Geochemical Journal, Vol. 25, pp. 31 to 44, 1991

Non-chondritic /holmium ratio and tetrad effect observed in pre-Cenozoic limestones

IWAO KAWABE, YOKO KITAHARA and KOH NAITO

Department of Earth Sciences, Faculty of Science, Ehime University, Bunkyo-cho 2-5, Matsuyama 790, Japan

(Received July 27, 1990; Accepted March 26, 1991)

REE and Y in limestones have been determined by ICP-AES method coupled with pre-concentra tion chemical procedures. The lanthanide tetrad effect has been clearly observed in the REE patterns of two Permian limestones. It is the W-type tetrad effect seen in . Y/Ho ratios for the limestones show large positive deviations from the chondritic ratio. Similar non-chondritic Y / Ho ratios are seen in a Precambrian limestone from South Africa, seawaters, and marine phosphorites. Their REE patterns are well characterized by the W-type tetrad effect. These facts and recent results of aquatic geochemistry of REE and Y strongly suggest that the positive Y anomalies are intimately associated with the W-type tetrad effect of REE in natural aquatic including seawaters and in some hydrogenous deposits from the solutions. Y is not a pseudo lanthanide that behaves like Ho in natural conditions where the lanthanide tetrad effect is operative. This must be a thermochemical effect related to the absence of 4f electron in Y3+ and the systematic differences in 4f electronic configurations in the REE3+ series. The REE plus Y patterns with positive Y anomalies and the W-type tetrad effect are important for the gochemical studies of limestones and other hydrogeneous deposits.

and related materials. The other one called the INTRODUCTION M-type is observed in materials that are Masuda and Ikeuchi (1979) first probably having remained after leaching by demonstrated that the lanthanide tetrad effect in aqueous media. They emphasized the com is recognized in peculiar REE abundance plementary relationship between the W and M patterns for seawaters and a marine phosphate tetrad effects. nodule. Subsequently, Masuda and his The tetrad effect itself was proposed original coworkers further found that such tetrad effects ly by Peppard et al. (1969 and 1970) on the basis are detectable in other natural substance in of their studies on solvent extraction of trivalent cluding limestones, valves of living shellfish, and . According to their fresh groundwaters, siliceous ores, leucogranites works, when the logarithmic distribution and so on (Kamioka and Masuda, 1986; Masuda coefficients for trivalent REE between organic et al., 1987; Masuda and Akagi, 1989). On the and aqueous phases are plotted against the other hand, detailed studies on REE distribution of the lanthanide, the plot ap in Pacific and Atlantic ocean water columns pears to consist of four distinct smooth curves. made by De Baar et al. (1983 and 1985a, b) pro The fifteen REE are grouped by the curves into vide further evidence for the lanthanide tetrad four tetrads with Gd being common to the sec effect in seawaters. Recently, Masuda et al. ond and third tetrads. The extended smooth (1987) proposed to distinguish two types of curves for the first and second tetrads and those tetrad effects in nature: The first one called the for the third and fourth tetrads intersect respec W-type by them is observed in natural waters tively between Nd and Pm and between Ho and

31 32 I. Kawabe et al.

Er. In a similar plot for trivalent actinides, they plasma-atomic emission spectrometry (ICP showed that an analogous tetrad effect is ap AES) after the separation of these parent with Cm being common to the second elements from matrix major elements. The and third tetrads. They wrote that, if the geochemical importance of intimate association lanthanide tetrad effect should be valid, the of non-chondritic Y/Ho ratios with the W-type "half -filled" shell effect would be joined by the lanthanide tetrad effect in limestones and "one -quater-filled" shell effect and "three seawaters is discussed. quaters-filled" shell effect because of the discon tinuities occurring respectively between the third and fourth and between the tenth and eleventh SAMPLES 4f electron additions. Two limestones (EL-1 and EL-2) were sam In response to the proposal of tetrad effect, pled from the Lower to Middle Permian lime Jorgensen (1970) and Nugent (1970) put forward stone body in Ohnogahara, Nonura-cho, Ehime quantum mechanical interpretations for the Prefecture. The limestone body occurs together tetrad effects in terms of the interelectron repul with greenstones, cherts, and clastic rocks. They sion energy of the q electrons in each 4f" or 5f' are collectively designated the Ohnogahara electronic configuration for trivalent lanthanides Group and one of the key pre-Cenozoic and actinides. geological units that constitute the Northern If the lanthanide tetrad effect is related to the Chichibu Belt in the western part of Shikoku ground state electronic configurations in which Island, Japan. The other is a calcareous gneiss the 4f sub-shell is successively filled by an addi sampled in Nishisetodani area along Ohnagatani tional electron, the relationship between Ho and River, Yatsuo-cho, Nei-gun, Toyama Prefec Y is important because the trivalent ionic radii ture. This is a typical sample of the most for Ho and Y are almost the same despite no 4f calcareous member of the calc-silicate gneiss of electron in Y. In this respect, Peppard et al. the Hida metamorphic rocks. The calc-silicate (1969) themselves noted that Y does not always gneiss is characterized as alternations of marbles behave as a pseudo lanthanide in solvent extract and quartzites. This sample was taken from a ions for REE. The distribution coefficient for Y marble-rich portion. It consists of calcite in a solvent extraction system involving H[DOP] mainly, and accessory amounts of quartz , (benzene) and HC1(aqueous) is comparable with clinopyroxene, and tremolite are present. This that for Ho or Er. However, in the other extrac sample differs petrographically from the two tion system utilizing DEH[CIMP] (benzene) and samples of the Ohnogahara limestone in having LiBr plus HBr (aqueous), such a coefficient for Y silicate impurities in it. differs greatly from any values of the coefficients for all the REE. In this context, it is interesting and important EXPERIMENTAL to study the lanthanide tetrad effect and the frac The aliquots of each powdered sample tionation between Y and Ho or other heavy REE weighing 10-30g have been used for repeated together even in natural substance. However, analyses of three to six times, because the REE almost all previous analytical results relevant to and Y contents were expected to be as low as the tetrad effects in natural samples include chondrite levels. Two methods for sample neither Y data nor the information as to Y/Ho decomposition were used in order to ensure the fractionations, because the mass spectrometric presence or absence of non-carbonate impurities dilution method (MSID) has been used that may have those elements and resist the principally. In this paper, we report REE and Y chemical dissolution by mineral acids or by HF determinations for some Japanese pre-Cenozoic plus mineral acids. The first method is the limestones using the inductively coupled Ar dissolution by HC1 alone. The is filtered Y/Ho ratio and lanthanide tetrad effect in limestones 33 to remove insoluble materials if present. Only rections for spectral interferences among the the filtrate is used in subsequent procedures for measured elements were made in calculating the the group separation of REE and Y. The other final results of the concentrations of REE and Y method includes additional decomposition pro in samples. The linear correction factors for spec cedures to make all the constituent materials tral interferences among REE, Y and Ca have soluble as completly as possible. The filtered in been re-determined for this study, and the soluble materials after dissolution by HC1 are previous set by Kawabe et al. (1988) was slightly ignited in a Pt-crucible, and then dissolved by revised. The blank tests for all the chemical pro HF + HNO3 + HC1O4 twice. It is filtered again, cedures indicate that the blank correction is un and this filtrate is combined with the first filtrate necessary. after dissolution by HCI. The filterd insolubles, As a result of REE and Y analyses in stan after being ignited in a Pt-crucible again, are fur dard rocks and clastic rocks enriched in heavy ther fused with 300 mg of (2:1) mixture of minerals, we confirmed that the residues after Na2CO3 and H3BO3. If necessary, this fusion is the dissolution by HF-mineral acids can be com repeated twice. The fusion product is then pletely dissolved by the fusion procedure using dissolved in HCl and added to the combined solu Na2CO3 and H3BO3 (Kawabe et al., in prepara tion of the first and second filtrates. This is the tion). So that we believe that all materials in each sample solution obtained by the second decom sample are satisfactorily decomposed into the position method. solution by the second method. By using the sam Both of the sample solutions by the two diges ple solution spiked with the standard solution tion methods are used in the same procedures for having 1001cg of La, Eu, Lu and Y each, the group separation of REE and Y, in which co recovery in the group separation was checked. precipitation with Fe(OH)3 is followed by the ca The recovery of more than 99 % was found for tion-exchange purification. The co-precipitation each spiked element. The alkali earth elements is made at pH = 6.5 by adding 20 mg Fe 3+ and and trivalent major elements are negligible in the ammonia water. This is repeated for the filterate final solution except for a minor amount of Ca. separated from the hydroxide precipitate in The level of Ca concentrations ranges from 70 to order to make the final recovery satisfactory. 200 ppm, but it is not high enough to cause The first and second precipitates are combined significant chemical interferences in the ICP and dissolved in 1.76 M HC1 for the cation-ex emission spectrometry for REE and Y. change separation of REE and Y from Fe and other matrix elements. The columns (1 cm x 11 RESULTS cm) of Dowex 50W X 8 (200-400 mesh) are used. 60 ml of 1.76 M HCl is used to load the sample Each sample has been analyzed three times or solution and to elute Fe" and other major ca more, in which the two methods for sample diges tions. REE and Y are eluted by 120 ml of 6 M tion described above were applied. No signifi HCI. After evaporation of the HCl solution con cant differences in REE and Y results between taining REE and Y to dryness, it is dissolved in the two methods were found even in the case of 50 ml of 0.34 M HCl and then used to determine the calcareous gneiss. So that the analytical REE and Y by ICP-AES. results of all runs for each sample were averaged The ICP emission spectrometer used in this irrespective of the decomposition methods, and study is a Shimazu ICPS-50. The operating con the mean values and one-sigma errors for the ditions and parameters of the spectrometer have respective means were calculated. These are been described elswhere (Kawabe et al., 1988). listed in Table 1 together with the number of Emission intensity data for all REE and Y repeated analyses. The data on the surface together with Ca in the final sample solutions and the C l chondrite are also listed in and standard solutions were acquired. The cor Table 1, since these data are used to discuss the 34 I. Kawabe et al.

present results in the next section. O!4 o o C 0 WI M O 0) The REE data for the surface seawater are N N +I +I M O d +I W taken from De Baar et al. (1985a). This set of 0 0 data is for the Pacific Ocean water at depth of 00 INS, N M C7 4 N 15 m from sea surface at the VERTEX II site. N U 0 +I N +I +I O ..oO Because Dy, Er and Y were not measured by De 0) 0 A N ~O Baar et al. (1985a), we estimated probable values d' 00 NN N 00 +I 00 Oyi +I "' +1 M for Dy and Er by interpolating in the REE abun O t N G dance pattern (see Fig. 1). The Y value which 0) C)0) could be consistent with the REE data by De -o N N N M 0\ H ' +1 M +I v, 0 Baar et al. (1985a) has been estimated by assum +1 0 ing a reasonable Y/Yb ratio as follows: We have

U r 00 Vn v, estimated the Y / Yb ratio from the previous 0 +I N +I o +1 00 .y C works that report not only REE concentrations N Q 0 q in seawater samples but also the Y concentra oor 0 E 0 o) 0 x w tions. Hogdahl et al. (1968) have reported such M N v M +I N +I O0 V1 0) NAA data for seawater samples from the central +1 0 0) 00 o~ Atlantic.Q Martin et al. (1976) have listed similar N U U ,~ U NAA determinations for an estuarine water with A N 0\ O 01 o '" +I K, +I M +I 0 N 28.3%o salinity in Gironde, France. Daidoji et al. Oy Z) (1985) have reported ICP-AES data of Y and M C) some REE for the surface seawater 100 m off the 00 0)0o 13 ES h H O M Pacific coast at Kumomi on the southwestern cor +1 O d ~ ~ 00 ner of Izu Peninsula, central Japan. The REE data by De Baar et al. (1985a, b), Martin et al. v1ONNN00 v M +I +I 0 V (1976) and Daidoji et al. (1985) show abundance patterns for heavy REE almost parallel to each ~"q 09 r-o M O W other. However, the heavy REE abundances in '" +I d• N C, O +I +I U the Atlantic seawater samples by Hogdahl et al. C) o O0 0 (1968) are found to be slightly different from the 0 N 00 above three. The Atlantic seawater samples have '"O N +1 -" +1 "" +1 -0 *G O S not been collected from the ocean surface but 0 0) 0) b V ~~-C from various depths greater than 900 m. N N o~ 0 z In N N M 0 O Therefore, from the data by Martin et al. (1976) N +I 8 +1 +I d and Daidoji et al. (1985), we have estimated an Y/Yb ratio of 18.2±0.6 for the surface ~F N '.0 O, N NN h N 0 a 00 00 0 +I N +I +1 O v: is seawater. This gives the Y value of 6.9 ppt listed O Q~ o ~ in Table 1. We did not use the Y / Ho ratio, O N 0) 00 (4 N 00 Ii M U In a, en W C' because Ho analyses by NAA and ICP-AES are N +I In +I "t +I v 0) subjected to larger analytical errors than Yb 0 0 N o r analyses. cO a 0 "t M a, '.0 N vi M v U PLI M+IN N The REE and Y data for the C 1 chondrite are +I +I N N wr) ts 0 C) h taken from the table of the CI chondrite mean M values by Anders and Grevesse (1989). Their * * 24 -a!'Z In' I U REE values for the C l chondrite are strikingly a a (8 Co CO rte.. !'O.h 0 parallel to the REE values for the ordinary con 0 6) * W W U 0U U * * drite of Leedey (L6) by Masuda et al. (1973) and

0N M Y/Ho ratio and lanthanide tetrad effect in limestones 35

10 Permian Limestone (EL-2) 5 fY

i

i

1

2 Permian Limestone (EL-1) tY 1

0 2 0.5 V C o 1 M C.) Y 0.2 0.5 f a. E 0 CO Bulawayan 0 20 Limestone Ul/ (Wildeman & Haskin, 1973) 0.1 10

Y 5

i

2

Surface Seawater(x 10 6) 1 La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu (Y) Fig. 1. REE plus Y abundance patterns for Permian and Precambrian limestone samples compared with the pattern for the surface seawater. The two Permian samples are from the Ohnogahara group (this study). The data for the Precambrian Bulawayan limestone are after Wildeman and Haskin (1973). The data and their sources on CI chondrite for normalization and surface seawater are listed in Table 1.

iT 36 I. Kawabe et al.

Masuda (1975) with a constant factor of Fig. 2 shows that the Y/Ho ratio is very close to 0.631±0.012 for all REE. Either set of chon the chondritic one. The REE pattern for the mar dritic REE reference values for normalization, ble is characterized by a straight line at least for therefore, can give essentially the same REE the segment from La through . Ho, suggesting abundance patterns. However, the set for the that the lanthanide tetrad effect is probably ab Leedey does not include the Y concentration. In sent. Haskin et al. (1966) have reported REE and order to discuss REE and Y together, we prefer Y data for nine carbonate samples including six red the Cl chondrite mean values to the Leedey limestones and two marbles. The REE plus Y values. abundance patterns for all their carbonates ap pear to resemble the pattern for the Red Creek DISCUSSIONAND CONCLUSIONS marble. However, fine features of their patterns are not so clear, probably because the analytical Characteristics of REE plus Y Abundance Pat uncertainties in their neutron activation analysis terns for these carbonates with low contents of REE The REE plus Y abundance patterns normal and Y are considerable. ized by C 1 chondrite for the three samples in this The close resemblance in REE and Y abun study are shown in Figs. 1 and 2, together with dance patterns between the limestones and the those for the surface seawater and the two surface seawater in Fig. 1 is quite expectable, if Precambrian carbonate rocks reported by the limestones have been formed in shallow Wildeman and Haskin (1973). The two Permian marine environments. The Ohnogahara limestone samples (EL-I and EL-2) show quite limestone body is thought to be the remains of a similar patterns commonly characterized by (1) Permian coral reef formed on the summit of a the W-type tetrad effect, (2) large Y / Ho ratios seamount at that time like other similar pre with positive deviations from the chondritic Cenozoic limestone bodies in southwest Japan value, and (3) negative anomalies of Ce and Eu. (Kanmera, 1987). This type of limestones does Their abundance patterns for heavy REE are not intercalate any terrigenous materials and very alike each other, but a noticeable difference always occurs overlying upon greenstones or can be seen in light REE, especially, in the first volcano-clastic rocks. These geological facts are tetrad. We like to emphasize that the three char consistent with the geochemical characteristics acteristics are also recognized in the REE plus Y of the Ohnogahara limestone from REE and Y abandance pattern for the surface seawater. The abundances. If significant amounts of ter pattern for the Precambrian Bulawayan rigenous materials are present in limestones , limestone by Wildeman and Haskin (1973) in their REE patterns become much similar to dicates a similar large Y/Ho ratio. The concave those of igneous rocks. The Y / Ho ratios also REE patterns for the first and second tetrads in become much close to the chondritic value as in the Precambrian Bulawayan limestone are igneous rocks. The source carbonate materials diagnostics of the W-type tetrad effect, although of the Red Creek marble, in this context, may no negative anomalies of Ce and Eu are ob have been formed under such an environment served. that some significant influx of terrigenous The REE and Y pattern for the calcareous materials was present. gneiss (CA-7) in Fig. 2 is quite different from all In view of the REE and Y patterns, the Hida the patterns in Fig. 1. The tetrad effect in heavy calcareous gneiss does not seem to resemble REE is not obvious and the fractionation be either the Ohnogahara limestone as a coral reef tween Y and Ho or other heavy REE is limestone or the Red Creek marble as a represen moderate, despite that a slightly concave pattern tative carbonate rock having terrigenous is recognized for the first tetrad. The Red Creek materials. As noted earlier, the Hida calcareous marble by Wideman and Haskin (1973) cited in gneiss and calc-silicate gneiss occur as alterna Y / Ho ratio and lanthanide tetrad effect in limestones 37

10

5 Calcareous Gneiss (CA-7J

Y

N 1 10 I-v C 0 V 0.5 5

r U 0 CL E «t CO

1

Y Q 0.5 Red Creek Marble --o (Wildeman & Haskin, 1973)

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu (Y) Fig. 2. REE plus Y abundance patterns for the Hida calcareous gneiss (this study) and for the Red Creek Mar ble (Wildeman and Haskin, 1973).

tions of marbles and quartzites. They are sub ed in REE and Y, because the analytical results jected to high- type regional do not differ significantly between the two sam metamorphism of upper amphibolite facies to ple solutions after the dissolution by HO alone granulite facies. Such marbles usually contain and after the complete dissolution by flux of small amounts of dolomite component up to 5 sodium carbonate plus boric acid. It seems mole%. These facts indicate that the calcareous difficult to judge whether the dolomitization and source materials initially deposited along with metamorphisms had altered the essential feature quartz-rich detritus materials and then suffered of the REE pattern or not. However, it is un from some weak dolomitization and regional likely that refractory minerals enriched in REE metamorphism. However, it is difficult to deter were originally present but completely decom mine whether the dolomitization occurred just posed during the dolomitization and metamor after depositional stage or during the metamor phism. Therefore, a unique depositional environ phism. The detritus materials do not contain ment, where quartz-rich detritus materials hav significant amounts of refractory minerals enrich ing no significant Y and REE stratificated 38 I. Kawabe et al. together with calcareous materials, is suggested. ocean boundary (Elderfield et al., 1990). Par Although the types of limestones examined ticulate and colloidal forms of REE are thought here are limited, it can be concluded that the to be removed considerably as the mixed colloids REE and Y abundances in limestones provide us of Fe hydroxide and organic matter coagulate important information on the depositional en rapidly in an early stage of the estuarine mixing vironments of source materials of the respective process (Hoyle et al., 1984; Elderfield et al., limestones. Besides the usual discussions about 1990). The remaining small fraction of partic Ce and Eu anomalies, the presence or absence of ulate and colloidal REE together with dissolved terrigenous materials in limestones can be shown REE are the continental supply to the oceans. by comparing the abundance patterns of REE However, the Y behavior in conjunction with plus Y for limestones with the unique pattern for REE cannot be inferred from the geochemical the surface seawater characterized by the W-type studies on estuarine mixing by Elderfield and tetrad effect and the non-chondritic Y/Ho ratio. coworkers, because they do not involve any Y data. "YAnomalies" in Seawaters and W-type Tetrad The results for the Gironde estuarine waters Effect by Martin et al. (1976) seem important, because We emphasized that the REE and Y abun they present the analytical results of Y and REE dance patterns for the surface seawater and for for particulates and water solutions in the the Permian limestone samples (EL-1 and EL-2) Gironde Estuary, France. They have filtered raw from Ohnogahara are commonly characterized water samples using a 0.45 am Millipore filter by the W-type tetrad effect, the unusually high and separated them into suspended particulates Y/Ho ratio greater than the chondritic value, and solutions. Both the solution and suspension and the negative anomalies of Ce and Eu (Fig. 1). from a water sample with the minimum salinity However, we also emphasize here that they are of S=0.1%o show REE abundance patterns com not exactly the same. For example, the abun parable with those of continental rocks having dance patterns of the fourth tetrad for the two no Y anomaly. The REE plus Y patterns for limestones can be shown by inclined smooth suspensions from other estuarine water samples curves, but heavy REE of the fourth tetrad for are almost the same as above irrespective of their the surface seawater displays a rather flat abun salinities. However, the solutions separated dance pattern with a slight concavity. Our esti from the waters with S = 7.0 and 28.3%o show W mate of the Y/Ho ratio for the surface seawater type REE patterns with positive Y anomalies is about twice of the chondritic value but it is still (Fig. 3). These characteristics are definitely clear smaller than the observed Y/Ho ratios in the in the solution with the maximum salinity of Ohnogahara limestones. In order to discuss 28.3%o. The results by Martin et al. (1976), these characteristics, we briefly refer to some therefore, are an important observation sug physico-chemical processes controlling the abun gesting that fine particles of terrigenous dances of REE and Y in seawaters. materials in estuarine waters with higher The Nd isotopic compositions in seawaters salinities become more likely to be filtered. As a and in oceanic hydrogenous sediments indicate result, the contribution of unfiltered terrigenous that continental input is the major source of REE solid particles to the Y and REE in filtrates in seawaters (Palmer and Elderfield, 1985 and becomes less important with increasing salinity, references therein). Major pathways of these and then the intrinsic features of abundances pat elements from the continents to the oceans are terns for dissolved Y and REE become visible. the rivers. Recent studies on REE in estuarine This interpretation is consistent with the con waters and their suspensions suggests that signifi clusion by Elderfield et al. (1990) as to the cant fractions up to 70-80% of riverine REE coagulation of particulates and colloids in the fluxes are removed in estuarine areas of the river estuarine mixing process. Therefore, estuarine Y/Ho ratio and lanthanide tetrad effect in limestones 39

500 Filtered Estuarine Waters Martin et al. (1976)

5=0.1 % o' 0 S=7.0 %o

N 100 O S=28.3 %o, L V c O 50 V Y

T V\~"0 Uco V O T Y i (D 10 0 0 0. E (0 5 Y I 0 ?r' I

1

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

(Y) Fig. 3. The enrichment patterns of REE and Yin the filtered Gironde estuarine waters reported by Martinet al. (1976).

waters as well as seawaters are to be understood shown for comparison. A and B correspond to as the natural environments in which terrigenous the averages for the seawaters at approximate fine particles are effectively transformed into the depths of 1,000 and 4,500 m, respectively. The particulates separable from the dissolved data sets by De Baar et al. (1983, 1985b) are chemicals. Such particulates are likely to be multiplied by respective constant factors of 0.71 separated from ambient solutions in the natural for A and 0.56 for B in order to make the rela environments, and they are easily separable by tive abundances of Sm in the two sets to be com artificial filtering procedure as well. mon. Both the relative REE patterns for the Figure 4 shows the REE plus Y abundance Atlantic deep seawaters by Hegdahl et al. (1968) patterns of deep seawaters in the central Atlantic and by De Baar et al. (1983, 1985b) are very by Hogdahl et al. (1968). The recent NAA data similar despite their diffrences in sample loca of REE by De Baar et al. (1983, 1985b) for deep tions and pre-concentration methods for seawaters in the western North Atlantic are also analyses. We believe that the reliability of REE 40 I. Kawabe et al.

Atlantic Deep Seawaters

10 Y Q

5 a) i -D r

U (A)

1 T

CD a 0

a) Q. Y E 10 I CI) f o 5 y

I

(s)

1 La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu (Y)

Fig. 4. REE and Y abundance patterns for deep sea waters at approximate depths of 1, 000 m (A) and of 4,500 m (B) in the Atlantic. Open circles: Average values of NAA data for seawaters in the Central Atlantic by Hegdahl et al. (1968). Solid circles: NAA data for sea water samples from the western North Atlantic Ocean by De Baar et al. (1983 and 1985b). In order to make the Sm abundances to be common between two sets of data , all the REE values for the deep sea waters from the western North Atlantic are multiplied in A and B by the con stant factors of 0.71 and 0.56, respectively.

and Y data reported by HOgdahl et al. (1968) is et al. (1968). endorsed by the recent data. Their Tb and Tm The analytical results by HOgdahl et al. values, however, seem to be subjected to large (1968) have been obtained for the unfiltered analytical errors as suggested by large apparent water samples, suggesting that any particulates variations in the original data. So that we do not in their water samples, if present, must have show the Tb and Tm data points by Hogdahl been analyzed together with dissolved Y and

0TU Y/Ho ratio and lanthanide tetrad effect in limestones 41

REE. In De Baar et al. (1983 and 1985b), on the waters from the seawater/sediment interface, other hand, sampled waters were pumped and the precipitation is just a reverse process of through a Chelex-100 column. Particulates must the dissolution. have been removed in the chromatographic col Keeping these points in mind, we have com umn. The reasonable agreement between the pared the variations of Y / Ho, Y / Yb and REE patterns based on the analyses with and Yb / Ho ratios in seawaters and marine deposits without associated particulates suggests that the including the limestones in this study (Fig. 5). presence of REE and probably Y in particulate The plotted data points are not many, because forms are insignificant in thier samples of the reliable data sets including REE and Y together Atlantic deep seawaters. Therefore, it is conclud are limited even today. Furthermore, because of ed that dissolved Y and REE in deep seawaters the lower abundance of Ho in the nature, the display the abundance patterns characterized by analytical uncertainty for Ho must be larger the positive Y anomalies and W-type tetrad than those for Yb and Y, whatever is the effect just like those in the surface seawaters. analytical method. Despite these difficulties, we can see the following points in Fig. 5: (1) Besides "Y -anomalies" in Limestones and Marine the three limestones shown in Fig. 1, the marine Deposits phosphorites as well as seawaters indicate large From the discussion about Y and REE data Y / Ho and Y / Yb ratios twice as great as the on estuarine waters and seawaters as above, we respective chondritic ratios, (2) The Red Creek can infer that natural waters having no ter marble shown in Fig. 2 and the argillaceous rigenous nor authigenic particulates show the limestones by Jarvis, I. and Jarvis, K. E. (1985) REE plus Y abundance patterns having the W are plotted close to the shale (NASC), but their type tetrad effect and positive Y anomaly. This Y/Ho and Y/Yb ratios are slightly higher the inference is partly justified by the observations chondritic ratios, (3) The deep-sea ferroma that W-type REE patterns are seen in fresh nganese nodules analyzed in our laboratory and groundwaters and valves of fresh water shellfish the marine sediments including one estuarine (Masuda et al., 1987). sediment reported by Jarvis, I. and Jarvis, K. E. We understand that the positive Y anomaly (1985) show the Y/Ho and Y/Yb ratios slightly and W type tetrad effect are originated from ther smaller than NASC, and (4) The deviations of modynamic processes related to the solubilities data points for the marine deposits from the of Y and REE in reactive phases like hydroxides chondrite or NASC in Fig. 5 appear to be consis involving Fe and Mn. The systematic differences tent with the observed variation of Yb / Ho ratios in 4f electronic configurations across the in seawaters by De Baar et al. (1983, 1985a, b). trivalent REE series and the absence of 4f elec We like to emphasize that the marine tron in Y can be reflected upon the ther phosphorites show positive Y anomalies but the modynamic process in the natural aqauatic en ferromanganese nodules and marine sediments vironments. If Y is a pseudo lanthanide which tend to show negative Y anomalies. Masuda and always behaves like Ho, the intimate association Ikeuchi (1979) have pointed out that the REE of the Y anomaly with W-type tertad effect in abundance pattern for the marine phosphorite seawaters cannot be explained. The reason why reported by Goldberg et al. (1964) exhibits a well the hydrogeneous deposits of deep-sea fer defined tetrad effect of the W type just like romanganese nodules display the REE patterns seawaters. This is also the case for the basically similar to those of the terrigenous phosphorites plotted in Fig. 5. The marine materials of shales is also explainable from this sediments and the deep-sea ferromanganese viewpoint: The deep-sea ferromanganese nodules shown in Fig. 5 display the REE pat nodules are formed as the precipitates from terns analogous to those of terrigenous materials seawater solutions partly influenced by pore like shales. Even though both of the phosphorite 42 I. Kawabe et al.

t

04. 100 at,~IIGrryCb ®s~sr2 a 0 0~01r0 O r00,gyp{ SG05'~p~ g G 1±,,, `rte Jtt `yp •/ A 80 Bulawayan EL-2

2 km A •W 60 5km EL-1 1 km

Surface Seawater 0 T Ic ACA-7 Deep Seawater x 40 AA Limestone /Red Creek Phosphorite Deep-sea Mn Nodule • 20 ., Marine Sediment O C 1 Chondrite 0 Shale (NASC)

0 0 10 20 30 40 50 (Y/Yb) by weight Fig. 5. Plots of Y/Ho ratio against Y/ Yb for limestones, marine deposits and seawaters in reference to C1 chondrite and NASC shale. Sources for the data points other than those in Table I and Figs. 1 and 2 of this study are as follows: NASC shale (Gromet et al., 1984; Haskin et al. 1966), Deep seawaters (Averages for three different approximate depths as indicated, based on the NAA data by Hegdahl et al., 1968), Limestones (two argillaceous limestones and a dolomitic limestone by Jarvis, I. and Jarvis, K. E., 1985), Phosphorites (Haskin et al., 1966; Jarvis, I. and Jarvis, K. E., 1985), Deep-sea Mn nodule (Average of eleven ferromanganese nodules from the central Pacific, unpublished data in our laboratory), and Marine sediments (two marine sediments, one estuarine sediment, and one marine mud by Jarvis, I. and Jarvis, K. E., 1985). The Yb/Ho ratios for shallow and deep seawaters are based on the results for the VERTEX II site in the Pacific (De Baar et al., 1985a).

and the ferromanganese nodule are marine tionations in Fig. 5 and the tetrad effects in authigenic deposits, they have contrasting char seawaters and limestones shown here may also acteristics of REE plus Y abundance patterns. suggest that such rate processes of scavenging Therefore, the intimate association of the non and regeneration of REE and Y in seawater col chondritic Y / Ho ratios with the W-type lan umns are important in addition to the admixing thanide tetrad effect in limestones and seawaters process of terrigenous and marine authogenic reported in this paper is merely one clue to recon materials. These processes in natural aquatic en sider the dissolution and precipitation reactions vironments, anyhow, make it visible that Y is involving seawaters and various authigenic not a pseudo lanthanide behaving like Ho. phases from the thermodynamic behaviors of REE and Y. The observed variety of Y/Ho frac Y/Ho ratio and lanthanide tetrad effect in limestones 43

Acknowledgments-We are grateful to Dr. M. Koma Hogdahl, O. T., Welsom, S. and Bowen, V. T. (1968) tsu (Ehime University) for his discussion and sugges Neutron activation analysis of lanthanide elements tions as to the Hida calc-silicate gneiss. Mr. M. Sudo in sea water. Adv. Chem. Ser. 73, 308-325. in our laboratory helped us in field and analytical Hoyle, J., Elderfield, H., Gledhill, A. and Greaves, works. We also thank Dr. T. Masuzawa (Water M. (1984) The behaviour of the rare earth elements Research Institute, Nagoya University) for his helpful during mixing of river and sea waters. Geochim. advice on literature data of seawater chemistry. This Cosmochim. Acta 48, 143-149. work was supported in part by a grant from Ehime Jarvis, I. and Jarvis, K. E. (1985) Rare-earth element University and by a grant for DELP from the Ministry geochemistry of standard sediments: a study using of Education, Science and Culture, Japan. inductively coupled plasma spectrometry. Chem. Geol. 53, 335-344. Jorgensen, C. K. (1970) The "tetrad effect" of Pep REFERENCES pard is a variation of the nephelauxetic ratio in the third decimal. J. Inorg. Nucl. Chem. 32, 3127 Anders, E. and Grevesse, N. (1989) Abundances of 3128. the elements: Meteoritic and solar. Geochim. Kamioka, H. and Masuda, A. (1986) Rare earth Cosmochim. Acta 53, 197-214. elements in Precambrian limestones. 1986 Ann. Daidoji, M., Tamura, S. and Matsubara, M. (1985) Meeting Geochem. Soc. Japan Abstr. 33. Determination of rare earth elements and thorium Kanmera, K. (1987) Carbonate rocks. Sedimentary in sea water by inductively coupled plasma atomic Rocks in Japan, Mizutani, S., Saito, T. and emission spectrometry. Bunseki Kagaku 34, 340 Kanmera, K. eds., pp. 85-142, Iwanami Shoten, 345. Tokyo. De Baar, H. J. W., Bacon, M. P. and Brewer, P. G. Kawabe, I., Okumura, M. and Ochi, M. (1988) (1983) Rare-earth distributions with a positive Ce Chemical analysis of silicate and rock samples anomaly in the western North Atlantic Ocean. by inductyively coupled argon plasma atomic emis Nature 301, 324-327. sion spectrometry. Mem. Ehime Univ. Sci. Ser. D. De Baar, H. J. W., Bacon, M. P., Brewer, P. G. and 11, 1-13. Bruland, K. W. (1985a) Rare earth elements in the Martin, J.-M., Hogdahl, O. and Philippot, J. C. Pacific and Atlantic oceans. Geochim. Cosmochim. (1976) Rare earth element supply to the ocean. J. Acta 49, 1943-1959. Geophys. Res. 81, 3119-3124. De Baar, H. J. W., Brewer, P. G. and Bacon. M. P. Masuda, A. (1975) Abundances of monoisotopic (1985b) Anomalies in rare earth distributions in REE, consistent with the Leedey chondrite values. seawater: Gd and Tb. Geochim. Cosmochim. Acta Geochem. J. 9, 183-184. 49, 1961-1969. Masuda, A., Nakamura, N. and Tanaka, T. (1973) Elderfield, H., Upstill-Goddard, R. and Sholkovitz, Fine structures of mutually normalized rare-earth E. R. (1990) The rare earth elements in rivers, patterns of chondrites. Geochim. Cosmochim. Acta estuaries, and coastal seas and their significance to 37, 239-248. the composition of ocean waters. Geochim. Masuda, A. and Ikeuchi, Y. (1979) Lanthanide tetrad Cosmochim. Acta 54, 971-991. effect observed in marine environment. Geochem. Goldberg, E. D., Koide, M., Schmitt, R. A. and J. 13, 19-22. Smith, R. H. (1963) Rare earth distribution in the Masuda, A., Kawakami, 0., Dohmoto, Y. and marine environment. J. Geophys. Res. 68, 4209 Takenaka, T. (1987) Lanthanide tetrad effects in 4217. nature: two mutually opposite types, W and M. Gromet, L. P., Dymek, R. F., Haskin, L. A. and Geochem. J. 21, 119-124. Korotev, R. L. (1984) The "North American Shale Masuda, A. and Akagi, T. (1989) Lanthanide tetrad Composite": Its compilation, major and trace ele effect observed in leucogranites from . ment characteristics. Geochim. Cosmochim. Acta Geochem. J. 23, 245-254. 48, 2469-2482. Nugent, L. J. (1970) Theory of the tetrad effect in the Haskin, L. A., Frey, F. A., Schmitt, R. A, and Smith, lanthanide (III) and (III) series. J. Inorg. R. H. (1966) Meteoric, solar and terrestrial rare Nucl. Chem. 32, 3485-3491. earth distributions. Phys. Chem. Earth 7, 167-321. Palmer, M. R. and Elderfield, H. (1985) Variations in Haskin, L. A., Wildeman, T. R., Frey, F. A., Collins, the Nd isotopic composition of foraminifera from K. A., Keedy, C. R. and Haskin, M. A. (1966) Rare Atlantic ocean sediments. Earth Planet. Sci. Lett. earths in sediments. J. Geophys. Res. 71, 6091 73, 299-305. 6105. Peppard, D. F., Mason, G. W. and Lewey, S. (1969) 44 I. Kawabe et al.

A tetrad effect in the liquid-liquid extraction order actinide (III) and lanthanide (III) tetrad effects. J. ing of lanthanides (III). J. Inorg. Nucl. Chem. 31, Inorg. Nucl. Chem. 32, 343-344. 2271-2272. Wildeman, T. R. and Haskin, L. A. (1973) Rare Peppard, D. F., Bloomquist, C. A. A., Horwitz, E. earths in Precambrian sediments. Geochim. P., Lewey, S. and Mason, G. W. (1970) Analogous Cosmochim. Acta 37, 419-438.