Effects of Biological Molecules on Calcium Mineral Formation

May 23, 2013 - Considering the data of mineral volume fraction obtained here as well as the solubility product of possible mineral polymorphs in the S...
0 downloads 11 Views 1MB Size
Download PDF
Article pubs.acs.org/Langmuir

Effects of Biological Molecules on Calcium Mineral Formation Associated with Wastewater Desalination as Assessed using SmallAngle Neutron Scattering Vitaliy Pipich,† Yara Dahdal,‡,§ Hanna Rapaport,∥ Roni Kasher,‡ Yoram Oren,‡ and Dietmar Schwahn*,§,⊥ †

Jülich Centre for Neutron Science JCNS-FRM II, Outstation at FRM II, D-85747 Garching, Lichtenbergstrasse 1, Germany Zuckerberg Institute for Water Research, Jacob Blaustein Institutes for Desert Research, Ben-Gurion University of the Negev, Sede Boqer Campus 84990, Israel § Forschungszentrum Jülich GmbH, Jülich Centre for Neutron Science JCNS and Institute for Complex Systems ICS, D-52425 Jülich, Germany ∥ Department of Biotechnology Engineering, Ben-Gurion University of the Negev, Beer-Sheva 84105, Israel ‡

S Supporting Information *

ABSTRACT: Calcium phosphate scale formation on reverse osmosis (RO) membranes is one of the main limitations on cost-effective desalination of domestic wastewater worldwide. It has been shown that organic agents affect mineralization. In this study, we explored mineralization in the presence of two biofilm-relevant organic compounds, the proteins bovine serum albumin (BSA) and lysozyme, in a simulated secondary effluent (SSE) solution using small-angle neutron scattering (SANS), and applied the results to analyses of mineral precipitation in RO desalination of secondary effluents of wastewater. The two proteins are prominent members of bacterial extracellular polymeric substances (EPSs), forming biofilms that are frequently associated with RO-membrane fouling during wastewater desalination. Laboratory experiments showed that both proteins in SSE solution are involved in complex mineralization processes. Only small portions of both protein fractions are involved in mineralization processes, whereas most of the protein fractions remain as monomers in solution. Contrast variation showed that composite particles of mineral and protein are formed instantaneously to a radius of gyration of about 300 Å, coexisting with particles of about μm size. After about one day, these large particles start to grow again at the expense of the 300 Å particles. The volume fraction of the 300 Å particles is of the order of 2 × 10−4, which is too large to represent calcium phosphate such as hydroxyapatite as the only mineral present. Considering the data of mineral volume fraction obtained here as well as the solubility product of possible mineral polymorphs in the SSE solution, we suggest the formation of protein-mineral particles of hydroxyapatite and calcium carbonate during scale formation.

■

INTRODUCTION A well-accepted treatment for domestic wastewater today is coarse filtration/sedimentation, followed by biological treatment using activated sludge and separation of the biomass from the treated effluents by ultrafiltration, which is termed secondary treatment. Treatment of secondary wastewater effluents by reverse osmosis (RO) or nanofiltration (NF) technologies is currently considered an important potential supplementary source of unlimited sustainable irrigation and of drinking water worldwide.1 One of the obstacles in the desalination of wastewater effluent by RO or NF is scale formation on the membrane surface. The precipitation of sparingly soluble salts, such as CaCO3 and CaSO4, is usually controlled by lowering the pH or using antiscalants; however, the effectiveness of available antiscalants for calcium phosphate is still questionable.2 Phosphate concentrations are high in the secondary treated © XXXX American Chemical Society

municipal wastewater effluents that are being used as RO feeds; for example, secondary treated effluents at the Shafdan Plant, Israel (Israel’s largest wastewater-reclamation plant), contain 85 mg L−1 (2.12 mM) calcium and 4.6 mg L−1 (0.048 mM) phosphate,3 which cause scaling followed by a substantial decline in the permeate flux.4,5 Today, a common practice for removing calcium phosphate salts is to add sodium aluminate at concentrations of 20 − 30 mg L−1 to the activated sludge phase, followed by microfiltration, as a pretreatment for RO. Sodium aluminate is an efficient coagulant for phosphates and turbidity reduction.6,7 Ning et al. showed that during ultrafiltration (UF) pretreatment of wastewater for desalination, calcium phosphate scaling Received: January 15, 2013 Revised: May 20, 2013

A

dx.doi.org/10.1021/la4001889 | Langmuir XXXX, XXX, XXX−XXX

Langmuir

Article

of millimeter thickness. Thus, sample preparation does not disturb the results, in situ experiments can be performed on identical samples over longer periods of time, and information such as the protein and mineral content is obtained from the aggregate components. The SANS technique has been recently introduced into environmental technology investigations, although it is widely applied in more basic studies of proteins in aqueous solutions, as well as in biomineralization research.16,17 The technique is useful not only for determining the size, shape, and structure of the hydration layer, but also the correlations between proteins and thereby the interaction potentials of the proteins in aqueous solutions. Thus, SANS technology is relevant for studies of protein crystallography as well as for investigating properties of diseases such as those of Parkinson or Creutzfeldt Jakob.18 In this study, we used diluted solutions of BSA and lysozyme in salt-free water and in SSE solution, which was about five times more concentrated than the bulk solution of secondary treated effluents in the Shafdan plant (for more details see Supporting Information Table S1 and ref 14). We changed the scattering contrast by varying the D2O/H2O ratio in the water, as well as in SSE solution. Mineralization in the SSE solution was examined over an incubation time of several minutes up to 20 h, and was compared with the results from the salt-free water. Concentrations of BSA and lysozyme were chosen in the range of 1−10 mg mL−1, and the SANS experiments were performed using a conventional and ultrasmall angle instrument, thereby covering a large angular analysis range to analyze particles from a few angstroms to micrometers.

originates from small calcium phosphate nanoparticles that pass through the UF membranes and accumulate on the RO membranes.6 Analysis of the dry foulants showed that it consisted of 20% by weight of organic matter and 80% of inorganic salts which was primarily calcium phosphate. Calcium phosphate minerals are normally found in several forms, such as amorphous calcium phosphate, Ca3(PO4)2·3H2O (ACP); tricalcium phosphate, Ca3(PO4)2 (TCP); tetracalcuim phosphate, Ca 4 (PO 4 ) 2 O (TTCP); octacalcium phosphate, Ca8(HPO4)2(PO4)4 (OCP); and hydroxyapatite, Ca10(PO4)6(OH)2 (HAP). Thermodynamically, HAP is the most stable calcium phosphate polymorph, and it can be considered as the most likely end product in many reactions; ACP plays a special role as a transition phase in the conversion of calcium phosphate to crystalline phases (OCP and HAP) over time, as predicted by Oswalds’ law of stages.8,9 Fouling and scaling are complex processes that can be categorized into four types: (i) biofouling, which occurs when microorganisms attach to the membrane and excrete extracellular polymeric substances (EPS), thus forming biofilms; (ii) scaling, which involves mineralization of sparingly soluble salts; (iii) organic fouling, which involves natural organic compounds, such as humic acid; and (iv) particulate or colloidal fouling.10−12 It has been shown that interdependencies may exist between scaling, organic fouling, and biofouling13 and also between fouling substances and the surface chemistry of the RO membrane.5 Y. Cohen and co-workers coated RO membranes with acrylate polymers and evaluated the effects of membrane surface chemistry on calcium sulfate dihydrate (gypsum) scaling.4 Calcium sulfate is often encountered in brackish water desalination processes. The effects of scaling were evaluated both by permeate flux changes and by real-time optical monitoring of the evolution of surface precipitates. In addition, they determined the influence of the foulants alginicacid and BSA on flux performance of the membranes in the presence of calcium sulfate.4 Previous study in our group investigated the effects of surface-exposed organic functional groups on calcium-phosphate mineral formation during RO wastewater desalination using surface pressure−area (Langmuir) isotherms.14 It was found that calcium-phosphate mineralization is accelerated by the surface functional groups in the order: PO4 > COOH ∼ NH2 > COOH/NH2 (1:1) > OH > ethylene glycol. Bacterial biofilms of EPSs consist of polysaccharides, proteins, and lipids, as well as RNA and DNA fractions in which microorganisms are embedded;15 the biomolecules and biopolymers are therefore widespread on both the membrane surface and within the feed effluent. In this study, we evaluated the effects of biofouling-borne biopolymers on calcium phosphate mineralization in solutions that simulate secondary effluents of wastewater. Scaling of calcium phosphate minerals during desalination of secondary and tertiary effluents is considered as a limiting factor in the performance of membranes.2 We selected two biofilm-relevant organic compounds, namely, the proteins bovine serum albumin (BSA) and lysozyme, which we studied in both salt-free water and in an ionic solution (a simulated secondary effluent, SSE). Small-angle neutron scattering (SANS) was used for the investigation, as it has the capacity to measure both the size and shape of nanometer- to micrometer-sized particles in solution, as well as rendering information on the nature (mineral or organic) of such particles. The SANS technique utilizes a nondestructive methodology that irradiates aqueous solutions

■

MATERIALS AND METHODS

We were interested in the mechanisms of mineralization induced by organic molecules to improve our basic understanding of fouling and scaling of the membranes in wastewater desalination using RO technology. A short account of the SANS technique, as well as a basic description of the equations of scattering theory and the relevant sample parameters, follows. SANS Experiments. The neutron experiments were performed using two SANS instruments, both in operation at the FRM II research laboratory in Garching, Germany.19 Some of the experiments were conducted using the classical KWS 1 SANS instrument, whereas others were conducted using the KWS 3 ultrasmall-angle scattering (USANS) instrument. The USANS instrument uses an elliptical mirror as an optical device, which focuses the primary beam onto a detector whose shape is determined by the few-mm2 area of the primary aperture at the entrance of the instrument. This instrument allows a smallest momentum transfer Q of the order of 10−4 Å−1, which is an order of magnitude less than that of classical pinhole instruments.20 The Q value is determined according to Q = 4π/λ sin(δ/2) from the scattering angle δ and the wavelength λ of the neutrons. The inverse value of Q gives the sizes of the objects, which mainly contribute to the range of scattering. Pin-hole instruments such as the KWS 1 are designed for a range of Q between 10−3 and 0.5 Å−1, thereby allowing full analysis of particles between 10 and 103 Å. Nowadays, SANS instruments with optical elements, such as refractive lenses21 and mirrors,20 allow Q values down to 10−4 Å−1 and thereby a full analysis of μm-sized particles. Instruments, such as the KWS 3, are particularly useful for our investigation. The scattering intensity is determined as the macroscopic crosssection dΣ/dΩ(Q), that is, the scattering intensity in absolute units of cm−1, which describes the scattering per unit sample volume. In its general form dΣ (Q ) = [ΦPVPΔρ2 ]F(Q )S(Q ) dΩ B

(1)

dx.doi.org/10.1021/la4001889 | Langmuir XXXX, XXX, XXX−XXX

Langmuir

Article

describes diffraction from scattering centers of the same type.22,23 The form factor F(Q) is caused by diffraction due to the particles itself, whereas the structure factor S(Q) is determined by spatial correlations between the particles (particles are denoted by subscripted P in the equations). In the case of dilute solutions, when the correlation between the particles becomes negligible (i.e., S(Q) ≅ 1), the scattering at Q = 0 from particles such as proteins in aqueous solution is determined according to dΣ (0) = ΦPVP[ρP − ρS ]2 dΩ

65%, 72%, and 77%, respectively (Figure 1, Table 2). The amorphous mineral polymorphs ACP and ACC, respectively match water at 51% and 64% D2O content. These numbers show that the D2O contents of matching of the proteins, amorphous and crystalline minerals are different and can therefore be distinguished. The theoretical basis for the SANS technique, and the tables characterizing the samples are summarized in Supporting Information. Sample Parameters. The two proteins used in this study, bovine serum albumin (BSA) and lysozyme, were purchased from SigmaAldrich (product nos A2153 and L7651, respectively) and were used as they were received. Values of parameters such as the molar weight, partial specific volume, and isoelectric point are from the literature (see Table 1). The molar volumes of both proteins were evaluated from their molar weights and partial specific volumes.

(2)

by the volume fraction ΦP and volume VP of the particles and by the scattering contrast, that is, the square of Δρ = ρP − ρS, determined as the difference between the coherent scattering length density ρ of the particle (P) and the solvent (S). The scattering length densities of water, the proteins BSA, and lysozyme, as well as those of the amorphous and crystalline minerals ACP, α-TCP, HAP, ACC, and calcite are plotted in Figure 1.

Table 1. Parameters of the Proteins BSA and Lysozyme

a

protein

M [kDa]

da [g cm−3]

volume [103 cm3 mol−1]

isoelectric point

BSA lysozyme25

66.43 14.4

1.36 1.40

48.76 10.25

pH = 4.7 pH = 11.5

Mass density or inverse partial specific volume ν ≡ 1/d

26

The main focus of this study was the exploration of BSA- and lysozyme-induced mineralization in a simulated secondary effluent (SSE) from the Shafdan wastewater reclamation plant. The concentration of phosphate in the SSE solution was 2.5 greater than that in the Shafdan plant effluent, whereas the other salts were about 5.2 times more concentrated than those in the plant effluent. These values introduce realistic estimates of the salt concentrations in RO desalination systems at a stage of 80% recovery, hence, a 5 times higher salt concentration. The extent of concentration polarization depends on the engineering parameters of the system, such as permeate flux and feed solution shear rate.14,27 The pH of the SSE solution was 7. Table 2 compiles the mineral solubility product KSP and saturation index SI, as well as the scattering length density ρ, the matching aqueous solution ΦD2O, and the largest possible volume fraction ΦMineral of possible calcium minerals in the SSE solution. The latter values were calculated from the concentrations of corresponding constituents. In particular the real concentrations of the phosphate species H2PO4−, HPO42−, and PO43− in pH 7 were: 7.80 × 10−5, 4.84 × 10−5, and 1.03 × 10−10 M, respectively. Some more parameters such as mass density are presented in Table S3 of Supporting Information, as well as the corresponding equations (Supporting Information eqs S12−S14) for determining the saturation index, SI, of the ions. Steiner et al.14 developed the SSE solution and checked the saturation index of several minerals. They found in consistence with our values in Table 2 that the SSE solution is oversaturated with respect to hydroxyapatite and close to saturation with respect to calcium carbonate. On the other hand, the tendency of sulfate minerals (such as gypsum CaSO4·2H2O and anhydride CaSO4) to precipitate is negative and lower than that of the phosphate and carbonate minerals.

Figure 1. Coherent scattering length densities ρ of water, the proteins BSA and lysozyme, the crystalline minerals tricalcium phosphate (αTCP), hydroxyapatite (HAP), and calcite, as well as of the amorphous polymorphs ACP and ACC versus D2O content. These numbers were calculated on the basis of their chemical structure and molar volume compiled in Table 2 and Supporting Information Table S2. The ρ values for the proteins were determined experimentally (Supporting Information Table S3). An important aspect of the neutron scattering technique involves the ability to change the scattering contrast without changing the chemistry of the system. This capability was used to identify particles with respect to their chemical composition. The most prominent example represents the ability to detect the exchange of hydrogen (H) and deuterium (D) in molecules, as the two hydrogen isotopes show very different coherent scattering lengths.24 The exchange phenomenon is explained by the interactions of the neutrons with the atomic nuclei of the elements. The dark blue line in Figure 1 depicts the ρ of water versus its D2O content; ρ values are slightly negative for pure H2O and increase linearly with increasing D2O content. The characteristic ρ values of BSA and lysozyme are represented by red and green symbols (and lines), respectively.16 The positive slope of the ρ-D2O relationship is caused by H−D exchange on the outer surfaces of proteins. From the blue, red, and green lines, it can be seen that at ΦM = 42% and 45% D2O content, the ρ of water and the ρ of proteins are the same, and therefore the scattering contrast determined by the difference [ρP − ρS] in eq 2 is zero. This indicates that under these conditions the proteins are not visible; that is, they are matched for neutrons in the water. On the other hand, the minerals calcium phosphate α-TCP and HAP as well as calcium carbonate have scattering length densities, which match a water D2O content of about

■

RESULTS OF BSA AND LYSOZYME IN SALT-FREE WATER The SANS data of the BSA and lysozyme proteins dissolved in salt-free water, and their analysis are represented in Supporting Information. Here we give only a summary of the relevant parameters, such as radius of gyration, molar volumes and scattering length densities. Figure 2 shows scattering patterns from lysozyme, dissolved in D2O at concentrations of 1.0, 2.5, 5.0, and 7.5 mg mL−1. All solutions show an interference peak as was also found for BSA (Supporting Information Figure S1) and which indicates ordering of the monomers with respect to a preferred averaged distance. The distances become smaller at higher protein concentrations as determined by N−1/3 where N C

dx.doi.org/10.1021/la4001889 | Langmuir XXXX, XXX, XXX−XXX

Langmuir

Article

Table 2. Parameters of Relevant Minerals in SSE Solutiona mineral ACP

KSP 2.0 × 10−26 M5 −26

SI

ρ [1010 cm−2]

ΦD2O (match)

ΦMineral [10−4]

−3.21

2.00 + 1.90 × ΦD2O

0.51

0.125

−3.41 6.68

3.93 4.19 + 0.391 × ΦD2O

0.646 0.724

0.068 0.067

α-TCP HAP

3.16 × 10 M 4.7 × 10−59 M9

dicalcium phosphate anhydrate dicalcium phosphate dihydrate (brushite)

1.26 × 10−7 M2 2.4 × 10−7 M2

−0.18 −0.46

3.86 2.11 + 3.39 × ΦD2O

0.636 0.75

0.058 0.093

monocalcium phosphate anhydrate ACC

7.24 × 10−2 M3 3.1 × 10−8 M2

−10.7 −0.18

2.65 2.24 + 1.72 × ΦD2O

0.462 0.535

0.067 7.25

vaterite aragonite calcite anhydrite gypsum

1.24 × 10−8 M2 5.0 × 10−9 M2 3.3 × 10−9 M2 4.37 × 10−5 M2 2.45 × 10−5 M2

0.21 0.61 0.79 −0.73 −0.48

4.40 5.11 4.76 4.06 2.22 + 3.37 × ΦD2O

0.714 0.816 0.765 0.664 0.774

3.91 3.37 3.66 2.13 3.48

5

ACP [Ca3(PO4)2·3H2O], α-TCP [Ca3(PO4)2], HAP [Ca5(PO4)3(OH)]; dicalcium phosphate dihydrate (brushite) [Ca(HPO4)·2(H2O)]; dicalcium phosphate anhydrate [Ca(HPO4)]; monocalcium phosphate anhydrite [Ca(H2PO4)2]; ACC [CaCO3·H2O];28 calcite (CaCO3); calcium sulfate (CaSO4). KSP (solubility product) see: p. 46 as well as Tables 10.1, 10.2, and 16.4 in ref 29. IAP (ion activity product). SI (saturation index): = log(IAP/KSP). Coherent scattering length density ρ, matching D2O content ΦD2O in water and largest possible volume fraction ΦMineral of minerals in SSE solution. a

Figure 2. SANS scattering patterns of lysozyme in D2O show interference peaks, which are caused by the ordering of the lysozyme monomers, that is, a preferred protein−protein distance because of long-range electrostatic interactions. Similar ordering is observed for BSA (Supporting Information Figure S2). The dashed lines represent the form factor obtained from a fit using the structure factor of Supporting Information eq S4.

Figure 3. Form factor at Q = 0 versus lysozyme and BSA volume fraction in H2O and D2O. The slope of the straight line delivers the product of VBSAΔρ2.

gyration Rg, were determined from fitting the experimental data with the form factor of BSA and lysozyme which are, respectively, consistent with values from literature, such as Rg = 30.530 and 12.4 Å.31

is the particle density. The data are fitted according to Supporting Information eq S1 in combination with Supporting Information eqs S5 and S10, represented by solid lines, and showing good agreement with the experimental data. The dashed lines represent the form factor expressed in Supporting Information eq S10; the whole set of parameters of this fit are compiled in Supporting Information Table S6. The important parameter to determine is the form factor of the protein, from which we can evaluate the radius of gyration, the molar volume, and the coherent scattering length density. To further analyze the scattering data, we plotted the form factor dΣ/dΩ(0) ∝ F(0) at Q = 0 of both proteins in Figure 3 versus protein concentration. According to eq 2, the data must conform to a straight line passing through zero concentration, as they do. The slope of this straight line delivers the product of the protein molar volume VProt and the square of the scattering contrast Δρ2 = [ρProt − ρwater]2. The different slopes of BSA in H2O and D2O solutions are caused by different values of ρwater in D2O and H2O (Figure 1). Other parameters, such as the radius of

Table 3. Parameters from the Form Factor of BSA and Lysozyme in Salt Free H2O and D2O BSA

lysozyme

solution

H2O

D2O

D2O

⟨Rg⟩ [Å] dΣ/dΩ(0) × ΦBSA−1 [cm−1] VBSA [103 cm3 mol−1]

30 ± 0.6 38.5 ± 0.8 65.6 ± 2

28 ± 0.3 67.6 ± 1.5 64.1 ± 2

12.7 ± 0.2 14.3 ± 0.2 10.2 ± 0.14

■

RESULTS OF SSE SOLUTION WITH AND WITHOUT BSA AND LYSOZYME This section contains the main results of the SANS experiments. After discussing the precipitation behavior of the SSE solution, we present results obtained for mixtures of the proteins BSA and lysozyme in SSE solution. D

dx.doi.org/10.1021/la4001889 | Langmuir XXXX, XXX, XXX−XXX

Langmuir

Article

SSE Solution. The SSE solution shows weak scattering from particles, as depicted in Figure 4a for the SSE−H2O and SSE−

Information Table S3) is only slightly larger than the SANS value. BSA in SSE Solution. Mixing BSA in SSE solution has an immediate strong effect on the formation of large particles, as is observed in the scattering pattern obtained immediately after mixing BSA in SSE−D2O over a Q range between 2 × 10−4 and 0.2 Å−1 (Figure 5a). Indeed, the USANS data obtained from

Figure 4. (a) Scattering from SSE−H2O and SSE−D2O solutions, showing scattering from particles of about 70 Å radius of gyration. The measurements were performed about 4 days after solution preparation. (b) An average scattering length density of ⟨ρ⟩ = 3.47 × 1010 cm−2 is determined for the particles The scattering of the 50% and 70% SSE− D2O solutions was too weak to give reliable data.

D2O solutions. Analyses of the data (solid lines) indicate a particle radius of gyration of about 70 Å (averaged over a sample volume of 0.1 mL), which corresponds to a radius of 90 Å in the case of spherically shape particles. Figure 4b shows the square root of the extrapolated intensity at Q = 0 measured at different contrasts, yielding ΦM = 0.59 ± 0.03. The square root of the ratio of the integrated intensity Q2, defined in Supporting Information eq S11, gives within error bars a consistent value of ΦM = 0.62. From ΦM, the coherent scattering length density of the particles, ρP = (3.54 ± 0.21) × 1010 cm−2, is derived. All parameters are compiled in Table 4.

Figure 5. (a) Scattering pattern of BSA in SSE solution immediately after mixing. Strong scattering because of the formation of micrometer-sized particles is observed at small Q values. At large Q values, scattering from BSA monomers is visible. (b) Scattering length density of BSA in water and SSE solution. There is no visible change of the BSA monomer scattering length density in SSE solution. (c) The square root of the amplitude P3.4 as a function of SSE−D2O, determined in the smaller Q regime of (a). The matching condition is fulfilled at ΦM = 0.62 ± 0.09 SSE−D2O content. This matching condition occurs at a higher D2O content than at that for BSA but below that of the mineral (Figure 1) and thereby represents composite particles consisting of mineral and protein.

Table 4. Parameters from Particles in SSE Solutiona dΣ/dΩ(0) [10−2 cm−1] Rg [Å] Q2 [10−7cm−1 Å−3] d Σ/d Ω(0)/Q2 ⇒ RP [Å] ΦM {I0} ({Q2}) ρP [1010 cm−2] Δρ [1010 cm−2] ΦP [10−6]

SSE−H2O

SSE−D2O

4.28 ± 0.1 70 ± 6 2.91 89 0.59 ± 0.03 (0.62) 3.54 ± 0.21 4.10 8.2

1.88 ± 0.04 1.11 93

2.85 7.7

BSA concentrations of 10 and 5 mg mL−1 show that the amount of precipitate only weakly depends on the amount of dissolved BSA. We observed strong scattering at small Q values due to the formation of Rg = 0.68 ± 0.01 μm large particles. These particles form quickly (within a few minutes) and then remain stable for several hours, as will be shown later. At large Q, the scattering from BSA monomers dominates; the fit gives a radius of gyration of 28 Å. The main effect of the salts in SSE solution on the monomers is the disappearance of the interference peak caused by the ions’ screening of the electrostatic long-range interactions, thereby leading to a loss of preferred distances of the BSA monomers. The parameters determined from these measurements are compiled in Supporting Information Table S8. The loss of protein order seems to be the only effect on the BSA monomers, as the scattering length density in Figure 5b is not visibly influenced by the salts. The scattering from the large aggregates at small Q follows a power law with an exponent of 3.38 ± 0.07. These data were derived from measurements using the KWS 1 instrument,

The parameters are as follows: Extrapolated scattering at Q = 0, d Σ/ d Ω(0) (eq 3); radius of gyration, Rg (Supporting Information eq S8); second moment, Q2 = ∫ d QQ2 d Σ/d Ω(Q) (Supporting Information eq S10); radius RP of spherically shaped particles; volume fraction of SSE−D2O, ΦM matching the particles’ scattering patterns; scattering length density, ρ; and the particle volume fraction, ΦP. a

The dimension of the particles is consistently determined from the ratio dΣ/dΩ(0)/Q2 (Supporting Information eqs S2 and S11) and, using Q2, we obtained a particle volume fraction of (8 ± 0.3) × 10−5; this value is slightly greater than the maximum possible volume fraction of HAP, given as 6.7 × 10−6 in Table 2 (Supporting Information Table S3); however, its scattering length density is too small for HAP but only slightly larger than the evaluated ACP and ACC values of, respectively, 3.12 and 3.27 × 1010 cm−2 at ΦM = 0.59. This means that an about 10% smaller mass density of ACP and ACC would exactly describe these particles as formed from ACP and ACC (Supporting Information Table S3). The maximum volume fraction of ACP with 12.5 × 10−6 in Table 2 (Supporting E

dx.doi.org/10.1021/la4001889 | Langmuir XXXX, XXX, XXX−XXX

Langmuir

Article

performed at a 20-m sample−detector position using neutrons with a wavelength of 19.7 Å. An exponent between 3 and 4 can be explained either by fractal or by large-size particle distributions. Particles with a large size distribution and of the sizes discussed here would have the effect in the Q range under consideration, such that the smaller particles are still in the Guinier regime whereas the larger particles are already in the Porod Q−4 regime, as expressed in Supporting Information eq S9. The larger particles in the Guinier regime become visible only in the Q range of USANS, as demonstrated in Figure 5a. In Figure 5c, the square root of the amplitude P3.4 is plotted as a function of the SSE−D2O content. The data follow a straight line and show zero scattering at ΦM = 0.62 ± 0.09 SSE−D2O content. This matching condition occurs at higher SSE−D2O contents than that observed for the BSA monomers and the amorphous minerals in Figure 1 and 5b, but below that of the crystalline minerals (Table 2, Figure 1). Therefore, these aggregates are identified as protein-mineral particles (PMP) whose mineral and protein volume fractions were each assessed to be of about 50% (see below). These composites may have some similarities with so-called calci-protein particles (CPP) built up of calcium phosphate and the glycoprotein fetuin-A. The formation of CPP is supposed to be important in bone formation.17,32 One scenario to explain these results is that a mineral matrix is homogeneously filled with protein monomers. As in this Q regime the BSA monomers scatter as point-like particles, this model would support the 3.4 power law as an approximate mass fractal. The scattering theory of fractals is reviewed in.22,23 The results of Figure 5 are the first clear evidence of stimulated mineralization by the protein BSA. Figure 6a−c shows the time evolution of the BSA monomers and that of PMPs in SSE−D2O solution from about 20 min to 3 h after sample preparation. These data were derived from in situ experiments using the same sample with an accumulation time of 10 min for each scattering curve. Figure 6a shows scattering from the monomers (Rg = 28 Å) and an upturn in intensity at small Q, as already shown in Figure 5a for even smaller Q values. The fit of this upturn once again yields a power law with an exponent of 3.4. The extrapolated scattering at Q = 0 from the monomers and the amplitude P3.4 of the small Q data are plotted versus time in Figure 6b and c, respectively. A decrease of about 15% in the monomer scattering is observed within the first 20 min, which is then followed by a small but continuous decrease in scattering, that is, of about 5.5% within 3 h. This decrease in scattering is caused by a reduction in the monomer number, as the radius of gyration of the monomer is always the same. The upturn of intensity at small Q is quantitatively described by the amplitude P3.4, as observed in Figure 6c; at about 30 min after mixing the solution, a sudden increase in scattering of about 45% is observed. The data in Figure 6b and c show that the reduction in the concentration of BSA monomers corresponds to an increase in concentration of the protein-mineral particles. To more completely characterize the mineral aggregates stimulated by BSA, we continued the SANS experiments using the USANS KWS 3, which allows measurements of an order of magnitude smaller Q, between 10−4 and 10−3 Å−1, as demonstrated in Figure 5a.19,20 This Q range is sensitive to aggregates of the order of μm size and permits full analysis in terms of the radius of gyration and the scattering intensity at Q = 0. In addition, the total observation time was extended to 20 h so as to allow a comparison of our results with previously published data.14 Three concentrations of BSA dissolved in

Figure 6. (a) “Large Q” SANS patterns of BSA in SSE−D2O. The red curve represents the solution immediately after mixing, and the other curves 30 min and more after the mixing. Scattering from BSA monomers dominates; the fit gives a radius of gyration of 28 ± 0.4 Å (see Figure 5b). The upturn of scattering at small Q is the result of large aggregates. b) Extrapolated scattering from monomers at Q = 0. A strong decrease in scattering is observed during the first 20 min [see red curve in panel a] and is followed by a continuous small linear decrease with time. The decreased intensity is caused by a decrease in the number density of the monomers. (c) The upturn at small Q follows a power law with an exponent of 3.38 ± 0.06 (Figure 5a). After 0.5 h, the amplitude of this scattering shows a sudden increase, in parallel with a decrease in the concentration of BSA monomers.

SSE−H2O and SSE−D2O were explored, 1.0, 2.5, and 10 mg mL−1. We expected that the measurements of the same processes in SSE−H2O and in SSE−D2O would deliver the scattering length density of the aggregated particles (as from the data in Figure 5c), from which information could be obtained about the internal structure of the particles, i.e., about the relative concentrations of mineral and protein. The results (see Figure 7a) show two scattering patterns in the low Q regime from a 10 mg mL−1 BSA solution in SSE−D2O, representing the properties of the precipitates observed during the first 16 and 19 h after mixing the sample. These data were fitted with Guinier’s law (Supporting Information eq S9), yielding the radius of gyration Rg and dΣ/dΩ(0); these two parameters are plotted versus time in Figure 7b and c. The large error bars of the scattered intensity imply that the scattering was extremely weak and that we are working at the lower limits of detection of mineral particles. Two stable conditions were observed for the 2.5 mg mL−1 BSA solution in SSE−D2O (Figure 7b); stable particles of 0.5 μm radius of gyration transform to 1 μm large particles after 9 h incubation time. The precipitated particles of the 10 mg mL−1 BSA solution (Figure 7c) are stable for about 13 h after 5 h of incubation; their average radius of gyration is 0.7 μm. After 18 h of annealing, the particles again start to grow, as is seen from the scattering pattern at 20 h. In the 1 mg mL−1 BSA sample (not shown here), we observe only a single stable situation, involving particles of about 0.5 μm in radius of gyration. F

dx.doi.org/10.1021/la4001889 | Langmuir XXXX, XXX, XXX−XXX

Langmuir

Article

therefore derived from the square root of A the coherent scattering length density ρP of the aggregates, as compiled in Table 5. Using this method, the matching condition of the aggregates with water is at 62.3% volume fraction of D2O (see Figure 1). This result is consistent with the matching conditions found in Figure 5c for the P3.4 amplitude. From the known scattering length densities of BSA, mineral, and water (Table 2 and Supporting Information Tables S3 and S4), we determined a 46% volume fraction of BSA and 54% volume fraction of mineral for the “300 Å” particles. These calculations were performed on the assumption that the equilibrium polymorph hydroxyapatite is one of the relevant minerals. However, as seen in Figure 1, the mineral could also be represented by nearly the same amount of the calcium carbonate polymorph calcite. The volume fraction of the “300 Å” particles can be determined from the second moment of scattering Q2 in eq A11. For the 2.5 mg mL−1 BSA solution in SSE−H2O at 11 h after mixing, we determine a Q2 = 1.023 × 10−5 cm−1 Å−3 from its fitted scattering pattern and ΔρPMP = 3.47 × 1010cm−2 from the BSA volume fraction c in Table 5 and the scattering length density ρ of BSA and HAP in Tables 2 and Supporting Information Tables S3 and S4. These parameters yield a particle volume fraction of ΦPMP = 4.3 × 10−4 (Supporting Information eq S11), about half of which is mineral and the other half protein. This indicates that 0.27 mg mL−1 BSA is bound in these particles, which represents 11% of the total amount of BSA. Another demonstration of consistent analysis of the “300 Å” particles is provided by the evaluation of the particle volume from the value d Σ/d Ω(0) = 120 ± 11.4 cm−1 of this sample. According to VP = d Σ/d Ω(0) × [ΦPΔρP2]−1 (eq 2), we determined that VP = 2.32 × 10−16 cm3 and an Rg = 295 Å, assuming that particle shapes are spherical, which is consistent with the fitted Rg parameter of 300 Å. Finally, we note that from d Σ/d Ω(0), a lower concentration of the PMPs is detected at higher BSA concentrations. Lysozyme in SSE Solution. Figure 8 shows large Q scattering of 2.5 mg mL−1 lysozyme in SSE−H2O and D2O. The fit of the data using Guiniers law (Supporting Information eq S9) is depicted by solid lines, and the corresponding parameters are compiled in Supporting Information Table S7. In both cases, we find a slightly larger Rg for the lysozyme monomer in H2O solutions than in D2O solutions. On the other hand, one finds slightly reduced scattering at Q = 0 because of the reduced volume fraction of the dissolved lysozyme monomers (by 15% in D2O and 14% in H2O). Because of this, only a small portion of the lysozyme monomers is involved in the process of mineralization, as was already observed in the BSA solutions. The scattering from these aggregates is already visible at small Q by the upturn in Q intensity (Figure 8b). This scattering will be quantitatively discussed on the basis of experiments at smaller Q obtained using the USANS KWS 3 instrument.

Figure 7. (a) Scattering pattern of the 10 mg mL−1 BSA solution from USANS measurements in the 10−4 Å−1 region. (b) Scattering at Q = 0 and Rg versus time for the 2.5 mg mL−1 BSA sample. Stable conditions occur for about 10 h. After 10 h, the particles start to grow again, reaching about twice their original size. (c) The same parameters for 10 mg mL−1 BSA, as derived from the scattering patterns in panel a. After 5 h of mixing, stable conditions are achieved (with 0.68 μm radius of gyration large particles), which last until about 18 h. After that time, the particles start to grow again. d) In addition to the 0.68 μm particles, there also exist particles with sizes of the order of the “300 Å” particles, whose extrapolated scattering at Q = 0 is depicted versus time for the SSE−H2O and SSE−D2O solutions. These particles are stable in the same interval as that of the 0.68 μm particles, and they disappear after 18 h when the larger particles start to grow again.

In parallel to these micrometer-sized particles, we observe at larger Q scattering from smaller particles of roughly 300 Å radius of gyration; as discussed below, particles of this size predominate in the SSE solutions of lysozyme. The scattering at Q = 0 of the “300 Å” particles in the 10 mg mL−1 BSA is plotted versus time for both SSE solutions in Figure 7d. The green dashed arrows indicate the interval of particle stability between 5 and 18 h. One clearly sees a correlation between the “300 Å” and the 0.68 μm large particles, especially after 18 h when the scattering from the “300 Å” particles disappears and the 0.68 μm particles start to grow again. We find stronger scattering in the SSE−H2O solution because of the larger scattering contrast of the particles. We defined the ratio A of dΣ/dΩ(0) from the SSE−H2O and SSE−D 2 O samples and evaluated the matching D 2 O concentration according to ΦM = √A(1 + √A)−1. We Table 5. Scattering Length Density of the Rg ≅ 300 Å Particles

a

BSA [mg mL−1]

time [h]

1 2.5 5 (P3.4) 10

6 ≤ t ≤ 15 6 ≤ t ≤ 12 12 ≤ t ≤ 15 5 ≤ t ≤ 19

ΦMatch

⟨ΦMatch⟩

ρP [1010 cm−2]

c [BSA]a

± ± ± ±

0.62 ± 0.02

3.77 ± 0.15

0.46 ± 0.03

0.61 0.64 0.62 0.62

0.02 0.01 0.09 0.03

c represents protein volume fraction of PMP. G

dx.doi.org/10.1021/la4001889 | Langmuir XXXX, XXX, XXX−XXX

Langmuir

Article

Figure 9b−d. Whereas Rg is approximately the same for all SSE solutions, the extrapolated scattering at Q = 0, d Σ/d Ω(0), is between 3 and 5 times larger in the SSE−H2O solutions than in the SSE−D2O solutions because of the larger scattering contrast in SSE−H2O. The mineral particles exist from the very beginning of the measurements and are stable for about 14−15 h before they disappear, as observed by a decrease in the Q = 0 intensity (Figure 9b−d). This behavior was observed for all lysozyme concentrations. In the BSA samples, we observed a similar scattering behavior and could correlate its disappearance with an increase in large-μm particles. Such particles of 0.83 ± 0.2 μm radius of gyration (d Σ/d Ω(0) = 4.65 ± 0.91 cm−1) were only detected for the 2.5 mg mL−1 SSE−D2O solution. It might be possible that in the other samples, these particles are even larger in size and therefore not visible in our experiments. The square root of A, defined as the ratio of d Σ/d Ω(0) from the SSE−H2O and SSE−D2O samples, delivers the matching SSE−D2O concentration of ΦMatch = √A(1 + √A)−1 and thereby the coherent scattering length density ρP of the aggregates, as compiled in Table 6. The 1 and 2.5 mg mL−1

Figure 8. (a) Lysozyme in SSE−H2O is dominated by scattering of monomers. The measurements were performed about 6 h after preparation. The corresponding data in salt-free H2O are shown in Figure 6b. (b) Lysozyme in salt-free D2O (blue dots; 7.5 h after preparation) and in SSE−D2O (red dots; 2 h after preparation). In SSE solution, a lower concentration of monomers and an enhanced scattering from aggregates is detected at small Q. The upturn at small Q can be described by a Q−4 power law with amplitude P4 = (3.19 ± 0.45) × 10−9 cm−1 Å−4.

Using the KWS 3, we followed (as for BSA) the scattering patterns of three lysozyme solutions of 1.0, 2.5, and 10 mg mL−1 in SSE−H2O and SSE−D2O over a period of 20 h (Figure 9). From the scattering pattern in Figure 9a, it is again

Table 6. Scattering Length Density of Rg ≅ 300 Å Particles and Content of Protein lysozyme [mg mL−1]

time [h]

ΦMatch

1 2.5 10

0 ≤ t ≤ 14 0 ≤ t ≤ 16 1 ≤ t ≤ 14

0.64 ± 0.02 0.64 ± 0.02 0.69 ± 0.05

ρP [1010 cm−2]

c [Lys.]

3.89 ± 0.14

0.39 ± 0.03

4.24 ± 0.35

0.19 ± 0.03

Figure 9. (a) USANS scattering pattern of 2.5 mg mL−1 lysozyme in SSE−H2O is dominated by aggregates of Rg of the order of 300 Å. (b− d) Extrapolated intensities at Q = 0 from 1.0-, 2.5-, and 10.0 mg mL−1 lysozyme solutions in SSE−H2O and SSE−D2O. For the fits, we fixed Rg = 300 Å, which was the averaged value from all data (and thus a reliable assumption). These particles were stable for about 14 h before they disappeared. The square root of d Σ/d Ω(0) from all three solutions shows matching between the 0.64 and 0.69 D2O contents, indicating that these are calci-protein particles.

lysozyme solutions show the same scattering length density, whereas the 10 mg mL−1 solution shows a slightly larger value of ΦMatch, yielding a mineral content of 81% instead of 61% (although the range of scattering length densities of the 1, 5, and 10 mg mL−1 solutions is within the range of the error bars); we assumed once again in our calculations that calcium phosphate (HAP) was the relevant mineral. The volume fraction of these protein-minerals can be estimated from the integrated intensity Q2 (Supporting Information eq S11), as we know the scattering length density of these particles. So, for the 2.5 mg mL−1 sample we determined Q2 = 9.4 × 10−6 cm−1 Å−3, from which we determined an aggregate volume fraction of 3.1 × 10−4. From the scattering contrast, we know that 39% of these particles consist of lysozyme, which means that a 1.2 × 10−4 volume fraction (0.17 mg mL−1) constitutes about 7% of the total amount of lysozyme. This amount of lysozyme is about half of the 15% detected reduction of lysozyme monomers in the SSE solution (Figure 8). The amount of precipitated aggregate appears to be inversely correlated with the amount of dissolved protein. The largest scattering from these aggregates is found in the 1 mg mL−1 solution whereas the smallest scattering is in the 10 mg mL−1 solution. These results confirm that only a small portion of the proteins is involved in the formation of protein−mineral aggregates (PMPs) of Rg ≅ 300 Å size.

clear that the scattering signals from these samples are extremely weak and are at the lower limit of reliable detection. The fit of the data with Guinier’s law (Supporting Information eq S9) yields a radius of gyration Rg which, on average, is 300 Å, and which we have chosen as a fixed value for the final fitting procedures of d Σ/d Ω(0), as depicted for all solutions in

DISCUSSION AND OUTLOOK This study addresses protein-induced aggregation of calcium minerals, in particular of calcium phosphate, using a model salt solution and analysis by small-angle neutron scattering (SANS) techniques. The proteins are bovine serum albumin (BSA) and lysozyme, and the model salt solution is a simulated secondary effluent (SSE) of the Shafdan wastewater reclamation plant in

■

H

dx.doi.org/10.1021/la4001889 | Langmuir XXXX, XXX, XXX−XXX

Langmuir

Article

Tel-Aviv, Israel. 14 The salt in the SSE solution was concentrated by a factor of 5 in comparison with the salts in the Shafdan wastewater effluent (Ca2+ by a factor of 5.2; excepting PO43−, which was concentrated by a factor of 2.6, see Supporting Information Table S1) to mimic a desalination stage of 80% wastewater recovery and polarization concentration. We explored in vitro the aggregation of minerals induced by the two proteins, which is expected to occur in the immediate neighborhood of the feed side of RO membranes and which is responsible for scaling. The BSA and lysozyme proteins are two prominent members of bacterial extracellular polymeric substances (EPS) involved in organic fouling.2 SSE Solution. The SSE solution itself shows weak scattering of neutrons from particles of about Rg = 70 Å (Figure 4a) and 8 × 10−6 volume fraction (Table 4). Their scattering length density is only slightly larger than for ACP and ACC but smaller than for the crystalline polymorphs such as α-TCP and HAP (Table 2). An about 10% smaller mass density of ACP and ACC (see values in Supporting Information Table S3) would exactly describe these particles as formed from ACP or ACC. The maximum mineral volume fraction of 12.5 × 10−6 for ACP is very near the ΦP of 8 × 10−6 determined from SANS (Table 4), whereas the ACC volume fraction is with 7.3 × 10−4 much larger. Formation of calcium phosphate is supported by a large saturation index SI of nearly 6.68 of HAP in comparison with the much smaller value of 0.79 for calcite (Table 2). So, in spite of the negative SI values for the amorphous polymorphs, we observe nanosized amorphous mineral polymorph particles which are stable for several days (stability of the SSE solution), thereby representing nuclei or precursor particles of inorganic crystalline particles.33 BSA in Salt-Free Water. BSA in salt-free water at concentrations of 2.5−10 mg mL−1 shows ordering with respect to a preferred mean distance determined from N−1/3 (N = particle density), which is caused by long-range electrostatic interactions (Supporting Information Figure S1). We determined the corresponding second virial coefficient A2 as (6.85 ± 0.78) × 10−4 mol cm−3, indicating the presence of repulsive interactions; this value is 2.6 times greater than that obtained using aqueous BSA solutions of lesser ionic strength, caused by the addition of 1 M NaCl.34,35 In SSE solution, the A2 of BSA becomes negligibly small, that is, the structure factor S(Q) = 1. We derived a molar volume of BSA which is about 33% greater than that derived from its molar weight, which is consistent with data in the literature, and which is explained by a hydration shell of densified water.34,35 Mixing of BSA in SSE. Mixing of BSA in SSE immediately induces strong formation of large aggregates, as depicted in Figure 5a. Half-micrometer-sized (radius of gyration) particles of fractal dimension 3.38 ± 0.06 are formed. We also found that most of the proteins remain as monomers and only a small but constant part of the BSA (∼11% in the 2.5 mg mL−1 solution) is involved in the process of aggregation. Another effect of mixing is that the correlation between the proteins is lost and only the form factor is measured, indicating that the long-range electrostatic interactions between proteins are shielded by the salt ions. The BSA monomers appear unaffected, as they show the same scattering length density as in salt-free water (Figure 5b). This result may point to the possibility that mineral clusters initiate the adsorption of proteins and not vise versa as observed for the serum protein fetuin-A monomers which coalesce with subnanometer large mineral cluster thereby inhibiting further mineral growth.17

The process of protein-induced mineralization is complex. In particular, the USANS experiments showed that particles of about 300 Å radius of gyration are formed spontaneously just after mixing. These particles were identified as protein-minerals (PMP) from the matching condition in the H2O/D2O SSE solution. Analysis of the scattering length density ρP shows that these particles consist of 46% BSA and 54% mineral; these values are similar to the results obtained by Ning et al.,6 who determined a volume fraction of organic matter of fouling layer on RO membranes of slightly less than 40% (weight percent of 20%); however, the similarity may be a coincidence. The 300 Å particles coexist and are correlated with much larger particles of the order of micrometer radius of gyration, as shown in Figure 7a. Their behavior appears to depend on the BSA concentration. In the 2.5 mg mL−1 BSA solution, 0.5 μmlarge (Rg) particles are formed within the first two hours, they then remain stable for 10 h, and then their size increases to 0.9 μm particles (Figure 7b). The 10 mg mL−1 sample achieved the first stable state of 0.68-μm-sized particles at a later time (after 5 h), and they remained stable for about 18 h (Figure 7c). A comparison of the 0.68 μm and 300 Å sized particles in Figure 7c and d clearly shows a correlation between these two classes of particles. Both particles were stable during the interval from 5 to 18 h. At less than 5 h (Figure 7d), we observe a decrease in concentration of the 300 Å particles, which after 18 h disappeared on account of their transformation into larger ones. We could not determine the scattering contrast of these larger particles as we obtained a clear signal from them only in the SSE−D2O solution. Lysozyme in Salt-Free Water. Lysozyme has been widely studied using SANS techniques, in most cases at concentrations greater than those used in this study to explore concentrationdependent effects between monomers18,31,36−38 and also to enhance the scattering signal for more precise data analysis.39 We determined the form factor of lysozyme following the same procedure as that for BSA. In contrast to BSA, we did not find a hydration shell surrounding lysozyme. The calculated molar volume is in agreement (within experimental accuracy) with the molar volume of 10.7 × 103 cm3 mol−1 determined in ref 39 but contrasts with values obtained using combined X-ray and neutron SAS experiments (see ref 31). The authors of ref 31 inferred the occurrence of a hydration layer from a comparison of the radii of gyration determined from the experimental curves with a fitting procedure using the CRYSON software program. We also performed such an analysis in the context of the data in Supporting Information Figure S5 and Table S7. The authors of,31 however, did not determine their scattering data in absolute units, and therefore could not determine the protein molar volume which is the basis of our conclusion. On the other hand, their radii of gyration (as derived from neutron data), determined as Rg = 13.8 ± 0.02 Å (H2O) and 12.4 ± 0.02 Å (D2O) (Table 1 of ref 31), are consistent with our results (see Supporting Information Tables S7). We also determined the second virial coefficient A2 = (3.2 ± 0.3) × 10−2 mol cm−3, which, on account of its large electrical load, is nearly 50 times greater than that of BSA. Lysozyme in SSE Solution. As observed for BSA, only a fraction (about 15% of 2.5 mg mL−1 solution) of lysozyme monomers is involved in mineral-aggregation processes. In all cases, we found stable conditions, that is, protein-mineral particles of constant number density and size from the start of the experiment until about 15 h, when the 300 Å sized particles disappeared. Thus, we observe patterns of stable particles in I

dx.doi.org/10.1021/la4001889 | Langmuir XXXX, XXX, XXX−XXX

Langmuir

Article

Fetuin-A acts as relevant inhibitor of uncontrolled calcium phosphate mineralization in mammals, i.e. the protein monolayer shields the CPPs against further growth. Such temporal stabilization of the protein-minerals against further growth is also observed in SSE. In conclusion, our results suggest that aggregates of calcium phosphate and calcium carbonate are formed within seconds to few minutes in the presence of organic molecules and biomolecules in SSE. These aggregates represent protein mineral particles (PMPs) stabilized over several hours against further increase. Today, calcium phosphate scaling is the limiting factor of achieving recovery higher than 80% and 50% in RO of, respectively, treated wastewater and seawater.2,6 Implications for the RO technique from this study could be further pretreatments of the feed effluent: (i) Reduction of organic matter and biomolecule content that were found to induce precipitation or (ii) reduction of “dangerous” ions, such as calcium, phosphate and carbonate by organic molecule induced mineralization followed by separation through a microporous membrane. Pretreatment (i) may reduce the scaling processes in RO desalination of secondary wastewater, whereas treatment (ii) applies organic molecules as coagulants in the concentrate of RO plants. Such procedure may reduce the salts content to such a low concentration that further RO treatment is allowed without dangerous scaling or fouling and thereby may achieve higher efficiency of RO desalination with appreciably lower volumes of disposed effluents.

lysozyme that are similar to those in BSA. A characteristic difference, however, is that μm-sized particles of lysozyme are not visible as in the case of BSA. It is likely that large particles formed, the sizes of which are outside the range of detection by the KWS 3. We determined a lysozyme volume fraction of 39% for the 1- and 2.5 mg mL−1 solutions and 19% for the 10 mg mL−1 solution, which is somewhat less that obtained for BSA. We also note that, in relation to the absolute values of d Σ/d Ω(0), the d Σ/d Ω(0) ratio is appreciably smaller in the 10 mg mL−1 solution than in the other solutions; this indicates that less 300 Å-sized aggregates are formed at larger protein concentrations. An explanation of this observation might be related to the small fraction of proteins involved in mineralization of the PMPs, that is, most of the proteins are dissolved as monomers in SSE. These protein monomers could take away some salt from SSE with the result of lower PMP volume fraction. In this case larger protein concentration would mean less salt for the PMPs, that is, a smaller PMP volume fraction. Such a behavior was observed for fetuin monomers attached with Posner cluster.17 A complexation of salt at the monomers increases their scattering length density which, if happened, was too weak in the present study to be detected as f.i. seen for BSA in Figure 5b. On the basis of our knowledge of aggregate composition and the second moment of scattering Q2 (Supporting Information eq S11), we determined the volume fraction of the 300 Å aggregates as ΦP = 4.3 × 10−4 and 3.1 × 10−4 for the 2.5 mg mL−1 BSA and lysozyme solutions, respectively. This indicates that 2.3 × 10−4 and 1.9 × 10−4 of the mineral volume fraction has been precipitated in the BSA and lysozyme solutions, respectively. We expect a maximal volume fraction of 6.7 × 10−6 and 3.7 × 10−4 for HAP and calcium carbonate, respectively (Table 2), based on concentrations of phosphate and carbonate in the SSE solution (Supporting Information Table S1). A comparison of these numbers with the total mineral volume fraction of the PMPs from SANS shows that calcium phosphate cannot be the only mineral precipitated as the 300 Å large mineral volume fractions are about 30 times greater than the maximum possible amount of HAP. It appears rather more likely to also explain the SANS results by a larger CaCO3 volume fraction on basis of the carbonate−bicarbonate equilibrium reaction. If precipitation of CaCO3 initiates a shifting of the equilibrium reaction of carbonate ions on account of bicarbonate, a large volume fraction of 3.7 × 10−4 of the carbonate (calcite) mineral is generated. Thus, HAP and calcite appear as the minerals formed in the PMPs in consistence with their maximum possible volume fraction and the positive SI values indicating supersaturation in SSE. This cannot be said for calcium sulfate which has a slightly negative SI parameter in SSE (Table 2). Another interesting aspect of this study is the formation of protein-minerals which are stabilized for several hours and which suddenly increase with no visible reason as shown in Figure 7b and c. This phenomenon reminds us to similar observation of calcium phosphate formation in the presence of the glycoprotein fetuin-A as reported in refs 17, 32, and 40. Stable CPPs of about 500 Å were observed for several hours after formation before they increased to about twice larger size. Contrast variation SANS experiments showed that the protein monomers were homogeneously distributed inside the smaller CPPs whereas they form a monolayer around the surface of the larger CPPs.17 This means that the increase of the CPPs is caused by a change of protein distribution inside the particles.

■

ASSOCIATED CONTENT

S Supporting Information *

Relevant scattering laws of small-angle neutron scattering (SANS) needed for this publication, as well as the SANS results from the proteins BSA and lysozyme in salt free water. We also give the equations for the saturation index SI together with the parameters of the SSE solution and the other solutions as determined from SANS. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Present Address ⊥

Technische Universität München, Physik-Department E13, Lehrstuhl für Funktionelle Materialien, D-85748 Garching, James-Franck-Str. 1, Germany Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS We would like to thank Marlies Hintzen, FZ-Jülich, for help in sample preparation and the reviewer for his suggestions which helped considerably to improve the content and readability of the manuscript. This work was funded by the Ministry of Science, Culture and Sport (MOST)and the Bundesministerium für Bildung und Forschung (BMBF) and was performed as part of the joint German−Israeli Research Program under the title “The Effect of Bioinspired Mineralization on the Scaling of RO/NF Membranes in Desalination and Water Treatment”. J

dx.doi.org/10.1021/la4001889 | Langmuir XXXX, XXX, XXX−XXX

Langmuir

■

Article

(25) Sophianopoulos, A. J.; Rhodes, C. K.; Holcomb, D. N.; van Holde, K. E. Physical studies of lysozyme, ICharacterization. J. Biol. Chem. 1962, 237, 1107−1112. (26) Mylonas, E.; Svergun, D. I. Accuracy of molecular mass determination of proteins in solution by small-angle X-ray scattering. J. Appl. Crystallogr. 2007, 40, s245−s249. (27) Sablani, S. S.; Goosena, M. F. A.; Al-Belushi, R.; Wilf, R. Concentration polarization in ultrafiltration and reverse osmosis: A critical review. Desalination 2001, 141, 269−289. (28) Gebauer, D.; Xingmin, L.; Aziz, B.; Hedin, N.; Zhao, Z. Porous tablets of crystalline calcium carbonate via sintering of amorphous nanoparticles. CrystEngComm 2013, 15, 1257−1263. (29) Brezonik, P. L.; Arnold, W. A. Water Chemistry: An Introduction to the Chemistry of Natural and Engineered Aquatic Systems Oxford University Press: Oxford, U.K., 2011. (30) Bendedouch, D.; Chen, S. H. Structure and interparticle interactions of bovine serum albumin in solution studied by smallangle neutron scattering. J. Phys. Chem. 1983, 87, 1473−1477. (31) Svergun, D. I.; Richard, S.; Koch, M. H. J.; Sayers, Z.; Kuprin, S.; Zaccai, G. Protein hydration in solution: experimental observation by X-ray and neutron scattering. Proc. Natl. Acad. Sci. U.S.A. 1998, 95, 2267−2272. (32) Heiss, A.; Jahnen-Dechent, W.; Endo, H.; Schwahn, D. Structural dynamics of a colloidal proteinMineral complex bestowing on calcium phosphate a high solubility in biological fluids. Biointerphases 2007, 2, 16−20. (33) Gebauer, D.; Cölfen, H. Prenucleation clusters and non-classical nucleation. Nano Today 2011, 6, 564−584. (34) Nossal, R.; Glinka, C. J.; Chen, S. H. SANS studies of concentrated protein solutions. I. Serum albumin. Biopolymers 1986, 25, 1157−1175. (35) Zhang, F.; Skoda, M. W. A.; Jacobs, R. M. J.; Martin, R. A.; Martin, C. M.; Schreiber, F. Protein interactions studied by SAXS: Effect of ionic strength and protein concentration for BSA in aqueous solutions. J. Phys. Chem. 2007, B 111, 251−259. (36) Stradner, A.; Cardinaux, F.; Schurtenberger, P. A small-angle scattering study on equilibrium clusters in lysozyme solutions. J. Phys. Chem. 2006, 110, 21222−21231. (37) Shukla, A.; Mylonas, E.; Di Cola, E.; Finet, S.; Timmins, P.; Narayanan, T.; Svergun, D. I. Absence of equilibrium cluster phase in concentrated lysozyme solutions. Proc. Natl. Acad. Sci. U.S.A. 2008, 105 (13), 5075−5080. (38) Liu, Y.; Fratini, E.; Baglioni, P.; Chen, W.-R.; Chen, S.-H. Effective long-range attraction between protein molecules in solution studied by small angle neutron scattering. Phys. Rev. Lett. 2005, 95, 118102. (39) Stuhrmann, H. B.; Fuess, H. A neutron small-angle scattering study of hen egg-white lysozyme. Acta Crystallogr. 1976, A32, 67−74. (40) Wald, J.; Wiese, S.; Eckert, T.; Jahnen-Dechent, W.; Richtering, W.; Heiss, A. Formation and stability kinetics of calcium phosphatefetuin-A colloidal particles probed by time-resolved dynamic light scattering. Soft Matter 2011, 7, 2869−2874.

REFERENCES

(1) Shannon, M. A.; Bohn, P. W.; Elimelech, M.; Georgiadis, J. G.; Mariñas, B. J.; Mayes, A. M. Science and technology for water purification in the coming decades. Nature 2008, 452, 301−310. (2) Greenberg, G.; Hasson, D.; Semiat, R. Limits of RO recovery imposed by calcium phosphate precipitation. Desalination 2005, 183 (1−3), 273−288. (3) Glueckstern, P.; Priel, M.; Gelman, E.; Perlov, N. Wastewater desalination in Israel. Desalination 2008, 222, 151−164. (4) Lin, N. H.; Kim, M.; Lewis, G. T.; Cohen, Y. Polymer surface nano-structuring of reverse osmosis membranes for fouling resistance and improved flux performance. J. Mater. Chem. 2010, 20, 4642−4652. (5) Contreras, A. E.; Steiner, Z.; Miao, J.; Kasher, R.; Li, Q. Studying the role of common membrane surface functionalities on adsorption and cleaning of organic foulants using QCM-D. Environ. Sci. Technol. 2011, 45, 6309−6315. (6) Ning, R. Y.; Troyer, T. L. Colloidal fouling of RO membranes following MF/UF in the reclamation of municipal wastewater. Desalination 2007, 208, 232−237. (7) Katz, I.; Dosoretz, C. G. Desalination of domestic wastewater effluents: phosphate removal as pretreatment. Desalination 2008, 222, 230−242. (8) Mathew, M.; Takagi, S. Structures of biological minerals in dental research. J. Res. Natl. Inst. Stand. Technol. 2001, 106, 1035−1044. (9) Mann, S. Biomineralization: Principles and Concepts in Bioorganic Materials Chemistry; Oxford University Press: Oxford, U.K., 2001. (10) Flemming, H. C. Reverse osmosis membrane biofouling. Exp. Therm. Fluid Sci. 1997, 14, 382−391. (11) Tang, C. Y.; Kwon, Y. N.; Leckie, J. O. Fouling of reverse osmosis and nanofiltration membranes by humic acidEffects of solution composition and hydrodynamic conditions. J. Membr. Sci. 2007, 290 (1−2), 86−94. (12) Vrouwenvelder, J. S.; van der Kooij, D. Diagnosis of fouling problems of NF and RO membrane installations by a quick scan. Desalination 2003, 15, 121−124. (13) Contreras, A. E.; Kim, A.; Li, Q. L. Combined fouling of nanofiltration membranes: Mechanisms and effect of organic matter. J. Membr. Sci. 2009, 327, 87−95. (14) Steiner, Z.; Rapaport, H.; Oren, Y.; Kasher, R. Effect of surfaceexposed chem-cal groups on calcium-phosphate mineralization in water-treatment systems. Environ. Sci. Technol. 2010, 44, 7937−7943. (15) Tarnacki, K.; Lyko, S.; Wintgens, T.; Melin, T.; Natau, F. Impact of extracellular polymeric substances on the filterability of activated sludge in membrane bioreactors for landfill leachate treatment. Desalination 2005, 179 (1−3), 181−190. (16) Jacrot, B. The study of biological structures by neutron scattering from solution. Prog. Phys. 1975, 39, 912−953. (17) Heiss, A.; Pipich, V.; Jahnen-Dechent, W.; Schwahn, D. FetuinA is a mineral carrier protein: Small angle neutron scattering provides new insight on Fetuin-A controlled calcification inhibition. Biophys. J. 2010, 99, 3986−3995. (18) Stradner, A.; Sedgwick, H.; Cardinaux, F.; Poon, W. C. K.; Egelhaaf, S. U.; Schurtenberger, P. Equilibrium cluster formation in concentrated protein solutions and colloids. Nature 2004, 432, 492− 495. (19) https://fzj.frm2.tum.de/ (20) Alefeld, B.; Schwahn, D.; Springer, T. New Developments of small angle neutron scattering instruments with focussing. Nucl. Inst. A Meth. Phys. Res. 1989, A 274, 210−216. (21) Frielinghaus, H.; Pipich, V.; Radulescu, A.; Heiderich, M.; Hanslik, R.; Dahlhoff, K.; Iwase, H.; Koizumi, S.; Schwahn, D. Aspherical refractive lenses forsmall-angle neutron scattering. J. Appl. Crystallogr. 2009, 42, 681−690. (22) Higgins, J. S.; Benoît, H. C. Polymers and Neutron Scattering; Clarendon Press, Oxford, U.K., 1994. (23) Roe, R. J. Methods of X-Ray and Neutron Scattering in Polymer Science; University Press, Oxford, U.K., 2000. (24) Neutron News, 1992, 3 (3), 29−37. K

dx.doi.org/10.1021/la4001889 | Langmuir XXXX, XXX, XXX−XXX