Anion Distributions in Sodium Silicate Solutions. Characterization by

of these anions varies with the concentration of dissolved silicate solids, the relative molar concentrations of cations and silica, and to a lesser e...
0 downloads 0 Views 175KB Size
Download PDF
10638

J. Phys. Chem. B 1997, 101, 10638-10644

Anion Distributions in Sodium Silicate Solutions. Characterization by Infrared Spectroscopies, and Vapor Phase Osmometry

29SI

NMR and

Jonathan L. Bass* The PQ Corporation, Research and DeVelopment Center, Conshohocken, PennsylVania 19428

Gary L. Turner Spectral Data SerVices, Inc., Champaign, Illinois 61820 ReceiVed: May 7, 1997; In Final Form: September 25, 1997X

Soluble silicates are complex mixtures of silicate anions. The larger anions are two- or three-dimensional condensation products of silicate monomer, SiO4-4. In more siliceous silicate solutions, i.e., SiO2:Na2O ratio >2.0, some of the silicate condenses to polymeric (colloidal) silica. Negative charges on the anions are balanced by protons or by cations, typically alkali metal or quaternary alkylammonium cations. The distribution of these anions varies with the concentration of dissolved silicate solids, the relative molar concentrations of cations and silica, and to a lesser extent, trace impurities. 29Silicon NMR spectroscopy of silicate solutions has been a powerful method for studying the connectivity of silicon and oxygen atoms in silicate solutions. In this paper we will show how variations in silicate band shape and peak location in the mid-infrared spectrum from 700 to 1300 cm-1 depend on concentration and silica:alkali ratio. We will interpret the infrared data in consonance with NMR results. We will also relate these variations in anion distribution to the average molecular weight of silicate solutions.

Introduction In his classic text on soluble silicates, Vail1 discusses several properties of silicate solutions which indicate that they are complex mixtures of anions. Of particular relevance to this report are the light-scattering results of Debye and Nauman,2 who showed that average molecular weight (AMW) of very dilute silicate solutions increased from about 60 to about 250 as the molar ratio of SiO2 to Na2O increased from 1 to over 3. Their results will be compared with average molecular weight measured by a vapor-pressure-lowering method. The trimethylsilylation method of Lentz,3 combined with gas chromatography-mass spectroscopy (GC-MS),4,5 was an important advance in detecting individual oligomeric silicate anions. This method has proven to be limited by the formation of nonvolatile trimethylsilyl derivatives of larger anions. Gel permeation chromatography was used4 to obtain molecular weight distributions of the nonvolatile derivatives. There appears to be evidence that some of the smaller anions condense during the derivatization process. 29Silicon nuclear magnetic resonance (NMR) spectroscopy has proven to be a powerful method for measuring the distribution of Si-O linkages in silicate anions where a silicon atom may be linked to one or more neighboring silicon atoms through Si-O bonds. The Qi notation, developed by Engelhardt et al.,6 in which i indicates the number of neighboring silicon atoms linked, through intermediate oxygen atoms, to the atom in question, is standard usage for describing 29Si spectra. These authors assigned 29 Si NMR peaks as follows: Q0 ∼ -72 ppm, Q1 ∼ -79 ppm, Q2(3R) (cyclic trimer) ∼ -82 ppm, Q2 and Q3(3R) (three ring) ∼ -87 to -91 ppm, Q3(4R)) (four ring) ∼ -96 to -98 ppm, and Q4 ∼ -108 ppm (broad). Svensson et al. 7 used 29Si NMR to characterize sodium silicate solutions with molar ratio of SiO2:Na2O ranging from 1 to 3.3 at concentrations typically found in commercial products. X

Abstract published in AdVance ACS Abstracts, December 1, 1997.

S1089-5647(97)01528-9 CCC: $14.00

Beard8 characterized concentrated alkali silicate solutions using transmission infrared spectroscopy. He studied silicates with different silica-to-alkali ratios and also different alkali cations. Since his samples were thin liquid films between two barium fluoride windows which totally absorb infrared radiation at about 900 cm-1, he was not able to observe some lower frequency Si-O vibrations. The alkali halides, such as KBr, which transmit below 900 cm-1, would dissolve in silicate solutions. Since the strongest peak in the Si-O vibration region, near 1000 cm-1, is very intense, he was limited to solutions which were 15% silica by weight, or less, which is approximately half the typical silica concentration found in commercial solutions. Couty and Fernandez9,10 compared 29Si NMR and FTIR of 1.4 M potassium and sodium silicate solutions with molar ratios ranging from 1.6 to 4.56. They reported that the disappearance of a shoulder at 1075 cm-1 in the FTIR spectrum corresponded with the disappearance of the Q4 band in the NMR spectrum. They resolved the observed band peaking near 1000 cm-1 into component bands using a Gaussian curve fit model.9 Roggendorf et al.11 reported NMR and FTIR results on 2.0 and 3.3 ratio silicates at commercial concentrations. They did not perform detailed analyses on the silicate absorption band. In this paper we use current infrared spectroscopic technology to correlate the features of silicate solution IR spectra with 29Si NMR spectra, as a function of SiO2:Na2O ratio and SiO2 concentration. All silicate solutions are prepared from silicate stock solutions which have consistent levels of impurities. The results will show that infrared spectroscopy serves as an important technique to gain insight into the complex chemistry of silicate solutions, especially since it is less expensive, with faster data acquisition, compared to 29Si NMR. Experimental Section In this paper, references to the molarity of a silicate solution, M, will mean the molarity of SiO2. All solutions were made using PQ Corporation N brand silicate, a commercial silicate © 1997 American Chemical Society

Sodium Silicate Solutions solution, 8.9% Na2O, 28.7% SiO2 by weight, as a stock solution. Trace impurities in this solution are 300 ppm Al2O3, 55 ppm Fe2O3, 60 ppm TiO2, 40 ppm CaO, and 15 ppm MgO. The solutions were diluted with high purity(>10 MΩ) water, and for those samples of lower ratio, reacted with J.T. Baker Analyzed Reagent 50% sodium hydroxide solution in appropriate amounts. Since the trace impurities in the NaOH solution total less than 10 ppm, impurity levels in the prepared solutions are proportionally reduced by dilution. Solutions were equilibrated at ambient temperature for at least 200 h prior to analysis. DuPont LUDOX SM-30 silica sol was obtained from Aldrich, Milwaukee, WI. The most concentrated solutions were made at 6M with SiO2: Na2O ratios of 3.3, 2.98, 2.48, 2.07, and 1.65. These are close to the concentrations of many commercial silicate solutions. Solutions with these ratios were also prepared at lower concentrations. When the SiO2:Na2O ratio is less than about 1.2, the solubility of silicate decreases sharply at room temperature as crystals of sodium metasilicate form. Therefore, solutions with SiO2:Na2O ratio of 1.03 and lower were prepared at concentrations of 1M or less. 29Si NMR spectra were obtained on a 363 MHz spectrometer, operating at a 29Si NMR frequency of 72.18 MHz. Data were collected with either Nicolet or Tecmag data acquisition systems, using a 10 mm Cryomagnet Systems, Inc., probe with a quartz insert. Quartz NMR tubes (Wilmad) were employed for all samples. Spectra were obtained as neat liquids with a ca. 80° pulse, with a 5 s recycle delay. Recycle delay tests showed that the 5 s recycle delay yielded quantitative results, even for Q4 silicon. Typically 700 to 6000 scans were accumulated for each sample; the number of scans was increased as the silica concentration decreased. FTIR spectra were collected on a Nicolet (Madison, WI) Magna 550 spectrometer using a SpectraTech (Shelton, CT) single/multibounce HATR accessory. A single bounce 45° ZnSe element was used for the 0.4 to 6 M solutions, while a multibounce 45° ZnSe element was used for the 0.2 M solutions. Three hundred scans were collected per sample at a resolution of 4 cm-1. The spectra shown in this paper were obtained by subtracting a pure water spectrum, from the sample spectrum and the base line was corrected in the region from 1300 to 650 cm-1 using Nicolet’s OMNIC software. Curve fitting was performed by transforming these spectra to the spc format used by Galactic Industries (Salem, NH) GRAMS/386 software and by using its curve fit program. A Gaussian model is used for the component bands. The fit is considered satisfactory if χ2 is less than 2. Osmolality measurements were taken using a Wescor (Logan, UT) 5500 Vapor Phase osmometer. It was calibrated using NaCl solutions of known osmolality. The average molecular weight was calculated from the osmolality based on dissolved solids assay obtained from alkali and silica titrations using a Brinkmann (Westbury, NY) 670 titration system. Results NMR. Figure 1 is an example of 29 Si NMR spectra illustrating features previously reported in the literature.5-12,14-16 These features include broader peaks, relatively more larger anions at higher ratio, and relatively more small anions with decreasing concentration for the same ratio. Figure 2 represents the quantitative distribution of the several types of Si-O linkages, based on integrated band intensities of the peaks in Figure 1. The relative stability of the larger anions in the 3.3 ratio silicate, compared to 2.07 ratio, is significant in many applications of silicates involving dilution.

J. Phys. Chem. B, Vol. 101, No. 50, 1997 10639

Figure 1. 29Si NMR of sodium silicate solutions. (a) 3.3 ratio, 6M; (b) 3.3 ratio, 1M; (c) 2.07 ratio, 6M; (d) 2.07 ratio, 1M.

Figure 2. Relative percentages of Si-O linkage types for solutions in Figure 1.

Because of slight differences in chemical shifts depending on concentration and cation type,12 chemical shifts in this paper are reported relative to the location of the monomer peak in each 29 Si NMR spectrum. Table 1 gives the assignments of peaks in 2.07 ratio, 1 M solution. When silicate solutions are diluted to 0.4 M, large three-dimensional anions and polymer predominate at high ratios, smaller three-dimensional and planar cyclics at intermediate ratios, and small anions in the most alkaline silicates (Figure 3). The resolution of 29 Si NMR spectra of 0.4 M solutions is sufficient to make estimates of the relative amounts of individual anions, based on peak height. These estimates are reported in Table 1S (Supporting Information). The data will be used later to help interpret infrared spectra. Although LUDOX SM-30 is a silica sol, rather than a silicate solution, its SiO2:Na2O ratio of about 15 represents a very high ratio siliceous material. Its relatively limited distribution of Si-O linkage types will aid in interpreting infrared spectra of the higher ratio silicate solutions. FTIR. Infrared absorption bands of sodium silicates are observed between 1250 and 700 cm-1. Although absorption

10640 J. Phys. Chem. B, Vol. 101, No. 50, 1997

Bass and Turner

TABLE 1: Assignment of 29Si NMR Chemical Shifts to Specific Anions, 2.07 Ratio, 1M Silicate Solution Si-O linkage

Si atom configurationa

anion

observed chemical shift, ppmb

chemical shift, ppmc ref 12

Q0 Q1 Q2(3R) Q3(3R) Q1 Q2 Q1 Q2 Q1 Q2(3R) Q2(4R) Q3(4R) Q2(4R) Q3(4R) Q2(4R) Q2(4R) Q3(3R) Q3(4R) Q3(3R) Q3(5R) Q3(4R)

Q11 Q11 Q22(3R) Q31(3R) Q11 Q22(3R) Q31(3R) Q11 Q22(3R) Q31(3R) Q12 Q21 Q12 Q21 Q12 Q22 Q12 Q22 Q12 Q23(3R) Q23(4R) Q32(4R) Q23(4R) Q32(4R) Q22(4R) Q34(4R) Q22(4R) Q34(4R) Q24(4R) Q22(4R) Q33(3R) Q31(4R) Q22(4R) Q33(3R) Q31(4R) Q22(4R) Q33(3R) Q31(4R) Q36(3R) Q310(5R) Q38(4R)

monomer monosubstituted cyclic trimer monosubstituted cyclic trimer monosubstituted cyclic trimer linear trimer linear trimer linear tetramer linear tetramer dimer cyclic trimer bridged cyclic tetramer bridged cyclic tetramer doubly bridged cyclic tetramers doubly bridged cyclic tetramers cyclic tetramer tricyclic hexamer tricyclic hexamer tricyclic hexamer prismatic hexamer prismatic decamer cubic octamer

-7.86 -9.8 -18.04 -7.96 -16.79 -8.18 -16.14 -8.62 -10.12 -14.1 -21.81 -14.5 -21.21 -15.95 -16.01 -17.47 -24.17 -17.12 -24.79 -24.97

-7.92 -9.78 -18.09 -8.04 -16.92 -8.25 -16.17 -8.51 -10.13 -14.2 -21.94 -14.57 -21.44 -15.99 -16.12 -17.51 -24.74 -17.08 -27.15 -27.31

a The notation for the configuration is defined as follows: Qi j(kR) where i is linkage type, j is the number of silicon atoms of that type in the anion, and k is the number of silicon atoms in the ring structure. b The chemical shift is calculated by subtracting from the location of the monomer peak at -72.19 ppm. c The chemical shift is calculated by subtracting from the location of the monomer peak at -71.3 ppm.

Figure 3. Relative percentages of Si-O linkage types for 0.4 M solutions, 0.05 to 3.3 ratio.

bands for sodium silicate glasses have been observed below 700 cm-1,17 the ZnSe ATR element becomes totally absorbing near 600 cm-1. Also, a broad, intense water band in this region makes observation of bands in the solution spectra far more difficult to observe. As mentioned in the Experimental Section, we use spectral subtraction to remove the edge of the water band from about 900 to 650 cm-1. This procedure has been used for all infrared spectra reported in this paper. Silicate FTIR spectra typically show a relatively sharp peak in the region from 1120 to 900 cm-1 superimposed on a broad band covering the range mentioned previously. The highfrequency edge of the broad band shifts to lower frequency as the silica:alkali ratio decreases, at constant silica concentration, or as the concentration decreases at constant ratio. Figure 4 compares the spectra of 0.4M solutions ranging in silica:alkali ratio from 3.3 to 0.2. The location of maximum absorption moves to lower frequency. The shape of the band also changes with ratio. Similar results are observed for other concentrations. Figure 5 compares 3.3 ratio solutions where the concentration varies from 6 to 0.4 M. In this figure the spectra are shown in the “full scale” display mode for each spectrum in order to show the band features more clearly. As the solutions are diluted the location of maximum absorption shifts to higher frequency. The band shapes also change with dilution. Figure 6 compares the location of maximum absorption as 3.3 and 2.0 ratio are diluted. The shift is somewhat less for the higher ratio silicate. Figure 7 shows the result of using the GRAMS/386 software on the infrared spectrum of 0.4 M, 2.07 ratio solution. The

Figure 4. FTIR spectra of 0.4M solutions. (a) 3.3 ratio, (b) 1.07 ratio, (c) 0.2 ratio.

silicate absorption band is resolved into nine component bands at this concentration and ratio. Although the curve fit shows bands below 850 cm -1, the strong absorption of water makes determination of band location and area very uncertain, so the significance of these bands will not be discussed in this paper. The number of component bands, band areas, and peak maxima locations vary with ratio and concentration. Table 2S (Supporting Information) summarizes peak locations and band areas for different ratios of 0.4 M solutions. The infrared spectrum of LUDOX SM-30 shows a very sharp, intense band near 1100 cm-1, with a broad shoulder near 1200 cm-1 (see Figure 2S). Weak bands are also observed near 960

Sodium Silicate Solutions

J. Phys. Chem. B, Vol. 101, No. 50, 1997 10641

Figure 8. Variation of average molecular weight. Dotted line: change with SiO2:Na2O ratio, 1M solutions. Solid line: change for 3.3 ratio solutions with dilution.

Figure 5. FTIR spectra of 3.3 ratio solutions. (a) 6M, (b) 2M, (c) 0.4M, “full-scale” mode.

Octahydridosilasesquioxane, H8Si8O12, a three-dimensional crystalline silicate, shows strong bands at 1130 and 880 cm-1.19 Average Molecular Weights. The average molecular weights (AMW) of sodium silicate solutions are highest for the most concentrated siliceous solutions, consistent with the higher relative levels of large anions observed in 29Si NMR and FTIR spectra. At a given concentration, AMW declines rapidly as SiO2:Na2O ratio decreases. At high concentrations for the same ratio, (3 to 6M) the decrease in AMW is detectable but is more rapid as the concentration drops below 2M. Figure 8 shows both the ratio effect at 1M SiO2 concentration and the dilution effect for 3.3 ratio solution. Since the osmolality range of the Wescor vapor phase osmometer extends from 0.1 to 3.4 mol/kg water, it is not possible to determine AMW of the more alkaline silicate solutions at high concentrations or the more siliceous solutions at very dilute concentrations. Fortunately, the instrument is capable of measuring AMW for the most important commercial silicates whose SiO2:Na2O ratios range from 3.3 to 1.0. Discussion

Figure 6. Shift in silicate band maxima, 3.3 and 2.07 ratio, with dilution.

Figure 7. Curve fit of observed silicate band, 2.07 ratio, 1M, to component bands using GRAMS/386 software.

and 780 cm-1. The spectra of sols of different ratio and silica content look similar, with slight shifts in peak locations. Crystalline silicates with different arrangements of siliconoxygen linkages have infrared bands at different frequencies within the 1250 to 700 cm-1 region.18,19 In his review paper, Griffith reports strong bands for crystalline orthosilicates around 1000-950 cm-1 and near 870 cm-1. Silicates with sixmembered Si-O rings, which are equivalent to Q2(3R) anions, have strong bands near 1015 and 815 cm-1, while silicates with eight membered Si-O rings, which are equivalent to Q2(4R) anions, have strong bands near 1075, 1005, and 875 cm-1.

Effect of SiO2:Na2O Ratio. Since 0.4M solutions have fairly well-resolved 29Si NMR peaks and several infrared component bands, they will be used for assignments based on anion types. Both types of spectra become more complex as SiO2:Na2O ratio increases from 0.05. This reflects the condensation of monomer to more complex anions, first linear, then planar cyclic, then three-dimensional anions, and finally polymeric silica. The infrared spectrum of the most alkaline silicate, 0.05 ratio, is dominated by three component bands near 975, 920, and 850 cm-1. These are mostly due to vibrations of monomer but there may be a small contribution from dimer. The 975 cm-1 band shifts slightly to higher frequency and drops sharply in band area above 2.07 ratio. A similar trend is observed for the 920 cm-1 band. The 850 cm-1 band shifts to about 885 cm-1 above 0.1 ratio and remains constant in location. The shifts may be due to larger contributions from dimer and other linear anions at higher ratios, with intensity finally decreasing at the higher ratios when these smaller anions contribute relatively less to the total anion distribution. The 885 cm-1 band is assigned to SiO- vibrations of these smaller anions. A band near 1000 cm-1 is observed in 0.1 and 0.21 ratios and is assigned to dimer, now present in greater concentration. It shifts to about 1010 cm-1 in 0.52 ratio. It is most intense between 0.52 and 1.65 ratio, where Q2(3R) and linear anions predominate. In higher ratio silicates, it also appears that threedimensional anions contribute, based on an assignment of Groenen et al.20 They assigned a sharp band at 998 cm-1 in the infrared spectrum of tetramethylammonium (TMA) silicate to SiO- of the cubic octamer. The solutions discussed in this

10642 J. Phys. Chem. B, Vol. 101, No. 50, 1997 paper have more complex distributions than TMA silicate. This complexity is manifested by the shifts in peak location and somewhat broader bands. Therefore, it appears that the 1000 to 1020 cm-1 band is a complex composite of contributing anions, with linear anions the major contributors at low ratio and larger anions contributing more as ratio increases. As ratio increases to 0.21, another component band centered near 1055 cm-1 is resolved. At this ratio we see Q2(3R) anions in the NMR spectrum. As ratio increases, the intensity of this band increases as the relative contributions of these anions increase. Another band is resolved near 1090 cm-1 at higher ratios where three-dimensional anions are observed by NMR. This latter band becomes more intense and broader in the highest ratio silicates (2.48 and higher). NMR spectra indicate that three-dimensional anions are most abundant in these more siliceous solutions. These solutions also have a broad band near 1150 cm-1. The assignments of these high-frequency bands is aided by comparing the infrared spectrum of very siliceous sol (LUDOX SM-30) (Figure 2S (Supporting Information)) with its 29Si NMR spectrum (Figure 1S (Supporting Information)). The sharp band near 1100 cm-1 in the sol is assigned to fourring three-dimensional anions, consistent with results reported by Groenen et al.20 who found this band in the infrared spectrum of 1M TMA silicate solution. The 29Si NMR spectrum12 of TMA silicate solution shows the anion type in this solution is more than 90% cubic octamer, Q38. The shoulder in the silica sol infrared spectrum is assigned to polymer. These two assignments are consistent with infrared component bands observed in 3.3 to 1.65 ratio silicates. In these solutions we see a small broad band centered near 1150 cm-1 and a more intense broad band centered near 1090 cm-1. The higher frequency band decreases in area in the case of 2.07 and 1.65 ratio and NMR indicates that Q4 polymer is absent. These results indicates that the 1150 cm-1 band is a complex composite of polymer and Q3(4R) anions in the more siliceous solutions (2.48 ratio and higher) with the band area (integrated intensity) decreasing when only Q3(4R) anions are present. This observation is consistent with the results of Couty and Fernandez9 who reported the absence of colloidal silica below 2.2 ratio. The 1090 cm-1 band appears to be a complex composite of Q3(4R), Q3(3R), and Q2(4R) anions. The 1040 cm-1 band becomes resolved from the 1090 cm-1 band when the contribution of Q3(3R) anions increases to a much higher level than the contribution of larger three-dimensional anions. Silicate solutions of higher concentrations exhibit similar trends when comparing infrared and NMR spectra. More concentrated, more siliceous solutions have more intense highfrequency shoulders in the 1300 to 1050 cm-1 portion of the observed silicate band. This reflects the greater relative contributions of larger anions to the anion distribution. However, 29Si NMR spectra are not as well resolved, making assignments to specific anions much more difficult. This, in turn, makes it difficult to develop firm correlations between NMR and infrared spectra of the more concentrated solutions. Effect of Concentration. The 29Si NMR spectra of 2.07 ratio solutions, Figure 1b (6M) and 1d (1M) show sharp, relatively well-resolved peaks. We observe a major increase in Q0 (-72 ppm in the experimental spectrum) and a major decrease in Q3(3R) (about -25 ppm relative to the monomer peak). Although the peak heights decrease relative to monomer when 2.07 ratio is diluted from 6 to 1 M, the relative integrated intensities (see Figure 2) increase for Q1 and Q2(3R). There is a slight decrease in the region from -16 to -19 ppm (relative to monomer), assigned to anions with Q2(4R) and/or Q3(3R) linkages. Bicyclic anions are relatively stable (∼-21 ppm) over the dilution range.

Bass and Turner

Figure 9. Normalized component band areas for 2.07 ratio, 1-6M solutions.

The dilution sequence of 3.3 ratio solutions ranging from 6 to 1 M shows that the larger cyclic anions are relatively more stable with dilution, compared to 2.07 ratio. The peak assigned to bicyclic anions cannot be seen above the noise. Figure 1c also shows a broad band due to polymer, Q,4 centered near -36 ppm (relative to monomer). Using integrated peak areas, as shown in Figure 2, the larger anions contribute more to the anion distribution in 1 M 3.3 than in 1 M 2.07 ratio. In the case of intermediate ratios, we find that the distribution of anions as solutions are diluted is a mix of the tendencies described above, with the 2.98 ratio solution similar to 3.33 and the 2.48 more like 2.07. More alkaline solutions, when diluted, show even more rapid depolymerization of large anions than 2.07 ratio. Application of the curve fitting program to FTIR spectra of these two dilution sequences results in resolving the broad silicate band of each solution into five relatively intense bands, plus several less intense ones, depending on concentration and ratio. For 2.07 ratio, the most intense of these five are centered near 1000 (a relatively narrow band) and 880 cm-1 (broader), with less intense bands centered near 980 (narrow), 1040 (broad) and 1100 (broad) cm-1. The 1000 and 880 cm-1 bands change proportionally to molar concentration while the others do not. Figure 9 compares the changes in three of these bands, normalized to their intensities in 6 M solution. The normalized areas for the 1000 cm-1 band fit reasonably well to a straight line as the solutions are diluted, with an R2 of 0.97. The decreases in normalized band areas for the 1040 and 980 cm-1 bands are less pronounced at higher concentrations. In 3.3 ratio solutions the most intense of these five is centered near 1070 (a broad band), somewhat less intense bands near 880 (broad) and 1010 cm-1, with much less intense bands centered near 985 (narrow) and 1135 (broad) cm-1. The 3.3 ratio solutions also show nearly proportional band area decrease for the 1010 and 880 cm-1 bands with dilution, while the others exhibit scatter as was seen for 2.07 ratio. Figure 3S (Supporting Information) compares normalized band area changes for three of the component bands in 3.3 ratio solutions. R2 for the 1010 cm-1 band line fit is 0.98. The more rapid decrease in intensity for the 1000 and 1010 cm-1 bands and their shifts to higher frequency account for the shift of the observed band peak to higher frequency as the solutions are diluted. The locations of the component bands in the high-frequency portion of the silicate band (1300 to 1000 cm-1) for 2.98 and 2.07 ratio at 1.5 M concentration are similar to those reported by Couty and Fernandez,9 except that we also find a band at 1052 cm-1 in 2.07 ratio, when the convergence criterion of χ2 < 2 is met. Since the NMR peaks in 6 and 1 M 2.07 ratio solutions are not as well resolved as the 0.4 M solutions, it is more difficult to correlate them with specific anion types. For example, there appears to be unresolved overlap between linear trimer and tetramer in the Q1 region (about -8 ppm relative to monomer) and between linear tetramer and tricyclic hexamer in the Q2(4R),

Sodium Silicate Solutions Q3(3R) region (about -16.2 ppm relative to monomer). Nevertheless, the chemical shifts of many of these peaks relative to the chemical shift of monomer are close to those found in 0.4 M solutions. Keeping these limitations in mind, it is still possible to use the assignments made for 0.4 M solutions to interpret the trends observed in infrared spectra of solutions of the same ratio diluted to lower concentration. On an absolute scale, infrared band areas will decrease because there are fewer silicate anions to absorb infrared radiation in more dilute solutions. The relatively slow decreases in band area for the 1040 and 1090 cm-1 reflect a redistribution of the relative levels of large anions as 2.07 ratio solutions are diluted, with the most significant relative changes occurring between 2 and 1 M. The 980 cm-1 band decreases more slowly because there are relatively more monomer and dimer anions in dilute solutions. The 1000 cm-1 band is a complex composite of contributions from all anion types, as discussed in the section on effect of ratio. Therefore, it can be expected to decrease in rough proportion to concentration. A possible explanation for the shift to higher frequency for the 1000 cm-1 band as 2.07 ratio is diluted is the relative stability of the prismatic hexamer, a large anion, observed at about -17.1 ppm relative to monomer in the NMR spectra. Figure 4Sa,b (Supporting Information) illustrate the relative increase in dimer and stability of prismatic hexamer as 2.07 ratio solutions are diluted. The 880 cm-1 band is assigned to SiO-, which in turn will be proportional to cation concentration. Therefore this band will decrease in a linear relation with dilution. The NMR spectra of 3.3 ratio solutions are even less resolved. However, the relative increase in dimer with dilution can be seen in Figure 5Sa,b, (Supporting Information) as well as the relative stability of prismatic hexamer. The shifts in band maxima toward 1135 and 1070 cm-1 from 1090 and 1040 cm-1 in 2.07 ratio reflect the relatively larger levels of larger anions in 3.3 ratio solutions due to their more siliceous compositions. The 1135 cm-1 band includes contributions from polymer. For a given molar concentration, the area of the 1070 cm-1 band is greater for 3.3 ratio than the 2.07 1040 cm-1 band, consistent with the relatively greater amount of Q3(4R) anions. On the other hand, the area of the 985 cm-1 band is less in 3.3 ratio since there is less monomer and dimer in the more siliceous solutions. Relationship between Anion Distribution and Average Molecular Weight (AMW). Because 29Si NMR spectra are complex at higher SiO2:Na2O ratios and/or higher concentration, it is beyond the scope of this paper to establish, quantitatively, a relationship between observed anion distributions and average molecular weight for such solutions. Figure 8 demonstrates that more siliceous and/or more concentrated solutions have higher average molecular weights. This is consistent with these solutions having a relatively larger proportion of larger anions as determined by NMR and infrared spectroscopy. The AMW results obtained by osmometry, 61 for 1.03 ratio, and 190 for 3.3 ratio in 1M solutions are comparable to the results reported by Debye and Nauman,2 60 and 250 respectively for 1 and 3.3 ratio, although their experiments were performed at much lower concentration. Data reported earlier in this paper indicate that the anion distributions in more alkaline (1.03 SiO2:Na2O ratio or less) 0.4 M solutions are less complex. Table 3S summarizes an estimation of relative linkage distribution in 1.03, 0.4 M solution from integrated peak areas of its 29Si NMR spectrum. In both the Q1 and Q2, Q3(3R) regions, several different anion types contribute to the integrated peaks, so estimates of each anion’s contribution are made from peak heights. For example, dimer,

J. Phys. Chem. B, Vol. 101, No. 50, 1997 10643 linear trimer, and tetramer all contain Q1 silicon atoms. The dimer is the major contributor at -8.6 ppm with the peaks of the latter two anions unresolved, contributing to a weak shoulder at -8.2 ppm. In the Q2, Q3(3R) region (-16 to -19 ppm), the Q2 contribution of the tetramer assigned as -16.1 ppm is well separated from the Q2 contribution of the trimer assigned as -16.8 ppm. Some anions whose peaks are observed but are weak are ignored in the estimate presented in Table 4S. The integrated peak intensities represent the contribution of each Si-O linkage type but individual anions may have more than one silicon atom of a given type. For example, the cyclic trimer has 3 atoms with Q2 linkages. This linkage type contributes 12% of the total integrated peak area, compared to 37% for monomer. However, since there are three Q2 atoms in the trimer, the relative amounts of anion are 0.04 for the cyclic trimer and 0.37 for the monomer. After the amount of each anion is determined relative to the monomer contribution of 0.37, the results are normalized to a total value of 1 for the anions. The normalized value for each anion is multiplied by its molecular weight to obtain its contribution to the AMW calculation. Another simplifying assumption made in the AMW calculation is that each silicon is balanced by the number of sodium ions in the ratio, e.g., in 1.03 ratio there are approximately two sodium atoms per silicon. Some anions which have relatively low values of connectivity would therefore be protonated, e.g., monomer may be represented as SiO4H2-2. Others, with high connectivity, such as cubic octamer, only have one nonbridging oxygen per silicon atom so there will be excess sodium ions which are balanced by hydroxyls. This assumption is supported by the higher pH of even more alkaline solutions such as 0.52 ratio, which has even more anions that cannot be balanced by 4 sodium atoms per silicon. The average molecular weight for 1.03 ratio using these calculations is 57, compared to 50.5 measured by vapor phase osmometry. For 0.52 ratio, the AMW is calculated to be 41, compared to 35. This agreement supports the assumptions made above. Conclusions The results reported in this paper demonstrate that differences in anion distributions in sodium silicate solutions depending on concentration and/or alkalinity of the solution can be characterized by 29Si NMR or infrared spectroscopy. 29Si NMR spectroscopy is important because it provides quantitative characterization of different Si-O linkage types. In more alkaline solutions it may be used to identify specific anion types. Table 1 summarizes these assignments. We have been able to estimate average molecular weights in dilute, very alkaline solutions with good agreement with vapor phase osmometry results. Although infrared spectroscopic data are a complex composite of absorption bands, contributed by several anion types, trends observed with changes in concentration and/or alkalinity closely follow the trends observed by 29Si NMR. Assignments of infrared component bands for different anion types, less specific than NMR assignments, are summarized in Table 2. The band locations are consistent with results reported for crystalline silicates and silicate glasses.17-20 Infrared spectroscopy shows great promise for observing changes in anion distribution on a much shorter time scale than is possible with 29 Si NMR. This can be useful for following reactions involving silicate solutions which have many important commercial applications. Our results also show that the extent of depolymerization of larger anions differs, depending on both the degree of dilution and alkalinity. The relative amounts of polymeric (Q4) silica

10644 J. Phys. Chem. B, Vol. 101, No. 50, 1997 TABLE 2: Assignments of Infrared Component Band Regions to Anion Types anion type polymer Q3(4R) rings Q3(3R), Q2(4R) rings, linear Q2 Q2(3R) rings SiO- cyclic anions linear Q1 (dimer, trimer, tetramer) + monomer monomer + dimer SiO- small anions

component band region (cm-1) 1300-1100 1120-1050 1070-1030 1050-1020 1020-1010 1005-995, 985-965 950-910 900-850

and larger three-dimensional (Q3(4R)) anions are greater for the more siliceous solutions (3.3 and 2.95 ratio) at high concentrations. These large anions continue to be major contributors to the anion distribution, relative to the smaller monomeric, linear chain and planar cyclic anions, in these solutions, even at 15to 50-fold dilution. In the case of more alkaline solutions, especially below 2.5 ratio, the relative contribution of the large anions to the distribution drops rapidly, even with 2- or 3-fold dilution. At low concentrations, in very alkaline solutions, we find that the anion distribution is mainly monomer, dimer and cyclic trimer. Acknowledgment. J. Bass thanks Professor James Falcone of West Chester University, West Chester. PA, Dr. Neil Miller of the PQ Corporation and Mrs. Tomasina Leone of the PECO Energy Corporation for stimulating discussions about the nature of sodium silicate solutions. Supporting Information Available: Two figures of supporting infrared and three figures of 29Si NMR spectroscopic

Bass and Turner data, three tables supporting NMR peak assignments to anion types, and one table detailing infrared component band data (9 pages). Ordering information is given on any current masthead. References and Notes (1) Vail, J. G.; Soluble Silicates. Their Properties and Uses; Reinhold Publishing: New York, 1952; Vol. 1. (2) Debye, P.; Nauman, R. J. Chem Phys. 1948, 17, 664. (3) Lentz, C. W. Inorg. Chem. 1964, 3, 574. (4) Dent-Glasser, L. S.; Lachowski, E. E.; Cameron, G. G. J. Appl. Chem. Biotechnol. 1977, 27, 39. (5) Ray, N. H.; Plaisted, R. J. J. Chem. Soc., Dalton Trans. 1983, 3, 475. (6) Engelhardt, G.; Zeigan, D.; Jancke, H.; Hoebbel, D.; Wieker, W. Z. Anorg. Allg. Chem. 1975, 418, 17. (7) Svensson, I. L.; Sjo¨berg, S.; L-O. O ¨ hman, J. Chem. Soc., Faraday Trans. 1, 1986, 82, 3635. (8) Beard, W. C. AdV. Chem. Ser. 1973, 112, 162. (Proceedings of the 3rd International Conference on Molecular Sieves. (9) Couty, R.; Fernandez, L. C. R. Acad. Sci., Ser. IIa 1995, 320, 823. (10) Couty, R.; Fernandez, L. C. R. Acad. Sci., Ser. IIa 1996, 322, 821. (11) Roggendorf, H.; Grond, W.; Hurbanic, M. Glass Sci. Technol. 1996, 69, 216. (12) Engelhardt, G.; Michel, D. High-Resolution Solid-State NMR of Silicates and Zeolites. John Wiley: New York, 1987; Chapter 3. (13) Iler, R. K.; The Chemistry of Silica; John Wiley: New York, 1979. (14) Hoebbel, D.; Ebert, R.; Pauli, J.; Kruschke, D. Z. Anorg. Allg. Chem. 1992, 614, 95. (15) Harris, R. K.; C. T. Knight, G. J. Chem. Soc., Faraday Trans. 2 1983, 79, 1525. (16) McCormick, A. V.; Bell, A. T.; Radke, C. J. Zeolites 1987, 7, 183. (17) Kamitsos, E. I.; Kapoutsis, J. A.; Jain, H.; Hsieh, C. H. J. NonCryst. Solids 1994, 171, 31. (18) Griffith, W. P. J. Chem. Soc. A 1969, 1969, 1372. (19) Bartsch, M.; Bornhauser, P.; Calzaferri, G.; Imhof, R. Vib. Spectrosc. 1995, 8, 305. (20) E. Groenen, J. J.; Emeis, C. A.; van den Berg, J. P.; de-JongVersloot, P. C. Zeolites 1987, 7, 474.