THE REMOVAL OF SOLUBLE SILICA FROM FRESH WATER. ENTERING THE SEPi'' 2

THE REMOVAL OF SOLUBLE SILICA FROM FRESH WATER ENTERING THE SEPi'' 2 GEORGE S. BIEN, DAVID E. CONTOIS, and WILLIAM H. THOMAS Scripps Institutin f Oceangraphy, University f Califrnia, La Jlla, Califrnia ABSTRACT Determinatins f sluble silica and chlrsity f water samples cllected in and arund the east Mississippi Delta shwed that mst f the sluble silica was remved frm the river water by sme prcess ther than dilutin with sea-water. While bilgical uptake by diatms can accunt fr part f this remval, it is mre likely that a majr prtin is remved by inrganic precipitatin. Labratry experiments with river-water and sea-water shwed that bth suspended matter frm the river water and electrlytes in the sea-water are necessary fr maximum inrganic precipitatin. It is cncluded that this prcess is an adsrptin f sluble silica n suspended matter as it cmes in cntact with electrlytes, rather than a simple frmatin f salts with electrlytes. The inrganic remval can increase the weight f sediment in the water which may be expected t reach the bttm by a maximum f 4.5 percent. INTRODUCTION The cncentratin f sluble silica in river-water is generally ten t fifteen times that f surface-water in the pen cean. The sluble silica cncentratin f Mississippi river-water averages 4.0-7.5 ppm, while the average cncentratin in Gulf f Mexic surface sea-water is 0.11 ppm. In spite f the large amunt f sluble silica that is brught int the cean, there is n sign f an accumulatin f sluble silica in surface seawater. Tw hyptheses have been prpsed t accunt fr the depletin f Cntributin frm the Scripps Institutin f Oceangraphy, New Series N. 986. This investigatin was supprted by a grant frm the Anierican Petrleum Institute, Prject 51. This paper is a reprint frm Gechimica et Csmchimica Acta, v. 14, p. 35-54, 1958 with a few mdificatins t make it cnfrm t the publicatin practice f the Sciety f Ecnmic Palentlgists and Mineralgists. The Sympsium n Silica in Sediments is thus a cmplete unit as presented in Ls Angeles. This imprtant part f the sympsium had been submitted fr publicatin prir t presentatin at the Sympsium and it was nt feasible t vvithdraw it. Permissin t reprint the article has been granted in writin,9-, by the Executive Editr f Gechimica et Csmchimica Acta and by the authrs. sluble silica : bilgical remval and inrganic precipitatin. Bilgical remval f sluble silica frm culture slutins by diatms has been shwn by JORGENSEN (1953). Als, ARMSTRONG (1954) and thers have shwn that fllwing the spring "blming" f diatms in the English Channel the cncentratin f sluble silica in surface water was much reduced. KRAUS- KOPF (1956) cncludes that because n disslved silica was precipitated when he added silica slutins t sea-water, bilgical remval must be the main factr in preventing the accumulatin f silica in sea-water. Mrever, during the initial mixing f river-water, befre bilgical remval can ccur, it might he expected that the cncentratin f sluble silica wuld decrease nly because f dilutin. If nly dilutin ccurred, the cncentratin f sluble silica wuld be a linear functin f dilutin, as measured by chlrsity determinatin. Any frm linearity wuld be a deviatin measure f the extent f remval. Such a linear relatinship was fund by MAEDA and his assciates (MAEDA 1952, 1953; MAKIMOTO et al., 1955). Hwever, their data nly cver the

[ 1,11 1 _N.1,.. 1111 1, REMOVAL OF SOLUBLE SILICA FROM FRESH WATER 21 _ - 89.'10 89 00 88.'50' I! N 1 2 3 4 5 6 7 8 *C-7 STATUTE MILES *C-6 ' 29' ' *** - 29* ' **.,... t 29 ' 10'. c -i - i 1410).'' V a. 6 Ile, ' t ' ',. i,.,... / 7 6 at,----. * - S lwv. cf. 10/11/,-aa C-2 IA t C-3 A-7.N. 'B II A+V),.., 8-I3 8-5N+ ii, 6 B-8 'B-II 0,4.8 A..13,A1+ 31'b 214 -.,...A.7,.-2 A-9A-ti? -9..'-p-2 A-12' A+-184 A-15 444 + B,10 IN., 4, C-4 + B-I5 +B-14 c. 5 0A-16 29 ' 10 0, C - 8 89'10' :: 1 89' 00' 88.'50' FIG. 1. Chart shwing statin lcatins in the vicinity f Pass a Lutre sub-delta f the Mississippi River. Statins B-.6, B-17, and B-18 are respectively 4, 8, and 12 statute miles east f Statin B-15. chlrinity range frm 5-171/4, which des nt represent the cmplete range. We have had the pprtunity t test the dilutin hypthesis at chlrinities f 0.1-21V in the area f the eastern Mississippi River Delta at Pass a Lutre, where the river empties int the Gulf f Mexic. We have als made measurements f phytplanktn prductin frm which we have attempted t calculate the magnitude f silicate remval by phytplanktn and furthermre we have studied pssible inrganic remval prcesses in the labratry, using water samples frm the Delta. Our results are reprted in this paper. METHODS AND MATERIALS Water samples were btained in the field at statins shwn in figure 1. One prtin f each sample was drawn ff int a glass citrate bttle, stppered, and returned t the labratry fr the determinatin f chlrinity by titratin with silver nitrate. Chlrsity was calculated frm this determinatin.

22 G. S. BIEN, D. E. CON TOIS, AND W. H. THOMAS A secnd sub-sample was drawn frm each Nansen bttle and filtered thrugh a mlecular filter (Millipre, type HA) having an average pre size f 0.45 R. The filtrate was placed in a plyethylene bttle and returned t the labratry fr analysis f sluble silica. Experiments with sdium silicate slutins f knwn cncentratin indicated that n significant increase in sluble silica cncentratin ccurred due t cntact f water samples with the glass f the filter hlders and evacuatin flasks. Sluble silica cncentratin was determined by the methd f DIENERT and WANDENBULCKE (1923). Light absrptin was measured with a Beckman Mdel-DU spectrphtmeter at a wavelength f 360 m[t and extinctin cefficients were crrected fr "salt errr". These "salt errrs" are different fr different salinities and are nt a linear functin f chlrsity. Beer's law slpes were, therefre, determined at different chlrsities and the "salt errr" crrectins applied at different chlrsities were determined frm these plts. Duplicate determinatins f sluble silica agreed within 5 percent, but because f a high and variable blank crrectin, the values given are accurate t nly t.tmles/1. Sluble silica was defined as the prtin f the ttal silica which frmed silicmlybdic acid in the presence f ammnium mlybdate and sulphuric acid. Accrding t ROY (1954) and ALEXANDER, HESTON and ILER (1954), nly the silicate that is in true slutin and in the inic frm will react with these reagents. The pssibility that, upn cntact with sulphuric acid during clr develpment, mre silica was disslved frm cllidal material which might pass the mlecular filter was tested by cmparing filtered river-water with dialysed riverwater. Bth the filtrate and the dialysate cntained the same amunt f sluble silica. Thus it is unlikely that such disslutin ccurred. Phytplanktn prductin was determined by the methd f STEEMAN-NIEL- SEN (1952) using C". Phytplanktn prductin in the Delta will be the subject f anther paper (THmAs and Sim- MONS, in preparatin). Only a range f values fr prductin btained immediately ff Pass a Lutre is used in this paper fr the calculatin f bilgical uptake f sluble silica. A sample f surface river-water fr labratry experiments n inrganic remval was cllected at Statin C-2 (see fig. 1). It has a chlrsity f 0.035 g. C171. and cntained 82.06 v.mles sluble silica/1. A surface-water sample frm the Gulf f Mexic was cllected east f the Delta beynd the area shwn in figure 1. The exact lcatin is nt knwn. This sample had a chlrsity f.61 g. C171. and cntained nly a trace f sluble silica. Bth river and Gulf samples were stred in 35 liter plyethylene carbys until used. Bilgical activity was stpped by the additin f 1 ppm HgCI, t these samples. Since sluble silica cncentratins did nt change during several mnths strage, treatment with HgC12 was successful in stpping bilgical activities. 80 I. 70 60 50 40 30 10 \ 1111 1,1, 0 5 i CHLOROS1TY (g.c1-/l.) FIG. 2. Sluble silica cncentratin as a functin f chlrsity f water samples. Data f June, 1953... \

0.2 A-15 40 2.8 i REMOVAL OF SOLUBLE SILICA FROM FRESH WATER 23 Statin Depth (ft) TABLE i.-silicate and chlrsity data f June 1953 Sluble silica knles/1. Chlrsity (g. C1-/1.) Statin Depth (ft) Sluble silica banles/1. Chlrsity (g. C1/i.) A-1 0 78.6 0.2 A-12 0 73.2 0.2 A-1 7.5 80.7 0.2 A-12 5 11.0 14.4 A-2 0 82.0 i A-12 10 7.7 16.0 A-2 12 70.2 0.2 A-12 25 7.3 18.8 A-3 0 77.5 0.2 A-12 36 11.0 13.9 A-3 16 76.0 0.2 A-13 0 72.8 0.4 A-4 0 76.3 0.2 A-13 5 31.3 8.4 A-4 7.5 76.8 0.2 A-13 10 6.2 14.8 A-5 0 76.0 0.4 A-13 25 4.2 18.2 A-5 3 64.3 2.0 A-13 40 3.8.2 A-5 6 35.2 9.0 A-14 36.0 8.4 A-6 0 72.8 0.3 H A-14 5 10.3 15.0 A-6 5 32.2 9.0 A-14 10 6.5 16.0 A-6 10 11.8 14.7 A-14 25 9.7 18.9 A-6 25 6.3 18.0 A-14 40 9.8 19.0 A-6 40 9.5 18.7 A-15 0 72.3 1.0 A-7 0 68.3 0.7, 5 19.7 12.4 A-7 5 28.8 8.9 A-15 10 7.3 15.6- A-7 10 12.0 14.1 A-15 25 4.0 18.3 A-7 25 H 2.8 18.0 A-15 40 7.0 H 18.5 A-7 40 2.8 18.9 A-15 85 8.3.3 A-7 104 8.2.2 A-16 0 71.2 J 0.7 A-8 0 51.0 4.9 A-16 5 4.5 16.7 A-8 5 18.5 13.1 A-16 10 2.7 18.1 A-8 8 17.7 14.5 A-16 25 4.0 19.4 A-9 0 66.7 1.4 A-16 i 2.8 '.2 A-9 5 29.3 10.0 -A-16 100 8.8.4 A-9 7 16.2 13.7 A-16 188 7.0.5 A-10 0 72.8 0.3 A-16 0 56.7 2.9 A-10 5 47.8 5.8 A-17 0 6.5 15.8 A-10 7 22.3 13.0 A-17 5 i 18.8 A-11 0 73.5 0.8 A-17 10 1.5 19.7 A-11 5 9.7 15.6 i A-17 25 2.0 19.9 A-11 10 7.5 16.6 A-17 40 2.3.1 A-11 25 7.3 17.7 A-17 100 3.0.5 A-11 40 8.5 18.7 A-17 188 5.0.3 RESULTS AND DISCUSSION Field data Figure 1 shws the statin lcatins at which water samples were btained during three different field trips. Tables 1, 2 and 3 summarize the results f the sluble silica and chlrsity determinatins. Plts f sluble silica versus chlrsity fr the varius grups f data are shwn in figures 2, 3 and 4. The straight, dashed line n each graph indicates the line arund which the pints shuld grup if the sluble silica f the river-water were merely diluted upn mixing with pen Gulf water. This line is btained in each instance frm average values f chlrsity and sluble silica in the pen Gulf water and "typical" riverwater. Determinatins f the chlrsity f pen Gulf water ver several field seasns have shwn that it varies relatively little averaging.4 g. C171., with a standard deviatin f 0.5 g/i. The sluble silica cncentratin f pen Gulf water averages 4.2 imles/1. and has a standard deviatin f 3.1 tmles/1. Vales f chlrsity and sluble silica f river-water are cnstant at any given

ec.)t 24 G. S. BIEN, D. E. CONTOIS, AND W. H. THOMAS 1 0 100 =90-, 70 - k' - 60 - N, N 50 - \ Lu 40- N 3 30 - N N -,t 8 s -. \..\ \ \ \ 1 I 0 110 5 CHLOROS1T( ( g.c1-/ L.) FIG. 3. Sluble silica cncentratin as a functin f chlrsity f water samples. Data f July, 1955. time, but vary cnsiderably thrughut the year. In general, they are prprtinal t the amunt f river discharge. As a result, the cnstants f the straight lines shwing the theretical relatinship between sluble silica and chlrsity are determined entirely by the characteristics f the river-water.3 A cnsideratin f figures 2, 3, and 4 leads t the cnclusin that the bserved distributin f sluble silica as a functin f chlrsity (dilutin) differs significantly frm that which wuld be expected frm the hypthesis that the sluble silica f the river-water is merely diluted upn mixing with Gulf water. The straight line represents the calculated dilutin curve assuming that the cncentratins in the sea-water are cnstant and using the average values presented here. Nte that the determined pints all lie belw the dilutin curve. The decrease f disslved silica cncentratin is much faster than dilutin alne. This Our values f sluble silica in Mississippi River water varied frm 65 t 1 micrnils f silica per liter (4.0-7.5 ppm) which is in excellent agreement with the values f SCHWARTZ (1938) and slightly lwer than the earlier values f CLARKE (1924). clearly indicates that the disslved silica is remved frm slutin by sme prcess. Even if the slpe f the theretical dilutin line shwn in these figures is changed t take the errr (th 2 micrmls) f the TABLE 2.-Silicate and chlrsity data f July 1955 Statin Depth (f t) Sluble silica (p.mles/1.) Chlrsity (g. C1-/1.) B-1 O 114 0.62 B-1 5 113 0.63 B-1 10 114 0.62 B-1 15 118 0.63 B-1 113 0.63 B-1 25 115 0.63 B-2 O 110 0.58 B-3 O 108 0.66 B-4 O 110 0.61 B-5 B-6 B-7 B-8 O O 110 0.66 109 1.02 110 1. 105 2. 101 2.24 99.9 2.66 B-9 O B-10 O B-11 79.7 5.08 B-11 5.9 13.2 B-11 10 22.9 13.9 B-11 15 6.0 16.6 B-11 3.0 17.8 B-11 B-12 B-13 25 6.5 96.7 89.6 18.2 3.33 3.87 B-14 75.1 5.92 B-15 O 85.0 4.66 B-15 5 24.1 13.6 B-15 10 4.8 16.9 B-15 15 7.6 16.8 B-15 5.2 16.4 B-15 25 3.0 18.0 B-16 0 61.9 7.68 B-16 5 12.3 15.4 B-16 10 7.0 16.0 B-16 15 2.8 17.2 B-16 1.2 18.0 B-16 25 1.4 18.3 B-17 O 27.4 11.6 B-17 5 4.8 16.6 B-17 10 5.0 16.5 B-17 0.6 17.8 B-17 25 0.8 18.2 B-17 37 0.4 18.4 B-18 O 7.4 17.0 B-18 5 1.8 17.0 B-18 10 1.8 17.1 B-18 0.6 17.8 B-18 40 0.6 18.5 B-18 70 2.0.4

REMOVAL OF SOLUBLE SILICA FROM FRESH WATER 25 80 70 60 3 s F2 40 (1, 30 10,\ N 5 10 15 Statin C-1 C-2 C-3 C-4 C-5 C-6 C-7 Depth (f t) N N CHLOROSITY (g.c1-/l., FIG. 4. Sluble silica cncentratin as a functin f chlrsity f water samples. Data f Octber, 1955. TABLE 3.Silicate and chlrsity data f Octber 1955 Sluble silica (./mles/1.) 63.9 59.3 56.1 41.9 7.5 8.2 4.6 Chlrsity (g. C1-/1.) 1.6 2.4 6.1 7.1 14.8 16.8 177.5 determinatin f the sluble silica int accunt, significant deviatins frm this straig-ht line still exist and the results cannt be explained n the basis f experimental errr. It has been suggested that minimum silica cncentratin is reached lng befre maximum chlrsity is reached. Since the determined pints d lie n a fairly straight line, nly dilutin tk place MAEDA (1952) in Japan, fund that it is exactly s. But Maeda's chlrinity range was frm 5 t 17 /. If we tk nly that prtin f ur plt, we wuld have arrived at the same cnclusin. By cmbining all data cllected during several seasns and ur labratry results we can shw schematically that remval by sme prcess must have taken place. The cncentratin f sluble silica as a functin f chlrsity in these waters (the data f figs. 2, 3 and 4 and tables 1, 2 and 3) can be described (with experimental errr) by the equatins (Si02), = (Si02) R B (CD C (Cl)2; B > C (1) where (Si02)1 is the sluble silica cncentratin fund, (Cl) is the chlrsity, (Si09)R is the initial cncentratin in the river and B and C are cnstants that vary seasnally (BIEN, 1953). These data can als be expressed (within experimental errr) by three lines 0,4, and BC in figure 5. Line OA represents the initial remval, AB, further remval, BC, the final dilutin f the water and OC, the theretical dilutin curve. A cnsideratin f this plt shws that nearly all f the sluble silica is remved. If nly a prtin were remved during the initial mixing (line OA) and thereafter nly dilutin ccurred, the experimental pints shuld fllw line AC. Instead, they d nt fall n such a line, but a minimum value fr sluble silica is reached befre maximum dilutin has ccurred (at pint CHLOROSITY FIG. 5. Illustrative relatinships f sluble silica t chlrsity.

26 G. S. BIEN, D. E. CONTOIS, AND W. H. THOMAS B). Any further dilutin with sea-water (line BC) results in n mre remval, since the sea-water already cntains this minimum amunt. Since remval ceases, the remval prcess must be cmplete, and all f the sluble silica riginally present must have been precipitated. This (les nt mean dilutin des nt ccur, but, that as fast as river-water is diluted, its sluble silica is remved. The pssible causes f the initial remval (line OA) and f further remval (line AB) are discussed later. Cncerning the initial drp in cncentratin, there is criticism that the disslved silica cncentratin determined by us at lw chlrsities is t high because cllidal silica will disslve after intrductin f cncentrated acid when the ph is belw 2.0. This is very unlikely since the cncentratin is very lw (cmpare 7.5 t 100 ppm) and cllidal silica is nt imprtant. Furthermre, the suspended slids which may cntribute mre silica had been remved by filtratin. T check ur results because f this criticism, we cmpared filtered river-water with dialyzed river-water. Bth the filtrate and the dialyzate cntained the same amunt f sluble silica. OA is very real. The explanatin f OA is that the aluminum ins in the river water are precipitated ut as the ph increased and carried with them much f the disslved silica as cprecipitate. Physical, chemical and bilgical prcesses tgether accunt fr this remval, and we tried t study these effects separately. Fr bilgical uptake the C14 methd was used and carried ut in the field, by measuring the C14 tagged carbn dixide taken up by the rganisms. The nn-bilgical prcesses were studied in the labratry using samples f river-water and Gulf water t which mercuric chlride was added t prevent bilgic activity. These experiment als served t clarify the suggestin that the decrease f disslved silica was due t dilutin alne. Labratry experiments n inrganic remval prcesses Several experiments were perfrmed which shw that results similar t the field data can be btained by mixing turbid river-water with sea-water. The experiments were carried ut by withdrawing an aliqut frm a sample f the river-water and adding an equivalent vlume f sea-water. The prcess was repeated at hurly intervals and each aliqut was then filtered and its cncentratin f sluble silica determined. Results btained using Mississippi riverwater and Gulf sea-water are shwn in table 4 and figure 6. The river-water used had a chlrsity f 0.035 g. Cl-/ liter and cntained 82.06 RAI f disslved silica per liter. The sea-water used had a chlrsity f.61 g. C-/ liter and nly a trace f disslved silica. These results are nt due t bilgical uptake because such activity was prevented by the additin f 1 ppm HgC12. It shuld be nted that the inrganic remval ccured within ne hur, which indicates that it will ccur quite rapidly in the field. Nte hw this plt resembles the schematic 10 15 CHLOROSiTY (9 FIG. 6. Sluble silica cncentratin as a functin f chlrsity as unfiltered riverwater is mixed with Gulf f Mexic water in the labratry. Black squares are pltted f rm data n table 4, Experiment I; black triangle shws data frm Experiment II; X-mark is data f rn Experiment III ; plusmark is data frm Experiment IV. The tw zers are end pint analyses.

5.12 11.4 3.92 17.52 ' REMOVAL OF SOLUBLE SILICA FROM FRESH WATER 27 TABLE 4.-Sluble silicate and chlrsity data f dilutin Experiment A Experiment Ns. I II III IV Chlrsity (g. C1-11.) Sluble SiO2 (umles/1.) Chlrsity (g. C1-11.) Sluble Si02 (,allles/1.) Chlrsity (g. C1-11.) Sluble Si02 (jumles/1.) Chlrsity (g. C1-11.) Sluble Si02 (Amles/1.) 0.035 82.06 0.035 82.06 0.035 82.06 0.035 82.06 2.06 65.4 0.82 76.8 2.06 69.8 2.06 70.1 4.12 61.7 1.62 70.4 3.92 68.1 61.8 ' 6.18 52.5 2.38 67.7 5.59 52.6 5.59 64.9 8.24 41.7 3.10 63.9 7.09 46.7 7.09 46.8 10.30 36.2 H 3.80 61.3 8.44 44.6 H 8.44 42.4 11.33 31.24.48 58.1 9.64 37.09.64 37.1 1 12.36 27.3 70.1 10.73 34.9 10.73 34.9 1 13.33 25.7 5.74 H 52.5 11.72 29.0 11.72 26.9 14.40.16.34 51.6 12.61 24.6 12.61 24.2 15.45 16.36.91 48.4 13.43.8 13.43.8 16.52 11.8 7.46 45.3 14.15 18.2 14.15 18.4 17.56 9.4 8.49 43.7 14.79 17.2 14.79 18.4 H 18.57 5.9 9.70 40.8 15.38 H 15.7 15.38 15.0 19.57 1.7 10.79 33.0 15.90 H 13.2 15.90 13.5 1.61 0 11.76 28.1 16.37 16.37 11.5. 12.65 28.1 16.79 9.4 16.79 9.1 13.44 22.1 17.17 8.3 H 17.17, 8.4 1 14.16.5 17.52 6.6 6.5 17.83 6.0 17.83 5.1 18.11 4.4 18.11 3.1 diagram n figure 5. Here the diluent has a definite cmpsitin and n quibbling abut the results is pssible. It is definitely shwn that AB n figure 5 is nt dilutin alne but the cmbinatin f dilutin and remval, while AC is the dilutin curve after the initial drp in cncentratin. Nn-bilgical remval, therefre, is quite imprtant in the prcess f remval f disslved silica. The same experiment was perfrmed by mixing Mississippi river-water with Pacific Ocean surface sea-water cllected 60 miles east f Guadalupe Island. This sample had a chlrsity f 18.04 g. Cl-/1., and cntained.1 [tmles sluble silica/1. The results are shwn in table 5 and figure 7, and are similar t thse btained with Gulf f Mexic water. Thus the phenmenn f inrganic remval is very likely a general ne and is nt due t anything especially characteristic f Gulf water. Tw pssible mechanisms by which inrganic remval f sluble silica might ccur are: (1) a reactin with electrlytes in sea-water resulting in plymerizatin r salt frmatin and (2) adsrptin r c-precipitatin with suspended slids r cllidal material in river-water as these cme in cntact with electrlytes in sea-water. We have used Milliprefiltered natural waters and artificial slutins in rder t distinguish between these tw prcesses. When filtered sea-water was diluted t varius chlrsities with distilled water and knwn amunts. f sdium metasilicate (105 ttm1es/1.) were added, n lss f sluble silica was bserved after strage f the slutins fr 1 week. Similar results were bserved using sdium chlride slutins rather than filtered seawater. These data are shwn in table 6. Successive dilutin f a sdium metasilicate slutin with a sdium chlride slutin als did nt result in inrganic remval (see Curve I, fig. 8). Thus, electrlytes alne d nt apparently cause inrganic remval. In a further test f this hypthesis, filtered river-water was successively di-

28 G. S. BIEN, D. E. CONTOIS, AND W. H. THOMAS TABLE 5.-Sluble silicate and chlrsity data f dilutin Experiment B Experiment Ns. I Chlrsity (g. C1-11.) Sluble Si02 (mles/l.) Sluble Chlrsity Si02 (g. C1-/1.) (mles/1.) 0.035 82.06 0.035 82.06 3.61 64.7 1.80 70.4 7.22 52.6 3.43 63.9 10.82 41.7 4.89 57.9 14.43 30.16.21 54.7 18.04.17.39 53.5 8.41 46.9 9.41 44.4 10.28 39.9 11.05 38.9 11.75 37.5 12.38 33.4 12.95 33.4 13.46 31.5 13.92 31.1 14.33 30.5 14.70 29.5 15.04 27.2 15.34 27.3 15.61 26.4 15.85 26.8 16.07 26.2 The sea-water used in this experiment had been standing in a glass bttle fr ver a year. II 80 TO 90 80 - - +\A :=1 6 0 70 ú 50 60 50 30 40 30 0 lifted with filtered Gulf water. These results are shwn in Curve II, figure 8. The deviatin frm linearity bserved was smewhat less than that bserved using unfiltered waters and may be due t incmplete remval f cllidal material by the Millipre filter. Such material has been shwn t be imprtant in inrganic remval by WHITE et al. (1956). It may als mean that sluble silica is precipitated by the presence f a small amunt f aluminium ins which passed the filter, frmed aluminium hydrxide and cprecipitated the silica (GT, 1956; AR- RHENIUS, 1954; OKAMOTO et al., 1957). In general, these results agree with thse f KRAUSKOPF ( 1956), which shwed that precipitatin f disslved silica des nt ccur when silica slutins are mixed with sea-water. If adsrptin f sluble silica culd ccur as suspended matter in the river reacted with electrlytes, inrganic remval wuld be bserved when suspended slids were added t filtered seai0 CH LOROS, 19 CL/ L.) 15 FIG. 7. Sluble silica cncentratin as a functin f chlrsity as unfiltered riverwater is mixed with Pacific Ocean water in the labratry. Black squares are pltted frm data n Experiment I and dts are frm data n Experiment II shwn in table 5. 10 CHLOROSITY 9. C1/ L.) FIG. 8. Sluble silica dilutin experiments. Curves pltted frm data n table 6. Curve I: sdium silicate slutin diluted with sdium chlride slutin ; Curve II: filtered riverwater diluted with filtered sea-water. (Slid lines are calculated dilutin curves.)

REMOVAL OF SOLUBLE SILICA FROM FRESH WATER 29 TABLE 6.-Amunts f sluble silica fund after strage fr 1 week in the presence f filtered Gulf water r sdium chlride slutins. Amunt added= 105 Amles/l. Filtered Gulf water plus distilled water Chlrsity (g. Cl-/I.) Si02 sluble fund (panles/1.) 0 1.81 3.61 5.41 7.22 9.03 10.84 12.64 14.45 16.75 18.06 105.4 103.0 105.0 105.0 105.2 104.0 105.5 103.6 105.0 104.5 106.0 Sdium chlride slutins Chlrsity (g. C1/l.) Si02 sluble fund (ianles/1.) 0 2.12 4.24 6.37 8.49 105.0 103.5 106.2 104.5 104.8 10.61 12.73 14.85 16.98 19.10.22 105.4 106.1 102.6 106.0 105.2 105.0 water enriched with sdium silicate. Such remval was bserved upn additin f bentnite aluminium xide, and particularly, suspended slids frm the Mississippi River, t enriched Gulf water. In these experiments, Gulf water was enriched, a cncentrate f suspended material was added, the sample was shaken, and allwed t stand vernight. After standing, the suspensin was filtered and sluble silica was determined. The results are shwn in table 7 and figure 9, and the data are crrected fr the amunt f sluble silica intrduced with the cncentrated suspended matter. It shuld be nted that suspended material frm the river is mre effective than either bentnite r aluminium xide. Als, since the amunt f remval increases with increasing amunt f suspended slids frm the river, inrganic precipitatin is very likely an adsrptin prcess. The necessity f electrlytes as well as suspended matter fr precipitatin was demnstrated by adding slid sdium chlride t river-water t increase chlrsity. These results are shwn in table 8 and figure 10. As the chlrsity increased, an initial drp f abut 9 percent in cncentratin was bserved, after which the cncentratin remaind cnstant. The cnditins f these experiments are different frm thse f KRAUSKOPF (1956), wh als added suspended matter t silica slutins. Little r n remval ccurred in his experiments when kalinite, mntmrillnite, calcite r Fe3 were added t silica slutins made up in 0 a. BENTONITE CLAY SUSPENSION 100 95 s c. SUSPENDED SOLIDS FROM RivER WATER 100 0---- W. 95 _. 8 : 90 111 IIIIIIII II 0.1.2.3 4.5 6.7 8.9 I 0 GRAMS BENTONITE / LITRE b. ALUMINIUM OXIDE SUSPENSION 100 --- - 95 90 I. 85 1111111 0.2 3 4 5.6 7 a 110 GRAMS Al2 / LITRE 3 I I 85 I I 0 2 3 4 5 6 7 1 8.9 1.9 GRAMS RIVER SUSPENDED SOLIDS / LITRE FIG. 9. The influence f suspended matter n the recvery f sluble silica f Tm enriched Gul f f Mexic water. a. bentnite clay, b. aluminum xide, c. suspended slids f Tm river water..9

30 G. S. BIEN, D. E. CONTOIS, AND W. H. THOMAS TABLE 7.-Remval f sluble silica by adding suspended slid matter t enriched sea-water (chlrsity= 16.27 g. Cl-/l.) Bentnite suspensin (suspensin cntains 0.102 g. slid/ml; ; filtered slutin cntains 47.4 Amles sluble silica/1.) Wt. clay added (g./1.) 0 0.04 0.0408 0.0612 0.0816 Si02 added (.mles/1.) 107.4 107.6 107.8 108.0 108.2 Si02 fund (um1es/1.) 107.4 101.6 97.6 96.5 97.6 Percent recvered 100.0 94.3 90.5 89.2 90.1 0.1024 108.4 95.2; 98.5 88.0; 90.5 0.1224 108.5 97.1 89.5 0.1530 108.7 95.6 88.0 0.4 109.3 102.5; 101.5; 97.5 93.0; 93.2; 90.5 0.306 110.4 101.2; 100.7 91.0; 90.0 0.408 111.2 102.0; 99.5 91.5; 89.5 0.510 112.1 104.3; 103.5; 101.2 98.0; 92.5; 90.0 0.612 113.0 105.0 93.0 0.714 114.0 105.3 92.5 0.816 115.0 104.8-91.0 0.918 115.9 104.8 90.5 1.02 117.0 104.8 89.5 Aluminium xide suspensin (suspensin cntains 0.1 g. slid/ml; sluble silica added =107 Amles/1.) Wt. Al3 added (g/1) Si02 fund (mles/1.) Percent recvered River mud suspensin (suspensin cntains 0.183 g. slid/ml; filtered slutin cntains 95,umles sluble silica/1.) Wt. Slid added (g./1.) O 107.0 100.0 0.024 104.0 97.4 0.048 100.0 94.0 0.072 95.5 89.0 0.096 101.0 94.5 0.12 99.5 93.5 0.144 100.3 94.1 0.168 100.3 94.1 0.192 99.7 93.5 0.216 100.0 94.0 0.240 99.7 93.5 Si02 added (umles/1.) Si02 fund (Amles/1.) Percent recvered 101.6 101.6 100.0 0.0732 101.97 98.3 96.5 0.1464 102.34 97.5 95.3 0.2196 102.72 97.5 95.0 0.2928 103.1 97.2 94.5 0.366 103.5 96.1; 95.2 93.0; 92.3 0.4392 103.8 95.2 92.8 0.5124 104.2 96.1 92.5 0.5856 104.6 96.1 92.0 0.6588 105.0 95.5 91.0 0.732 105.3 95.0; 92.5 90.2; 89.0 1.098 106.3 92.0 87.0

83.1;82.8 REMOVAL OF SOLUBLE SILICA FROM FRESH WATER 31 distilled water. Our experiments shw that bth electrlytes (sea-water) and suspended matter are necessary. On the ther hand, in Krauskpf's experiments with silica-enriched artificial sea-water, Fe3 remved abut the same amunt f sluble silica as in ur experiments (6-9 percent), but calcite was withut effect. It shuld be nted that the cncentratins f sluble silica used in Krauskpf's experiments are 25-35 times greater than thse used in ur experiments r fund in river-water. Als, KRAUSKOPF des nt state hw much suspended matter was added t his slutins. It is prbable that the cncentratins f silica and suspended matter are critical in such experiments. In ur experiments bth the cncentratin f silica and suspended matter frm the river apprximate thse fund in the river. A prblem which remains fr investigatin is the rle f irn and aluminium in systems which apprximate the natural river-cean cmplex. Bilgical remval f sluble silica by phytplanktn It is very difficult t estimate whether significant remval f sluble silica due t bilgical uptake by diatms ccurs during the initial mixing f river-water. Abut 15 percent f the sluble silica in river-water r abut 10-18 ttmles/1. are remved by prcesses ther than dilu- TABLE 8.Sluble silica remval frm river-water by sdium chlride Chlrsity Si02 fund Percent (g. Cl/1.) (mles/l.) recvered 0.034 90.0 100.0 0.65 84.9 94.5 1.25 83.2;85.6 92.5;94.5 1.86 83.0;82.8 92.3;92.2 i 2.47 92.4;92.2 3.07 83.2;83.0 92.5;92.3 3.68 82.8;80.5 92.0;89.5 4.29 83.0 92.3 * Original sluble silica cntent f this water was 82.06 i.tmles/1. This was increased t 90 limles/1. fr the experiment. 2 - CHLOROSII, 9. CL, FIG. 10. The effect f sdium chlride n the sluble silica cncentratin f Mississippi River water. tin during- mixing t a chlrsity f 8 g. C171. Our -measurements (THmAs and SIMMONS, 1957) shw that frm 1 t 7 timles rganic carbn are frmed per liter per day by surface phytplanktn ff Pass a Lutre. Mst f the phytplanktn ff Pass a Lutre are diatms (SimmNs and THOMAS, 1957). Using a factr fr the rati between carbn and silica atms in diatms (100 : 40) calculated frm SVERDRUP et al. (1942, p. 234), we can calculate that between 0.40 and 2.80 [Lmles f silica are taken up per liter per day by diatms. Thus abut 2 percent f the ttal sluble silica in riverwater might be taken up in 1 day by ffshre diatms r abut percent f the initial remval prcess might be due t diatm uptake if the assumptin is made that it takes 1 day fr 1 1. f fresh riverwater t be diluted t a chlrsity f 8 g. CI-A This assumptin is almst certain t be true during perids f high river discharge. Accrding t SCRUTON (1959), the average surface current at the muth f Nrth Pass during June 1953 was 2 knts. This velcity wuld decrease t less than 1 knt as the water mved seaward. Thus 1 1. f water might mve seaward sme 24 miles during 1 day. The existence f a thin layer f relatively fresh turbid water (chlorsity 1.3 g. Cl-/1.) during the seasn f high river discharge even tens f miles ff the delta (SCRUTON and MOORE, 1953; THOMAS and SIMMONS, 1957) indicates that mixing t a chlrsity f 8 g. C171. des nt ccur

32 G. S. BIEN, D. E. CONTOIS, AND W. H. THOMAS during the mvement f the water fr 24 miles. At a perid f lw discharge, the abve assumptin is prbably nt true. The current velcity is less than half that measured during high river discharge (Scrutn, 1956), s that 1 1. f riverwater might mve nly 12 miles during 1 day. Hwever, ur measurements shw that a layer f relatively fresh water des nt extend that far ffshre (THmAs and SIMMONS, 1957) during the fall mnths. Mixing t the final chlrsity prbably takes several days at all seasns. Thus bilgical uptake may accunt fr an appreciable amunt f silica in the cmplete prcess. When the field results (table 9) and the labratry results (table 10) are expressed n a cmparable basis, there is a difference between the tw sets f data, which may be attributed t bilgical uptake, since bilgical activities were stpped by the additin f mercuric chlride in the labratry experiments. In cmparing the sets f data, vve have used the fllwing equatins : (SiO2) = ( Si02)R K(C1) (2) is the equatin describing the theretical dilutin f sluble silica in river-water with Gulf water. In this equatin (Si02)1 is the theretical cncentratin which shuld be fund at a given chlrsity (Cl) ; (Si02)R is the initial cncentratin in the river-water, and K is a cnstant. Anther equatin represents the sluble silica actually fund, at a given chlrsity (5i02)1, and was given previusly within equatin ( 1). The final equatin used is that expressing the theretical dilutin if riverwater is diluted with Gulf water cntaining n silica (SiO2)'e1 = (Si02)R K'(C1) (3) in which K' is slightly greater than K. This is used because we are nly interested in the remval f silica frm riverwater and thus can neglect pssible remval frm Gulf water. The percent remval is then calculated frm the difference between (Si02)1 and (Si02)1 by dividing this value by (Si02)1 and multiplying by 100. Table 9 shws the abve calculatins fr the field data ; table 10, thse fr the experiments in which river-water was successively diluted with Gulf water. The field and experimental results are cmpared in figure 11. The difference between the tw sets f data is shwn by crsshatching. It shuld be nted that belw a chlrsity f 8 gm C111. little r n difference is fund, but beynd this pint the difference is cnsiderable. This supprts previus estimates that bilgical activity will ccur nly in waters f high chlrsity in which sufficient time fr uptake has elapsed. An alternative explanatin fr the difference between the tw curves is that in the labratry experiments there may nt be enugh time fr maximum remval t ccur, since the river-water and Gulf water were nly mixed fr 1 hur befre sluble silica was determined. Hwever, if this is the reasn fr the difference, the initial percentage remval fr ne set f data shuld nt agree s well with that fr the ther set. CONCLUSION When fresh water enters the sea, it wuld be expected that the high cncentratin f sluble silica in fresh water culd be reduced by dilutin with seawater having a lw cncentratin f sluble silica. Hwever, the sluble silica cncentratin is reduced by sme prcess ther than dilutin. Tw such prcesses can ccur : bilgical uptake r inrganic percipitatin. Calculatins based n phytplanktn prductin measurements and estimates f mixing rates f water masses in the Mississippi Delta area shw that bilgical remval may pssibly accunt fr sme f the initial remval during perids f high river discharge. During perids f lw river discharge such remval is less pssible. Hwever, after the initial mixing, at chlrsities greater than

TABLE 9.-The difference between calculated and bserved values fr sluble silica as a functin f chlrsity. Data frm field samples A-Dates f June Chlrsity- (g. Cl-/1.) 1953 (Si02)e1 (ymles/1.) (Si 0 2) f (mles/1.) A (Si02) (Si 02) ci_ (Si02» (mm1es/1.) (SiO2)'ci (mles/l.) % Remval 100 (A SiO2) (Si )'c 0.2 76.4 76.4 76.4 0 3.0 66.4 61.8 4.6 68.8 7.0 6.0 55.7 48.0 7.7 54.5 14.0 9.0 44.9 35.0 9.8 43.1 23.0 10.0 41.4 31.2 10.2 39.3 25.7 12.0 34.2 22.5 11.7 31.8 H 36.8 13.5 28.8 17.0 11.8 26.0 H 45.0 15.0 23.5 11.5 12.0.9 H 59.0 16.0 19.9 7.5 12.4 16.6 H 74.7 17.0 16.4 5.6 10.8 12.8 84.3 18.0 12.8 4.4 8.9 9.1 H 92.3 19.0 9.2 4.4 4.8 5.3 90.6 B-Data f July 1955 0.6 113 113 O 0 0 3.0 99.6 96.0 3.6 101.3 3.5 6.0 82.8 74.0 8.8 82.2 10.7 9.0 66.0 53.0 13.0 65.0.0 10.0 60.4 46.0 14.4 59.3 24.3 12.0 49.1 31.0 18.1 47.9 37.9 13.5 40.8.0.8 39.3 53.0 15.0 32.3 12.0.3 30.8 53.5 16.0 26.8 8.0 18.8 25.1 75.0 17.0 21.2 5.0 16.2 19.4 83.5 18.0 15.5 3.0 12.5 13.7 91.3 19.0 9.4 2.0 7.4 8.1 91.5 C---Data f Octber 1955 0 70 70 O 70 0 5 54 50 4 53-7.5 7.0 47.4 42 5.4 46 12.0 10.0 37.7 30 7.7 36 21.0 15.0 21.6 11 10.6 18.5 57.0 17.0 15.0 5.6 9.4 12.0 78.0 18.0 12.0 5.0 7.0 8.0 87.0 TABLE 10.-The difference between calculated and bserved values fr sluble silica as a functin f chlrsity. Data frm labratry experiments Chlrsity (g. C1-/1.) (Si02) (S102) 'C (AMOleS/1.) (SiO 2) f (UM OleS /1.) A (Si02) (Am1es/1.) %Remval 1000s -SR O 82.1 82.1 O O 3 70.1 64.4 5.7 8.1 6 58.2 51.2 7.0 12.0 9 46.2 39.3 8.9 19.3 10.5 40.2 33.1 7.1 17.7 12.0 34.3 26.2 8.1 23.6 13.5 28.3.0 8.3 29.3 15.0 22.3 15.0 7.3 32.8 16.0 18.3 11.3 7.0 37.8 17.0 14.3 7.5 6.8 47.5 18.0 10.4 3.75 6.6 63.4 19.0 6.3 1.25 5.0 79.5

34 G. S. BIEN, D. E. CONTOIS, AND W. H. THOMAS 100 90 80 (7) 70 J 6d 50 40 30 O 1. FIELD DATA i 4 16: X LABORATORY EXPERIMENT 5 íl 10 15 CHLOROSITY (g.ci-/l.) FIG. 11. The percentage remval f sluble silica as a functin f chlrsity. Plus-sign is data f June, 1953; square fr data f July, 1955; black triangle fr data f Octber, 1955. See tables 9 and 10. 8 g. 011., bilgical remval may be appreciable, since enugh time culd elapse fr uptake by diatms t ccur. Much mre extensive wrk wuld be required t distinguish definitely between bilgical uptake and inrganic precipitatin in the field. Tw mechanisms f inrganic precipitatin f sluble silica have been prpsed: (1) reactin f sluble silica with electrlytes in sea-water resulting in salt frmatin r (2) adsrptin r c-precipitatin f sluble silica with suspended X slids r cllidal material in river-water as these cme in cntact with electrlytes. Our data shw that bth electrlytes and cllidal r suspended slids are necessary fr maximum inrganic remval. The fact that increasing remval ccurs when the amunts f suspended slids are increased indicates that adsrptin r cprecipitatin f sluble silica n these slids prbably ccurs. These precipitatin prcesses result in the remval f almst all f the sluble silica frm river-water. Since this riverwater cntains frm 4.0 t 7.5 mg/l. sluble silica and 160 t 1500 mg/l. f suspended slids are present in the riverwater, the weight f suspended material in these waters can be slightly increased by these prcesses, and there fre presumably sme f the silica in the deltaic sediments wuld als be derived frm these prcesses. In cnclusin, we have shwn that remval f all r mst f the disslved silicn in river-water ccurs when this water mixes with sea-water. Bilgical uptake can accunt fr an appreciable amunt f this remval during the final mixing f the tw water masses, but prbably nt initially. Silica is als precipitated inrganically by cprecipitatin and adsrptin in which suspended slids in the river-water and electrlytes in the sea-water are required. REFERENCES ALEXANDER, G. B., HESTON, W. M., and ILER, R. K., 1954, The slubility f amrphus silica in water: Jur. Phys. Chem., v. 58, pp. 453-455. ARMSTRONG, F. A. J., 1954, Phsphrus and silicn in sea water ff Plymuth during the years 1950 t 1953: Jur. Marine Bil. Assc. United Kingdm, v. 33, pp. 381-392. ARRHENIUS, G., 1954, Origin and accumulatin f aluminsilicates in the cean: Tellus, v. 6, N. 3, pp. 215-2. BIEN, G. S., 1953, Chemistry f sediments and interstitial waters. In "Study f nearshre sediments and their envirnment in the nrthern Gulf f Mexic": Prgress Reprt Am. Petrleum Institute Prject 51: N. 11, 51 pp. CLARKE, F. W., 1924, The data f gechemistry: 5th Ed., U.S. Gel. Survey Bull. 770, p. 80. DIENERT, F. and WALDENBULCKE, F., 1923, Stir la dsage de la silice dans les eaux: Cmpt. Rend. Acad. Sci., Paris, v. 176, pp. 1478-1480. GOTO, Katsumi, 1956, Precipitatin f silica in the presence f aluminum: Bull. Chem. Sc., Japan, v. 29, pp. 741-742. JORGENSEN, E. G., 1953, Silicate assimilatin by diatms: Physil. Plantarium, v. 6, pp. 301-315. KRAUSKOPF, K. B., 1956, Disslutin and precipitatin f silica at lw temperatures: Gechim. et Csmchimica Acta, v. 10, pp. 1-26.

REMOVAL OF SOLUBLE SILICA FROM FRESH WATER 35 MAEDA, H., 1952, The relatin between chlrinity and silicate cncentratin f water bserved in sme estuaries: Publ. Set Marine Bil. Lab., v. 2, pp. 249-255. MAEDA, H., 1953, The relatin between chlrinity and silicate cncentratin f water bserved in sme estuaries, II: Jur. Shimnseki Cll. Fisheries, v. 3, pp. 167-180. MAKIMOTO, H., MAEDA, H., and ERA, S., 1955, The relatin between chlrinity and silicate cncentratin f water bserved in sme estuaries, III: Recrds Oceang. Wrks Japan, v. 2, pp. 106-112. ROY, C. J., 1945, Silica in natural waters: Am. Jur. Sci., v. 243, pp. 401-402. SCHWARTZ, M. C., 1938, The remval f silica frm water fr biler feed purpses: Jur. Am. Water Wrks Assc. v. 30, pp. 659-678. SCRUTON, P. C., 1956, Oceangraphy f Mississippi Delta sedimentary envirnments: Am. Assc. Petrleum Gel., Bull., v. 40, pp. 2864-2952. SCRUTON, P. C. and MOORE, D. C., 1953, Distributin f surface turbidity ff Mississippi Delta. Am. Assc. Petrleum Gel. Bull., v. 37, pp. 1067-1074. SIMMONS, E. G. and THOMAS, W. H., 1957, The phytplanktn f the Mississippi Delta: In preparatin. STEEMAN-NIELSEN, E., 1952, The use f radiactive carbn (C") fr measuring rganic prductin in the sea: Jur. Cns. Internat. Explr. Mer., v. 43, pp. 117-140. SVERDRUP, H. V., JOHNSON, M. W., and FLEMING, R. H., 1942, The ceans: their physics, chemistry, and general bilgy: Prentice-Hall, N.Y. THOMAS, W. H. and SIMMONS, E. G., 1957, Phytplanktn prductin in the Mississippi Delta: In preparatin. WHITE, D. E., BRANNOCK, W. W., and MURATA, K. J., 1956, Silica in ht-spring waters: Gechin, et Csmchimica Acta, v. 10, pp. 27-59. RECORD OF DISCUSSION T. R. Walker: Dr. Bien claims that electrlytes d remve sme sluble silica frm river waters. Dr. Krauskpf claims that they d nt. Can these divergent views be recnciled? Dr. Bien: There is n difference frm Dr. Krauskpf's results. When suspended matter is present, any electrlyte will remve sluble silica (see fig. 10). Whether it is cprecipitatin with aluminium xide r simply salt effect is pen t questin. Anyway, the prblem is very cmplicated and n definite answer can be given at the present.