You are on page 1of 32

&oehimicaet Cosmochimica

Acts, 1971.Vol. 35,pp. 567to 598. PergsmonPress. Printedin NorthernIreland

Geochemistry and origin of formation waters in the western Canada


sedimentary basin-III. Factors controlling chemical composition*
BRIAN HITCHON
Research Council of Alberta, Edmonton, Canada

GALE K. BILLINGS
Department of Geosciences, New Mexico Tech., Socorro, N.M.

and

J. E. KLOVAN
Department of Geology, University of Calgary, Calgary, Canada

(Received 18 September 1970; accepted in revised form 6 January 1971)

Abstract-Twenty major and minor chemical components are reported for 78 formation waters
from oil fields and gas fields in Alberta, Canada. Using published pore volume and chlorinity
distribution data, a volume-weighted mean composition of formation waters in the western
Canada sedimentary basin is presented. The results of Q-mode, R-mode, and second-order
R-mode factor analyses are tabulated and interpreted. The volume-weighted mean composition
is similar to that of present day sea water and is compelling evidence for an ultimate origin from
sea water of the major portion of the dissolved salts in the formation waters. Dilution by fresh
water recharge and concentration by membrane filtration are the major factors controlling
chemical composition. Together, they produce a chemical population ranging from freshwater to
brines which is confirmed by the Q-mode analysis. Compared to sea water, the volume-weighted
mean formation water has gained NaCI, which occurs as a separate factor in the R-mode analysis,
and quantitative calculations demonstrate that sufficient halite has been dissolved from Middle
Devonian evaporite beds to account for the observed gain in NaCl, and that the balance of
halite dissolved from bedded evaporites in the basin has been, and is being, lost to the surface.
The presence of Mg and Ca in separate factors, together with quantitative calculations, indicate
that the ionically balanced loss of Mg and gain of Ca in the volume-weighted mean formation
water cannot be attributed to dolomitization. Similarly, the total loss of SO, is not due to
conversion of H,S. Chlorite formation could account for the loss of Mg. Other factors con-
trolling chemical composition and suggested by the factor study include cation exchange on
clays, a probable contribution of Br and I by desorption from the clay fraction of argillaceous
rocks, as well as from organic matter, and solubility control of C&CO, and Sr-SO, concentrations.

INTRODUCTION

THIS paper is the third in a series of reports (HITCHON and FRIEDMAN, 1969 ;
BILLINGS et al., 1969) on the isotopic and chemical characteristics of a suite of
formation waters from oil fields and gas fields in the western Canada sedimentary
basin. We are concerned here with the inter-relations among 20 major and minor
chemical components in 78 formation waters from Alberta, Canada. Details of
sample collection techniques and analytical methods are given in HITCHON and
FRIEDMAN (1969) and BILLINGS et al. (1969), respectively. We indicate the implica-
tions of the volume-weighted mean composition of formation waters in western
Canada, briefly reported by BILLINGS et al. (1969), and then outline the Q-mode
and R-mode factor analyses that were carried out on the chemical data. The results

* Joint publication of the Research Council of Alberta (Contribution No. 520), Department
of Geosciences, New Mexico Tech., and Department of Geology, University of Calgary.
567
568 B. HITCHON,GALE K. BILLINGSand J. E. KLOVAN

of the factor analyses are interpreted in terms of a series of factors that control the
chemical composition of the formation waters. The paper concludes with a brief
summary of the origin of the formation waters in western Canada based on our
knowledge of their hydrology, and of their isotopic and chemical composition.

VOLUME-WEIQHTED MEAN COMPOSITION


Any study that seeks to interpret the chemical composition of formation waters
on a quantitative basis must take cognizance of the lack of knowledge concerning
both the composition of the original water in the sediments, and the composition
of the present formation waters in the shales.
From knowledge of the rock volume in each stratigraphic unit (HITCHON, 1968)
and the conditions of deposition (MCCROSSAN and GLAISTER, 1964), we estimate
that about 85 per cent of the strata of the western Canada sedimentary basin were
deposited under marine conditions. The remaining 15 per cent were deposited
under mainly brackish water, or possibly fresh water, conditions. Using comparable
rock volume data for Alberta (HITCRON and HOLTER, 1971) and the geological atlas
edited by MCCROSSAN and GLAISTER (1964) we estimate that 80 per cent of the
strata in Alberta were deposited under marine conditions. This latter value compares
favorably with an average of 78 per cent for the data reported by LAYER (1958),
based on less complete information. However, evaporites form a volumetrically
important part of the succession in several stratigraphic units (HITCHON, 1968),
and BILLINGS et al. (1969) have demonstrated the possible presence, in some strati-
graphic units, of bitterns left after halite precipitation but before potassium-salt
precipitation. Consideration of these various factors leads the authors to the
conclusion that if all strata are assumed to have originally contained sea water the
error would be negligible, in view of our present state of knowledge of the history
of formation waters in sedimentary basins. This assumption was used by HITCHON
and FRIEDMAN (1969) in their calculations for the deuterium mass balance in the
formation waters of the western Canada sedimentary basin, and is used in this paper.
The summary of pore volumes in sedimentary rocks of the western Canada
sedimentary basin compiled by HITCHON (1968, Table V) was obtained through the
use of assigned porosity values, including estimates of decrease in porosity of shales
with depth of burial (HITCHON, 1968, Table IV). The pore volume data, together
with maps showing regional variations in chlorinity of the formation waters based
on more than 10,000 analyses (HITCHON, 1964), allows determination of the volume
of formation waters of differing salinities, based on knowledge of the relation of
chlorinity and salinity. From the known relations of chloride to the other chemical
components a volume-weighted mean chemical composition can be calculated
(Table 1). More details of the method of calculation are provided by BILLINGS et al.
(1969, pp. 212-214). Throughout these computations we have assumed that the
composition of the formation waters in the shales is similar to that in the adjacent
sandstones and carbonates. Variations of the SP-curve for shales suggests that this
assumption is valid. Using these assumptions the volume-weighted mean com-
position of formation water in the western Canada sedimentary basin given in
Table 1 can be used for quantitative calculations.
The ionic balance of the volume-weighted mean formation water is good (cations
Origin of formation waters in the westan Canada sedimentary basin-111 569

Table 1. Comparison of sea water and volume-weighted mean formation


water from tht, western Canada sedimentary basin

sea. water* Formation water? Net chsnge


Formation water
Component m&$/l. melt: m&l. mel$ Sea water m&l. me1

Li 0.17 0.02 10.7 1.64 62.9 -I- 10.83 + 1.52


NB 10,760 468.06 14,340 623.88 1.3 + 3’680 + i55.73
K 387 9.90 St31 14.36 1.4 + 174 + 4-46
Rb 0.12 0.00 0.88 O*Ol 7.3 iO.76 + 0.009
ME 1290 106*09 317 26.07 0.25 -973 - so*02
CC 413 20.61 2210 110.38 6.3 + 1797 + 89.67
Sr 8 0.16 108 2.46 13.5 + 100 +2*28
Mn 0.002 0.00 0.32 0.01 160.0 j-O*318 +0.01
Zn 0.01 o*oo 0.30 0.01 30.0 10.29 + 0.01
Cl 19,360 545-W 26,920 759.06 I-4 + 7$70 +i13.47
Br 67 0.84 II4 I.42 1.7 +47 + 0.59
I 0.06 0.00 9 0.07 160-O + 8.94 + 0.07
HCO, 142 2.33 1600 24.62 IO.6 + 1356 + 22.26
SO, 2710 66.42 350 7.29 0.13 -2360 -49.14
Total dissolved
solids 35,100 - 46,400 - 1.3 + 11,300 -

* Modified after CULTS (1965) and @OLDBERG (1968).


t Modified after BILLIXGSet al. (1969).
$ Sea water: cations = 604.86; anions = 606.26. Formation water: cations = 778.71; anions = 792.36.

778-71 mel; anions 792.36 mel), considering the method used to obtain the mean
contents of the various components. The salinity is 1.3 times that of sea water,
implying a net gain of dissolved salts. With the exception of Mg and SO,, all
components show a net gain, compared to a sea water starting composition. The
most reasonable primary source for these extra. ions is the rock matrix through
which the formation waters are flowing. The values given in the net change column
in Table 1 indicate that the major gain may be attributed to Na and Cl, with Ca
and Mg showing an apparently balanced gain and loss respectively. Consideration
of the total gains and losses implies that the gain and loss of HCO, and SO,, respec-
tively, may be related, although the individual material balance is not good. The
other components represent only minor gains from the sedimentary rocks. The
conclusion drawn from our simplified model is that, relative to the present pore
volume, there has been negligible removal of dissolved solids, although there has
been a major re~st~bution. We surmise that during compaction some water was
expelled in the initial stages, probably without change in composition, and during
the later stages, probably only slightly modified, if at all. Membrane filtration
probably does not take place to any significant extent until compaction has reached
some, as yet undefined, critical stage. While recognizing the loss of unmodified
or only slightly modified water up to the stage of effective membrane filtration,
there has been negligible removal of dissolved solids since that stage, as evidenced
by the mass balance data. The factor analysis therefore provides us with an indica-
tion of the chemical and physical controls that have been exerted on the formation
waters remaining after that, as yet undefined, critical stage.
In the deuterium mass balance calculations (HITCHOX and FRIEDMAX, 1969)
the model essentially assumed instantaneous mixing in arriving at a. value of not
570 B. HITCHON,GALE K. BILLINGSand J. E. KLOVAN

more than 2.9 times the present pore volume of fresh water. GORDON RITTENHOUSE
(personal communication) has pointed out that by using the concept of incremental
mixing models of BRENNER et al. (1965) the fresh water required to account for the
deuterium mass balance may be as low as 1.13 of the present pore volume, depending
on the model chosen. This suggested new approach to the problem of the mixing
of fresh water and sea water in sedimentary basins invalidates neither the mixing
concept of HITCHON and FRIEDMAN (1969) nor the present factor study, but does
allow mixing models to be postulated which require lower quantities of fresh water
and are more consistent with known low permeabilities of the rocks in many basins.
The incremental mixing model does however imply the gradual loss of water from
the basin-but to what extent the composition of the dissolved solids have been
modified is not known. We may expect that in the initial stages of incremental
mixing the lesser degree of compaction will be associated with relatively easy mixing
and loss of essentially unmodified water. During the later stages of incremental
mixing with a greater degree of compaction the mixing process will be retarded,
the permeability decreased, and the loss of water reduced-with consequent negli-
gible removal of dissolved solids since that, as yet undefined, critical stage. The
suggestion of the use of an incremental mixing model is thus an important contribu-
tion to our understanding of these processes, though, clearly, more work is required
on the validity of the various models which might be proposed.
The chemical composition and physical properties of the 78 formation waters
from oil fields and gas fields in Alberta are shown in Table 2. The sample numbers
correspond to those used by HITCHON and FRIEDMAN (1969, Table 4), where addi-
tional information on the salinity, stable isotopes, production, and reservoir param-
eters are presented. Chemical data are not available for sample No. 43 from the
Carboniferous (HITCHON and FRIEDMAN, 1969, Table 4). The need to ensure that
the samples analysed are representative of the water in the formations precluded
the use of drillstem test samples. The relative frequency of salinity for this sample
of 78 formation waters is similar to that obtained by considering the entire western
Canada sedimentary basin. Two discrepancies are that near-surface fresh waters
are under represented and calcium- and magnesium-rich brines from some of the
deeper stratigraphic units (HITCHON and HOLTER, 1971) are absent in the sample.
Apart from these limitations, the samples analysed are chemically and volumetrically
essentially representative of formation waters in the western Canada sedimentary
basin.

FACTOR ANALYSIS
Factor analysis is a statistical technique designed to explain complex relations
among many variables in terms of a few “factors,” which themselves represent
simpler relations among fewer variables. Factor analysis only demonstrates the
relations, it does not explain them. The explanation of the factors must be in the
context of known information about the variables. There are several types of factor
analysis. Q-mode factor analysis consists of a comparison of the samples in terms of
the variables. This technique essentially evaluates the homogeneity of the sample
or population being studied. R-mode factor analysis comprises a comparison of the
relations among the variables in terms of the samples, and requires a homogeneous
Origin of formation w&em in the western Canada sedimentary basin--III 571

population to be meaningful. Three main types of solution are available. The


principal components solution derives orthogonal reference factor axes located in
n-dimensional space such that the total amount of information, or variance, con-
tained in the data is systematically extracted. The first factor axis explains the
most variance; the second maximised the remaining variance, and so on. This
method has been used in the study of formation waters by LEE (1969) and COLLINS
(1970). In the varimax solution, orthogonal factor axes are located such that the
variance of each factor is maximized. This “rotation” of orthogonal axes results in
a more equitable distribution of factor loadings than that produced by the principal
components method. COLLINS (1967), DAWDY and FETH (1967), EGLESON and
QUERIO (1969) and KRAMER (1969) use the varimax solution in their examination of
formation waters. Oblique factor solutions differ significantly from the first two
methods in that the reference factor axes are not constrained to be orthogonal. A
variety of analytical criteria are available to locate the exact positions of the reference
axes ; biquartimin, binormamin and promax are examples of these varieties.
Oblique solutions are appropriate where underlying causal factors are suspected to
be correlated among themselves. SPENCER (1967) has demonstrated that this is
often the case in geochemical studies. SAMPSON(1968) further illustrates the use of
oblique solutions in the analysis of formation water data. SPENCER(1966a, b; 1967)
and CAMERON (1967) provide pertinent background information with particular
application to geochemical problems.
A basic assumption of factor analysis is that the factors are linearly related-a
condition not met very often with natural data. Log transformation of the data
makes it more amenable to factor analysis. Because many geochemical data
approximate a lognormal distribution it is important that the logarithmic transforms
of the geochemical data be factored and not the raw data, although the use of raw
data does not preclude an interpretable suite of factors. All the authors cited
previously who applied factor analysis techniques to the study of formation waters
used raw data, although DAVIS (1966) has shown that at least the dissolved solids
content of ground-waters commonly approximate a lognormal distribution.

FACTOR RESULTS
Consideration of the many possible inter-related physical and chemical controls
that might be operative led to the conclusion that the chemical and physical data
should be factored separately. The only exception was the inclusion of salinity
(total dissolved solids) in some of the Q-mode and R-mode analyses because of the
need to distinguish formation waters of the same ion ratios but with different ion
concentrations. Initial examination indicated that all components were essentially
lognormally distributed, and consequently the logarithm (base e) of the concentra-
tion of each component was used to form the correlation matrix for the factor
analyses. Some components (Co, Cr, Ni, CO,) were present in so few of the samples
that they were not included in the final factor study. However, in some preliminary
factor ana yses, including Co, Cr, Ni and CO,, we found that Co and Ni were associ-
ated with Fe, and Cr and CO, tended to be independent.
The Q-mode varimax factor score matrix of the chemical data and salinity is
shown in Table 3. Three factors account for more than 98 per cent of the information
572 B. HITCWON, GALE K. BILLINGS and J. E. KLOVAN

Table 2. Major and minor chemical components, and selected physical properties of

Sample N8 K Li Rb Ce Mg Sr Fe Mn
NO. Field (pool) (mg/l.) (mgll.) (mg/l.) (mg/l.) (mg/l.) (mg/l.) (mgll.) (mg/l.) (mg/l.)

upper Cret8ceous

81 Pembina (Keystone Belly


River M) 4400 22 0.52 0.08 65 21 13.3 2-7 0.10
82 Pembina (Belly River I) 6360 25 0.37 0.18 55 87 13.3 3.8 0.12
83 Pembina (Belly River G) 1670 8 0.18 0.10 22 * 2.2 0.55 0.1
28 Pouce Coupe South (Doe
Creek A) 3750 14 16 0.06 41 43 17.4 11.7 0.4
52 Crossfield (Cardium A) 1990 5.6 0.6 o-10 208 * 1.oo 0.52 0.04

Lower Cretaceous-Lower Colorado Group

27 Gordondale (Peace River) 8290 50 2.2 0.10 249 79 38 2.7 0.10


85 Comrey (Bow Ishmd) 1510 10 0.28 0.04 17 8 0.52 1.8 0.1
D Pendant D’Oreille (Bow
Island) 3290 7.6 0.5 0.14 14 10 O-96 0.47 <0.04
60 Countess (Bow Island A) 8450 38 2.0 0.06 178 46 32 0.58 0.2
E Countess (Bow Island A) 9340 36 2.0 <0.02 197 69 24.9 5.5 2.50
66 Cessford (Basal Colorado A) 10,300 36 2.4 0.04 274 94 60 1.90 0.3
68 Cessford (Viking C) 7260 40 1.8 <O-O4 71 32 20 7.5 0.1
11 Legal (Viking) 22,800 98 4.2 0.14 1230 409 219 27.2 0.7
29 Joarcam (Viking) 21,000 82 3.6 0.10 469 304 141 20 0.3
36 Chigwell (Viking B) 10,800 25 2.0 0.08 72 36 29 0.46 0.10

Lower Cretaceous-Mannville Group

26 Gordondale (Gething A) 6340 54 3.4 0.10 209 68 51 27 0.9


35 Cold Lake (Colony A) 11,500 60 2.2 0.04 562 296 55 3-l 0.40
58 Gilby (Basal Mannville B) 16,800 440 14 o-70 489 104 11.4 54 0.6
1 Glen Park (Glauconitio A) 33,800 680 19.2 1.10 2770 1140 420 7.7 O-61
10 Campbell-Namao (Namao
Blairmore D) 34,000 840 21.6 1.2 4640 1000 276 0.08 3.18
89 Pendant D’Oreille (Mann-
ville A) 3130 36 0.94 0.06 12 18 2.4 0.94 0.1
71 Horsefly Lake (Mannville) 2260 74 2.6 o-14 38 15 1.21 0.56 <0.04
70 Wildcet (S. Roy. Hays #S-S) 2710 70 2.8 0.12 32 4 O-27 0.05 <0*04
C Bantry (Mannville A) 8710 132 8.0 0.38 39 52 3,22 0.08 <0.04
55 Aerial (Mannville) 6930 58 2.0 0.06 223 30 26.7 0.09 0.09
34 Cheuvin (Mennville A) 28,200 760 16.2 0.72 2670 1310 205 13.0 0.19
33 Bellshill Lake (Sk&more) 27,000 1280 26 l-6 2620 927 149 0.25 08
30 Malmo (Blairmore A) 42,900 1820 42 2.0 9740 1760 481 0.41 1.4
Origin of formation waters in the western Canada sedimentary basin-III 573

78 formation waters from oil fields and gas fields of Alberta, Canada

Refrac-
Resis- tive
tivity Index
CU Zn Cl Br I so, HCOIl B Density (ohm m; (25°C;
(mgll.) tmg/l.) (mgll.) (mg/l.) (mg/l.) (mg/l.) (mgll.) (ppm) PH (6O’F) 25°C) Nab)

0.12 0.08 6730 76 36 3 661 5 8.39 1.0091 0.498 l-3344


0.26 3.05 8650 92 39 4 1220 - 8.24 1.0120 0.3990 1.3351
0.05 0.09 2250 14 5 7 413 3 7.73 I.0042 1.4900 1.3368

0.11 0.42 5370 35 10 3 784 2 7.69 1.0087 0.7290 1.3330


0.07 0.09 798 13 4 24 3560 11 8.30 1.0050 1.5000 1.3344

0.04 0.49 13,100 33 11 5 1030 4 7,76 1.0124 0.3080 1.3356


0.07 0.07 994 2 2 6 1870 5 8.12 1.0042 1.8300 1.3329

0.08 0.06 1530 6 1 3 2330 - 9.50 1.0059 0.9590 1.3332


0.07 0.04 13,800 98 30 6 418 8.7 7.62 1.0153 0.2550 1.3366
0.13 0.00 14,600 95 24 218 357 11 8.00 1.0178 0.2530 1.3364
0.06 0.05 16,000 89 14 2 258 8.2 7.94 1.0175 0.2350 1.3367
0.09 0.05 11,400 81 28 6 437 8.8 7.92 1.0116 0.3190 1.3366
0.05 0.44 38,200 169 28 11 761 8 6.97 I.0455 0.1350 1.3426
o-12 0.40 34,300 179 36 4 176 8 7.10 1.0414 0.1390 1.3413
0.12 0.06 14,900 124 38 4 2450 8.19 1.0182 0.2380 1.3371

0.10 0.47 9780 20 10 8 1560 3 7.25 1.0091 0.3770 1.3349


0.08 0.40 18,400 66 13 4 379 7 7.34 1.0210 0.2320 1.3365
0.19 0.06 25,400 74 14 901 2080 18 6.77 1.0315 0.1610 1.3400
0.16 0.18 60,900 243 10 45 432 - 6.62 1.0680 0.0724 1.3495

0.16 0.11 61,100 208 17 441 509 - 6.93 1.0687 0.0736 1.3496

0.11 0.05 2790 5 1 46 3660 5 8.03 1.0103 0.6310 1.3346


0.12 0.11 1600 tr. tr. 11 2280 - 8.10 1.0058 1.1800 1.3333
0.08 0.05 1210 lx. tr. 265 1460 5.8 7.95 1.0075 1.1500 1.3336
0.07 0.03 7540 32 3 7 7750 - 8.60 1.0176 0.3290 1.3369
0.09 0.13 9130 36 10 539 2640 - 7.70 1.0119 0.3750 1.3366
0.21 0.08 53,200 207 12 6 425 - 6.99 1.0591 0.0826 1.3476
0.08 0.15 48,800 171 9 7 685 - 6.83 1.0538 0.0882 1.3461
0.06 0.35 95,800 376 17 466 473 54 6.20 1.1018 0.0700 1.3566
674 B. HITWON, GALE K. BILLINGS and J. E. KLOVAN

Table 2

Sample Na K Li Rb Ca Fe Mn
No. Field (pool) (mgll.) (mg/l.) (mg/l.) (mg/l.) (mg/l.) (31.) ($11.) (mg/l.) (mg/l.)
-
Jurassic

48 Gilby (Jurassic F) 17,000 480 18.1 1.30 523 140 28.2 0.65 0.11
88 Aden (Swift) 730 68 0.76 0.16 55 43 3 0.24 0.6
93 Conrad (Ellis) 1850 62 1.6 0.04 43 19 1.1 0.05 0.1

Triassic

24 Worlsey (Triassic A) 42,600 920 32 0.84 1970 682 156 0.51 0.8
20 Sturgeon Lake South
(Triassic A) 53,700 1660 30 1.6 3070 827 152 21 2.0

Carboniferous

5 Paddle River (Rundle) 37,400 1680 37.5 3.90 5380 982 398 1 .“‘J
_. 0.14
8 Glen&s (Banff) 27,300 840 13.2 1.3 2360 661 355 14.2 0.35
51 Harmattan Elkton
(Ruudle A) 26,300 800 31.0 I.20 2220 388 539 3.9 0.15
49 Medicine River (Pekisko I) 14,700 440 15.9 0.84 227 101 25.7 2.0 0.12
87 Aden (Rundle A) 510 70 0.80 0.10 96 58 6.1 0.57 0.2
72 Del Bonita (Rundle) 3420 188 7.8 0.18 44 60 0.27 0.79 0.5
69 Enchant (Elkton A) 10,600 108 6.4 0.10 21 40 15 48 0.60
64 Jenner (Pekisko A) 9230 90 4.6 0.14 35 2 11.3 26.2 0.3
67 Cessford (Pekisko A) 6090 52 1.8 0.06 73 26 18.6 8.6 0.3

Upper Devonian-Wabamun Group

17 Wildcat (Shell Simonette


# 12-28) 73,800 5800 72 16.6 23,500 2490 900 0.42 I.0
4 St. Albert-Big Lake (D-1B) 34,500 1920 34 2.4 7180 1650 289 5.8 0.20

Upper Devonian-Winterburn Group

2 Wizard Lake (D-2A) 50,900 4200 56 4.8 14,000 2130 481 0.15 0.6
14 Excelsior (D-2) 33,000 1980 36 2.4 7770 1620 222 0.05 0.6
12 Fairydell-Bon Accord (D-2B) 31,100 1700 30 2.2 5950 1450 207 co.01 0.2
44 Wimborne (D-2A) 48,900 8400 74 10.8 22,700 2870 925 0.83 <0.04
B Joffre (D-2) 58,800 2480 29.5 2.7 6910 1010 232 0.58 co.04
56 Alix (D-2) 50,200 7200 76 11.0 23,500 3560 720 0.16 0.6
37 Chigwell (D-2A) 61,800 8000 72 Il.0 21,600 2890 815 0.07 0.7
41 Stettler (D-2A) 41,300 4600 44 4.8 12,800 2280 392 0.31 0.1
42 Fenn-Big Valley (D-2A) 29,700 4200 48.6 6.4 10,000 1730 278 0.26 0.05
46 West Drumheller (D-2A) 28,300 3300 44 4.4 10,700 2170 300 2.3 0.6
73 Youngstown (Arcs) 9360 740 9.8 0.64 1380 428 36 12.9 1.5
Origin of formation waters in the western Canada sedimentary basin-III 575

(continued)

Refrac-
Resis- tive
tivity Index
Cl HCOI B Density (ohm m; (25’C;
(m”,;;., (mzp;l., (mg/l.) (marl.) (mill.) (n$.) (mg/l.) (wm) PH (60°F) 25V) Nan)

0.12 0.09 26,500 76 14 675 1600 - 7.60 1.0329 0.1530 I.3404


0.17 0.06 496 tr. * 20 1090 8 7.90 I.0023 2.9200 1.3327
0.08 0.08 912 9 * 10 3700 7.93 1.0055 1.4500 1.3335

0.16 0.08 71,500 151 21 761 474 - 6.67 1.0795 0.0695 1.3520

0.04 0.22 91,400 193 21 558 784 29 6.67 1.1039 0.5780 1.3568

0.08 0.45 69,300 219 2 370 694 89 7.15 I.0785 0.0601 1.3523
0.11 3.05 48,900 201 9 6 536 7.12 I.0564 0.0855 1.3464

0.20 0.53 46,900 128 26 27 728 66 6.50 1.0542 0.1080 1.3455


0.44 0.11 22,200 65 12 711 1840 - 7.61 I.0275 0.1820 1.3392
0.09 0.07 320 tr. * 9 1070 1 7.54 1.0021 3.6200 1.3326
0.05 0.38 2970 7 1 448 2770 16 8.28 I.0089 0.6420 1.3339
0.26 0.19 12,100 58 5 1290 1820 14 8.28 1.0202 0.2530 1.3359
0.10 0.34 8440 42 2 32 3270 - 8.43 1.0156 0.3460 1.3359
0.24 0.55 7370 49 14 18 3480 15 8.15 I.0119 0.4010 1.3353

0.07 0.33 171,000 466 22 276 474 103 6.43 I.1673 0.0418 1.3783
0.07 0.57 72,200 355 14 765 368 80 7.04 I*0812 0.0760 1.3512

0.02 0.49 121,000 484 23 580 433 104 6.54 1.1287 0.0630 1.3633
0.12 0.49 69,100 363 19 1050 581 - 6.44 1.0816 0.0850 1.3524
0.07 0.05 63,500 300 13 611 794 - 6.60 1.0715 0.0724 1.3505
0.07 0.15 134,000 956 18 362 770 - 6.70 1.1448 0.0466 1.3695
0.13 0.05 112,000 574 30 937 593 103 6.50 1.1191 0.0620 1.3606
0.03 0.07 129,000 961 21 520 578 234 6.48 1.1448 0.0442 1.3690
0.07 0.14 153,000 970 21 452 192 - 6.29 1.1443 0.0468 1.3690
0.03 0.21 98,000 642 16 778 157 - 6.90 1.1109 0.0539 1.3601
0.17 0.09 70,300 494 12 1020 293 - 6.71 1.0805 0.0671 1.3533
0.20 0.13 77,000 514 11 1040 592 99 6.29 1.0814 0.0658 1.3535
0.14 0.06 15,800 20 1 3910 1030 - 7.34 I.0225 0.2210 1.3377
576 B. HITCBON, GALE K. BILLINGS and J. E. KLOVAN

Table 2

Sample N8 K Li Rb Ca
No. Field (pool) (mgll.) (mgll.) (mg/l.) (mg/l.) (mg/l.) (m?l.) (mzl.) (m21.) Cm%.)

Upper Devonian-Woodbend Group

25 Worsley (D-3G) 56,100 3520 46 5.0 17,700 1880 900 41 4.1


90 Normandville (D-3B) 65,000 2580 40 6.8 25,200 3380 820 0.89 1.4
16 Simonette (D-3) 92,800 5200 74 15.4 24,300 2160 900 0.58 1.2
19 Little Smoky (D-3) 74,300 7400 84 18.8 26,300 2710 1020 9.0 11.0
80 Pine Creek (D-3) 60,200 7600 100 15.4 17,600 1590 1060 0.04 0.1
79 Windfall (D-3A) 45,900 8000 86 11.2 16,200 1720 850 0.65 0.1
50 Homeglen-Rimbey (D-3) 39,900 6000 54 7.8 38,700 3370 1280 <O.Ol 1.2
3 Bonnie Glen (D-3A) 53,100 4400 46 5.2 32,000 3990 1320 1.8 1.1
9 Acheson (D-38) 49,800 2940 38 5.4 22,900 3200 730 1.27 0.9
13 Skaro (Cooking Lake) 47,100 840 20 0.84 8840 2980 300 0.19 0.7
A Wimborne (D-3A) 39,600 8600 74.1 12.0 24,000 3210 1170 0.42 0.22
45 Wimbone (D-3A) 48,800 8800 76 12.0 22,800 2590 945 34 1.4
38 Clive (D-3A) 49,200 7800 76 9.2 20,600 3070 740 0.01 0.6
32 B8shaw (Ireton A) 48,600 7800 78.1 12.0 20,100 3270 680 0.06 0.21
31 Malmo (D-3A) 37,800 5400 60 5.4 27,200 2930 560 0.16 0.9
H Duhamel (D-3B) 55,100 5200 60.4 4.22 18,600 2880 608 1.5 0.40
40 Stettler (D-3A) 32,200 5200 45.5 6.8 11,500 1970 298 0.04 0.07
54 Fenn-Big Valley (Fenn
D-3E) 29,700 3600 48.0 6.0 9500 1960 255 0.50 0.14
47 West Drumheller (D-3A) 31,600 5000 52 5.0 9360 1930 280 0.03 0.1

Upper Devonian-Beaverhill Lake Fm.

92 Snipe Lake (Beaverhill Lake) 73,600 1560 32 2,4 10,900 1140 460 3.9 0.10
F Judy Creek (W. Beaverhill
Lake) 65,000 1560 26.4 2.00 4690 607 232 42 0.13

Granite Wash

23 Wildcat (Triad Iroquois


#15-16) 85,300 2380 30 8.8 24,900 1430 735 82.00 52.00
91 Red Earth (Granite Wash A) 63,700 840 9.8 0.68 20,300 2450 565 87 13

-Not determined.
tr. trace.
l Below detection limits.

1. Cobalt below detection limits (0.005 mg/l.) except in sample number 3(0.009); ll(O.009); 19(0.014);
F(0.015).
2. Chromium below detection limits (0.01 mg/l.) except in sample number 5(0*012); lg(O.016); 23(0.019);
3. Nickel below detection limits (0.025 mg/l.) except in sample number l(O.089); ll(O.066); 19(0.036);
72(0.028); 79(0.026); 91(0*101); F(0.042).
4. C8rbOn8te absent except in sample number 28(186); 29(319); 36(370); 48(206); 49(164); 52(356);
82(667); 83(108); 85(359); 87(189); 88(244); 89(1230); C(1510); D(1910). Values in parenthesis in mg/l.
5. Boron was determined only on selected samples, and was not included in the factor study.
6. Analysts: Major elements (Ca, Mg, Cl, Br, I, SO,, HCOII, COI) and physical proport,ies: D. R. Shaw, Oil
Minor elements (K, Li, Rb, Sr, Fe, Mn, Cu, Zn, Co, Cr. Ni): G. K. Billings.
NE: Chemistry Leboratory, Geology Division, Research Council of Alberta.
B: Dr. A. A. Levinson, Dept. of Geology, University of Calgary, Calgary, Alberte.
Origin of formation watersin the westernCanada sedimentary
basin-111 577

(continued)
Refrac-
Resis- tive
tivity Index
cu Zn Cl BI! I so1 HCO, B Density (ohm/m; (25°C;
(m&) (m&) (mgll.) (mg/l.) (mph) (m&J (mph) (ppm) PH (6O’F) 25’C) NaD)

0.05 o-49 128,000 393 23 295 200 66 6.06 1.1366 0.0494 1.3666
0.06 0.11 168,000 616 12 512 138 47 6.36 1.1719 0.0416 1.3769
0.28 0.47 167,000 396 16 228 767 116 7.08 1.1688 0.0406 1.3742
0.06 0.40 172,000 465 19 269 412 118 6.27 1.1778 0.4640 1.3761
0.06 0.19 119,000 319 21 120 1100 178 7.31 1.1362 0.0476 1.3666
0.11 0.32 116,000 469 16 319 696 - 6.96 1.1378 0.0481 1.3669
0.07 0.11 148,000 1120 22 314 348 173 6.41 1.1616 0.0476 1.3738
0.20 27.6 162,000 848 26 266 130 5.98 1.1681 0.0413 l-3748
0.11 0.16 131,000 898 21 454 261 - 6.36 I.1446 0.0476 1.3694
0.49 2.12 99,900 296 16 790 139 33 6.78 1.1096 0.0561 1.3696
0.12 0.14 126,000 424 19 343 960 283 6.60 1.1466 0.0600 1.3688
0.07 0.11 126,000 961 20 367 767 240 7.01 1.1449 0.0480 1.3696
O-06 o-17 131,000 996 21 426 456 224 6.75 1.1446 0.0461 1.3692
0.20 0.11 123,000 820 20 651 613 148 6.84 I .1429 0.0411 1.3686
0.06 0,77 122,000 660 26 696 392 145 6.42 1.1366 0.0630 1.3646
0.08 1.66 126,000 679 21 637 161 81 6.60 1.1371 0.0600 1.3661
0.19 0.12 76,800 537 12 831 416 - 6.64 1.0862 0.0636 1.3646

0.18 0.16 73,800 480 11 664 - 6.31 1.0796 0.0696 I.3632


0.11 0.09 77,000 473 13 1120 439 - 6.76 I.0814 0.0668 1.3636

0.09 10.60 139,000 267 20 688 261 99 7.06 1.1492 0.0436 1.3686

0.13 0.06 113,000 291 28 914 218 123 6.90 1.1238 0.0620 1.3616

0.26 2.73 183,000 262 5 248 216 - 6.69 1.1788 0.0436 1.3762
0.12 0.76 127,000 636 10 690 18 29 4.32 l-1416 0.0449 1.3676

20(0.000);
23(0*02g);
26tO.013);
26(0*006);
46(0*011);
60(0.006);
68(0.029);
64(0.008);
73(0.012);
gl(O.028);
44(0*034);
46(0*016).
20(0.033);
23(0.3gO);
26(0*042);
26(0*034);
29(0*047);
46(0*046);
68(0.110);
64(0*071);
07(0.028);
eg(O.078);

66(161);
60(82);
64(3430);
OO(lO8);
67(74);
68(217);
69(2290);
70(1770);
71(880);
72(1020);
81(12g);

and Gas Conservation Board, Edmonton, Alberta.


578 B. HITOHON, GALE K. BILLINGS and J. E. KLOVAN

Table 3. Q-mode varimax factor score matrix of chemical data and salinity for
78 formation waters from oil fields and gas fields of Alberta, Canada

Factor
Variable
1 2 3

Ch 1.626 0.019 0.027


cu 0.029 -0.545 0.226
Fe -0.514 0.480 3.013
K 1.471 -0.139 -0.590
Li 1.095 0.545 -0.431
Mg 1.598 0.517 0.265
Mn 0.280 0.967 0.723
Na. 0.812 -1.505 0.705
Rb 1.170 1.228 -0.852
Sr 1.292 0.755 0.951
Zn 0,142 0,935 0,733
Br 0.974 0.204 0.831
Cl 1.059 - 1.250 0.908
HCO, -0.218 -2.220 -0.147
I 0.320 0.080 0.883
SO, 1.154 -0.315 -1.172
Total dissolved solids 0.909 - 1.761 0.625
Per cent of information explained
by factor 54.65 37.17 6.21

Cumulative per cent of information 54.65 91.82 98.03

among the samples. The first factor, which accounts for nearly 55 per cent, has
high positive scores for the alkali metals, the alkaline earth metals, halogens (except
I), SO, and salinitiy. The second factor is characterized by high negative scores for
HCO,, salinity, Na and Cl, and a high positive score for Rb. The first two factors
account for nearly 92 per cent of the information among the samples. The third
factor accounts for slightly more than 6 per cent and has a very high positive score
for Fe and a low negative score for SO,. When the varimax factor components
for these three factors are normalized and the data points plotted on a ternary
diagram (Fig. l), nearly all samples fall in a narrow band along one side of the
diagram, between factor 1 and factor 2. Sample points on the same flow path have
been joined together, and the arrows indicate the flow direction, based on hydro-
dynamic studies (HITCHON, 1969a, b). The Q-mode analysis thus indicates that the
data we are examining are relatively homogeneous and, as we demonstrate in the
next section of this paper, represent a single source (sea water) which is being
subjected to two major processes, which we can identify as concentration by
membrane filtration (factor 1) and dilution by freshwater recharge (factor 2). Other
processes, for example that represented by factor 3, are relatively minor. An
R-mode analysis of these data is therefore justified.
An R-mode factor analysis, using a varimax solution, was carried out on the
logarithmic transforms of the same data used for the Q-mode. The varimax factor
Origin of formation waters in the western Canada sedimentary bmin-III 579

FACTOF ;
FACTOR 2

Fig. 1. Ternary diagram of normalized varimax factor components for Q-mode


on 78 formation waters from oil fields and gas fields of Alberta, Canada. Arrows
indicate flow directions based on hydrodynamic studies (HITCHON, 1969a, b).

matrix is shown in Table 4. The communality is a measure of the fraction of the


variance of each ion that is explained by the factors that have been extracted.
The size of the eigenvalue represents the variance of the original data that has been
extracted on to each factor. KAISER (1960) has suggested that factors with eigen-
values less than 1.0 are generally not significant. However, SPENCER (1966a) has
observed that in dealing with geochemical data eigenvalues greater than O-5 can
often be given physical meaning (upon subsequent rotation). CAMERON (1967)
found it convenient with geochemical data to rotate all factors with eigenvalues
greater than 0.1, and then on a second run to discard all those factors that do not
appear to be significant. The low communalities that are observed for Mn and SO,
if only six factors are considered, confirms the observation (SPENCER, 1966a) that
eigenvalues less than O-5 may sometimes be pertinent when dealing with geochemical
data. The high communalities indicate that most of the variance of each of the
ions and of the salinity is explained by the eight factors extracted, which account
for nearly 97 per cent of the total variance among the samples. Loadings less than
0.2 have been omitted from the table, since they correspond, approximately, to
less than 5 per cent of a variable.
When the same data are factored using the oblique biquartimin solution the
resulting factor matrix (Table 5) indicates that much of the “noise” present in the
varimax factor matrix has been removed. A matrix of correlations between the
eight factors extracted indicates that they are virtually uncorrelated, with no
580 B. HITCHON,GALE K. BILLINGSand J. E. KLOVAN

Table 4. R-mode varimax factor matrix of chemical data and salinity for 78 formation
waters from oil fields and gas fields of Alberta, Canada*

Factor
Variable
1 2 3 4 5 6 7 8 Communality

CS 0.884 -0.210 -0.260 0.956


cu 0.993 0.996
Fe 0.954 0.996
K 0.959 0.979
Li 0.954 0.969
Mg 0.874 -0.207 0.878
Mil 0.261 0.201 0.915 0.998
NE% 0.851 -0.396 0.957
Rb 0.933 0.924
Sr 0.834 -0.389 - 0.240 0.954
Zn 0.951 0.999
Br 0.757 -0.670 0.950
Cl 0.847 -0.430 0.982
HCO, -0.387 0.238 0.847 - 0.202 0.995
I 0.359 -0.893 0.960
SO4 0.619 O.i55 0.983
Total dissolved
solids 0.888 -0.343 0.980

(Principal
component) 10.495 1.848 1.127 0.961 0.689 0.569 0.403 0.360
Per cent of variance exolained bv factor

50.18 6.22 II.06 6.03 6.43 8.26 6.23 4.37

Cumulative DA* cent of variance

50.18 66.40 67.46 73.49 79.92 86.18 92.41 96.78

* Loadings <0.2 omitted.

correlation coefficients exceeding 6.25. Further, all eigenvalues are greater than 0.5.
The factors from the oblique solution may be illustrated (Fig. 2) in a similar manner
to that used by SPENCER (1966b), where rectangular boxes represent the factors
and the centre line of each box is a zero loading for the ion. Positive loadings occur
above the centre line and negative loadings below. The further an ion is from the
centre line the higher is its loading. Factor 1 is characterized by high positive
loadings for the alkali metals, the alkaline earth metals, halogens, salinity and SO,,
with a small negative loading for HCO,. Factor 5 is a halogen factor, but represents
particularly I and Br. The other factors indicate independent controls for Cu, Fe,
Zn, Mn, HCO, and SO,. The fact that the halogens, SO, and HCO, appear in two
factors each implies multiple controls.
Examination of the data in Table 5 and Fig. 2 indicated that many of the
physical and chemical controls that are generally believed to be operative in mineral-
solution reactions were not revealed by the factors extracted. That is, the mechanism
causing concentration of the formation waters was the dominant process but the
increased concentration masked some of these other processes. Accordingly, an
R-mode factor analysis was carried out with the salinity (total dissolved solids)
partialled out by the process of running a partial correlation matrix and adjusting
Origin of formation waters in the western Canada sedimentary basin-111 581

Table 5. R-mode biquartimin factor matrix of chemical data and salinity for 78
formation waters from oil fields and gas fields of Alberta, Canada*

Factor
Variable
1 2 3 4 5 6 7 8

Ca 0.877
cu 0.992
Fe 0.925
K 0.944
Li 0.955
Mg 0.907
Mn 0.867
Na 0.871
Rb 0.950
Sr 0.851
Zn 0.912
Br 0.796 0.363
Cl 0856 0.207
HCO, -0.334 0.759
I 0.425 0.738
so.4 0.575 0.664
Total dissolved
solids 0.891
Eigenvalue 8.639 1.020 0.940 0.870 0.843 0.810 0.642 0.514

* Row vectors normalized for 8 factors. Loadings <0.2 omitted.

3 4 5 a
Fe in

Fig. 2. Diagrammatic representation of factors from R-mode biquartimin oblique


solution of chemical data and salinity for 78 formation waters from oil fields and gas
fields of Alberta, Canada.
582 B. HITCEON, GALE K. BILLINCW and J. E. KLOVAN

Table ti. R-mode varimax factor matrix of chemical data, with sdinity partialfed
out, for 78 formation watws from oil fields and g&s fields of Alberta, Canada*

Factor

Variable 1 2 3 4 5 6 7 8 9 10 Communafity

Cs, 0.796 -0*410 - 0.207 0.890


cu O-985 0.993
Fe -I+289 0.226 0.883 O*Q81
K 0.807 -0.232 - 0.326 0.288 0.949
Li 0.596 -0.278 - 0.292 0.420 0.822 - 0.206 0.887
WI - 0.960 0.972
Mn 0.952 0.981
Nrt -0.722 -0.579 0.931
Rb 0.835 -0.259 0.204 0.853
SF 0.384 0.497 - 0.222 -0.268 -0~529 0.884
Zn 0.972 0.998
Br 0.909 0.219 0.915
Cl -0.268 0.708 -0.505 0.904
HCO, 0.941 0.943
I -0.281 0.806 -0.20” 0,810
SO* O-268 0.893 0.933

Eigenvalue 4.979 2.472 I.890 1.312 0.974 0.909 0.733 0.593 0.344 0.416

(Principal
OOIIlpO”-
td)

Per cent of vrarisnce exdained bv factor


21.02 15.51 6.87 6.51 6.54 7.82 9.03 8.53 4.75 6.06

Cumrdstive per cent of variance


-.
21.02 36.53 43-40 49.91 56.45 64.27 73.30 81.83 86-58 92*64

* Loadings CO.2 omitted.

the allal~~i~al values accordingly. The varimax f&or matrix for the B-mode with
salinity partia’lled out is shoun in Table 6. Only the eigenvalue for the tenth factor
is less than 0.5, and the high communalities indicate that most of the variance of
the ions is accounted for by the ten factors extracted, which cun~~~latively ex-
plain nearly 93 per cent of the variance among the variables.
The corresponding factor matrix using the oblique biqunrtimin solution is shown
in Table ‘i and diagrammatically illustrated in Fig. 3. All eigenvalues are close to,
or more than 1.0. A. matrix of correlations between the nine fa,ctors extracted
indica,tes that they are virtually uneorrelated, with no correlation coefficients
exceeding 0*30, Upon rotation to oblique solution, the tenth factor contained no
relevant information and was therefore dropped from further consideration. Factor
1 shows high positive loadings of some of the alkali metals and alka*line earth metals
opposed to a high negative loading for Na. Factor 3 is essentially a halogen factor,
but represents particularly Br and I. Factors 4 and 6 are characterized by high
positive loadings for an anion and high negative loadings for a cation. Factor 3 is
essentially an independent Mg factor, which is opposed by a low negative loading
for Fe. Both Cu and Zn form independent factors. The appearance of Fe in four
factors indicates complex chemical controls. It is dominant only in factor 9, where
it is associated with a moderate loading for Na and Cl. In factor 5 Fe is closely
associated with Mn.
Origin of formation waters in the western Canada sedimentary basin-III 583

Table ‘7. R-mode biquartimin factor matrix of chemical data, with salinity
partialled out, for 78 formation waters from oil fields and gas fields of
Alberta, Canada*

Factor
Variable
1 2 3 4 5 6 7 8 9

Ca 0886 -0.342
cu 0.979
Fe -0.250 -0.226 0.422 0.640
K 0.663 0.252 0.260
Li 0.403 0.276 0.392 0.298 0.204
Mg 0.946
Mn 0.938
Na -0.800 0.342
Rb 0.808 -0.213 0.221
Sr 0.512 0.469 -0.388 0.272
Zn 0.953
Br 0.936
Cl -0.258 0.731 0.338
HCO, 0.911
I 0.866
SO, 0.893

Eigenvalue 3.140 2.524 1.215 1,213 1.173 1.155 1.020 0.962 0.872

* Row vectors normalized for 9 factors. Loadings <0.2 omitted.

I 2

r Br
I
Cl
Mg

K 1
I
sr Sr

Li K Rb

‘,
~ i
Rb

Fig. 3. Diagrammatic representation of factors from R-mode biquartimin oblique


solution of chemical data, with salinity partialled out, for 78 formation waters from
oil fields and gas fields of Alberta, Canada.

The factor results show that although some ions, such as Cu and Zn, occur
independently of all other components, the majority of ions exhibit direct or inverse
relations, to a major or minor extent, with at least one other ion. This strong
interdependence of most of the ions therefore precludes factoring the physical
properties and basic chemical data together if the most simplistic result is desired.
However, the factor scores are a reduced set of new uncorrelated variables that have
584 B. HITCHON,GALE K. BILLINGSand J. E. KLOVAN

essentially the same information content as the original variables. The factor
scores are best thought of as the “amounts” of the factors present in each sample.
In this light, it is evident that the factors can be considered as properties of the
formation waters-properties which combine the complex inter-relations between
the ions in the waters. To determine the relationship between the factors and the
physical properties of the waters, factor scores and physical properties for the 78
samples were entered into a second-order factor analysis. Unfortunately, the use
of a partial correlation matrix to generate the factors for the analyses with salinity
partialled out does not permit the calculation of factor scores for these particular
runs. Therefore the second-order R-mode analysis was carried out using the factor
scores from the varimax solution with salinity (Table 4), pH from Table 2, and the
rest of the physical properties from HITCHON and FRIEDMAN (1969, Table 4). The
resulting R-mode varimax factor matrix is shown in Table 8 and illustrated diagram
matically in Fig. 4. The high communalities indicate that most of the variance
of the physical properties and factor scores is explained by the nine factors extracted
which together account for more than 91 per cent of the variance among the variables.
811 eigenvalues are close to, or more than 1.0. Factor 1 accounts for more than one
third of the cumulative variance and shows high positive loadings of the interpreted
sea water factor with temperature, depth, pressure, ~30~8,6D and Pn2s, and a high
negative loading for pH. We recognize, with RITTENHOUSE et al. (1969), that pH
as measured in the laboratory may not be representative of pH at reservoir condi-
tions, but believe that because pH is statistically related to composition (factor 1)

5 6
! 8
-
‘0. -i C” -ii *In -FACTOR

P”

Fig. 4. Diagrammatic representation of factors from second R-mode of physical


properties and factor scores for 78 formation waters from oil fields and gas fields of
Alberta, Canada.
Table 8. R-mode varimax factor matrix of physical properties and factor scores for 78 formation waters from oil
fields and gas fields of Alberta, Canada*

Factor
Variable Com-
1 2 3 4 5 6 7 8 9 munality

PH - 0.585 0.506 - 0.405 - 0.359 0.912


6D (yc SMOW) 0.842 -0.276 0.836
601s (%, SMOW) 0.870 -0.287 0.880
Depth (feet) 0.924 0.918
Pressure (psi) 0.917 - 0.303 0.956
Temperature (“C) 0.947 0.928
Fluid Potential (feet) 0.251 -0.857 0.287 0.901
- 0.303 - 0.254 0.789
3sccz (psi)
(psi) 0.722
0.418 0.433 0.424 -0.314 -0.307 - 0.203 0.833
Factor 1 (sea water) 0.800 0.423 0.232 0.910
Factor 2 (Fe) 0.995 0.973
Factor 3 (Halogens) 0.901 0.890
Factor 4 (Cu) 0.992 0.987
Factor 5 (Zn) 0.979 0.967
Factor 6 (HCO,) 0.976 0.964
Factor 7 (Mn) 0.984 0.970
Factor 8 (SO,) 0.967 0.956

Eigenvalue (Principal component) 5.907 1.977 1.446 1.239 1.150 1.029 1.002 l*OOO 0.819

Per cent of variance explained by factor

34.55 7.78 8.75 7.67 6.66 6.57 6.06 7.13 6.41

Cumulative per cent of variance

34.55 42.33 51.08 58.75 65.41 71.98 78.04 85.17 91.58

* Loadings to.2 omitted.


586 B. HITCRON,GALE K. BILLINGSand J. E. KLOVAN

the changes in pH have been systematic, as demonstrated by RITTENHOUSE et al.


(1969, Fig. 8). The HCO,-factor is loaded with pH. The SO,-, Zn-, Mn-, and
Fe-factors show high positive loadings, and are opposed by moderate negative
loadings for Puzs ; Pcoz ; Pco, and pH; and PEIZs and Pcoz, respectively. The
halogen factor exhibits a high positive loading, together with a moderate positive
loading for Pco,. The Cu-factor is independent. Factor 3 indicates that fluid
potential has moderate positive loadings with pH and pressure, together with
moderate negative loadings with Pco,, and the interpreted sea water factor,

FACTORS CONTROLLINGCHEMICAL COMPOSITION


The results of the factor analyses are now interpreted in terms of factors that
control the chemical composition of the formation waters.

Sets water origin


The most compelling evidence for an ultimate origin from sea water of the major
portion of dissolved salts in the formation waters of western Canada is the fact that
in the sedimentary basin 85 per cent of the rocks were deposited under marine
conditions, and the volume-weighted mean formation water has a composition
similar to that of present day sea water (Table 1). In the Q-mode ternary diagram
(Fig. 1) we indicate the condition of deposition of the strata from which the formation
waters were obtained. Only one formation water (No. 36) from strata deposited
under marine conditions and which was not from an active recharge region, had a
salinity less than that of present day sea water. We interpret this forma’tion water
as an extreme membrane filtered water. This interpretation is supported by the
very low K/Na ratio of 0.002 as expected for the effluent of a selective membrane
filter (BILLINGS ek al., 1969). Formation waters with salinities more than present
day sea water and which are from strata deposited under brackish water conditions
(No. 10, 30, 33, 34, 58) are all from the Lower Cretaceous Mannville Group in regions
where hydrodynamic study (HITCHON, 1969a, b) indicates that invasion of saline
waters from the underlying Devonian strata has taken place. In other words,
except where fresh water recharge has taken place, or the formation water can be
recognized as an extreme membrane filtered water, strata deposited under marine
conditions contain formation waters more saline than present day sea water.
Further, except where invasion by saline waters has taken place, strata deposited
under brackish water conditions contain formation waters less saline than present
day sea wa’ter. This single sea water source for the major portion of the dissolved
salts in the formation waters of western Canada accounts for the dominance of
factor 1 and factor 2 (nearly 92 per cent of the variance) in the Q-mode analysis
(Table 3) and for the homogeneity of the samples (Fig. I), as well as for the importance
of factor 1 in the R-mode analysis (Table 5; Fig. 2).

Dilution by fresh water recharge


Study of fluid flow in the western Canada sedimentary basin (HITCHON, 1969a, b)
and of the deuterium content of formation waters (HITCHON and FRIEDMAN, 1969)
indicates extensive mixing of modified sea water with fresh water at the same
latitude, and that the movement of fresh water through the basin has redistributed
Origin of formation waters in the western Canada sedimentary basin-111 587

the dissolved salts of the modified sea water into the observed salinity variations.
In the Q-mode ternary diagram (Fig. 1) those arrows which show flow from factor 2
towards factor 1 represent fresh waters entering the basin from recharge regions.
As the fresh water mixes with the saline waters of the basin the content of dissolved
salts and deuterium decreases (Fig. 1, this paper; HITCHON and FRIEDMAN, 1969,
Fig. 8).

Membrane Jiltration
Abundant and convincing evidence for membrane filtration in geologic environ-
ments has been thoroughly reviewed by VAN EVERDINGEN (1968) and BERRY (1969).
With respect to the western Canada sedimentary basin, BILLINGS et al. (1969)
discussed selective mobility of alkali metals during membrane filtration, and HITCHON
and FRIEDMAN (1969, Fig. 8) presented qualitative evidence for isotopic fractionation
of deuterium on passage of water through micropores in shales. In the Q-mode
ternary diagram (Fig. 1) those arrows which show flow from factor 1 towards factor
2 are for formation waters moving from the deeper parts of the Alberta basin,
through shale ultrafilters updip, out of the basin. Along each flow path the more
saline formation waters, that is those closer to factor 1, are found on the upflow
side of the shale ultrafilters and have consequently increased contents of many ions,
as well as of total dissolved solids. The formation waters that have passed through
the ultrafilters, that is those closer to factor 2, have lower salinities, and in the case
of sample No. 36, which is an extreme membrane filtered water, a salinity consider-
ably below that of sea water. Q-mode analysis is insufficiently sensitive to allow
distinction of relative mobility of ions in the membrane filtration process. In the
second-order R-mode analysis (Table 8 ; Fig. 4), high positive loading for the sea
water factor, temperature, depth, pressure and 6D (% SMOW) are opposed by a
high negative loading for pH in the first factor, which accounts for over one third
of the cumulative variance. This bipolar factor is consistent with increased
membrane efficiency with depth, pressure and temperature, with the mixing
hypothesis for deuterium presented by HITCHON and FRIEDMAE (1969), and with
the observations of WHITE (1965, p. 352) on the decrease of pH in membrane
concentrated formation waters. The high positive loading for SOla (%, SMOW)
reflects the temperature dependent exchange of 0 l8 between formation water and
carbonates (HITCHON and FRIEDMAN, 1969) and is unrelated to the membrane
filtration process.

Solution of halite
The major gain to the volume-weighted mean formation water (Table 1) may
be attributed to Na and Cl. Inspection of the ionic balance in the net change
column suggests the solution of halite. Evaporites comprise 5.7 per cent of rocks
in the western Canada sedimentary basin (HITCHON, 1968), of which nearly 75 per
cent are of Middle Devonian age, consisting predominantly of halite. Many authors
have demonstrated evidence for salt solution, and much of the pertinent literature
has been reviewed and expanded on by HOLTER (1969). In the main deep area
affected by solution in southern Saskatchewan, approximately 10,000 square miles
have been delineated which were originally underlain by salt at an average thickness
588 B. HITCHON,GALE K. BILLINGSand J. E. KLOVAN

of about 250 ft (HOLTER, 1969, Fig. 10). This represents a loss of about 475 cubic
miles of salt, which would yield a solution of 16,160 “g/l. NaCl if dissolved in the
pore volume of the western Canada sedimentary basin (63,600 cubic miles). The
gain to the volume-weighted mean formation water corresponds to 9100 “g/l. NaCl
indicating that excess dissolved evaporites have left the basin and are not in the
formation waters. There is abundant evidence to demonstrate that salt solution
took place over long periods of time (HOLTER, 1969, Table II) with maximum
solution probably occurring in post-Jurassic time, after Laramide erogenic move-
ments initiated the present hydraulic system. Further, chemical, isotopic and
hydrodynamic data have been evaluated to demonstrate solution of the shallow
updip margin of Middle Devonian evaporites at the present day (HITCHON et al.,
1969). This latter study supports the authors’ contention that, at least at the present
day, brines from dissolved evaporites and not formation waters are being lost from
the basin. The net change of ions other than NaCl suggests that this may have
been a long term phenomenon, and that, in fact, much of the dissolved salts of the
original sea water are still present in the basin, unlike the original deuterium of the
same formation waters, which is gradually being replaced by deuterium from local
precipitation (HITCHON and FRIEDMAN, 1969). Summarizing the geological evidence
for addition of NaCl to formation waters in the western Canada sedimentary basin
we conclude that sufficient halite has been dissolved from Middle Devonian eva-
porite beds to account for the observed gain in NaCl of the volume-weighted mean
formation water, and that the balance of halite dissolved from bedded evaporites
in the basin has been, and is being, lost to the surface.
A weak Na-Cl loading occurs in factor 9 of the R-mode with salinity partialled
out, and accompanies a moderate loading for Fe (Table 7, Fig. 3). The development
of channel flow in the low fluid-potential drain lying above Middle Devonian strata
(HITCHON, 1969a), in which the majority of halite is found, minimizes the effects
of halite solution in post-Middle Devonian rocks. Also, only in the main area of
salt solution are the formation waters in the strata overlying the Middle Devonian
halite strongly influenced by brines from the dissolved salt (HITCHON, 197 1, Fig. 4).
Outside the main area of salt solution fluid flow in Middle Devonian and Lower
Paleozoic strata is updip, and tends to be parallel to the Precambrian basement in
accordance with the T&h-Freeze-Witherspoon fluid flow model [see HITCHON
(1969a, b), HITCHON (1971) and HITCHON and HAYS (1971) for further details of the
model]. The weak NaCl loading thus results in part from the hydrodynamic
situation but also because we did not include analyses from Middle Devonian and
Lower Paleozoic strata in our study, for reasons noted previously. Further confirma-
tion that the effect of halite solution in the samples studied is minimal may be
obtained by examination of a scatter diagram showing the relation of Br and total
dissolved solids (BILLINGS et al., 1969, Fig. 1). Only one sample (No. 23) fell suffi-
ciently far inside Group III brines of RITTENHOUSE (1967) to possibly justify its
classification in that group. Yet the presence of a weak Na-Cl loading could not
be due to only one sample. This would suggest that plots of Br and total dissolved
solids are sufficiently sensitive to distinguish saline formation waters that originate
through solution of halite by fresh water [Group III; for example, samples 74A
and 75A in Fig. 1 of BILLINGS et al. (1969), are the two saline springs described by
Origin of formation waters in the western Canada sedimentaq basin-III 589

HITCHONet al. (1969, Table l)], but generally cannot distinguish between a membrane
concentrated formation water and the same formation water with a minor addition
of brine from solution of halite by that formation water. The moderate loading
for Fe may indicate a source from the red beds closely associated with Middle
Devonian halite. BOIKO (1966) found that Li and Sr were concentrated in salt-
bearing clays and marls associated with evaporites. Solution of halite would be
accompanied by flushing of the associated clays and marls, thus possibly giving
rise to the weak Sr-Li loading in factor 9 (Fig. 3).

Ilolomitixation

Many authors have observed an increase in Ca and decrease in Mg in individual


saline formation waters, relative to present day sea water, and attributed this effect
to the part played by the formation water in dolomitizing adjacent limestones. In
factor studies carried out by COLLINS(1967) DAWDY and FETH (1967) and SAMPSON
(I968), using a variety of techniques, Ca and Mg were found to be consistently
loaded together. LEE (1969) found 3 factors with Ca and Mg loaded together, and
one strongly bipolar Ca-Mg factor which he did not attempt to interpret. From his
study of Louisiana formation waters COLLINS (1970) concluded that dolomitization
probably had occurred and that this mechanism was important in controlling the
amounts of Ca, Mg and Sr in the formation waters. However, his factor analysis
failed to yield a clear indication of the relation of Ca and Mg. Although KRAMER
(1969) factored formation waters from different rock types separately, he found the
groupings to be very similar regardless of rock type, and that Ca and Mg were
loaded together against HCO,. He interpreted these loadings in terms of carbonate
equilibria. The present study shows high positive loadings for Ca and Mg in the sea
water factor (Table 5, Fig. 2), but in the R-mode with salinity partialled out (Table
7, Fig. 3), Ca is opposed by Na in factor 1, and by HCO, in factor 4, with Mg weakly
opposed by Fe in factor 3. Clearly, Ca and Mg take part in several reactions, and
their distribution in formation waters cannot be attributed solely to dolomitization
or factor analysis would indicate opposite loadings on the same factor.
The ionically balanced gain and loss of Ca and Mg, respectively, in the volume-
weighted mean formation water (Table 1) suggests the dolomitization of limestone,
If we assume that all Mg lost from the volume-weighted mean formation water
(63,600 cubic miles with 973 mg/l. Mg) has been used in the dolomitizing reaction,
then about 190 cubic miles of limestone could be converted to dolomite. The volume
of dolomite in the western Canada sedimentary basin is 62,950 cubic miles (HITCHON,
1968, Table III), of which an unspecified volume may be primary dolomite. There-
fore, even if loss of Mg has resulted from dolomitization of limestone, the volume of
limestone dolomitized represents only an insignificant fraction of the total volume
of dolomite now present in the basin. This also means that the amount of secondary
calcite caused by such reaction is insignificant. Since much of this dolomite was
originally deposited as limestone we conclude that either the process of dolomitization
was essentially a penecontemporaneous reaction, or it took place after deposition at
a time when the limestone was exposed at the surface by erosion, for example, in
Early Cretaceous time. We surmise that in both these cases Ca from the dolomitizing
590 B. HITCHON, GALE K. BILLINGS and J. E. KLOVAN

reaction would not enter the deeper formation waters in significant amounts. In the
first case it would be lost to the ancient sea water, and in the second case would form
part of a near-surface groundwater flow regime, and be lost in local disoharge
regions on the paleosurface.
Summarizing the evidence for dolomitizing reactions from the volume-weighted
mean formation water we conclude that if loss of Mg and gain of Ca is due to dolomit-
ization, then the amount of dolomite formed is an insignificant portion of the dolomite
now present in the basin, and therefore whenever most of the dolomitization did
occur it must have been at times and in situations where Ca from the dolomitizing
reaction was not incorporated into the deeper formation waters.

Bacterial reduction of sulphate

Reduction of SO, through bacterial action in the subsurface has been cited by
many authors to account for low SOa contents of many formation waters and the
presence of H,S in some natural gases. If we assume that all SO, lost from the
volume-weighted mean formation water (63,600 cubic miles with 2360 mg/l. SO,)
were to be bacterially reduced to H,S, the reaction would generate 5.1 x 1021 cubic
feet of H,S. In the subsurface, H,S is present as free gas and in solution in formation
water. In Alberta, known reserves of sulphur from natural gases rich in H,S have
been officially estimated at 120 million tons, equivalent to 2.4 x 1Ol5 cubic feet of
H,S. HEnl (1959) states that less than 1 mg/l. H,S in water imparts a strong odor.
Many formation waters do not contain H,S, but even if we assume that, on average,
all formation waters contained 1 mg/l. H,S, this accounts for only 4.8 x lo’* cubic
feet H,S. These calculations indicate that even using the most generous and reason-
able allowances for ultimate sulphur reserves and adjusting for the amount of H,S
held in solution through increased pressures in deep strata, the total volume of H,S
present falls short by three to five orders of magnitude of that which might be
generated from the total SO, loss. HOSLER and KAPLAN (1966) indicate substantial
changes in the SO4 content of sea water over geologic time. Even though their
percentage changes range from 45 per cent more SO4 in Permian seas to 25 per cent
less SO, in Early Paleozoic seas, these amounts are insufficient to significantly affect
the calculations. It is also possible that the SO, content of the original sea water
incorporated into the sediments was less than that of ancient open sea water because
of prior precipitation of gypsum in bedded form or as early authigenic cements.
In the R-mode with salinity (Table 5, Fig. 2) SO, has a moderate positive loading
with other ions in the sea water factor (1) but also occurs independently with a high
positive loading in factor 8. The second-order R-mode (Table 8, Fig. 4) shows a weak
bipolar relation between factor 8 and PHZS. These relations suggest that only a minor
part of the SO4 loss must be attributed to conversion to H,S, possibly by bacteria
which, however, would appear to be sufficient, as shown above, to account for the
known H,S reserves. However, most of the H,S in natural gases probably originates
by conversion from S-bearing organic compounds. The remaining SO, is controlled
by solubility relations (Table 7, factor 5) which probably account for the major part
of the SO, loss, and which we discuss below.
Origin of formation waters in the western Canada sedimentary basin-III 591

Formation of chlorite
Quantitative calculations for the dolomitizing reaction, and the loading of Ca
and Mg on separate factors in the R-mode with salinity partialled out (Table 7,
Fig. 3), imply that we must seek separate controls on the content of Ca and Mg in
formation waters. The independent, high positive loading of Mg in factor 3 (Table
7, Fig. 3), together with the loss of Mg determined from the volume-weighted mean
formation water (Table 1) suggests the removal of Mg from formation water through
generation of minerals such as chlorite. Few other reactions could explain the loss
of Mg independent of other ions measured. The formation of chlorite from mont-
morillonite, and especially from kaolinite (which is devoid of magnesium), requires a
considerable addition of Mg. Starting with the average composition of montmoril-
lonite (1~7A1,0,.0~6Mg0.8Si0,.2K,O) observed by Ross and HENDRICKS (1945),
and including the maximum amount of @3Mg that can be absorbed, ECKHARDT
(1958) obtained the following equation :

1~7A1,0,.0~9Mg0.8SiO~.2H~O + 9*2MgO + 6H,O =


10~1Mg0.1~7A1,0,.6~4Si0,.8H,O + 1*6SiO,.

Thus for every molecule of chlorite formed by this reaction, 9.5MgO have been added.
In terms of Mg, this corresponds to 20.8 per cent by weight of the chlorite molecule.
If all the loss of Mg from the volume-weighted formation water is attributed to the
conversion of montmorillonite to chlorite, then the weight of Mg lost (973 mg/l. in
63,600 cubic miles) will form 106 cubic miles of chlorite with a mean density of 2.8.
This corresponds to less than 0.94 per cent by volume of shales in the western
Canada sedimentary basin. Cretaceous shales of Alberta contain only low amounts of
chlorite, but some Lower Cretaceous sandstones in the foothills contain considerable
quantities of early Mg-rich authigenic chlorite (CARRIGY and MELLON, 1964 ; MELLON,
1964; MELLON, 1967). Up to 15 per cent of probably detrital chlorite has been
found in Upper Devonian shales of central Alberta (CAMPBELL and OLIVER, 1968),
although in correlative strata PELZER (1966) felt that the clays, including chlorite,
may have an origin other than detrital. It is clear that loss of Mg from formation
waters may easily be accounted for by generation of authigenic chlorite from mont-
morillonite. HILTABRAND (1970) showed clearly that modern argillaceous sediments
remove at least 100 mg/l. Mg from sea water at 100°C to create Mg-clays.
MILLER (1967) observes that chlorites formed by halmyrolysis in the first stages
of diagenesis should be Mg-rich chlorites or sudoites, but that most sedimentary
chlorites have high Fe contents. ECKHARDT (1958) calculates that the Mg content of
the interstitial water (essentially sea water) is not sufficient for the formation of pure
Mg-chlorites, but if the divalent Fe which is present in the sediment in sufficient
quantities is included, then the formation of chlorites containing Fe in the later
stages of diagenesis can easily be explained. This suggests that factor 3 (Fig. 3)
represents the conversion of Mg-chlorites to Fe-chlorites, as well as the generation
of new chlorite from montmorillonite.

Cation exchange on clays


Cation exchange characteristics of clay minerals depend on type of clay, avail-
ability of exchange sites, strength of bonding of the exchangeable ions, activity of
592 B. HITCHON, GALE K. BILLINGS and J. E. KLOVAN

ions in solution, type of adsorbed ions, degree of compaction, pressure, and tem-
perature. For these reasons there is no universally applicable series of preference in
cation exchange. VAN EVERDINGEN (1968) has reviewed various preference series
and concluded that Ca is preferred over Na in dispersed-low concentration systems,
and Na preferred over Ca in dispersed-high concentration systems and in compacted
systems. In either case, the bipolar nature of Ca and Na in factor 1 of the R-mode
with salinity partialled out (Table 7, Fig. 3) is probably best explained through
cation exchange on clays. The much higher concentrations of Na and Ca in formation
waters may account for their dominance in cation exchange reactions in sedimentary
rocks, and hence for their high loadings in factor 1. Smaller non-hydrated ionic
I . -1

. l...
.
. .

. .
. . l

. .*

..

l .

.- .m l
i
.
. . l .

1
.

.
.* l .
l .
. .* :
.
. l

. .
p .
0.
.
0. l
!
. . ..

.
,o L--L ,
-33 ~20 -10

FACTOR SCORE

Fig. 5. Scatter diagram illustrating the relation of formation temperature and


factor score for the cation exchange factor in 78 formation waters from oil fields
and gas fields of Alberta, Canada.
radius, larger hydrated ionic radius, and larger polarization accompany lower
replacing power. The order Rb-K-Sr-Li-Mg is one of generally decreasing non-
hydrated ionic radius, and generally increasing hydrated ionic radius and polarization.
Thus, with the exception of the dominant exchange cations Ca and Na, the factor
loadings decrease with decreasing replacing power, and this relation may explain
the extremely low positive loading for Mg (<O-2 and therefore not shown in Table 7
or Fig. 3). Cation exchange proceeds more readily with increasing temperature and
a scatter diagram (Fig. 5) indicates increasing factor score (based on R-mode in
Table 6), and hence increasing degree of cation exchange, with formation temperature.
Origin of formation waters in the western Canada sedimentary basin-111 593

Contribution from organic matter


The relative importance of organic matter or the clay fraction of argillaceous
rocks as a primary source for Br and I in formation waters has been a subject of
debate for many years. Some authors recognize organic matter as the source for
both Br and I (COLLINS and EQLESON, 1967; KREJCI-GRAF, 1963a, b; NIKANOROV,
1966; WALTERS, 1967), while KOZIN (1960) believes desorption from clays is im-
portant for both halogens. SMIRNOVA(1966, 1969) explained Br in formation waters
by leaching of its disseminated form in rocks, and like COLLINS (1969) found an
association of I and organic matter in formation waters. Unfortunately, our factor
study does little to resolve this dilemma.
In the R-mode with salinity (Table 5, Fig. 2), I, Br and Cl are present in the sea
water factor (1) and also form an independent halogen factor (5), but with the order
of factor loadings reversed. In the sea water factor, the order of factor loadings may
represent the order of abundance of these halogens in sea water (Table l), whereas
in the halogen factor (5) the order of factor loadings could be interpreted in terms of
relative concentration in the volume-weighted formation water, compared to sea
water, thus accounting for the high positive loading for I (concentration factor 150)
compared to those for Br and Cl (concentration factors 1.7 and 1.4, respectively).
When salinity is partialled out (Table 7, Fig. 3), the halogens exhibit high positive
loadings in factor 2, associated with a moderate positive loading for Sr, and opposed
by a low negative loading for Rb. It is difficult to interpret the loadings for Sr and
Rb, other than to suggest that their presence in the halogen factor indicates that
desorption of halogens from clays is accompanied by expulsion of Sr and fixation of
Rb. The effect of partialling out salinity has increased the loadings for Br and I,
which probably reflects the increased release of Br and I from clays into formation
waters when the Cl concentration of the formation water becomes sufficiently high
(KOZIN, 1960). Leaching of different rock types by water was studied experimentally
by BUDZINSKII (1967) who found that the action of carbonated water facilitated the
movement of halogens from the rocks to the solution. This may explain the moderate
positive loading of Pco, with the high positive loading of the halogen factor in the
second R-mode (Table 8, Fig. 4).
Evidence from our factor study seems to support an origin of Br and I primarily
from the clay fraction of argillaceous rocks. However, we did not include in our
variables any component representative of the organic matter in formation waters,
nor of organic matter in the rock matrix for the second-order R-mode analysis. We
conclude, therefore, that the clay fraction of argillaceous rocks probably makes an
important contribution to the Br and I contents of formation waters, but a significant
contribution from organic matter, either in the rock matrix or the formation water,
cannot be ruled out.

Solubility relations
The mutual presence of some ions in formation waters exerts control on their
concentration through a mineral solubility product. The carbonates and sulphates
of some of the alkaline earth metals are particularly susceptible to this control.
KRAMER (1969) found factors which he interpreted as representing the following
594 B. HITCHON,GALE K. BILLINGSand J. E. KLOVAN

equilibria, :
CaMg(CO,), + 2H+ = Ca2+ + Mg2+ + 2HCO,-
CaCO, + H+ = Ca2+ + HCO,-
(Ba - Sr)SO, = (Ba - Sr)2+ + SOa2-
(Ba - Sr)CO, + H+ = (Ba - Sr)2+ + HCO,--.

We interpret the high positive loadings of HCO, and SO,, and the moderate
negative loadings of Ca and Sr, respectively, in factors 4 and 5 of the R-mode with
salinity partialled out (Table 7, Fig. 3) in terms of the above equations. Factors 7
and 8 in the R-mode with salinity (Table 5, Fig. 2) generally correspond to factors
4 and 5 in Table 7 and Fig. 3. KRAMER (1969) noted that failure to include Hf in his
factor analysis may have resulted in the relatively weak factor he found relating Ca,
Mg and HCO,. We find the same situation in our R-mode with salinity, but upon
second factoring with physical properties observe high loadings of the HCO,-factor
and pH in factor 2 (Table 8, Fig. 4). Unfortunately Ba was not determined, but had
it been included in the factor study, the work of KRAMER (1969) indicates that it
would be present with a high negative loading in factor 5 (Table 7, Fig. 3), and
possibly present with a negative loading in factor 4. Barite (BaSO,) is present as an
interstitial mineral in sedimentary rocks in western Canada, whereas celestite
(SrSO,) has not been reported. This situation reflects the much lower solubility
product of BaSO,, relative to SrSO,.
Another possible explanation for factor 5 in the R-mode with salinity partialled
out (Table 7, Fig. 3), and for the large gain in Sr of the volume-weighted formation
water, compared to Ca, is the introduction of Sr from the diagenetic recrystallization
of aragonite to calcite, and the replacement of aragonite by anhydrite. This may
explain the opposed loadings of Sr and SO, in factor 5; and Ca, occurring in both
phases, would not show on this factor. Unfortunately factor scores were not available
for the R-mode with salinity partialled out, but one might anticipate high scores for
this factor would be associated with carbonate rocks.

Other controls
The trace metals Cu, Fe, ML and Zn tend to occur as separate factors, although
Mn and Fe, which behave in a similar geochemical manner, are loaded together in
some factors (Table 7, Fig. 3). In the second-order R-mode, the Zn-, Mn-, and Fe-
factors are opposed by a weak Pcoz loading, which may indicate control through
stability relations in the system R-CO,-CO, 2--HCO,-, where R represents Zn, Mn
or Fe. The low negative loadings for pH and P nzS opposite the Mn- and Fe-factors
suggests possible control of Mn through hydroxides and Fe through sulphides. The
complete independence of Cu remains unexplained.

COMMENT
Throughout our study we have assumed that the dissolved species are present
as individual free ions. This may be true in extremely dilute aqueous solutions, but
there is abundant evidence for extensive formation of ion-pairs in our starting
material-sea water (GARRELS and CHRIST, 1965, Chap. 4). Theoretical considera-
tions, and examination of natural brines (TRUESDELLand JONES, 1969), demonstrate
Origin of formation waters in the western Canada sedimentary basin-111 595

enhancement of the tendency to form ion-pairs with increased salinity. It was


beyond the intent of the present paper to calculate or measure the degree of formation
of ion-pairs in the formation waters under study, and thus no consideration was
given to the effect of ion-pair formation on the factors. It is sufficient to observe
that ion-pairs undoubtedly exist in the formation waters but that their influence in
controlling the chemical composition is not known.

ORIGIN OF FORMATIONWATERS IN WESTERNCANADA


The final section presents a succinct, possibly dogmatic, summary of this series of
papers on the geochemistry and origin of formation waters in the western Canada
sedimentary basin. Ancient sea water is the source of the majority of deuterium,
oxygen-18 and dissolved salts in formation waters of western Canada. The deuterium
of the original sea water is gradually being mixed with deuterium from surface
water under the present hydraulic system. The distribution of deuterium has been
derived through mixing of the diageneti~ally modified sea water with not more than
2.9 times, and possibly as little as I.13 times, as much fresh water at the same latitude.
Movement of this fresh water through the basin has redistributed the dissolved salts
into the observed salinity variations. There has been fractionation of deu~rium on
passage of water through micropores in shales, and extensive enrichment of oxygen-
18 through exchange of oxygen isotopes between water and carbonate minerals.
The volume-weighted mean formation water has a composition similar to sea
water. The salinity is l-3 times that of sea water with the major part of the gain
resulting from solution of Middle Devonian bedded halite. With the exception of
Mg and SO,, all components show a net gain, compared to sea water. Relative to the
present pore volume there has thus been negligible removal of dissolved salts, although
there has been a major redistribution. Saline springs at the regional discharge result
from solution of halite and gypsum by meteoric water, and not from loss of formation
water. The major factor causing concentration of the dissolved salts is membrane
filtration. Together with dilution by fresh water recharge this results in a chemical
population ranging from fresh water to membrane concentrated brines. Dolomitiza-
tion caused by formation waters has not been an important factor in controlling
the relative propo~ions of Ca and Mg in the formation waters, and neither has
bacterial reduction of SO, played a major role in the loss of SO, from the formation
waters. Rather, Mg may have been used in the generation of chlorite and the amount
of SO, controlled by the ~lubility of SrSO, and BaSO,, based on the work of KRAMER
(1969). Further, the gain in Sr in the volume-weighted mean formation water is
possibly from the aragonite to calcite diagenetic recrystallization. Redistribution
of some alkali metals and alkaline earth met&Isbetween rock matrix and formation
water has taken place through exchange on clays. Desorption of Br and I from
clays also occurred, although a contribution of these halogens from organic matter
cannot be ruled out. CaCO, solubility equilibria have been demonstrated. The trace
metals Cu, Fe, iUn and Zn have all been concentrated in formation waters but their
mode of concentration remains equivocal.
In summary, the formation waters of western Canada represent ancient sea
waters in which deuterium is being changed by mixing with deuterium from surface
water, oxygen-18 has been extensively exchanged with carbonates of the rock
596 B. HITCHON, GALE K. BILLINOS and J. E. KLOVAN

matrix, and following redistribution by membrane titration and dilution by fresh-


water recharge, the dissolved salts have attained equilibrium with the rock matrix.
The processes include solution of evaporites, formation of new minerals, cation
exchange on clays, desorption of ions from clays and organic matter, and control by
solubility of minerals.

Acknowledgements-We reiterate our appreciation to those who assisted in the collection end
analysis of the samples and who are duly acknowledged in the two previous parts of this series
of reports. Critical and constructive reviews of the manuscript were provided by Dr. D. W.
SPENCER, Woods Hole Oceanographic Institution, and Dr. R. GREEN, Research Council of
Alberta. We thank these reviewers for their valuable comments.

REFERENCES
BERRY F. A. F. (1969) Relative factors influencing membrane filtration effects in geologic
environments. Chena. Geol. 4, 295-301.
BILLINGS G. K., HITCHON B. and SHAW D. R. (1969) Geochemistry and origin of formation
waters in the western Canada sedimentary basin, 2. Alkali metals. Chem. Geol. 4, 211-223.
BOIKO T. F. (1966) Rare element distribution in halogenic deposits. Dokl. Akad. Nauk SSSR
171, 457-460.
BRENNER M., NIEDERWIESERA., PATAKI G. and WEBER R. (1965) Theoretical aspects of thin-
layer chromatography. In Thin-layer Chrcwnatogr~hy. A Laboratory Handbook, Chap. H.
Springer-Verlag.
BUDZINSKII Y. A. (1967) Water-soluble rocks of the northern Caucasus in the light of experi-
mental data. Izv. Akad. Nauk SSSR, Ser. Geol. No. 4, 116-126.
CAMERONE. M. (1967) A computer program for factor analysis of geochemical and other data.
Geol. Surv. Can. Paper 67-34.
CAMPBELL F. A. and OLIVER T. A. (1968) Mineralogic and chemical composition of Ireton and
Duvernay Formations, central Alberta. Bu.ll. Can. Petrol. Geol. 16, 40-63.
CARRIGY M. A. and MELLON G. B. (1964) Authigenic clay mineral cements in Cretaceous and
Tertiary sandstones of Alberta. J. Sediment. Petrol. 34, 461-472.
COLLINS A. G. (1967) Geochemistry of some Tertiary and Cretaceous age oil-bearing formation
waters. Environ. Sci. Technol. 1,725-730.
COLLINS A. G. (1969) Chemistry of some Anadarko basin brines containing high concentrations
of iodide. Chem. Geol. 4, 169-187.
COLLMS A. G. (1970) Geochemistry of some petroleum-associated waters from Louisiana. U.S.
Bur. Minea Rept. Invest. 73%
COLLINS A. G. and EGLESON G. C. (1967) Iodide abundance in oilfield brines of Oklahoma.
Science 156, 934-935.
CULKIN F. (1965) The major constituents of sea water. In Comical Oceanography, Vol. 1, Chap.
4. Academic Press.
DaVIS G. H. (1966) Frequency distribution of dissolved solids in ground water. Ground Water
4, No. 4, 5-12.
DAWDY D. R. and FETH J. H. (1967) Applications of factor analysis in study of chemistry of
groundwater quality, Mojave River valley, California. Water Redour. Red. 3, 605-510.
ECKHARDT F. J. (1958) tuber chlorite in sedimenten. Geol. Jahrb. 75, 437-474.
EGLESONG. C. and QUERIOC. W. (1969) Variation in the composition of brine from the Sylvania
Formation near Midland, Michigan. Environ. Sci. Technol. 3, 367-371.
GARRELSR. M. and CHRIST C. L. (1966) Solutione, Minerals, and Equilibria, Harper & Row.
GOLDBERGE. D. (1965) Minor elements in sea water. In Chemical Oceanogrqhy, Vol. 1, Chap. 4.
Academic Press.
HEM J. D. (1959) Study and interpretation of the chemical characteristics of natural water.
U.S. Beol. Survey Water-Supply Paper 1473.
HILT-RAND R. R. (1970) Experimental diagenesis of argillaceous sediment. Unpublished
dissertation, Louisiana State Univ., 162 pp.
Origin of formation waters in the western Canada sedimentary basin-III 597

HITCHON B. (1964) Formation fluids. In Cfeological H&tory of Western Canada, Chap. 15.
Alberta Sot. Petrol. Geol., Calgary, Alberta.
HITCHON B. (1968) Rock volume and pore volume data for plains region of western Canada
sedimentary basin between latitudes 49’ and BOON. Amer. Assoc. Petrol. Geol. Bull. 52,
2318-2323.
HITCHON B. (1969a) Fluid flow in the western Canada sedimentary bmin. 1. Effect of topography.
Water Reeour. Rea. 5, 186-195.
HITCHON B. (1969b) Fluid flow in the western Canada sedimentary basin. 2. Effect of geology.
Water Reeour. Rec. 5,460-469.
HJTCHON B. (1971) Origin of oil: geological and geochemical constraints. In Advalb. Ghem Ser.
101.In press.
HITCHON B. and FRIEDMAN I. (1969) Geochemistry and origin of formation waters in the western
Canada sedimentary basin--I. Stable isotopes of hydrogen and oxygen. Geochim. Cosntochim.
Acta 33, 1321-1349.
HITCHON B. and HAYS J. (1971) Hydrodynamics and hydrocarbon occurrences, Surat Basin,
Queensland, Australia. Water Resow. Res. In press.
HITCHON B. and HOLTER M. E. (1971) Calcium and magnesium in alberta brines. Res. Council
of Alberta, in press.
HITCHON B., LEVINSON A. A. and REEDER S. W. (1969) Regional variations of river water
composition resulting from halite solution, Mackenzie River drainage basin, Canada. Water
Besour. Rea. 5, 139ti-1403.
HOLSER W. T. and KAPL~LNI. R. (1966) Isotope geochemistry of sedimentary sulfates. Chem.
Geol. 1,93-135.
HOLTER M. E. (1969) The Middle Devonian Prairie Evaporite of Saskatchewan. Saskatchewan
Dept. Mineral Resources, Rept. 12%
KAISER H. F. (1960) The application of electronic computers to factor analysis. Educ. Peychol.
Me-t. 20, 141-151.
KOZIN A. N. (1960) Geochemistry of bromine and iodine of formation waters of the Kuybyshev
area on the Volga. Petrol. Geol. 4, 110-113.
KFLAMERJ. R. (1969) Subsurface brines and mineral equilibria. Chem. Geol. 4, 37-50.
KREJCI-GR~ K. (1963a) mer Geol. Mitt. 2, (4) 351-391.
Rumiinische 6lfeldwlisser.
KREJCI-GRAX K. (196313) Diagnostik der salinitiitsfazies der Glwiisser. Fortschr. Geol. Rheinid.
We&f. 10,367-448.
LAYER D. B. (1958) Characteristics of major oil and gas accumulations in the Alberta basin. In
H&tat of Oil, pp. 113-128. Amer Assoc. Petrol. Geol. Tulsa, Oklahoma.
LEE P. J. (1969) Principal component analysis on the connate water of gas and oil fields in
northern Taiwan.Proc. Geol. Sot. China No. 12 for 1968, 121-128.
MCCR~SSAN R. G. and GLAISTER R. P. (Editors) (1964) Geological History of Westerr, Canada.
Alberta Sot. Petrol. Geol., Calgary, Alberta.
MELLON G. B. (1967) Stratigraphy and petrology of the Lower Cretaceous Blairmore and ~nnn-
ville Groups, Alberta Foothills and Plains. Res. Council of Alberta, Bu,ll. 21.
MUELLERG. (1967) Diagenesis in argillaceous sediments. In DiUgene8i8 i)~ Sedimenta, Chap. 1.
Elsevier.
NIKANOROV A. M. (1966) Accumulation and preservation of bromine and iodine in ground
waters of Mesozoic and Cenozoic deposits in the eastern Caucasus. Sowet. Geol. NO. 10, 102-l 10.
PELZER E. E. (1966) Mineralogy, geochemistry and stratigraphy of the Besa River shale,
British Columbia. Bull. Can. Petrol. Geol. 14, 273-321.
H’ITTENHOUSE G. (1967) Bromine in oil-field waters and its use in determining possibilities of
origin of these waters. Amer. Assoc. Petrol. Geol. Bull. 51, 2430-2440.
RmENHoUsE G., FULTON R. B., GRABOWSKI R. J. and BERNARD J. L. (1969) Minor elements in
oil-field waters. Chem. Geol. 4, 189-209.
Ross C. Se and HENDRICKS S. B. (1945) Minerals of the mont,morillonite group. U.S. Geol.
#un,. Prof. Paper 205B, 23-79.
598 B. HITCHOS, GALE K. BILLINGS and ,J. E. KLOVAN

SAMPSONR. J. (1968) R-mode factor analysis program in Fortran II for the IBM 1620 computer.
In Computer Programs for M~t~~ri~te An&&s im theology, 13-38. Computer Contribution 20,
State Geol. Survey, Univ. of Kansas, Lawrence, Kansas.
SMIRNOVAA. Y. (1966) Soumes of bromine and iodine in ground waters of Lower Cretaceous
deposits in western Predkavkaz’e. Izv. Vyssh. Ucheb. Zavedenii, Geol. Razved. 10, 88-90.
YMIRNOVAA. Y. (1969) Genesis of boron, bromine end iodine in ground waters of western
Ciscaucasian Cretaceous deposits. Sovet. i&l. No. 8, 101-111. (Translated in Int. Geol.
Rev. 12, 703-710.)
SPENCERD. W. (1966%) Factor analysis. lFoo& HoEe Oce~~og~~phic I~stit~t~o~ Ref. 66-39.
SPENCERD. (1966bf Factors affecting element d~tributions in e Silurian greptolite bend. Chem.
Geol. 1, 221-249.
SPENCERD. W. (1967) Factor analysis II: A description of the application of oblique solutions,
ll’oods Hole Oceanographic Irastitution Ref. 67-59.
TRUESDELLA. H. and JONESB. F. (1969) Ion association in natural brines. Cherra.GeoE.4,51-62.
VAN EVERDINOENR. 0. (1968) Mobility of main ion species in reverse osmosis and the modifica-
tion of subsurface brines. Can. J. hurts SC& 6, 1253-1260.
WATERS L. J. (1967) Bound heIogens in sediments. Mess. Inst. Tech. AD-663,583.
WHITE D. E. (1965) Saline waters of sedimentary rocks. In Fluids i?&Subsurface _&wironme?ats-
a Sympoeium, pp. 342-366, Mem. 4, Amer. Assoc. Petrol. Geol., Tulsa, Oklahoma.

You might also like