JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1991 VOL. 6 129 Pre-concentration by Coprecipitation Part 1. Rapid Method for the Determination of Ultra-trace Amounts of Germanium in Natural Waters by Hydride Generation-Atomic Emission Spectrometry* Ian D. Brindle Mary E. Brindle and Xiao-chun Let Chemistry Department Brock University St. Catha fines Ontario L2S 3A 1 Canada Hengwu Chen Department of Chemistry Hangzhou University Hangzhou Zhejiang 3 7 0028 People 3 Republic of China A rapid method for the determination of ultra-trace amounts of germanium in sea-water surface water and ground- water has been developed. This method initially involved coprecipitation of gallium and magnesium with hydroxide ions. Subsequent investigations have shown that gallium is unnecessary for quantitative recovery if calcium and carbonate ions are added with the magnesium and the precipitation is carried out at a high pH.The formation of a rather coarse precipitate aids in the filtration step by reducing the filtration time from hours to minutes without any loss of analyte. The control of pH is crucial to the successful use of this method. Concentrations of germanium in waters have been determined by hydride generation atomic emission spectrometry with a detection limit for the method of 0.6 pg ml-' (30) for a 1 I sample. Determinations of germanium concentrations in ground sea and surface waters range from 0.5 to 17 pg mi-' (3 I sample). Initial experiments into the application of the method to samples spiked with 20 ng ml-1 concentrations of copper colbalt nickel and zinc have proved to be effective.Keywords Pre -concen tra tion; coprecipita tion; germanium determination; hydride genera tion; natural water Analytical methods used for the determination of trace levels of germanium have included spectrophotometry1-3 and atomic spectrometry."I3 Among the spectrophotometric determina- tions of germanium a recently reported method3 showed the best detection limit to be approximately 48 ng ml-I of germa- nium. Hydride generation coupled with atomic spectro- metry,I@l3 however has generally provided a further improvement in the detection limit for the determination of germanium. Andreae9 trapped germane in a liquid nitrogen U- tube before releasing it to an atomic absorption spectrometer for detection obtaining a detection limit of about 200 ng ml-I.Zhang et a/.' I recently demonstrated in situ pre-concentration of germanium in a palladium coated graphite furnace. Upon the increase of temperature of the graphite furnace the trapped germanium was released and then determined by atomic ab- sorption spectrometry. With this in situ pre-concentration sen- sitivities expressed in terms of characteristic mass for arsenic antimony and selenium of 10.0 13.1 and 14.7 pg per 0.0044 A respectively were reported. A gas flow batch type hydride generation system for the determination of hydride-forming elements in the presence of L-cysteine has been developed.12-15 With this method high sensitivity was obtained without pre-trapping. A detection limit of 12 pg ml-I was obtained for germanium.'3 This is the best detection limit reported to date.However the method was not sensitive enough for the direct determination of germani- um in surface-water samples as concentrations of germanium in these samples were at sub-pg ml-l levels. A pre- concentration method was therefore required in order to bring the concentrations within the measurable range. Coprecipitation by hydroxides of metals such as Al Cd Fe Ga La and Mg has been reported for the pre-concentration of trace Although these pre-concentration systems have been widely used Akagi et al." claimed a superior copre- cipitation system for the determination of trace metals by induc- tively coupled plasma atomic emission spectrometry (ICP- * Presented st the Fifth Biennial National Atomic Spectroscopy Symp- t Present address Department of Chemistry.University of British sium (BNASS). Loughborough UK 18th-20th July. 1990. Columbia. Vancouver. British Columbia V6T I WS. Canada AES). They reported that precipitation of Ga(OH) in the pres- ence of magnesium gave good recoveries for Al Co Cr Fe La Mn Ni Pb Ti V Y and Zn. With this system these workers noted that spectral interferences from gallium in the determina- tion by ICP-AES were negligible and the contamination was minimal. Akagi and Harapchi*> have recently reported a modification of the method that included a centrifugation step. Another method for the determination of elements in water which involves a reductive precipitation has been described. Tetrahydroborate( HI) was used to reduce palladium and iron24 or palladium alone2s which allowed efficient collection of trace concentrations of various elements from water.Since iron con- tains germanium as a ubiquitous impurity and as palladium in- terferes with the generation of hydrides," this method cannot be used for the determination of those elements by hydride generation AES. The combination of gallium and magnesium was thought to be an appropriate medium for the collection of ultra-trace concen- trations of elements in surface waters where magnesium would be added to the water to aid in the coprecipitation process. By using the coprecipitation approach for the pre-concentration of germanium the gallium-magnesium system was investigated. Experimental Instrumentation Data were acquired with a Spectraspan V d.c.plasma atomic emission spectrometer equipped with a Beckman hydride gen- erator modified as described elsewhere.14 Signals were record- ed on a Fisher Recordall Series 5000 chart recorder. A Brinkmann variable volume Macro-Transferpettor was used for all analyte injections with the volume fixed at 5.0 ml. Determinations of pH were performed with a Radiometer pH meter (PHM 82 standard pH meter). Reagents The germanium standard L-cysteine and sodium tetrahydro- borate( 111) were prepared as previously described." l 7 Gallium metal (Aesar Toronto Ontario Canada) was dissolved in con-130 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1991 VOL. 6 centrated nitric acid and diluted with distilled water to the re- quired concentration. Magnesium (Aesar) was dissolved in 1+1 nitric or hydrochloric acid.Calcium was prepared from calcium carbonate (Aesar) by dissolving in 1 mol dm-3 hydro- chloric acid. Sodium hydrogen carbonate [BDH (AnalaR) Toronto Ontario Canada] was dissolved in distilled water. Solutions of metals were prepared from dilutions of atomic ab- sorption standards or from the elements which were of analyti- cal-reagent grade or better. Purification of the reagents was performed by adding freshly prepared magnesium hydroxide to a solution of the reagent at the appropriate pH to prevent dissolution. The solution was stirred and heated and the precipitate allowed to settle after which the supematant solution was decanted off; thus the blank values for germanium were reduced by approximately half.This prwedure was described by Doroshkov and Chuiko26 for the purification of magnesium sulphate. Precipitation Procedure Method one To 1 1 of water were added magnesium and gallium solutions to give final concentrations of 500 and 25 pg ml-I respective- ly. The sample was then treated by the same method as used by Akagi et a1.22 For the determination of germanium the so- lution was prepared as described in previous work.I2 Method two Solutions of magnesium calcium and sodium hydrogen car- bonate (all 1% m/v) were added to a 1 1 volume of sample to give final concentrations of 200 60 and 200 pg ml-' re- spectively. The pH was then raised to 9.75 * 0.25 with sodium hydroxide solution ( 1 mol dm-3) and the solution heated to near boiling point and then allowed to cool over- night.After cooling the solution was filtered through a 5 pm filter membrane. The membrane was removed from the appa- ratus and placed in a beaker. The precipitate was dissolved in 1 mol dm-' nitric acid and the solution treated appropri- ately for the determination. For the determination of germa- nium the solution was prepared as described in previous work.'* Results and Discussion Preliminary Studies Method One For the coprecipitation of trace amounts of germanium similar conditions to those reported by Akagi et al." were used. A satisfactory recovery of 95% from a solution containing 0.1 ng ml-1 of germanium was obtained. However the procedure was very time consuming with the filtration step requiring up to 8 h to recover the precipitate from 1 1 of solution.In order to overcome this problem attempts were made to reduce the filtration time whilst retaining the good recovery of ger- manium. Overnight digestion produced a precipitate that was more tractable. The supernatant solution (approximately 900 ml) passed through the filter rapidly and slow filtration occurred only as the last 100 ml of the mixture of precipitate and solu- tion from the bottom of the beaker were filtered. Thus the total filtration time was shortened from 8 to approximately 3 h. A recovery of 97% from 0.1 ng ml-I of germanium in 1 .O 1 of so- lution was obtained by use of this technique. Various digestion procedures were applied in an attempt to reduce the filtration time further. Filter membranes of various pore sizes were also investigated.The results are shown in Table 1. The results in Table 1 also indicate that slight changes in the concentrations of magnesium and gallium did not significantly affect the recovery of germanium. With a higher concentration of magnesium and gallium however the time required for filtration was also longer owing to the larger amount of precipitate formed. Obviously variation of sample volume changed the filtration time as this depends on the amounts of sample and precipitate. Since the 5 pm filter mem- brane gave similar results to 0.4 and 1.0 pm filters 5 pm filter membranes were used in further studies. Coprecipitation of Germanium in the Presence of Mg2+ Ga3+,Ca2+ and HC03- Based on the preliminary studies described above coprecipita- tion and determination of germanium in water samples were performed following method one as stated under Experimen- tal.During the analysis of a particular ground-water sample the filtration time for the sample precipitate was much shorter than that for the precipitate formed in a blank. A visibly coarser precipitate was obtained from the sample than from the blank which contained distilled water and the reagents for the coprecipitation. The sample and the blank differed only in their matrix. The ground-water sample contained a high level of calcium (approximately 200 pg ml-I). The formation of CaC0 was thought to be responsible for the larger particle size of the precipitate which eased the filtration. Therefore further investigations involving the addition of reagents to allow the formation of CaC03 were carried out.Instead of the direct addition of CaCO to the sample solu- tion the addition of Ca2+ [in the form of CaC& or Ca(N03)2] Table 1 Recovery of Ge and filtration time under different experimental conditions Concentration of Mg/ Concentration of Ga/ Overnight Sample pg ml-' pg ml-' digestion 1 .0 I of distilled water 0.1 ng ml-' of Ge in 1 .0 1 of 0.1 ng ml-' of Ge in 1 .0 I of 0.1 ngml-'ofGein l.0lof 0.02 ng ml-' of Ge in 0.5 1 of 0.1 ng ml-' of Ge in 1 .0 1 of 0. I ng ml-I of Ge in I .0 I of 0.05 ng mf-' of Ge in I .O 1 of 0.01 ng ml-I of Ge in 1.0 I of distilled water distilled water distilled water distilled water distilled water distilled water distilled water distilled water 500 500 550 480 5 10 640 560 500 500 25 25 30 25 40 38 25 25 25 No Yes No Yes Yes Yes Yes Yes Yes Heating digestion No No Yes Yes Yes Yes Yes Yes Yes Membrane size/pm 1 1 1 1 1 0.4 0.4 5 5 Filtration time/h 8 3 3 1 0.8 3 2 0.8 0.7 Recovery of Ge (96) - 97 97 98 93 91 94 99 97131 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1991 VOL.6 g 120 c I (a) A 1 .- 5 100 E C 80 & 60 r $ 40 20 0 a - 0 100 200 300 400 500 600 700 Filtration time/min Fig. 1 (a) Recovery of germanium as a function of magnesium at two concentrations of calcium. ( h ) Filtration time as a function of magnesium concentration at two concentrations of calcium. A 60; and B 180 pg ml-l of calcium a 0 g 7 7.5 8 8.5 9 9.5 10 10.5 11 PH Fig. 2 Recovery of germanium as a function of pH and NaHC03 solutions to the sample was chosen. Upon the addition of NaOH used to adjust the pH of the sample solu- tion (e.g.to pH lo) CaC03 Mg(OH)? and Ga(OH)3 were formed. This precipitate quantitatively collected the analyte and formed larger precipitate particles. With this modification an initial set of two experiments gave recoveries of 98-104% for 100 ng of germanium spiked in 1.6 1 of distilled water and less than 10 min were required for the filtration. Optimizations were carried out with this modified coprecip- itation system by varying the concentration of Mg2+ at calcium concentrations of 60 and 180 pg ml-i in order to maximize the recovery of germanium and minimize the filtration time while the concentrations of Ga3+ and NaHC03 were kept constant at 25 and 150 pg ml-l and the pH of the solution was kept at 9.Fig. l(a) shows the recovery of germa- nium and Fig. l(h) the filtration time as a function of the con- centration of Mg2+ in the presence of 60 and 180 pg ml-I of Ca”. As can be seen from Fig. 1 at a Ca2+ concentration of 60 pg mki complete recovery of germanium was obtained when the Mg?+ concentration reached approximately 30 pg ml-I. Below this level germanium was not completely recov- ered. Within the range of Mg2+ concentrations 30-380 pg ml- I good recoveries of germanium were obtained. With a higher concentration of Mg” a longer filtration time was also re- quired as indicated in Fig. I(h). probably because of the for- mation of a larger amount of Mg(OH) precipitate. When a concentration of Ca2+ of 180 pg ml-i was used a higher minimum concentration of Mg’+ was also required to provide good recovery of germanium.As shown in Fig. I(a) the re- covery of germanium was incomplete at Mg2+ concentrations below 180 pg ml-i in the presence of 180 pg m1-I of Ca2+. However trace amounts of germanium were completely re- covered when the Mg2+ concentration was greater than 250 pg ml-I in the solution containing 180 pg ml-i of Ca2+. The filtration time was generally shorter when more Ca2+ was present in the solution as illustrated by the two curves in Fig. 1. The results shown in Fig. I suggest an interdependence between the concentrations of Mg2+ and Ca2+. This interdepen- dence can be ascribed to competitive precipitation and copre- cipitation among the various species forming the precipitate. The presence of Ca2+ did not enhance the recovery of germa- nium probably because CaC03 could not efficiently collect the germanium; but its presence allowed the precipitate to be more easily removed hence a more rapid filtration time was achieved.The basicity of the solution was also optimized as it is another important factor which influences the quality of copre- cipitation. Fig. 2 demonstrates the effect of pH on the recovery of germanium. As shown in Fig. 2 at a pH value of 8. I only 22% of the germanium was recovered. At this pH little precip- itate was observed. Over the pH range of 9.0-10.5 germanium was quantitatively recorded. Thus a pH of between 9.5 and 10.0 was chosen for the precipitation. Coprecipitation of Germanium With Mg and Ca Method Two Suprisingly when the Ga3+ concentration was varied from 1.5 to 60 pg ml-I complete recovery of germanium was obtained at all of the gallium concentrations studied.Therefore further studies were carried out with a coprecipitation system in the absence of gallium. It was found that in the absence of gallium trace amounts ( 100 ng) of germanium were quantitatively recovered from a 1.6 1 solution containing 60 180 and 600 pg ml-I of Ca2+ Mg and NaHC03 respectively after the pH had been adjusted to 10. For comparison a recovery of 9 1 % was also obtained from 100 ng of germanium in 1.6 1 of a solution containing only 120 pg ml-i of Mg2+ after the pH had been adjusted to 10. These results suggest that Ga(OH) is not necessary as a coprecipita- tion carrier. It appears that Mg(OH)2 is the primary carrier in the coprecipitation of trace amounts of germanium. The proposed coprecipitation system developed in this work consists of Mg” Ca2+ and HC03-.The optimum condi- tions for this system were briefly re-examined based on those previously studied where gallium was present. It was found that the concentrations of Mg2+ and Ca2+ showed the same in- terdependence and their optimum concentrations were similar to those shown in Fig. 1. However with the modified system without gallium the filtration time was further reduced to between 1 and 5 min for a sample of approximately 2 1. Quan- titative recoveries of germanium were obtained at a similar pH range as shown in Fig. 2. Thus at pH values between 9.0 and 10.5 recoveries of 90-1 10% were obtained and hence a pH of 9.75 f 0.25 was used.Samples prepared by the three methods were investigated by electron microscopy. The precipitate formed from gallium and magnesium hydroxides verges on being amorphous. The addition of calcium and carbonate to the initial solution causes the formation of dendritic crystals which probably act as a filter medium to support the relatively amorphous Ga-Mg precipitate. Crystals of possibly cubic habit are also observed covered with the amorphous precipitate. For the precipitate formed with magnesium and calcium rosettes of dendrites and framboidal twinned crystals possibly of cubic habit are apparent with little evidence of an amorphous pre- cipitate. Once the optimum concentrations of Mg2+ and Ca2+ had been chosen the concentration of NaHC03 was varied from 25 to 750 pg ml-I and showed little effect on germanium re- covery.An NaHCO? concentration in the range of 100-200 pg ml-I was chosen.129 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1991. VOL. 6 Table 2 Recoveries of spiked germanium from water samples Concentration of Ge determined/ Ge spiked/ Ge found/ Ge recovery Sample pg ml-I ng ng (%) 500.0 ml of sea-water 4.5 0 10.0 ml of 1 .O ng ml-I of Ge spiked into 500.0 ml of sea-water 24.0 - - 10.0 9.75 - 0 98 3.0 I of Smith- 10.0 ml of 2.5 ville well water 15.7 ng ml-I of Ge spiked into 3.0 1 of Smithville well water 23.6 25.0 23.7 95 Table 3 Concentration of germanium in water samples Concentration of Ge/pg ml-l ~ Source of water sample Method one Method two A well in Smithville. Ontario 17 17 ? 1.5 (n=3) Welland River Ontario 1.5 1.4 Port Colborne Lake Erie 1.3 Port Weller Lake Ontario 0.7 Queenston Spring water Niagara Falls Ontario 0.5 Sea- water 4.5 - - - - Recovery of Germanium Spiked Into Water Samples Recoveries of the germanium spiked into real water samples were also studied the results are listed in Table 2.The recov- eries of germanium spiked into a sea-water sample was obtained by method one where gallium was present. Recovery results for 25 ng of germanium spiked into a Smithville (Ontario Canada) well-water sample was obtained by method two in the absence of gallium. As can be seen from Table 2 satisfactory recoveries of germanium were obtained in both in- stances indicating that the use of gallium is unnecessary. Determination of Germanium in Water Samples A number of water samples were analysed for germanium ini- tially by method one; the results are summarized in Table 3.For comparison two of the samples were also analysed by method two and the results are in good agreement with those obtained by method one. Germanium concentrations in all the water samples analysed are at low pg ml-I levels. One groundwater sample (Smithville well water) contained a higher concentra- tion of germanium. The reason for this higher germanium con- centration in the well water is currently being investigated. The concentration of germanium in the reagent blank a 3.0 1 solution containing 170 65 and 130 pg ml-I of Mg Ca and NaHC03 respectively at a pH of 10 was found to be approxi- mately 1 pg ml-I. Blank values for germanium from solutions of 6% NaBH 0.4% L-cysteine and 0.01 mol dm-.> HNO? that had not undergone pre-concentration were not detectable.When the reagents were pre-purified simply by stirring with Mg(OH) the blank was reduced to approximately half of the original amount as indicated under Experimental. Pre-concentration of Transition Elements A preliminary study of transition elements (Cu Co Ni and Zn) spiked into distilled water at 20 ng ml-I gave effective re- coveries using method two but higher pH values were required than for the collection of germanium. Essentially quantitative recoveries ( 100 f 10%) of the four elements tested were obtained in the pH range 10-10.5. At the lower pH values smaller amounts of precipitate were obtained. Thus at pH 9 recoveries of 20-30% were obtained.The detection limits for the elements are given by the manufacturer as Cu 0.002 pg ml-I at 324.754 nm; Co not given at 341.474 nm; Ni 0.006 pg ml-I at 305.082 nm; and Zn 0.006 pg ml-1 at 202.548 nm. Conclusion A rapid quantitative method for the pre-concentration of ele- ments from various waters has been developed. Results for the pre-concentration of germanium are excellent. Further work on the pre-concentration of other elements from surface and ground waters will be undertaken in order to develop a useful method for the determination of ultra-trace concentrations of elements in waters. The authors thank the Ontario Ministry of the Environment for funding this research (Grant 4346). H.C. gratefully ac- knowledges the assistance of Hangzhou University for provid- ing funds for a visiting scholarship.G. Hooper of Norton Advanced Ceramics is thanked for the electron microscope work. I 2 3 4 5 6 7 8 9 10 1 1 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 References Hemandis V. Macia L. and Sala J. V. Analyst 1987 112 1007. Shen H. Wang Z. and Xu G.. Analyst 1987 112,887. Nukatsuka I. Takahashi K. Ohzeki K. and lshida R. Analyst 1989,114 1473. Eckhoff M. A. McCarthy J. P. and Caruso J. A. A d . Chrm. 1982,54 165. Thompson M. and Pahlavanpour B. Anal. Chim. A m 1979 109 251. Fazakas J. Tulunta 1984,31 573. Nakata F. Sunahara H. Fujimoto H. Yamamoto M.. and Kumena- ru T.. J. Anal. At. Specwom. 1988 3 579. Hambrick G. A.. Ill Froelich P. N. Jr. Andreae M. O. and Lewis B. L. Anal. Chem. 1984,56,421 Andreae M. O. and Froelich P. N. Jr. Anal. Cheni. 1981 53 287. Halicz L. Analvst 1985 110,943. Zhang L. Ni Z. M. and Shan X.-q. Spwtwdiirn. Acta. Part B 1989,44,339. Brindle 1. D. Le X.-c. and Li X.-f. J . Anal. A?. Spectt.om. 1989. 4 227. Brindle 1. D. and Le X.-c. Anal. Chern. 1989.61 1175. Brindle 1. D. and Le X.-c. Anulwt 1988. 113 1377. Brindle 1. D. and Le X.-c. Anal. Chim. A m . 1990 229 239. Bruninx E.. P h i l i p J. Res. 1979,33,264. Novikov. A. I. and Shchekoturova E. K. Radiokhimiw 1972 14 152. Andrianov A. M. and Poladyan V. E. Zh. Anul. Khini.. 1975 30 1622. Sarkisov. E. S. Lidin R. A.. and Krymskaya. E. B.. I:).. Akucl. Nard. S.S.R.R. Neor<q. Muter .. 1970,6 281. Tsuyama A. and Nakashima S. Birnseki Kagakrr 1980.29 8 I . Shigelomi. Y. Birnseki K a p k i r 1975 24 699. Akagi. T.. Fuwa K.. and Haraguchi H. A n d . Chin?. A m . 1985. 177 139. Akagi. T. and Haraguchi H. A w l . Ckeni.. 1990 62 8 1. Skogerboe. R. K. Hanagan. W. A. and Taylor H. E.. A n d . Chem. 1985.57.28 15. Nakashima. S. Sturgeon R. E. Willie. S. N.. and Berman S. S.. Anal. Chin?. A r w 1988. 207 29 1. Doroshkov V. Ya.. and Chuiko V. T.. /:I-. Khini. Khini. TrA.hnol.. 1967. 10 1087. Puper- 0/03345F Rec.ei\*ed July 24th. 1990 Accepted No\vmher 12th. 1990