inorganic interface - American Chemical

May 17, 1993 - Box 8, The Heath, Runcorn Cheshire WA7 4QD, U.K.. Brigid R. Heywood. Department of Chemistry and Applied Chemistry, University of ...
0 downloads 0 Views 3MB Size
3594

Langmuir 1993,9, 3594-3599

Interactions at the Organic/Inorganic Interface: Molecular Design of Crystallization Inhibitors for Barite Lindsay A. Bromley, Denis Cottier, Roger J. Davey,* Brian Dobbs, and Simon Smith Crystal Chemistry Team, Research and Technology Department, ZENECA Fine Chemicals Manufacturing Organisation, P.O. Box 8, The Heath, Runcorn Cheshire WA7 4QD, U.K.

Brigid R. Heywood Department of Chemistry and Applied Chemistry, University of Salford, Salford M5 4WT, U.K. Received May 17, 1993. In Final Form: September 14, 199P The effect of additives on synthetic barite crystals, prepared under conditions chosen to mimic those encountered in off-shore oil fields, has been extensively studied. The addition of certain stereochemically related diphosphonatesproduces a characteristicmorphologychange which has been shown to be consistent with a binding model in which the diphosphonate ion replaces two sulfate ions in the (011) surface. A detailed understanding of the recognition processes involved has now led to the rational design of new surface-activemoleculeswith greatly improved efficacy, and a greater appreciationof the factorsinfluencing morphological changes resulting from inhibitor interference.

Introduction The addition of specific molecules to crystallizing systems is well known as a means of inhibiting crystallization processes and modifying the morphology of crystalline products.' The structural basis for such effects has received attention over a number of years since the early studies on inorganic materials by Whetstone2 and more recently includes extensive studies on molecular crystal^.^ As a result of these and other reports on biologically related material^,^ factors controlling recognition at crystal surfaces are now well appreciated. In a previous paper,5 the specific case of the interaction of simple phosphonates with surfaces of barium sulfate was considered and the specific recognition factors leading to morphological changes were elucidated. This paper extends this work and shows how such data may be used in conjunction with kinetic measurements to direct the design of more powerful crystallization inhibitors for the barium sulfate system.6 Previous Studies In previous work5 the experimental morphology of barium sulfate crystals prepared from synthetic sea and formation waters was found to be simple (001) rhombic plates bound by (210) faces. It was discovered that addition of molecules containing two phosphonic acid groups linked by a three-atom chain, such as (iminodimethy1ene)diphosphonic acid, caused a significant change in morphology. Combining morphological prediction with *Abstract published in Advance ACS Abstracts, November 15, 1993. (1) Davey,R. J.;Polywka,L.A.; Maginn,S.J. In AdvancesinIndustrial Crystallisation; Garside, J.; Davey, R. J.; Jones, A. G. Eds.; Butterworth-Heinemann: London, 1991; p 150. (2) Whetstone, J. Discuss. Faraday SOC.1954,16, 132. (3) Weissbuch, I.; Addadi, L.; Lahav, M.; Leiserowitz,L. Science 1991, 253, 637. (4) Mann,S.;Didymus,J.; Sanderson, N. P.; Heywood, B.; Aso-Samper, E. J. J. Chem. SOC.,Faraday Trans. 1990,86, 1873. ( 5 ) Black, S. N.; Bromley, L. A.; Cottier, D.; Davey, R. J.; Dobbs, B.; Rout, J. E. J. Chem. SOC.,Faraday Trans. 1991,87,3409. (6) Davey, R. J.; Black, S. N.; Bromley, L. A,; Cottier, D.; Dobbs, B.; Rout, J. R. Nature (London) 1991,353, 549.

Q743-7463/93J2409-3594$04.001Q 0

experimental and analytical results showed that the first new faces to appear were (Oll}. This result indicated that the binding site for the diphosphonate must be located on the (011)surface, which is built up of alternate layers of barium and sulfate ions. The binding site was identified as any position containing two like-oriented sulfate ions,which had an S-S separation of 5.6 8, (Figure 1). This corresponded well to the lowenergy conformations of the active diphosphonate molecules in their deprotonated states. Observations of the additive efficiency at varying pH levels showed that only the doubly charged anions were recognized by the growing surfaces. It was also shown that this binding site lies not only in the (011) plane but also in the (210) plane.5 This has consequences for the effect of increasing additive concentrations on the crystal morphology. In previous reports in which morphological changes have been used to indicate specific interactions between the surface and additive, it has been accepted that the morphological importance of the affected face willincreasewith increasingconcentration of the a d d i t i ~ e . ~In? ~ the barium sulfate/diphosphonate system this is not the case;the (011)face appears and then disappears as the additive concentration increases.6 This is not surprising if the influence of the relative growth rates of the (0011, (2101, and (011) faces on the overall morphology is considered. For example, a ratio of 1:5:6.8yields a (001) rhombus of appropriate aspect ratio and with (011) faces absent. This is shown in Figure 2a and can be compared to the experimental morphologies of Figure 1,in ref 5, and Figure 5a, in this paper. Reduction of the (011) face growth rate to a relative value of 6 then introduces this face (Figure 2b). If the additive were to affect this face alone, then further reductions in its growth rate would yield hexagonal crystals with (011) faces of increased size as shown in Figure 2c. Since, however, the phosphonate also lies on the (210)face, it seems reasonable to suppose that the growth rate of this face will also be depressed. Reduction of the growth rate of the (210) face to a relative value of 4.4 has the effect (Figure 2d) of reducing the size of the (011) face; further reductions remove it completely. Hence, the experimentally observed 1993 American Chemical Society

Interactions at the Organicllnorganic Interface

Langmuir, Vol. 9, No. 12,1993 3595

Figure 1. View of the (011)surface showing the diphosphonatebinding site and relevant intersite separations. NrNn= 8.95 A; NpNs = 9.00 A.

Q 10011

Q

...

1011)

,_I.

1210)

a

b

(JQ Ion)

j:

c

d

Figure2. BaSOIcrystalmorphologieswithrelativegrowthrates (001):(’21ok(o11)= (a) k56.8,(h) 1:56,(C) k54,and (d) k4.44.

morphologicalchangesscanbeseenastheresultofchanges in the relative (011)/(210) face growth rates as the diphosphonate concentration increases. Structurally,this behavior may be expected for two reasons. the layer structure of the (011) face is likely to make it more accessible to the additive at low concentrations since for this surface the steric hindrance of the in-plane hydrated cations does not have to he Overcome as it does for the (210) face. Secondly, on the (210) face there are two binding sites per 100 A*, while on the (011) face this falls to 1.4, suggestingthat even when the more accessible (011)

face is saturated with diphosphonate, further adsorption can occur on the (210) face and reduce ita growth rate further. Thus,asdemonstrated, it is pwiblefor seledivebmding at a specific site to lead to morphological changes which depend sensitively on the relative growth rates of faces and hence, as observed, on the concentration of the additive. In the study described here, we utilize this knowledge of the binding site and morphological data together with molecular modeling to aid the process of designing novel inhibitors with superior properties. It is alsoshown how complimentary kinetic measurements can to assess the relative effectiveness of a series of such be used inhibitors. The above understanding was used to formulate a strategy for molecular design. This was based on the premise that molecules containing more than one active motif, capable of simultaneously binding to several sites, wouldhaveenhanced bindingenergiesandofferasuperior steric harrier. them more effectiveinhibitors than molecules containing only a single motif. The graphical molecular package Chem-XT was used to visualize the (011)surface in order to gain undenrtanding interadions at this organic/~organic interface. The surface contains many activesites, related by the symmetry of the structure (Figure 1). According the design outlined above, any molecule conbini,,gtwo (iminodimethy~ene)djp~osp~on~~ groups should he more effective than (iminodimethy1ene)diphos(7) Chon-X mol& madeling suite, developed and distributed by Chemical Design ~td.,Oxford.

3696 Langmuir, Vol. 9, No.12, 1993

t a')-

*\e A 1

t

Bromley et al.

a.

. . I t&...* v I +--* 6 d i'

a.

e i

..

'I 1

A

e'

4

Figure 3. Representation of additive 5 occupying two adjacent binding sites on the (011) surface.

phonate itself-provided that both of the active groups can simultaneously accem the same surface. It was necessary, therefore, to consider how the two active motifs could be linked together and to understand the requirements of this link. Structural analysis of the (011)face revealed that the most efficient design for new molecules was one that allowed the two active groups to access a separation of ca. 9.0 A (Figure 1). Further modeling revealed that links at least 7.4 A long would allow this, (Here the length of the link is defined as the tion between the two terminalcarbon atoms bonded to th two nitrogen atoms of the active groups.) The minimum link length represents the situation where the link would be fully extended for the activegroups to access two adjacent sites. A slightly longer link, e.g., 8.5A,would allowfor some minimization of steric interactions because of the extra flexibility, and so these molecules might be more effective. Figure 3 demonstrates these ideas, for a 10-carbon link, showing how the alkyl chain can adopt a low-energy conformation above the surface, hence enhancing the performance of the additive as a growth inhibitor. It is conceivable that there is also a maximum link length above which a combination of solubility and conformational factors act to reduce efficacy. This is not easily determined by molecular modeling but could be assessed using experimental data. Themoleculestobe synthesized,therefore,had tosatisfy the following requirements: (1)the twoactivegroups must be able to access a separation of 9.0 A and (2) both groups must be able to easily access the same surface simultaneously. These criteria had to be satisfied regardless of

sepT

whether the intergrouplink was rigid or flexible. Molecular modeling was then used to identify suitable candidate molecules. A series of possible intergroup links were constructed, and only those satisfying the above criteria were selected. Then retrosyntheticanalysis was performed on the potential additives that contained these intergroup links. Only those compoundswith feasible syntheticroutes were considc-ed for synthesis;some examples are displayed in Figure 4. These design criteria relate specifically to binding on the (011) face. As noted earlier, however, these molecules are also likely to influence the (210) face. The separation between adjacent binding sites on this surface is ca. 7.1 A. This suggests that molecules designed to simultaneously access two adjacent sites on the (011) face may also bind to the (210) face in a similar way. It is further evident that the more complex geometry of the sulfate array on the (21.0) face compared to the (011) face may allow these molecules to access alternative binding sites and hence further enhance their potency. Experimental Section Synthesis. Thecompoundsidentifiedby theabove work were synthesized from their respective diamines following the Mannich-typereaction reported by Moedritzerand Iran.8 The general procedure for this reaction was as follows: H,NRNH, + 4H,PO, + 4HCl-b 8HCHO

-

(H20,PH,C),NRN(CH,P0,H,), R repreRontR a flexible link. A wide variety of chemicals were wed. For hydrocarbon chains the etarting diamines were (8) Moedritzor, K.; Irani,

R J . Org. Chem. ISM, 32, 1603.

Interactions at the Organicllnorganic Interface

Figure 4. Examples of additives synthesized.

Langmuir, Vol. 9, No. 12,1993 3597 (ii) Data for (H~O~PH~C)~N(CHI)IN(CE~O~H*)~,~: yield 69.5 g, 72% (based on 0.19 mol of diamine); isolated as a colorless oil; lH NMR 6 3.50 (8H, d, PCHZN),3.30 (4H, m, NCHz), 1.66 (4H, m, NCHZCHZ),1.25 (6H, m, (CHZ)~). (iii)Data for (HrOgHtC)zN(CHi)eN(CHQO,H,)e 3 yield 35.9 g, 40% (based on 0.17 mol of diamine); isolated as a white crystalline solid; 'H NMR 6 3.35 (8H, d, PCHsN), 3.30 (4H, m, NCHZ),1.60 (4H, m, NCHZCHZ),1.25 (8H, m, (CHd4). (iv) Data for (HtOsPH&)2N(CHz)JV(CHQOIHl)t,4 yield 58.9 g, 69% (based on 0.16 mol of diamine); isolated as a pale cream crystalline solid; lH NMR 6 3.45 (8H, d, PCHzN), 3.35 (4H,m, NCH32), 1.65 (4H,m, NCHZCHZ), 1.25 (lOH, m, (CHds). (v) Data for (HaOgH~)2N(CHa)iaN(CHQO,Hl)r,6 yield 50.8 g, 64% (based on 0.15 mol of diamine); isolated as a white crystalline solid; lH NMR 6 3.60 (12H, m, PCHzN, NCHz), 1.90 (4H, m, NCHzCHz), 1.40 (12H, m, (CHZ)~). (vi) Data for (H2OsPH&)a(CH2)&J(CHIPO,Hl)r, 6 yield 77.4 g, 60% (based on 0.23 mol of diamine); isolated as a white crystalliie solid; 'H NMR 6 3.55 (4H, s, NCHz), 3.50 (8H, d, PCHzN), 1.85 (4H, m, NCHZCHZ),1.45 (8H, m, ~CHZCHZ), 1.40 (4H, m, (CHd4). Data for (HaOQHIC)zNN(CHa)rO(CHa)~(CH~O~a)& 7: yield 130 g, 71% (based on 0.35 mol of diamine); isolated as a pale yellow oil; 'H NMR 6 3.77 (4H, m, NCH&H,O), 3.60 (8H, m, NCHZ,OCHZ),3.52 (8H, d, PCHZN). (viii) Data for (H2OQH#)rN(CH&O(CHa)aO(CH&NN(CHQOSH~)~, 8 yield55.9g,68% (basedon0.15molofdiamine); isolated as a pale yellow oil; lH NMR 6 3.55 (8H, d, PCHzN), 3.45 (12H, m, NCHz, OCHZ),2.00 (4H, d, CHZ). (is) Data for (HaOSHIC)2N(CHa)~(CHICHa)rN(CH,),N(CHzPOsH&, 9: yield 49.3 g, 86% (based on 0.10 mol of diamine); isolated as a red brown oil; 'H NMR 6 3.55 (8H, m, N(CHZCHZ)ZN), 3.50 (8H, d, PCHZN),3.35 (4H, m, NCHd, 3.20 (4H, m, NCHz), 2.15 (4H, m, CHz). Testing. The ability of these molecules to inhibit B&04 precipitation and modify crystal morphologies was tested using essentially the same experimental procedure described in ref 5. After mixing the formation and sea waters, the mix was stirred at 70 OC for 4 h. A sample was then removed and filtered, and a 10-mLaliquot was pipetted into 10mL of 1% aqueous ammonia solution/l % EDTA. The residual BaW concentration was determined using ICP spectrometry. Experimenta with additives were performed in a manner identical to that of the controls with the additives dissolved in the sea water solution to yield final concentrations in the range 0.0006-0.96 mmol dm". All experiments were performed at 70 OC.

purchased from Aldrich. For hydrocarbon chains containing heteroatoms, 1,2-bis(2-aminoethoxy)ethaneand 1,2-bis(3-aminopropoxy)ethane were obtained from Fluka and Fluorochem Results and Discussion was Ltd., respectively, while N,"-(3-aminopropyl)piperazine Morphological Changes. As previously discussed: obtained from Aldrich. the experimental morphology of barium sulfate crystals The diamine (0.25 mol) was dissolved in an aqueous solution prepared from synthetic sea and formation waters is simple (500 mL of water) of orthophosphorus acid (1mol) and hydrochloric acid (0.75-1.25 mol, 37% solution). The reaction was rhombic plates with their longest dimension ca. 6 pm. This brought to reflux, 100-110 "C, and an aqueous solution of morphology is independent of temperature (between 25 formaldehyde (2 mol, 36% solution) was added dropwise over a and 70 "C)and pH (between pH 3 and pH 7). period of 1 h, with stirring. The reaction was then refluxed for All the additives synthesized contained the (iminodia further 4-6 h, under an atmosphere of nitrogen, stirring methy1ene)diphosphonate motif, and hence, not surpriscontinuously. all had some effect on the precipitated barium ingly, After this time, the reaction was allowed to cool and the water sulfate. For compound 7, typical changes are seen in Figure was removed under reduced pressure. In most cases,this yielded 5, which illustrates a number of features. Firstly, the a viscous oil but the longer chain compounds precipitated out. molecules that can access two binding sites are effective Some oils were converted to solids by partitioning between at significantly lower concentrations than single motifs. ethanol/water mixes. All solidswere recrystallizedfrom ethanol/ Thus, as shown in Figure 5b, the disruption of growth in water. The compounds which could not be obtained as solids were analyzed as oils. the [OOl] zone is evident a t 0.6 pM,whereas for the single The compounds ware analyzed by 'H NMR, using DzO as the motif the first changes only become evident at 20 pM,6 an solvent, and the d values were recorded in parts per million. The enhancement of approximately 30-fold. Secondly, the ratios of relevant peak integrals were used to give an estimate nature of the morphological changes, although specific to of compound purity. In most cases the purity was at least 95%. the [OOl] zone, differ from those of the single motif. The Those compounds isolated as solids were all found to decompose appearance of new { O l l ) faces has not been observed, at a temperature within the range 200-250 "C, in agreement with presumably, as discussed earlier, because the additives the literature values.* are able to bind to both (011) and (210) faces. (i) Data for (H~O~PH~C)~N(CHZ)JV(CH~POSHI)~; 1: yield Disruption of growth on the {210)faces is evident from 52.6 g, 50% (based on 0.22 mol of diamine); isolated as a white levels as low as 0.3 p M , while at 0.6 p M and higher, the crystalline solid; lH NMR 6 3.50 (8H, d, PCHzN), 3.45 (4H, m, crystals appear to become increasingly bifurcated (Figure NCHZ),1.75 (4H, m, NCHZCHZ),1.40 (4H, m, (CH&).

3598 Langmuir, Vol. 9, No.12, 1993

a

Bromley et al.

L

a

d

e

C

Figure 5. BaSO, crystal morphologies 88 a result of the addition of additive 7 (a) no additive (X5K); (b)0.6 pM (X12.5K); (c) 0.9 FM (X12.5K); (d) 1.2 p M (X5K); (e) 2.1 p M (x5K);(00.96 mM (X6K) (all reproduced at 55% of original size).

5b,c) as if twinned a c r m the (001) plane. Electron diffraction indicates that all samples are single crystals (the texture evident on the (001) faces in the SEMs is a result of washing and has no structural basis) but that the direction of elongation shifta as the additive level rises. Thus, in Figure 5b,0.6pM, thecrystalsretaintheirrhombic shape with some rounding along the [OlO] direction and pitting of the (210) faces. A t 0.9 pM (Figure 5c) the situation is reversed, with the most pronounced rounding along the [lo01 direction and the [OlO] direction facetted. The side faces of the rhombus now exhibit some discontinuity such that the (210)face has been replaced by higher index planes. It appears also that the [OlOI direction growthis moreseverely inhibited than the [lo01 direction leading to the change in the a/b aspect ratio of 0.75 from 1.2. At higher loadings (Figure 5d, e) the [lo01 direction curvature becomes more pronounced, with formation of dish and ultimately spheres (Figure 50. There is a significant effect of the additives on crystal size, with loadings as high as 0.96 mM yielding crystals with sizes

below 1 pm, compared to the native 6-pm rbombuaes (Figure 50. Under these conditions it seems likely that the concentration profile experienced by the system changes such that crystallization takes place at higher levels of supersaturation. This may lead to growth proceases increasingly dominated by mass transfer which may in part be responsible for the loss of crystal facets. The observation of morphological changes at additive loadings as low as 1 pM makes it possible to exclude chelation of dissolved Ca2+as an important factor in this system. The precipitating system used6 contains approximately 10 mM Ca*+ as CaClr6HnO. If the worst case were assumed, in which each phosphonate bound one calcium ion, then it follows that a 10 pM solution of an additivewithtwomotifs wouldbind4 X 10pMCa2+which would representaOA%decrease ineffectiveconcentration of available additive. Independent experiments indicated that a reduction in the Ca2+concentration by 25 % had no effect on the rhombic morphology. Overall these data were consistent with a mode of binding in agreement with the proposed binding model,

Langmuir, Vol. 9, No.12, 1993 3599

Interactions at the Organicllnorganic Interface

80

.-.n 3

3

.c

I

s?

0.0

0.2

0.4

0.6

0.8

1.0

Concentration of Additive (mM)

Figure 6. Inhibition of BaSO, precipitation as a function of additive concentration for three additives: ( 0 )(iminodiiethy1ene)diphosphonicacid, ( 0 )compound 1, and (A)compound 7.

in which two diphosphonate motifs simultaneouslyaccess adjacent surface binding sites. This validates our strategy. Kinetic Effects. The efficiency of the additives as barium sulfate scale inhibitors was quantified in terms of percent inhibition, defined as % inhibition =

ICP reading - 1.89

56 - 1.89 The ICP reading is in milligrams per liter, 1.89 mg/L = blank reading at pH 7, and 56 mg/L = concentration of Ba2+immediately after mixing but before precipitation. A comparison of the effects of three additives is shown in Figure 6. The single motif additive, (iminodimethylene)diphosphonate, has a very limited effect at the concentrations investigated. The molecule hexamethylenebis[bis(phosphonomethyl)aminel ,which according to our model cannot span two binding sites because ita link length is less than 7.4 A, initially showed some improvement over the single motif molecule,but was in fact slightly worse at higher concentrations. This presumably reflects the balance of two opposing effects: the first should improve the performance because the unbound part of the molecule acts as asteric barrier, and the second should effectively reduce the performance by half, as two motifs are removed for each molecule bound. An additive, however, that can span two sites with ease, such as compound 7, shows a dramatic improvementin efficiency at far lower concentrations.

401

0

.

I

2

.

I

4

.

0 I

6

. , . , . 8

Maximium Link Length

1

0

1

2

(A)

Figure 7. Relationship betweenthe inhibiting power of additives at 0.96 mM and the maximum length of the linkage: (m) (iminodimethy1ene)diphosphonic acid, ( 0 ) compound 1, (+) compound 2, (0)compound 7, (0)compound 3, (A)compound 4, and (A)compound 8.

Figure 7 compares the efficacy of all the additives tested at an equivalent concentration of the active motif of 0.96 mM, with respect to link length. These data confirm the important of the 7.4-A link length, with effectively all precipitation being inhibited above this critical value.

Conclusions This work has again demonstrated the power of combined morphological data and molecular modeling as a means of probing the nature of interactions of organic molecules at inorganic surfaces. The need to simultaneously satisfy geometric, stereochemical,and electrostatic requirementsis clear, and it is only with this understanding that we were able to design novel additives for enhanced efficacy. Additionally, however, this work also demonstrates the need for additionaltest data (inthis case kinetic) as a means of discriminating between target molecules and checking the validity of the design strategy. Rigid links, between motifs, should lead to even further improvements because entropy effects, with respect to random occupation of sites, would be reduced. A further extension of this work, currently under investigation, is to consider how this approach can be used to tailor-make polymers with even higher degrees of specificity.