.D., 1975. The chalk groundwater tritium anomaly - A possible explanation.
Journal of Hydrology, v. 25, p. 159-165.

71

Freeze, RA, and IA. Cherry, 1979. Groundwater. Prentice-Hall, Englewood Cliffs,
New Jersey, 604 p.

Fritz, P., 1982. Environmental isotopes and ground water at test sites for waste
repositories. Societa Italiana di Mineralogia e Petrologia, v. 38, p. 1165-1174.

Gillham, R.W., MJ.L. Robin, and DJ. Dytynyshyn, 1984. Diffusion of nonreactive and
reactive solutes through fine-grained barrier materials. Canadian Geotechnical
Journal, v. 21, p. 541-550.

Grisak, GE, 1975. The fracture porosity of glacial till. Canadian Journal of Earth
Sciences, v. 12, p. 513-515.

Grisak, G.E., J.A. Cherry, J.A. Vonhof, and J.P. Blumele, 1976. Hydrogeologic and
hydrochemical properties of fractured tillin the interior plains region. Glacial
Till, edited by R.F. Legget, The Royal Society of Canada Special Publications No.
12, p. 304-335.

Grisak, GE, and J.F. Pickens, 1980. Solute transport through fractured media, 1. The
effect of matrix diffusion. Water Resources Research, v. 16, p. 719-730.

Grisak, G.E., J.F. Pickens, and LA. Cherry, 1980. Solute transport through fractured
media, 2. Column study of fractured till. Water Resources Research, v. 16, p.
731-739.

Hendry, MJ., 1982. Hydraulic conductivity of a glacial till in Alberta. Ground Water,
v. 20, p. 162-169. '

Hendry, MJ., 1983a. Groundwater recharge through a heavy-textured soil. Journal of
Hydrology, v. 63, p. 201-209.

Hendry, MJ., R.W. Gillham, and J.A. Cherry, 1983b. An integrated approach to
hydrogeologic investigations - A case history. Journal of Hydrology, v. 63, p. 211-
232.

Hendry, MJ., 1988. Hydrogeology of clay till in a prairie region of Canada. Ground
Water, v. 26, p. 607-614.

Hoffman, C.M., and G.L. Stewart, 1966. Quantitative determination of tritium in natural
waters. United States Geological Survey Water-Supply Paper 1696-D, p. D1-
D18.

Holcomb, S., 1972. Soil survey of Genesee County, Michigan. Soil Conservation
Service, US. Department of Agriculture.

Horberg, L., 1952. Pleistocene drift sheets in the Lethbridge region, Alberta, Canada.
Journal of Geology, v. 60, p. 303-330.

72

Hufen, T., R. Duce, and L. Lau, 1969. Some measurements of the tritium content in
the natural waters of southern Oahu, Hawaii. Water Resource Research Center
Technical Report Number 34, University of Hawaii.

Hut, G., 1986. Intercomparison of low-level tritium measurements in water.
International Atomic Energy Agency, Vienna.

Hvorslev, MJ., 1951. Time lag and soil permeability in groundwater observations. US.
Army Corps of Engineers Experimental Station Bulletin, no. 36, 50 p.

International Atomic Energy Agency, 1969. Environmental isotope data no. 1: World
survey of isotope concentration in precipitation (1953-1963). IAEA, Vienna,
Technical Report Series Number 96.

International Atomic Energy Agency, 1970. Environmental isotope data no. 2: World
survey of isotope concentration in precipitation (1964-1965). IAEA, Vienna,
Technical Report Series Number 117.

International Atomic Energy Agency, 1971. Environmental isotope data no. 3: World
survey of isotope concentration in precipitation (1966-1967). IAEA, Vienna,
Technical Report Series Number 129.

International Atomic Energy Agency, 1973. Environmental isotope data no. 4: World
survey of isotope concentration in precipitation (1968-1969). IAEA, Vienna,
Technical Report Series Number 147.

International Atomic Energy Agency, 1975. Environmental isotope data no. 5: World
survey of isotope concentration in precipitation (1970-1971). IAEA, Vienna,
Technical Report Series Number 165.

International Atomic Energy Agency, 1979. Environmental isotope data no. 6: World
survey of isotope concentration in precipitation (1972-1975). IAEA, Vienna,
Technical Report Series Number 192.

International Atomic Energy Agency, 1983. Environmental isotope data no. 7: World
survey of isotope concentration in precipitation (1976-1979). IAEA, Vienna,
Technical Report Series Number 226.

International Atomic Energy Agency, 1986. Environmental isotope data no. 8: World
survey of isotope concentration in precipitation (1980-1983). IAEA, Vienna,
Technical Report Series Number 264.

Javandel, I., C.Doughty, and CF. Tsang, 1984. Groundwater Transport: Handbook of
Mathematical Models. Water Resources Monograph Series 10, American
Geophysical Union, Washington, DC, 228 p.

73

Kazi, A., and 1.1.. Knill, 1973. Fissuring in glacial lake clays and tills on the Norfolk
coast, United Kingdom. Engineering Geology, v. 7, p. 35-48.

Keller, C.K., G. van der Kamp, and LA. Cherry, 1986. Fracture permeability and
groundwater flow in clayey till near Saskatoon, Saskatchewan. Canadian
Geotechnical Journal, v. 23, p. 229-240.

Knott, J.F., and J.C. Olimpio, 1986. Estimation of recharge to the sand and gravel
aquifer using environmental tritium, Nantucket Island, Massachusetts. United
States Geological Survey Water-Supply Paper 2297, 25 p.

Larson, GJ., M.R. Delcore, and S. Offer, 1987. Application of the tritium interface
method for determining recharge rates to unconfined drift aquifers, 1.
Homogeneous case. Journal of Hydrology, v. 91, p. 59-72.

Levrett, F., and RB. Taylor, 1915. The Pleistocene of Indiana and Michigan and the
history of the Great Lakes. United States Geological Survey Monograph 53, 529

p.

Libby, W.F., 1953. The potential usefulness of natural tritium. Proceedings of the
National Academy of Science, v. 39, p. 245-247.

LLoyd, J.W., 1983. Hydrogeology in glaciated terrains. Glacial Geology: An
Introduction for Engineers and Earth Scientists, edited by N. Eyles, Pergamon
Press, Oxford, p. 349-368.

Long, D.T., T.P. Wilson, MJ. Takacs, and DH. Rezabek, 1988. Stable-isotope
geochemistry of saline near-surface ground water; east-central Michigan Basin.
Geological Society of America Bulletin, v. 100, p. 1568-1577.

Long, D.T., and GJ. Larson, 1988. Evidence for hyperfiltration in a shallow
drift/sandstone aquifer system (abstract). National Water Well Association,
AGWSE meeting, Las Vegas, December 13-14.

Long, EC, 1976. Liquid scintillation counting: Theory and techniques. 915-NUC-76-
7T, Beckman Instruments, Inc., Irvine, California, 56 p.

Mann, W.B., M.P. Urterweger, and BM. Coursey, 1982. Comments on the NBS
tritiated-water standards and their use. International Journal of Applied
Radiation and Isotopes, v. 33, p. 383-386.

McDonald, M.G., and AW. Harbaugh, 1988. A modular three-dimensional finite-
difference ground-water flow model. Techniques of Water Resources
Investigations, Report Number TWI 06-Al.

MILLRWC, 1987. Midwest Interstate Low-Level Radioactive Waste Commission, Host
State Selection Resolution, St. Paul, Minnesota, passed June 30, 1987.

74

Monaghan, G.W., and GJ. Larson, 1986. Late Wisconsinan drift stratigraphy of the
Saginaw ice lobe in southcentral Michigan. Geological Society of America
Bulletin, v. 97, p. 324-328.

Nir, A., 1964. On the interpretation of tritium ’age’ measurements of groundwater.
Journal of Geophysical Research, v. 69, p. 2589-2595.

Norris, SE, 1963. Permeability of glacial till. United States Geological Survey
Professional Paper 450-E, p. E150-E151.

Ostlund, H.G., and E. Werner, 1962. The electrolytic enrichment of tritium and
deuterium for natural tritium measurements. Tritium in the Physical and
Biological Sciences Volume 1, International Atomic Energy Agency, Vienna, p.
95-104.

Packard, 1987. Appendix C: Statistics of count measurements. Tri-Carb Liquid
Scintillation Analyzer Operation Manual, Packard Instrument Company, no. 169-
3059.

Payne, B.R., 1972. Isotope hydrology. Advances in Hydroscience, v. 8, p. 95-138.

Prudic, DE, 1982. Hydraulic conductivity of a fine-grained till Cataraugus County, New
York. Ground Water, v. 20, p. 194-204.

Rabinowitz, D.D., G.W. Gross, and GR. Holmes, 1977. Environmental tritium as a
hydrometeorologic tool in the Roswell basin, New Mexico. Journal of Hydrology,
v. 32, p. 3-46.

Regalbuto, DR, 1987. An isotopic investigation of groundwater in Leelanau County,
Michigan. M.S. thesis, Michigan State University.

Rehm, B.W., S.R. Moran, and G.H. Groenewold, 1982. Natural groundwater recharge
in an upland area of central North Dakota, USA. Journal of Hydrology, v. 59,
p. 293-314.

Robertson, W.D., and J.A. Cherry, in press. Tritium as an indicator of recharge and
dispersion in a groundwater system in central Ontario. Water Resources
Research.

Sharp, Jr., J.M., 1984. Hydrogeologic characteristics of shallow glacial drift aquifers in
dissected till plains (north-central Missouri). Ground Water, v. 22, p. 683-689.

United States Geological Survey, 1980. Water resources data for Michigan. United
States Geological Survey Water-Data Report MI-80-l.

Vinogradov, A.P., A.L. Devirts, and EL Dobkina, 1968. The current tritium content of
natural waters. Geochemistry International, v. 5, p. 952-966.

75

Vogel, J.C., LThilo, and M. Van Dijken, 1974. Determination of groundwater recharge
with tritium. Journal of Hydrology, v. 23, p. 131-140.

von Buttlar, H., and I. Wendt, 1958. Ground-water studies in New Mexico using tritium
as a tracer. Transactions of the American Geophysical Union, v. 39, p. 660-668.

von Buttlar, H., 1959. Ground-water studies in New Mexico using tritium as a tracer, II.
Journal of Geophysical Research, v. 64, p. 1031-1038.

Wiitala, S.W., K.E. Vanlier, and RA. Krieger, 1963. Water resources of the Flint area,
Michigan. United States Geological Survey Water Supply Paper l499-E, 86 p.

Williams, RE, and RN. Farvolden, 1969. The influence of joints on the movement of
ground water through glacial till. Journal of Hydrology, v. 5, p. 163-170.

Woodward & Clyde, 1989. Draft remedial investigation report for the Berlin & Farro
Liquid Incineration, Inc. site, Swartz Creek, Michigan. Draft report.

Wyerman, T., 1978. Laboratory facility of the analysis of natural levels of tritium in
water. United States Geolgical Survey Open File Report.

nIcuIceN STATE UNIV. LIBRARIES
1|1111111llli1|1111111111illiillllillliiliilliii
31293005674381