Organic–mineral interfacial chemistry drivesheterogeneous nucleation of Sr-rich (Bax, Sr1−x)SO4from undersaturated solutionNing Denga, Andrew G. Stackb, Juliane Weberb, Bo Caoa, James J. De Yoreoc,d, and Yandi Hua,1

aDepartment of Civil and Environmental Engineering, University of Houston, Houston, TX 77004; bChemical Science Division, Oak Ridge NationalLaboratory, Oak Ridge, TN 37831; cPhysical Sciences Division, Pacific Northwest National Laboratory, Richland, WA 99354; and dDepartment of MaterialsScience and Engineering, University of Washington, Seattle, WA 98195

Edited by François M. M. Morel, Princeton University, Princeton, NJ, and approved April 30, 2019 (received for review December 10, 2018)

Sr-bearing marine barite [(Bax, Sr1−x)SO4] cycling has been widelyused to reconstruct geochemical evolutions of paleoenvironments.However, an understanding of barite precipitation in the ocean,which is globally undersaturated with respect to barite, is missing.Moreover, the reason for the occurrence of higher Sr content in marinebarites than expected for classical crystal growth processes re-mains unknown. Field data analyses suggested that organic mol-ecules may regulate the formation and composition of marinebarites; however, the specific organic–mineral interactions are un-clear. Using in situ grazing incidence small-angle X-ray scattering(GISAXS), size and total volume evolutions of barite precipitateson organic films were characterized. The results show that bariteforms on organic films from undersaturated solutions. Moreover,from a single supersaturated solution with respect to barite, Sr-richbarite nanoparticles formed on organics, while micrometer-sizeSr-poor barites formed in bulk solutions. Ion adsorption experimentsshowed that organic films can enrich cation concentrations in theadjacent solution, thus increasing the local supersaturation andpromoting barite nucleation on organic films, even when the bulksolution was undersaturated. The Sr enrichment in barites formed onorganic films was found to be controlled by solid-solution nucleationrates; instead, the Sr-poor barite formation in bulk solution was foundto be controlled by solid-solution growth rates. This study provides amechanistic explanation for Sr-rich marine barite formation and offersinsights for understanding and controlling the compositions of solidsolutions by separately tuning their nucleation and growth rates viathe unique chemistry of solution–organic interfaces.

Sr-rich marine barite | organic–mineral interactions | solid solution |nucleation and growth | paleoenvironments

Sr-bearing barite [(Bax, Sr1−x)SO4] is an important marineauthigenic mineral present in the seawater column and marine

sediments (1–5). Marine barite formation, essential to the globalcycles of organic carbon, sulfate, and oxygen, can be used to recon-struct the changes in geochemical cycles and paleoproductivity, andSr incorporation may reflect the oceanographic conditions (2, 5–7).Despite the geochemical importance of marine (Bax, Sr1−x)SO4formation, two paradoxes still exist. First, the global oceans areundersaturated with respect to barite, while barite is extensivelypresent in seawater column and marine sediments (5, 7, 8). Second,Sr-rich marine barite (>50 mol% strontium) formation (7, 9) is notthermodynamically favorable (7, 10), as the solubility of SrSO4(Ksp,SrSO4 = 10−6.63) is three orders of magnitude higher thanBaSO4 (Ksp,BaSO4 = 10−9.98) (3, 10). Field and laboratory studiesprovide evidence for the association of organics with both paradoxes(7, 11); however, the specific roles of organics in these thermo-dynamically unfavorable mineralization processes are not clear(7, 11, 12).Organic–mineral interactions play essential roles in biomineraliza-

tion processes (13, 14), which produce minerals with well-controlledmorphologies, structures, and compositions not readily explainedby traditional inorganic mineral nucleation and growth processes

(13, 15). Therefore, understanding how organic–mineral interactionsdirect mineral nucleation and growth could advance the use ofbiominerals as chemical archives of paleoenvironments (7, 16,17). In addition, manipulating organic–mineral interactions offersstrategies for the synthesis of crystals with tunable properties for avariety of applications (13).Despite the importance of organic–mineral interactions, the

impact of the organic interface on the thermodynamics and kineticsof mineralization is poorly understood (14). Here, we advance thatunderstanding by using in situ grazing-incidence small-angle X-rayscattering (GISAXS) to measure heterogeneous nucleation andgrowth rates of (Bax, Sr1−x)SO4 precipitated on organic filmsfrom Sr-rich solutions. As previously demonstrated, GISAXSmeasures nucleation and growth rates at solid–liquid interfaceswith sufficient temporal and spatial resolution (18–26) to provideunique information about substrate–mineral interactions (24). Inour recent study, a range of organic coatings (terminated with –COOH,–SH, and mixed –SH-and-COOH functional groups) were foundto significantly affect the heterogeneous nucleation and growthkinetics of barite by controlling the local ion concentrations andthe interfacial energies for heterogeneous barite nucleation (24).However, barite precipitation on organics from undersaturatedsolutions and with Sr enrichment during coprecipitation was notinvestigated, leaving the paradoxes described above unresolved.

Significance

The wide occurrence of Sr-rich marine barite in undersatu-rated seawater presents a paradox. Here, in undersaturatedsolution, we observe barite nucleation on organic films andshow it is enabled by cation enrichment. In supersaturatedsolution, this enrichment generates nanometer-sized Sr-richnuclei on organic films, while Sr-poor barite grows quicklyto micrometer-sized crystals in bulk solutions. Theoreticalsolid-solution calculations explain the distinct Sr incorporationin barites on organic films and in bulk solutions. The findingsresolve the barite paradox and provide insights into manipu-lating solid-solution nucleation and growth through theunique chemical environment near organic–mineral interfaces,which may revise our understanding of many biomineraliza-tion processes and allow strategies for tailoring material syn-thesis to achieve desired sizes and compositions.

Author contributions: Y.H. designed research; N.D., A.G.S., and J.W. performed research;N.D., J.W., B.C., and Y.H. analyzed data; and N.D., J.J.D., and Y.H. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Published under the PNAS license.

See Commentary on page 13161.1To whom correspondence may be addressed. Email: yhu11@uh.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1821065116/-/DCSupplemental.

Published online May 21, 2019.

www.pnas.org/cgi/doi/10.1073/pnas.1821065116 PNAS | July 2, 2019 | vol. 116 | no. 27 | 13221–13226

EART

H,A

TMOSP

HER

IC,

ANDPL

ANET

ARY

SCIENCE

SEN

VIRONMEN

TAL

SCIENCE

SSE

ECO

MMEN

TARY

Dow

nloa

ded

by g

uest

on

May

20,

202

0

In this study, we use GISAXS (Fig. 1) to quantify in situ hetero-geneous nucleation of (Bax, Sr1−x)SO4 on bare or organic-coatedSiO2 substrates from undersaturated or slightly supersaturatedsolutions (Table 1). Ex situ scanning transmission electron micros-copy coupled with energy-dispersive X-ray spectroscopy (STEM-EDX), X-ray photoelectron spectroscopy (XPS), and inductivelycoupled plasma–mass spectrometry (ICP-MS) are also employed tocharacterize the morphology and composition of the (Bax, Sr1−x)SO4on organic films (Fig. 1). As a comparison, homogeneous (Bax,Sr1−x)SO4 from the same slightly supersaturated solutions arecharacterized by small-angle X-ray scattering (SAXS) (Fig. 1),scanning electron microscopy coupled with energy dispersive X-rayspectroscopy (SEM-EDX), XPS, and ICP-MS (Fig. 1). To under-stand the organic–mineral interactions, ion adsorption onto bare ororganic-coated SiO2 was also measured through batch adsorption/desorption experiments. To understand the distinct morphologiesand compositions of (Bax, Sr1−x)SO4 on organic films and in bulksolutions, theoretical calculations of solid-solution nucleation andgrowth rates were conducted.

Results and DiscussionHeterogeneous (Bax, Sr1−x)SO4 Precipitation on Organics from Under-saturated Solution. The solutions (Table 1, Ba100Sr0, Ba50Sr50,Ba30Sr70, Ba10Sr90, Ba05Sr95, and Ba01Sr99) had fixed initialtotal cation and anion concentrations equal to 10−4 M with varyingaqueous Ba/Sr ratios. The solutions’ saturation indices [SI = log(Q/K), where Q and K represent the ion product and solubilityconstant] with respect to BaSO4 and SrSO4 end members are shownin Table 1. Most solutions were undersaturated with respect toSrSO4 and slightly supersaturated with respect to BaSO4. SolutionBa01Sr99, which was undersaturated with respect to both endmembers, represented undersaturated seawater (27).Homogeneous (Bax, Sr1−x)SO4 precipitates were only detected

from relatively more supersaturated solutions (Table 1, Ba100Sr0and Ba50Sr50 with SI values of 1.82 and 1.53 with respect to BaSO4).No homogeneous (Bax, Sr1−x)SO4 was detected from other solutionswith SI ≤ 1.31, consistent with a previous study, which showed thathomogeneous barite precipitation cannot be easily achieved withan SI of ∼1.2 (28).Under all solution conditions, heterogeneous precipitates were

detected on organics by GISAXS (SI Appendix, Fig. S1), and theevolution of average size and total volume of the precipitates werecalculated (Fig. 2). Continuous increases in total particle volume(Fig. 2A) were measured as the reactions progressed, indicatingcontinuous heterogeneous precipitation on organic films evenfrom undersaturated solution (Ba01Sr99).To understand heterogeneous (Bax, Sr1−x)SO4 precipitation

on organic films from undersaturated bulk solution, we hypoth-esized that organic films may alter the local supersaturation to besignificantly different from the bulk solution. Cation adsorptiononto organics, which could increase the local SI and promoteheterogeneous precipitation on organics, has been reported inprevious studies (11, 15, 24). To test our hypothesis, Ba2+, Sr2+,and SO4

2− ion adsorption onto the organic films were measured.

While no adsorption of sulfate ions onto the organics was de-tected, significant amounts of cation adsorption were measured(Table 1), which could increase the local dissolved cation con-centration near the organic films (7, 11, 29). Typically, adsorbed ionson substrates, including inner-sphere and outer-sphere complexes,can form an electrical double layer (EDL) with a thickness of∼1 nm (30, 31). As it is very challenging to measure cation con-centrations in the local solution near organic films, here we assumethat the local cation concentration can be approximated by dis-tributing the adsorbed Ba2+ and Sr2+ ions in a 1- to 10-nm waterfilm covering the organic film surface. Using such estimated local ionconcentrations near the organic-coated substrates, all local solutionsnear the organic–water interfaces were calculated to be highlysupersaturated (SI Appendix, section 1.1 and Table S1). Forcontrol experiments with bare SiO2 substrates without organiccoatings, no Ba2+ or Sr2+ ion adsorption was detected by ICP-MS,and no heterogeneous precipitates were measured by GISAXS (SIAppendix, Fig. S2). Our findings demonstrate that the role of theorganic films is to enrich Ba2+ and Sr2+ ions in the vicinity of theorganic–water interfaces, greatly increasing the local saturationindex to the point of significant supersaturation and thus inducingheterogeneous (Bax, Sr1−x)SO4 precipitation from undersaturatedbulk solutions. These direct observations of continuous hetero-geneous precipitation on organic films from undersaturated bulksolutions along with the identification of the underlying mecha-nism resolves the first paradox of marine barite formation fromglobally undersaturated seawater.

Heterogeneous (Bax, Sr1−x)SO4 Nucleation on Organic Films VersusHomogeneous (Bax, Sr1−x)SO4 Growth. In slightly supersaturatedsolutions (Ba100Sr0, SIBaSO4 = 1.82, and Ba50Sr50, SIBaSO4 = 1.53;Table 1), the homogeneous (Bax, Sr1−x)SO4 precipitates were toolarge to be measured by SAXS (measurable range, 1–50 nm) andwere measured to be micrometer-sized by SEM (Fig. 3, Left Insetand SI Appendix, Fig. S3) at the end of the 1-h experiment, indi-cating fast growth. Interestingly, the sizes of heterogeneous pre-cipitates on organic films were measured to only increase slightlyfrom 2 to ∼4 nm within 1 h by GISAXS (Fig. 2B), indicating anucleation-dominated precipitation process without much growth.The heterogeneous (Bax, Sr1−x)SO4 precipitates were also measuredto be nanometer size by STEM (Fig. 3, Right Inset).Generally, in a supersaturated solution, precipitation starts with

nucleation, which increases particle number, followed by growth,which increases particle size. Fig. 3 schematically illustrates the generaldependence of nucleation (red line) and growth rates (black line)on a given solution’s supersaturation level (32). When the solutionis slightly supersaturated (stage I), growth rate is much higher thannucleation rate; while in a highly supersaturated solution (stageII), nucleation rate can be much higher than growth rate (32, 33).Such insights could explain the observed size difference for thehomogeneous and heterogeneous precipitates. For homogeneousprecipitation in bulk solution with low supersaturation (SIBaSO4 =1.82 and 1.53 for Ba100Sr0 and Ba50Sr50 solutions), growth was thedominant process leading to the formation of large micrometer-sized

Fig. 1. Schematics of in situ (A) SAXS and (B) GISAXSsetups for measuring (Bax, Sr1−x)SO4 particle formationin bulk solutions (homogeneous precipitation) and atorganic–water interfaces (heterogeneous precipitation),respectively. The 2D SAXS and GISAXS scatteringimages are analyzed to calculate the evolutions ofaverage particle size and total particle volume (Fig. 2).

13222 | www.pnas.org/cgi/doi/10.1073/pnas.1821065116 Deng et al.

Dow

nloa

ded

by g

uest

on

May

20,

202

0

crystals (Fig. 3, Left Inset); while for heterogeneous precipitationat organic–water interfaces where local supersaturation levels arehigh (SIBaSO4 of ∼6–8 with the 1-nm water film assumption men-tioned in earlier section), nucleation was the dominant process,leading to the formation of small nanometer-sized particles (Fig. 3,Right Inset) on the organic films. Thus, the GISAXS, SEM, andSTEM data reveal yet another role of the organics: Heterogeneousbarite precipitation on organic films is a nucleation-dominant processand is distinct from growth-dominated homogeneous precipitationin bulk solutions.

Sr Enrichment in Heterogeneous (Bax, Sr1−x)SO4 Precipitates on OrganicFilms. Our results also show that the compositions of the homo-geneous and heterogeneous (Bax, Sr1−x)SO4 precipitates are re-markably different. For homogeneous precipitates formed inBa50Sr50 solution, the molar percentage of Sr (i.e., 1 − x) wasmeasured to be only 0.04 by ICP-MS (SI Appendix, Table S2).XPS and SEM-EDX measurements could detect only Ba, whileSr is under the detection limit of XPS and SEM-EDX. Nevertheless,ICP-MS, XPS, and SEM-EDX measurements all indicated limitedSr incorporation in the homogeneous (Bax, Sr1−x)SO4 precipitates.In contrast, Sr enrichment in heterogeneous (Bax, Sr1−x)SO4 pre-cipitates were measured to have molar percentages of Sr (i.e., 1 − x)ranging from 0.20 to 0.99 (SI Appendix, section 1.2 and Table S2).For homogeneous (Bax, Sr1−x)SO4 precipitation, because growth

is the dominant process, the dependence of (Bax, Sr1−x)SO4 growthrate on the composition (i.e., x) of the (Bax, Sr1−x)SO4 solid solutionis considered to understand the limited Sr incorporation in theprecipitates. The saturation indices [SI(x)] of the solutions withrespect to the (Bax, Sr1−x)SO4 precipitates can be calculated asfunctions of the mole fractions of Ba, x, in the solid solutionsaccording to Eq. 1 proposed by Pina and Putnis (34):

SIðxÞ=Log

0BB@

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi�αxBa2+

��α1−xSr2+

� �αSO2−

4

��Ksp,BaSO4αBaSO4

�x�Ksp,SrSO4αSrSO4

�1−xvuuut

1CCA, [1]

where αBa2+ , αSr2+ , and αSO42− are the activities of Ba2+, Sr2+, and

SO42− ions in solutions. Ksp,BaSO4 and Ksp,SrSO4 are the solubility

constants of BaSO4 and SrSO4 end members, being 10−9.98 and10−6.63 at 25 °C based on the spec E8 database in Geochemist’sWorkbench (GWB, release 9.0; Aqueous Solutions). αBaSO4 andαSrSO4 are the activities of BaSO4 and SrSO4 end members in(Bax, Sr1−x)SO4 solid solution. Recent studies indicated that(Bax, Sr1−x)SO4 is nonideal solid solution (35, 36), and αBaSO4

and αSrSO4 can be calculated according to Thompson–Waldbaumequations (SI Appendix, section 1.3) (35, 36).

As higher supersaturation leads to faster growth rate, thecomposition (i.e., the x value) leading to the maximum value ofSI(x) represents the solid solution with the fastest growth rate.Under all solution conditions (Table 1, Ba50Sr50, Ba30Sr70,Ba10Sr90, Ba05Sr95, and Ba01Sr99 solutions), SI(x) reachedmaximum with x ∼ 1 (x = 0.99995, 0.99990, 0.99950, 0.99900, and0.99500; Fig. 4A and SI Appendix, Table S2), due to the muchhigher solubility of celestite (Ksp,SrSO4 = 10−6.63) over barite(Ksp,BaSO4 = 10−9.98). The calculated x values suggest that limitedSr incorporation should occur in the homogeneous precipitates,which agrees, in general, with ICP-MS, XPS, and SEM-EDXmeasurements (Fig. 5A and SI Appendix, Table S2). These resultsindicate that the low level of Sr incorporation in the homogeneousprecipitates, in which growth is the dominant process, is associatedwith the (Bax, Sr1−x)SO4 solid solution with the largest SI(x)—i.e.,the highest stability.The differences in compositions of homogeneous and het-

erogeneous (Bax, Sr1−x)SO4 precipitates suggest an importantrole for the organics in modulating Sr incorporation during (Bax,Sr1−x)SO4 precipitation. Previous studies hypothesized that thepreferential interactions of organics with Sr over Ba may pro-mote Sr incorporation in barite (7). However, here ion adsorp-tion/desorption experiments showed a preferential accumulationof Ba over Sr toward the organic films, as the partitioning co-efficients of Sr/Ba (calculated by dividing the adsorbed Sr/Baamounts by aqueous Sr2+/Ba2+ ion concentration ratios) weremuch lower than 1 (Table 1).Another hypothesis is that the different nucleation and growth

processes for homogeneous and heterogeneous precipitationmay affect the compositions of (Bax, Sr1−x)SO4 precipitates. As

Table 1. Solution conditions for precipitation and ion adsorption experiments

Initial conditions for precipitation* Ion adsorption†

Sample BaCl2, mM SrCl2, mM Na2SO4, mM SI‡ for BaSO4 SI‡ for SrSO4 Ba2+, % Sr2+, % Kd§

Ba100Sr0 0.10 0 0.10 1.82 20.5 ± 0.9Ba50Sr50 0.05 0.05 0.10 1.53 −1.80 28.7 ± 0.4 10.1 ± 0.3 0.32 ± 0.04Ba30Sr70 0.03 0.07 0.10 1.31 −1.65 30.1 ± 2.7 10.3 ± 1.6 0.32 ± 0.02Ba10Sr90 0.01 0.09 0.10 0.83 −1.54 35.8 ± 4.6 13.7 ± 2.4 0.34 ± 0.02Ba05Sr95 0.005 0.095 0.10 0.53 −1.51 39.8 ± 0.8 15.7 ± 0.7 0.35 ± 0.08Ba01Sr99 0.001 0.099 0.10 −0.17 −1.50 34.9 ± 1.3 17.5 ± 1.0 0.55 ± 0.02

*Initial conditions for precipitation: for all conditions, the ionic strength and pH values are 0.58 mM and 5.61 as calculated by GWB.†Ion adsorption: the percentage of Ba2+ or Sr2+ adsorption onto the organic films are calculated as amounts of ions desorbed in 2%nitric acid divided by the total amounts of ions in the initial solutions.‡SI: saturation indices with respect to BaSO4 and SrSO4 end members.§Kd: the partitioning coefficients are calculated by dividing the molar ratios of Sr/Ba adsorbed on the organic surfaces with aqueousSr2+/Ba2+ molar concentration ratios.

Fig. 2. Total particle volume (A) and average particle size (B) of hetero-geneous (Bax, Sr1−x)SO4 precipitates on organic films are calculated andplotted vs. reaction time based on in situ GISAXS measurements. Details ofthe calculations can be found in our previous publications (19–26).

Deng et al. PNAS | July 2, 2019 | vol. 116 | no. 27 | 13223

EART

H,A

TMOSP

HER

IC,

ANDPL

ANET

ARY

SCIENCE

SEN

VIRONMEN

TAL

SCIENCE

SSE

ECO

MMEN

TARY

Dow

nloa

ded

by g

uest

on

May

20,

202

0

heterogeneous (Bax, Sr1−x)SO4 formation is a nucleation-dominatedprocess with little growth (sizes maintained at 2–4 nm), we hy-pothesized that the compositions of heterogeneous (Bax, Sr1−x)SO4 precipitates are related with the highest nucleation rates,instead of growth rates, as determined by (Bax, Sr1−x)SO4 solidsolutions under the local solution conditions near the organicfilms. In contrast to growth rates, which are controlled only bySI(x), the nucleation rates [J(x)] of a solid solution are controllednot only by saturation index [SI(x)], but also by interfacial energy[σ(x)] and molecular volume [Ω(x)] of the solid solution, accordingto Eq. 2, proposed by Pina and Putnis (34):

JðxÞ= D

Ω53ðxÞ

exp

"−Bσ3ðxÞΩ2ðxÞ

K3T3½lnð10SIðxÞÞ�2#, [2]

where D (∼10−9 m2/s) is the mean diffusion coefficient for ions inwater. Ω(x), σ(x), and SI(x) are the molecular volumes, interfa-cial energies, and saturation indices of the solid solution, whichcan be calculated using the molecular volume, interfacial energy,and saturation index of the end members as well as the compo-sition of the solid solution (i.e., x). Detailed calculations can befound in the publication by Pina and Putnis (34).According to Eq. 2, the composition of the solid solution (i.e., x)

can affect the nucleation rate of the solid solution, J(x), in dif-ferent ways. On one hand, the increase in x will result in higherSI(x) due to the lower solubility of BaSO4 (Ksp,BaSO4 = 10−9.98) overSrSO4 (Ksp,SrSO4 = 10−6.63), thus leading to higher J(x); on the otherhand, the increase in x can result in higher σ(x) due to the higherinterfacial energy of BaSO4 (σBaSO4 = 0.125 J/m2) over SrSO4(σSrSO4 = 0.097 J/m2), as well as higher Ω(x) due to the highermolecular volume of BaSO4 (ΩBaSO4 = 8.60 × 10−29 m3) over SrSO4(ΩSrSO4 = 7.70 × 10−29 m3), leading to lower J(x). Thus, the maxi-mum nucleation rate J(x) is a balanced outcome of changes in SI(x),σ(x), and Ω(x) as functions of x. Using the local ion concentrations

for heterogeneous (Bax, Sr1−x)SO4 nucleation from Ba50Sr50,Ba30Sr70, Ba10Sr90, Ba05Sr95, and Ba01Sr99 solutions (cal-culated with 1-nm water film assumption as discussed before),J(x) was calculated to reach a maximum at x = 0.92, 0.82, 0.34,0.12, and 0.014, respectively (Fig. 4B). The calculated x valueswith maximum J(x) matched the measured x values (Fig. 5B andSI Appendix, Table S2) in general, indicating that Sr enrichmentin the heterogeneous precipitates, in which nucleation is thedominant process, is associated with (Bax, Sr1−x)SO4 solid solutionswith the highest nucleation rate. As shown in SI Appendix, Table S2,Sr mole fractions in the solid solution formed on organics fromdifferent solution conditions could be higher or lower than 50%,but all are much higher than the Sr mole fractions in the homog-enous precipitates formed under the same solution conditions. So,the phrase “Sr-rich barite” was used here to describe the hetero-geneous precipitates on organics as opposed to “Sr-poor barite”formed in the bulk solution. It does not mean that the mole fractionof Sr in heterogeneous barite was always higher than 50%.Studies on the thermodynamics of (Bax, Sr1−x)SO4 solid solutions

are still under active development (37), and one of the uncertaintiesis the interaction parameter of Ba–Sr ions in (Bax, Sr1−x)SO4 solidsolutions, which determines the ideality (SI Appendix, section 1.3)and the free energy of mixing (SI Appendix, section 1.4) of the solidsolution (35–39). Using the interaction parameters developed bydifferent models (35, 36), the calculated compositions (i.e., x) variedto some extent but were all in general consistent with the measuredcompositions of (Bax, Sr1−x)SO4.Here, we assumed that the bulk solution and interfacial fluid

differed only in ion concentrations due to cation enrichment,which promoted the heterogeneous nucleation of Sr-enrichedbarite. However, there are many other differences between theinterfacial environments and bulk fluid environments (e.g., waterstructure and dynamics, the presence of a surface contributing tothe interfacial energy, etc.) (14). However, measuring the waterstructure and dynamic at the organic–water interfaces is tech-nically difficult. Here, the interfacial energies of end members ashomogeneous nuclei in bulk solution (σ in Eq. 2) were used inthe calculations. For heterogeneous nucleation of a sphericalparticle on a flat surface, the effective interfacial energies wouldbe determined by not only the liquid-nucleus interfacial energybut also the contact angle (θ) of the nuclei on the flat surface (40,41). Detailed calculations of compositional (i.e., x) dependenceof (Bax, Sr1−x)SO4 heterogeneous nuclei on contact angle (θ) areavailable in SI Appendix, section 1.5, Fig. S4, and Table S3. Based

Fig. 4. (A) Calculations of optimum x values (varied over the range of0.99500–0.99995 in SI Appendix, Table S2, shown clearly in the Inset of A) ofhomogeneous precipitates with the highest solid solution SI(x) under bulksolution conditions (Table 1) according to Eq. 1. (B) Calculations of optimumx values (varied over the range of 0.01–0.92, SI Appendix, Table S2) of het-erogeneous precipitates with the highest solid-solution nucleation rate [J(x)]under local solution conditions according to Eq. 2.

Fig. 3. Schematic showing cation enrichment at organic-water interfaces,which altered the solution saturation index (SI), as well as the correlations ofnucleation (red line) and growth (black line) rates with SI (32). Under slightlysupersaturated bulk solutions (Table 1, Ba50Sr50, bulk SI = 1.53), growth isthe dominant precipitation process (stage I); thus, barite precipitates in bulksolution (SEM image, Left Inset) grow fast to micrometer-sized crystals,which are Sr-poor [(Ba0.96Sr0.04)SO4]; stage II indicates a nucleation-dominatedprecipitation process at a highly supersaturated solution, which can beachieved near organic–water interfaces due to significant cation enrichment.The STEM image (Right Inset) shows the nanometer-sized, Sr-rich heteroge-neous precipitates [(Ba0.72Sr0.28)SO4] formed on organic films from Ba50Sr50solution (local SI = 8.10).

13224 | www.pnas.org/cgi/doi/10.1073/pnas.1821065116 Deng et al.

Dow

nloa

ded

by g

uest

on

May

20,

202

0

on the calculations, using liquid–nucleus interfacial energy torepresent the heterogeneous interfacial energies, i.e., assumingθ = 180° as homogeneous nucleation, would result in the upperlimit of calculated x values. With lower contact angles (θ = 45,90, and 135°) between the nuclei and the substrates for hetero-geneous nucleation, the calculated x values were lower andmatched better with the measured compositions. Therefore, althoughthe exact contact angle (θ) was unknown, it would not affect theconclusion that the measured compositions of the heterogeneousprecipitates on organics matched with the calculated composi-tions correlating with the fastest nucleation rates on organics.The robustness of the conclusion was further validated by varying

the thickness of the water film near the organic films (31), whichwere assumed to be 1, 2, 5, or 10 nm. The dependence of SI(x)and J(x) on the different thickness of water film was calculatedusing the local ion concentrations vs. water film thicknesses (SIAppendix, Table S4). For all assumed values of the water filmthickness, we found that our conclusion was valid: The compositionsof the heterogeneous precipitates (SI Appendix, Table S4) were ingeneral consistent with (Bax, Sr1−x)SO4 solid solution having thefastest nucleation rates [J(x)] given the local solution conditionsnear the organic films (SI Appendix, section 1.6 and Table S4).

ConclusionsThis study provides direct observation of continuous nucleationof Sr-rich (Bax, Sr1−x)SO4 from undersaturated bulk solution andthe formation of barites with Sr levels that are thermodynami-cally unfavorable. Organics films are found to enrich cationsfrom bulk to the local solution near the films. This enrichmentincreases the local supersaturation and promotes formation ofSr-rich barite nuclei. Therefore, the two paradoxes of marinebarite formation are resolved. Furthermore, our findings uncoverthe role of organics in controlling the local supersaturation level,which is found to greatly affect the balance between nucleationand growth rates and thus the distinct sizes and compositions ofthe heterogeneous precipitates on organic films and the homo-geneous precipitates in bulk solutions.These findings have a number of implications that can advance

discoveries in the general field of solid-solution precipitation inmany natural and engineered settings. First, the general consis-tency between measured compositions and those predictedusing solid-solution nucleation and growth rates suggest thatmolar fractions of Sr in barite could serve as a signature reflecting

the local paleoenvironments of (Bax, Sr1−x)SO4 formation. Second,this knowledge can also be used for predicting and controlling toxicstrontium removal in industrial processes. Third, the roles of theorganic films discovered here in regulating the sizes and composi-tions of the precipitates may allow better understanding of manybiomineralization processes. Finally, this understanding suggestsapproaches to material synthesis for achieving tunable sizes and fortailoring impurity levels by controlling the balance between solid-solution nucleation and growth rates.

Materials and MethodsPreparation of Organic Coatings on Glass. Glass slides (Superfrost) were cutinto small pieces (1 cm × 1 cm) and cleaned. The glass samples were im-mersed in toluene for 24 h, with 10 wt% (3-triethoxysilyl)propyl succinic acid(Gelest) and 10 wt% (3-mercaptopropyl)trimethoxysilane (Sigma-Aldrich)added as –SH and –COOH precursors. The silanol head groups of the precursorchemicals bond with glass samples, leaving the –SH and –COOH functionalgroups exposed in solution. Mixed thiol and carboxylic acid (–SH-and-COOH)functional groups were used here as proxies for organic-rich microenviron-ments, as these functionalized groups are abundant in soil, groundwater,and marine water (12, 42). The coated glasses were rinsed with acetone toremove loosely attached organics. Characterization of –SH-and-COOH coat-ings on the glass substrate was conducted in our previous study (24). Also, atomicforce microscopy measurements were conducted on the SH-and-COOH–coatedsubstrate, and the phase image (SI Appendix, Fig. S5) showed that the surfacewas fully coated without the presence of steps or hillocks.

Solution Conditions for (Bax, Sr1−x)SO4 Precipitation and Ion AdsorptionExperiments. Barium chloride (BaCl2), strontium chloride (SrCl2), and so-dium sulfate (Na2SO4) powders were used to prepare solutions listed inTable 1. For all precipitation experiments, BaCl2 and SrCl2 solutions weremixed first, and precipitation started right after mixing with Na2SO4 solution.The total cation concentrations ([Ba2+] + [Sr2+]) were fixed as 0.1 mM withvarying [Ba2+]/[Sr2+] ratios (Table 1, Ba100Sr0, Ba50Sr50, Ba30Sr70, Ba10Sr90,Ba05Sr95, and Ba01Sr99 solutions), and the concentration of Na2SO4 was0.1 mM, so the total cation/anion ratio was 1 in all solutions. The initial solutions’ionic strength, pH, and SIs with respect to BaSO4 and SrSO4 end members werecalculated using spec E8 database in GWB, release 9.0 (Aqueous Solutions).Batch adsorption/desorption experiments were also conducted to investigateBa2+, Sr2+, and SO4

2− ion adsorption on –SH-and-COOH–coated substratesunder those solution conditions (details in SI Appendix, section 2.1).

In Situ SAXS and GISAXS Measurements. In situ observations of homogeneous(Bax, Sr1−x)SO4 precipitation in solution and heterogeneous (Bax, Sr1−x)SO4

precipitation on bare and organic-coated SiO2 substrates were performedwith SAXS and GISAXS measurements at Beamline 12 ID-B of the AdvancedPhoton Source (APS), Argonne National Laboratory (Argonne, IL). The en-ergy of the incident X-ray was 14 keV. Accordingly, an incident angle of0.10° was selected to achieve >99% reflectivity of the incident X-rays, so onlyscattering generated from nanoparticles at the substrate surfaces wereprobed (43). Before each measurement, a piece of bare or coated SiO2

substrate was placed into a homemade GISAXS cell (24, 25), with the surfacealigned vertically with the center of incident X-ray beam. After alignment,0.7 mL of freshly mixed solution (Table 1) was injected into the homemadecell and the in situ measurement started. Extreme care was taken in theoperation during solution injection for each set of GISAXS measurement, toavoid bubble formation in solution and on walls of the GISAXS cell. GISAXSmeasurements were recorded every 2 min with an exposure time of 20 s anda sleeping time of 100 s for 1 h. For SAXS measurements, a piece of bare SiO2

substrate was placed into a homemade GISAXS cell, and the incident X-raybeam was aligned to be 5 mm above the substrate surface. SAXS measure-ments were also recorded every 2 min with an exposure time of 20 s and asleeping time of 100 s for 1 h. Preliminary tests showed no heterogeneousprecipitation or ion adsorption occurred at the bare SiO2 substrate; thus, thepresence of bare SiO2 substrate does not alter the solution conditions. ForSAXS measurement, the sizes of homogeneous (Bax, Sr1−x)SO4 precipitateswere larger than the SAXS detectable range (1–50 nm) (22, 25), so SEM wasemployed to characterize the morphology of the homogeneous precipitatesat the end of the 1-h precipitation experiments.

For GISAXS data analysis, the first 2D scattering image was used asbackground (SI Appendix, Fig. S6) and was subtracted from later images, tosubtract the scattering caused by solution and the substrates. Then thesubtracted 2D images were converted to 1D scattering curves of intensity (I)plotted vs. scattering vector (q) (44). With the assumption of noninteracting

Fig. 5. Compositions (x) of homogeneous (A) and heterogeneous (Bax, Sr1−x)SO4

(B) precipitates measured by ICP-MS, XPS, and SEM/STEM-EDX, plottedagainst the molar percentages of aqueous Ba2+ ions over the total cations[xBa2+ , Ba

2+/(Ba2+ + Sr2+)] in the initial solutions (Table 1). Detectable ho-mogeneous precipitates only formed from solutions Ba100Sr0 and Ba50Sr50(Table 1). The limited Sr incorporation (i.e., x ∼ 1) in the homogeneous (Bax,Sr1−x)SO4 precipitates (A) agrees with the calculated x value associated withthe maximum SI(x) under bulk solution condition. The measured composi-tions (i.e., x) of heterogeneous (Bax, Sr1−x)SO4 precipitates (B) agree with thecalculated x values associated with the maximum J(x) under local solutionconditions.

Deng et al. PNAS | July 2, 2019 | vol. 116 | no. 27 | 13225

EART

H,A

TMOSP

HER

IC,

ANDPL

ANET

ARY

SCIENCE

SEN

VIRONMEN

TAL

SCIENCE

SSE

ECO

MMEN

TARY

Dow

nloa

ded

by g

uest

on

May

20,

202

0

spherical nanoparticles with lognormal distributions, the average particlesizes were calculated by fitting the 1D scattering intensity curves. All dataanalysis was performed with Igor Pro-6.34 (WaveMetrics), and more detaileddata analysis can be found in our previous publications (19–26).

Ex Situ Characterization of Homogeneous (Bax, Sr1−x)SO4 Precipitates in Solution.To collect sufficient amounts of homogeneous precipitates for analysis, par-ticles precipitated in an hour from 1 L of solutions (Table 1) were collectedusing centrifugal filters (Millipore; 100k) and were transferred to a clean glassslide for characterization. The morphology of homogeneous (Bax, Sr1−x)SO4

was characterized using SEM (JEOL 6010 LA). The compositions of homoge-neous precipitates were measured by EDX, XPS (SI Appendix, section 2.2), andICP-MS (Perkin Sciex Elan Drc II). For ICP-MS measurements, the collectedprecipitates on filters were dissolved in 4 mL of 2% HNO3 overnight, and thedissolved Ba2+ and Sr2+ concentrations were measured.

Ex Situ Characterization of Heterogeneous (Bax, Sr1−x)SO4 Precipitates on Substrates.At the end of the 1-h precipitation experiments, the organic-coated SiO2 substratewas rinsed quickly with ultrapure water to remove the residue solution and

was dried with nitrogen gas gently. Morphologies of the heterogeneousprecipitates on organic film were measured using STEM (Hitachi HF-3300).Compositions of the precipitates were measured using XPS, STEM-EDX (SIAppendix, section 2.3), and ICP-MS. For ICP-MS measurements, the precipi-tates formed on substrates were dissolved in 4 mL of 2% HNO3 overnight, andthe dissolved Ba and Sr concentrations were quantified using ICP-MS. Thesuccessful dissolution of heterogeneous precipitates was discussed in SI Ap-pendix, section 2.4.

ACKNOWLEDGMENTS. We greatly appreciate Dr. James D. Kubicki’s helpwith the local solution condition estimation. We also thank Dr. Bo Chenfor operating the XPS instrument at Rice University, and Dr. Xiaobing Zuofor valuable discussion on SAXS/GISAXS experiments and data analysis atbeamline 12 ID-B. Use of the facilities at beamlines sector 12 ID-B at APSwas supported by the US Department of Energy (DOE), Office of Science,Office of Basic Energy Science, under Contract DE-AC02-06CH11357. STEM/EDX characterization was conducted at the Center for Nanophase Materials,a DOE user facility. This work was supported by the US DOE, Office of Science,Office of Basic Energy Sciences, Chemical Sciences, Geosciences, and BiosciencesDivision.

1. J. McManus et al., Geochemistry of barium in marine sediments: Implications for itsuse as a paleoproxy. Geochim. Cosmochim. Acta 62, 3453–3473 (1998).

2. A. Paytan, M. Kastner, E. Martin, J. Macdougall, T. Herbert, Marine barite as a monitorof seawater strontium isotope composition. Nature 366, 445–449 (1993).

3. C. Monnin, C. Jeandel, T. Cattaldo, F. Dehairs, The marine barite saturation state ofthe world’s oceans. Mar. Chem. 65, 253–261 (1999).

4. M. A. Bertram, J. P. Cowen, Morphological and compositional evidence for bioticprecipitation of marine barite. J. Mar. Res. 55, 577–593 (1997).

5. M. T. Gonzalez-Muñoz, F. Martinez-Ruiz, F. Morcillo, J. D. Martin-Ramos, A. Paytan,Precipitation of barite by marine bacteria: A possible mechanism for marine bariteformation. Geology 40, 675–678 (2012).

6. E. M. Griffith, A. Paytan, Barite in the ocean—occurrence, geochemistry andpalaeoceanographic applications. Sedimentology 59, 1817–1835 (2012).

7. T. J. Horner et al., Pelagic barite precipitation at micromolar ambient sulfate. Nat.Commun. 8, 1342 (2017).

8. D. M. Singer, E. M. Griffith, J. M. Senko, K. Fitzgibbon, I. H. Widanagamage, Celestinein a sulfidic spring barite deposit—a potential biomarker? Chem. Geol. 440, 15–25(2016).

9. F. Dehairs, R. Chesselet, J. Jedwab, Discrete suspended particles of barite and thebarium cycle in the open ocean. Earth Planet. Sci. Lett. 49, 528–550 (1980).

10. P. Van Beek, J.-L. Reyss, P. Bonte, S. Schmidt, Sr/Ba in barite: A proxy of barite pres-ervation in marine sediments? Mar. Geol. 199, 205–220 (2003).

11. F. Martinez-Ruiz et al., Barium bioaccumulation by bacterial biofilms and implicationsfor Ba cycling and use of Ba proxies. Nat. Commun. 9, 1619 (2018).

12. R. Keren et al., Sponge-associated bacteria mineralize arsenic and barium on in-tracellular vesicles. Nat. Commun. 8, 14393 (2017).

13. D. Wang, A. F. Wallace, J. J. De Yoreo, P. M. Dove, Carboxylated molecules regulatemagnesium content of amorphous calcium carbonates during calcification. Proc. Natl.Acad. Sci. U.S.A. 106, 21511–21516 (2009).

14. O. Branson et al., Nanometer-scale chemistry of a calcite biomineralization template:Implications for skeletal composition and nucleation. Proc. Natl. Acad. Sci. U.S.A. 113,12934–12939 (2016).

15. P. J. Smeets, K. R. Cho, R. G. Kempen, N. A. Sommerdijk, J. J. De Yoreo, Calciumcarbonate nucleation driven by ion binding in a biomimetic matrix revealed by in situelectron microscopy. Nat. Mater. 14, 394–399 (2015).

16. N. Vidavsky et al., Initial stages of calcium uptake and mineral deposition in sea urchinembryos. Proc. Natl. Acad. Sci. U.S.A. 111, 39–44 (2014).

17. P. Gilbert, M. Abrecht, B. H. Frazer, The organic-mineral interface in biominerals. Rev.Mineral. Geochem. 59, 157–185 (2005).

18. T. Li, A. J. Senesi, B. Lee, Small angle X-ray scattering for nanoparticle research. Chem.Rev. 116, 11128–11180 (2016).

19. Y. Hu, Q. Li, B. Lee, Y.-S. Jun, Aluminum affects heterogeneous Fe(III) (Hydr)oxidenucleation, growth, and ostwald ripening. Environ. Sci. Technol. 48, 299–306 (2014).

20. Y. Hu, C. Neil, B. Lee, Y.-S. Jun, Control of heterogeneous Fe(III) (hydr)oxide nucle-ation and growth by interfacial energies and local saturations. Environ. Sci. Technol.47, 9198–9206 (2013).

21. Y. Hu, B. Lee, C. Bell, Y.-S. Jun, Environmentally abundant anions influence the nu-cleation, growth, Ostwald ripening, and aggregation of hydrous Fe(III) oxides.Langmuir 28, 7737–7746 (2012).

22. C. Dai, Y. Hu, Fe(III) hydroxide nucleation and growth on quartz in the presence ofCu(II), Pb(II), and Cr(III): Metal hydrolysis and adsorption. Environ. Sci. Technol. 49,292–300 (2015).

23. C. Dai, M. Lin, Y. Hu, Heterogeneous Ni-and Cd-bearing ferrihydrite precipitation andrecrystallization on quartz under acidic pH condition. ACS Earth Space Chem. 1, 621–628 (2017).

24. C. Dai et al., Heterogeneous nucleation and growth of barium sulfate at organic-water interfaces: Interplay between surface hydrophobicity and Ba2+ adsorption.Langmuir 32, 5277–5284 (2016).

25. C. Dai, X. Zuo, B. Cao, Y. Hu, Homogeneous and heterogeneous (Fex, Cr1-x)(OH)3precipitation: Implications for Cr sequestration. Environ. Sci. Technol. 50, 1741–1749

(2016).26. C. Dai, J. Liu, Y. Hu, Impurity-bearing ferrihydrite nanoparticle precipitation/de-

position on quartz and corundum. Environ. Sci. Nano 5, 141–149 (2018).27. C. Monnin, D. Cividini, The saturation state of the world’s ocean with respect to

(Ba,Sr)SO4 solid solutions. Geochim. Cosmochim. Acta 70, 3290–3298 (2006).28. I. H. Widanagamage, E. A. Schauble, H. D. Scher, E. M. Griffith, Stable strontium

isotope fractionation in synthetic barite. Geochim. Cosmochim. Acta 147, 58–75

(2014).29. L. Li, A. J. Fijneman, J. A. Kaandorp, J. Aizenberg, W. L. Noorduin, Directed nucleation

and growth by balancing local supersaturation and substrate/nucleus lattice mismatch.

Proc. Natl. Acad. Sci. U.S.A. 115, 3575–3580 (2018).30. R. Rahnemaie, T. Hiemstra, W. H. van Riemsdijk, Inner- and outer-sphere complexa-

tion of ions at the goethite-solution interface. J. Colloid Interface Sci. 297, 379–388

(2006).31. E. Mamontov et al., Dynamics and structure of hydration water on rutile and cassit-

erite nanopowders studied by quasielastic neutron scattering and molecular dynamics

simulations. J. Phys. Chem. C 111, 4328–4341 (2007).32. M. Haruta, B. Delmon, Preparation of homodisperse solids. J. Chim. Phys. 83, 859–868

(1986).33. G. Cao, Y. Wang, Nanostructures and Nanomaterials: Synthesis, Properties, and

Applications (Imperial College Press, 2004), pp. 57–58.34. C. Pina, A. Putnis, The kinetics of nucleation of solid solutions from aqueous solutions:

A new model for calculating non-equilibrium distribution coefficients. Geochim.

Cosmochim. Acta 66, 185–192 (2002).35. M. Klinkenberg et al., The solid solution–aqueous solution system (Sr, Ba, Ra)SO4

+ H2O: A combined experimental and theoretical study of phase equilibria at Sr-rich

compositions. Chem. Geol. 497, 1–17 (2018).36. V. Vinograd et al., Thermodynamics of the solid solution–aqueous solution system

(Ba, Sr, Ra)SO4 + H2O: I. The effect of strontium content on radium uptake by barite.

Appl. Geochem. 89, 59–74 (2018).37. F. Heberling, D. Schild, D. Degering, T. Schäfer, How well suited are current thermodynamic

models to predict or interpret the composition of (Ba, Sr)SO4 solid-solutions in geo-

thermal scalings? Geotherm. Energy 5, 9 (2017).38. M. Prieto, Thermodynamics of solid solution–aqueous solution systems. Rev. Mineral.

Geochem. 70, 47–85 (2009).39. J. Weber et al., Unraveling the effects of strontium incorporation on barite growth—

In situ and ex situ observations using multiscale chemical imaging. Cryst. Growth Des.

18, 5521–5533 (2018).40. C. Dai et al., Heterogeneous lead phosphate nucleation at organic–Water interfaces:

Implications for lead immobilization. ACS Earth Space Chem. 2, 869–877 (2018).41. A. Fernandez-Martinez, Y. Hu, B. Lee, Y. S. Jun, G. A. Waychunas, In situ determi-

nation of interfacial energies between heterogeneously nucleated CaCO3 and quartz

substrates: Thermodynamics of CO2 mineral trapping. Environ. Sci. Technol. 47, 102–

109 (2013).42. B. Rao, C. Simpson, H. Lin, L. Liang, B. Gu, Determination of thiol functional groups on

bacteria and natural organic matter in environmental systems. Talanta 119, 240–247

(2014).43. B. L. Henke, E. M. Gullikson, J. C. Davis, X-ray interactions: Photoabsorption, scattering,

transmission, and reflection at E = 50–30,000 eV, Z = 1–92. At. Data Nucl. Data Tables

54, 181–342 (1993).44. B. Lee et al., Anomalous grazing incidence small-angle x-ray scattering studies of

platinum nanoparticles formed by cluster deposition. J. Chem. Phys. 123, 074701

(2005).

13226 | www.pnas.org/cgi/doi/10.1073/pnas.1821065116 Deng et al.

Dow

nloa

ded

by g

uest

on

May

20,

202

0