Forces Between Silica Particles in Isopropanol Solutions of 1: 1

Home , Van der Waals force

arXiv:2003.04058v2 [cond-mat.soft] 5 Jun 2020 Dojcsfo opst aeil ree integrated print even 19]. to or [18, used batteries materials were Li-ion composite inks from ten- Such objects surface 3D viscosity. control or to drying, phases large sion, non-aqueous contain ei- can be of can or portions inks based Material non-aqueous shapes. completely 3D ther or printing 2D is in solvents materials non-aqueous of using ex- process Another a of 17]. ample [16, complex materials high-quality structural or of mix- ferroelectric production powder permit used ceramic which are of tures, ketones, homogenization or and alcohols milling as example for or- such where An media, processing, polar ceramics processes. ganic includes natural process a some such techno- as of of well view as in non- interesting logical in equally processes are solvent, media natural aqueous important most the water Although be- is [13–15]. mechanisms forces or structural oscillatory ions [10–12] hind multivalent interactions mea- of such effects electrostatic of the on study examples to Some aimed surements systems. aqueous in done ex- and precision [7–9]. reproducibility high cellent with routine measurements enable force force tweezers atomic surface optical of and based (AFM), technique microscopy apparatus probe force surface colloidal as (SFA), such the techniques in advancement probing Recent force [1–6]. design particle print- and ink-jet ing, processing, systems, ceramic biological treatment, in water processes waste such of We examples processes. find technological can and natural many in portant ∗ E-mail: atmjrt ftesraefremaueet are measurements force surface the of majority vast A im- are liquids in immersed surfaces between Forces ocsbtenSlc atce nIorpnlSltoso Solutions Isopropanol in Particles Silica between Forces [email protected] rniinfo oooi odme siltr interacti oscillatory surface. damped the to to monotonic from close transition molecules adso isopropanol ion of strong by structuring alc induced forces heterogeneities non-DLVO in charge repulsive by pairing and caused ion attractive pronounced additional calcu by nm, 10 lengths caused Debye are than deviations meas larger These The attrac substantially observed. double-layer, is are Waals, repulsion p interactions der oscillatory A van damped Namely, and investigated. is systems. particular, particles these In silica microscope. between force interactions atomic on based technique neatosbtenslc ufcsars spoao s isopropanol across surfaces silica between Interactions .INTRODUCTION I. eateto nrai n nltclCeity Univers Chemistry, Analytical and Inorganic of Department cecsI,3 uiEns-nemt 25Gnv,Switzer Geneva, 1205 Ernest-Ansermet, Quai 30 II, Sciences ijn tjmrv´,MroGli n rgrTrefalt Gregor and Galli, Marco Stojimirović, Biljana Dtd ue8 2020) 8, June (Dated: iia ooe esrdi ae r lopeeti non- in present media. also polar are water aqueous in measured ones forces cited to solvation above similar and double-layer The that shown [25]. has research glycol were ethylene surfaces across fluorocarbon interacting when inter- solvation measured Attractive were actions measured. hydration- was distances, re- repulsion short at long-range like and distances, observed, large was At pulsion colloidal with [24]. silica measured technique between were forces probe glycol The molecules ethylene 23]. solvent in [20, surfaces of surface oscil- solid structuring the These by near formed forces. are oscillatory be- lations short-range included the which and havior, force, double-layer dominated repulsive was were by which regimes behavior, two long-range measurements, the and observed, these Christenson In by [20–22]. measured Horn first were SFA glycol and with ethylene methanol, acetone, sheets carbonate, mica propylene between polar mix- across Forces their water. and with media tures polar solid non-aqueous only between across is interactions on surfaces there literature the However, in data applications. scarce practical many in cosinclqisadhgl ocnrtdauossalt aqueous concentrated highly and liquids ionic across surface. the the near form molecules stem alcohol forces the short-range of These short ordering alco- 29]. At pure 27, in [26, observed [26]. hols was contents repulsion and step-like alcohol a mixtures high distances the at in association constant the ion the dielectric of to the attributed variation of was The variation forces double-layer mixture. of length the decay in content range chang- water the by change ing systems, force double-layer these the of In magnitude and 26–29]. [2, technique colloidal a probe with investigated were mixtures alcohol-water sdsrbdaoe o-qeu ovnsaeused are solvents non-aqueous above, described As aey hr a enalto neeti ufc forces surface in interest of lot a been has there Lately, and alcohols across surfaces silica between Forces ltosaemaue ihclodlprobe colloidal with measured are olutions n suncovered. is ons ae rmnmnlsl concentrations. salt nominal from lated ehr fdffrn ocsaefudin found are forces different of lethora ial,a nrae ocnrtosthe concentrations increased at Finally, pin hra h atroiiaefrom originate latter the whereas rption, h nuneo : lcrltso the on electrolytes 1:1 of influence the r bevd h omraepossibly are former The observed. are rddcylnt ftedouble-layer the of length decay ured iennDV,rpliesolvation, repulsive non-DLVO, tive hlsltos tsprto below separation At solutions. ohol t fGeneva, of ity land ∗ : Electrolytes 1:1 f 2 solutions. It has been shown that in very concentrated Cantilever and pure ionic liquids long-range exponential repulsion Silica Particle exists [23, 30, 31]. These repulsions have decay lengths OH much larger compared to the Debye length and furthermore the decay length increases with increasing concentration in highly concentrated systems. This behav- Isoprop yl Alcohol ior is opposite to the behavior observed in dilute electrolytes [31]. It was further observed that at high concen- Silica Particle trations of electrolytes the transition from monotonic to oscillatory forces is present [31] and that the wavelength of these oscillations can be abruptly changed by varying solvent composition [23]. These experiments sparked a FIG. 1. Schematic representation of a colloidal probe exper- renewed interest in theoretical description of ionic flu- iment. Force measurements between two silica particles were ids and a variety of theoretical approaches were utilized done in isopropanol solutions. to describe these new exciting experimental data [32–38].

Measurements mentioned above have been performed us- ◦ ing SFA, where one typically uses mica as a surface. Us- 1200 C for 2 hours, to burn the glue, achieve firm at- ing colloidal probe AFM would enable to use other sur- tachment and decrease surface roughness of the parti- faces during similar experiments, which would test the cles. Solutions were made in isopropanol (99.8%, Extra hypothesis that the phenomena observed in ionic liquids Dry, AcroSeal, Acros Organics) with addition of tetra- are interface independent. butylammonium bromide (TBAB, 99+%, Acros Organ- ics) or lithium chloride (LiCl, BioXtra, >99.0%, Sigma- Here, we investigate forces between silica colloids in Aldrich). isopropanol solutions of 1:1 electrolytes. Due to a lower dielectric constant of isopropanol as compared to water, Before force measurement, cantilevers and substrate the electrostatic coupling is stronger in these solutions. were cleaned in Milli-Q water and ethanol, and then The stronger electrostatic coupling induces a very rich treated in plasma for 20 minutes. When mounted into the behavior of these simple systems. Variety of different AFM, the fluid cell was filled with solution, and the par- type of forces are found. In addition to double-layer and ticle on the cantilever was centered above another one on van der Waals interactions, attractive non-DLVO and the quartz disk with a precision of around 100 nm. The short-range solvation forces are present. At increased speed of the approach of the cantilever to the substrate concentrations the transition from monotonic exponen- during the measurement was 500 nm/s for all experi- tial to damped oscillatory interactions is observed. ments, except for 50 mM solutions, where the approach speed of 100 nm/s was used. For a selected pair, the cantilever deflection was recorded in 150 approach-retract II. MATERIALS AND METHODS cycles. For each salt concentration, measurement was done on 3-5 different pairs of particles. The approach parts of the curves were averaged and used for analy- A. Force Measurements sis. Hookes law was used to convert deflection to force. Cantilever spring constant was determined by the Sader The surface force measurements were performed on method [39]. a closed-loop atomic force microscope (MFP-3D, Asy- lum Research) using the colloidal probe technique at ◦ room temperature 23 ± 2 C, see Fig. 1. The AFM B. Analysis of the Force Curves is mounted on an inverted optical microscope (Olym- pus IX70). Spherical silica 4 µm particles (Bangs Lab- The extended DLVO theory is used to analyze the force oratories Inc., USA) were attached to tipless cantilevers curves [7, 40] (MikroMasch, Tallin, Estonia) with the help of a small amount of glue (Araldite 2000+). Some particles were F = FvdW + Fdl + Fatt, (1) spread onto quartz polished disk (Robson Scientific, Saw- bridgeworth, UK) which was used as a bottom of a liquid where the total force between two particles is a superposi- cell in which measurements were done. The quartz disk tion of van der Waals, FvdW, and double-layer, Fdl, forces was beforehand cleaned in piranha solution (3:1 mixture as in classical DLVO and we add an additional attractive of H2SO4 98% and H2O2 30%). Both cantilevers with exponential term, Fatt. The van der Waals force is cal- particles and the substrate were heated in an oven at culated with a non-retarded expression for two spherical 3 particles with radius R [7, 40] The pressure is then integrated to obtain energy per unit area for two plates HR 1 − · FvdW = 2 , (2) ∞ 12 h ′ ′ Wdl = Π(h )dh . (11) where H is the Hamaker constant and h is the surface- Zh surface separation. The double-layer force between two spherical particles The double-layer forces are calculated by solving the of radius, R, is then obtained by using the Derjaguin Poisson-Boltzmann equation in the plate-plate geometry approximation, which connects sphere-sphere and plate- 2 d ψ(x) 2e0c plate geometries [7, 40] 2 = sinh(βe0ψ), (3) dx εε0 Fdl =2πReff Wdl, (12) where e0 is the elementary charge, c is the number concentration of the 1:1 electrolyte, β = 1/(kBT ) is the in- where Reff is the effective radius and is equal to R/2 for verse thermal energy, ε0 is the vacuum permittivity, and two identical spheres. ε = 17.9 is the dielectric constant of the isopropanol. The non-DLVO additional attractive term defined in ψ(x) is the electric potential, and x is the coordinate nor- Eq. (1) is modeled with an exponential function [12, 41] mal to the plates. The plates are positioned at x = −h/2 − F = −AR e qh, (13) and x = h/2. Due to symmetry, the Poisson-Boltzmann att eff equation is solved in the 0 ≤ x ≤ h/2 half-space with the where A is the amplitude and q−1 is the decay length of following boundary conditions this additional force. dψ At higher concentrations of salt, damped oscillatory = 0 and (4) dx forces are present and they are modeled as x=0 dψ −h/ξ εε = σ − C [ψ(h/2) − ψ ] , (5) Fosc = BReff e cos(2π/λ + φ), (14) 0 dx in dl x=h/2 where B is the amplitude, ξ is the decay of damping, λ where σ and ψdl are surface charge density and diffuse- is the wavelength, and φ is the phase shift of the oscilla- layer potential of the isolated surface, respectively. These tions. two parameters are connected through 2κεε βe ψ σ = 0 sinh 0 dl , (6) III. RESULTS AND DISCUSSIONS βe 2 0 where κ is the inverse Debye length Colloidal probe technique based on AFM is used to measure the forces between silica colloids across alco- 2βe2c hol solutions. Specifically, we study the interactions in κ = 0 . (7) s εε0 tetrabutylamonium bromide (TBAB) and LiCl solutions. Both salts are 1:1 electrolyte dissolved in isopropanol. C is the inner-layer capacitance. The regulation pa- in First, we look at the interactions between silica par- rameter, p, is used for the interpretation of capacitances. ticles in isopropanol without added salt, which are pre- This parameter interpolates between constant potential sented in Fig. 2. Without added salt forces are repul- (CP) with p = 0 and constant charge (CC) with p = 1, sive and long-ranged with a decay length of ∼ 80 nm. and it is defined as They can be accurately fitted with DLVO theory with C p = dl , (8) constant regulation approximation down to separation Cdl + Cin distance of about 10 nm, see Fig. 2a. This fit allows to extract the diffuse-layer potential, the regulation param- where diffuse-layer capacitance, Cdl, is calculated as eter, and the electrolyte concentration. The extracted βe ψ C = εε κ cosh 0 dl . (9) diffuse-layer potential is equal to 101 mV and can be dl 0 2 converted through Eq. (6) to diffuse-layer surface charge 2 The solution of the Poisson-Boltzmann Eq. (3) yields the density of 0.34 mC/m . The latter value is about 10-20 electric potential profile between two surfaces, ψ(x), from times lower as compared to silica in water [42–44]. The lower charge density of the silica surface can be explained which its value at the mid-plane can be extracted ψM = ψ(0). The disjoining pressure is then calculated as by longer-range electrostatics in solvents with lower dielectric constants. Bjerrum length, which estimates the Π(h)=2kBTc [cosh(βe0ψM) − 1] . (10) distance at which electrostatic interaction is equal to the 4

Constant (a) (b) (a) (b) Charge 0.10 0.10 DLVO Li + Cl Charge Br non-DLVO + Regulation N 0.05 0.05 Constant Potential 0.00 0.00

5 mM TBAB 50 mM LiCl -0.05 -0.05

Force/Effective Radius (mN/m) Force/Effective

Force/Effective Radius (mN/m) Force/Effective 0 10 20 30 0 10 20 30 Separation Distance (nm) Separation Distance (nm)

FIG. 2. Forces in isopropanol without added electrolyte. FIG. 3. Van der Waals forces between silica in isopropanol (a) DLVO fits with constant charge (CC), constant poten- solutions with (a) 5 mM tetrabutylamonium bromide (TBAB) tial (CP), and constant regulation (CR). (b) Comparison of and (b) 50 mM of lithium chloride (LiCl). DLVO and non-DLVO fits with additional exponential attraction. Note that in both cases the Hamaker constant −21 H = 1.0 · 10 J is fixed in the fits. TBAB (a) LiCl (b) 0.15 0.15

0.10 0.10 thermal energy, is equal to 0.71 nm in water, while it 0.05 0.05 equals to 3.1 nm in isopropanol. These values show, 0.00 0.00 that more energy is needed to separate a negative and 50 mM -0.05 -0.05

Force/Effective Radius (mN/m) Force/Effective a positive charge in alcohol and therefore it is harder Radius (mN/m) Force/Effective to charge surface in alcohol solutions. The regulation 0 10 20 30 40 50 0 10 20 30 40 50 parameter determined from the force, p = 0.52, sug- Separation Distance (nm) Separation Distance (nm) gests that the charge regulation of silica surfaces upon FIG. 4. Forces at different concentrations of (a) tetrabutyla- approach is considerable. Further, we assume 1:1 elec- monium bromide (TBAB) and (b) lithium chloride (LiCl). trolyte to be present in the alcohol solution, where the fitted concentration is equal to 3.2 µM. These traces of ions present in the solutions are possibly coming from the This constant is slightly lower than the one measured small amount of water in alcohol. Note that we did not across aqueous solutions for similar particles [43]. This tried to remove traces of ions before the measurements. difference is due to higher refractive index of isopropanol A more detailed graph of interaction in pure iso- as compared to water. The low value of the Hamaker propanol, shown in Fig. 2b, reveals that DLVO theory constant is probably also a consequence of some residual overestimates the force at distances lower that ∼ 10 nm. nanoscale roughness of the particles. The Hamaker con- Therefore below 10 nm attractive non-DLVO forces are stant has been shown to decrease with increasing rough- present. This additional attraction can be modeled with ness. [46, 47]. The van der Waals interaction does not simple exponential attraction described in Eq. (13). The depend on the type of added salt, which is consistent improved non-DLVO model is accurate down to sepa- with earlier observations [9, 12]. rations of about 1 nm. The extracted decay length The forces for the transition from low to high salt are and amplitude of the additional attraction are equal to shown in Fig. 4. The experiments were performed in −1 q = 2.2 nm and A = 0.16 mN/m, respectively. These two different salt solutions, namely tetrabutylamonium additional non-DLVO forces will be addressed in more bromide (TBAB) and lithium chloride (LiCl). These detail below. measurements enable us to study the influence of ion Let us now look at forces at high concentrations of size on double-layer interactions, since the tetrabutyla- added salt. We refer here to high salt concentration, for monium (TBA+) ion is bulkier compared to the lithium conditions when the double-layer interactions are com- ion. In both cases the transition from repulsive to at- pletely screened and van der Waals attraction is dom- tractive forces is observed by increasing salt concentra- inant. In isopropanol these conditions are reached al- tion. The repulsive forces are slightly longer-ranged in ready at 5 mM of added salt, which is about a factor the case of LiCl as compared to TBAB. Furthermore, of 10-100 lower as compared to aqueous systems [42–45]. a higher concentration of LiCl is needed to completely In Fig. 3 the van der Waals interaction between silica is screen the repulsion as compared to TBAB. In order to shown for the two salts. In both situations, the attraction extract more details about these systems, we have fitted can be accurately fitted with Eq. (2) and the Hamaker the experimental curves with extended DLVO theory, see constant of H = (1.0 ± 0.1) · 10−21 J can be extracted. Eq. (1). In these fits the Hamaker constant was fixed 5

(a) Isopropanol (b) Water where K is the association constant and square brack- ets denote molar concentrations. The concentration of NaCl + − 1:1 free ions is equal to cfree = [A ] = [B ] and total con- TBAB + MgSO4 centration is ctot = [A ] + [AB]. The ionization frac- 2:2 LiCl tion is finally defined as a ratio cfree/ctot. The solid lines LaFe(CN)6 3:3 in Fig. 5a are calculated with the chemical equilibrium model Eq. (16), where the equilibrium constant K is the only adjustable parameter. The fitted association constants for TBAB and LiCl are 1.5 mol/L and 5.0 mol/L, respectively. In literature these values are typically rep- resented as logarithmic form log K, which gives values FIG. 5. (a) Ionization fractions of TBAB and LiCl salts in 10 isopropanol as determined from AFM force measurements. of 3.2 and 3.7 for TBAB and LiCl, respectively. The asso- (b) Ionization fractions of 1:1, 2:2, and 3:3 salts in water, data ciation constant can be independently determined from taken from [48]. The solid lines are calculated with chemical electrical conductivity measurements by analysis devel- equilibrium model shown in Eq. (16). oped by Fuoss and Onsager [50, 51]. For TBAB association constants were determined in methanol and ethanol mixtures [52]. The log K values for the dielectric con- · −21 10 to the value of H = 1.0 10 J, which is consistent stant corresponding to the present isopropanol system with high salt measurements. Diffuse-layer potential, (ε = 17.9) are 2.4 and 2.8 for methanol and ethanol based electrolyte concentration, regulation parameter, and ad- solutions, respectively [52]. Our value for TBAB in iso- ditional attraction amplitude and decay were determined propanol of log10 K = 3.2 is therefore perfectly consis- by least square fitting. The extended DLVO theory ac- tent with the published results on TBAB methanol and curately describes the experimental curves down to the ethanol solutions. ∼ separation distances of 1 nm. Ion pairing cannot be completely understood by ac- Regulation parameters were observed to be indepen- counting only for electrostatic interactions and addi- ± dent of concentration and equal to 0.51 0.08 and tional solvent specific interactions must be taken into ac- ± 0.62 0.14 for TBAB and LiCl, respectively. For both count [52]. However, Bjerrum theory which includes only salts regulation parameters are similar to the values ob- Coulombic and hard-sphere interactions still gives rea- tained in pure isopropanol and the surfaces regulate fairly sonable estimates for the the extent of ion association. strongly. According to Bjerrum theory, the association constant The concentration of free ions can also be determined can be calculated as [53, 54] from the double-layer fitting. All the extracted concen- rmax trations were smaller than nominal ones for both salts − K =4πN e βU(r)r2dr, (17) investigated. Therefore the measured decay lengths of A Zrmin the double-layer forces are larger than expected based on nominal salt concentrations. These results suggest where NA is the Avogadro number, r is the center that the salts are not fully dissociated and that some to center distance of the ions, and U(r) = −ℓB/r is fraction of ions form ion pairs [48, 49]. The ionization the electrostatic energy between cation and anion with 2 fractions for both salts in isopropanol are plotted as a ℓB = e0/(4πεε0) being the Bjerrum length. The bounds function of nominal salt concentrations in Fig. 5a. The of the integral are the minimal distance the two ions can ionization fractions are approaching unity only for very approach, rmin, and the maximal distance at which we dilute isopropanol solutions and are rapidly dropping at consider ions to be paired, rmax. While the minimal ap- concentrations above 0.1 mM. Above 1 mM more than proach distance is determined by ion size, the maximal 50 % of ions form ion pairs. This behavior can be very distance is less defined, however the precise value of the well explained by the chemical equilibrium model, which upper bound does not affect the results drastically [53]. accounts for ion pair formation The Bjerrum theory permits us to estimate the ion sizes based on the constants extracted from ionization frac- − A+ +B ⇋ AB, (15) tions. The calculated values of minimal approach are ˚ ˚ + − 2.5 A for LiCl and 3.0 A for TBAB. The former value where A and B are cations and anions, respectively, − agrees perfectly with the sum of the Li+ (0.7 A)˚ and Cl while AB represents a neutral ion pair. This equilibrium (1.8 A)˚ radii [55]. While the reported values of minimal can be quantified by the following mass action law approach distance for tetrabutylamonium salts in non- [A+][B−] aqueous solvents vary substantially [53], the average of K = , (16) [AB] ∼ 3 A˚ agrees well with our result for TBAB. The dif- 6

(a) (b) TBAB (a) (b) 1.5

LiCl 1.0

LiCl 0.5 TBAB

0.0

0 1 2 3 4

FIG. 6. Surface potentials extracted from AFM force mea- FIG. 7. (a) Amplitude of the additional non-DLVO expo- surements and electrokinetic measurements in alcohol solu- nential attraction in isopropanol solutions. (b) Short-ranged tions of (a) TBAB and (b) LiCl. repulsive non-DLVO forces in TBAB solutions, due to isopropanol structuring near the surface. Arrows indicate the steps in the force profile. ference in the ionization fraction for LiCl and TBAB in alcohol is therefore due to the difference in ion size. The bulkier TBAB salt forms less ion pairs compared to more attractions in isopropanol without added salt, shown in compact LiCl. Fig. 2b, such attractions are also present in both TBAB One can further compare ion association in isopropanol and LiCl solutions. These attractions can be described − with pairing in aqueous systems, see Fig. 5b. In water, by the decay length, q 1, of 2.2 nm and 2.5 nm for the 1:1 electrolytes do not show any ion pairing, and TBAB and LiCl, respectively. The fitted amplitudes association only becomes prominent in the 2:2 and 3:3 of the attractions, A, defined in Eq. (13), are shown in electrolytes [48, 56]. The ionization fractions measured Fig. 7a. The attractions are the strongest at low salt lev- for 1:1 electrolyte in isopropanol is somewhere between els and they disappear at concentrations above ∼ 1 mM. the values measured for 2:2 and 3:3 electrolytes in wa- Non-DLVO attractions between silica surfaces are not ter. This fact can be understood by comparing Bjerrum present in the aqueous solutions of simple monovalent lengths in water (0.71 nm) and in isopropanol (3.1 nm). electrolytes, like KCl [43, 44, 57]. On the other hand Since the Bjerrum length in water is about 4-5 times they were observed in aqueous solutions of hydrophobic smaller than in isopropanol, the ions have to be more monovalent ions [58], and in the presence of multivalent charged to achieve the same electrostatic attraction en- counterions [12, 41, 44]. In aqueous solutions of monova- ergy at contact. lent hydrophobic ions the decay lengths of these attrac- The diffuse-layer potentials extracted from the force tions are between 1.5 and 3 nm, while they are around curves are show in Fig. 6. In both solutions the poten- 1 nm in the solutions of multivalent counterions. Simi- tials increase with increasing concentration as they are lar to the present case of alcohol solutions, these attrac- progressively screened by adding more ions in the solu- tive non-DLVO forces also disappear in water at high tion. The values for both salts are similar at low concen- salt concentrations. Currently the source of these attrac- trations, while at higher concentrations the screening of tive forces in aqueous media is not clear as they might TBAB is more effective. This difference stems from the be connected to ion-ion correlation [41], lateral charge more effective dissociation of TBAB in solutions. One heterogeneities [9], spontaneous charge fluctuations [59], can convert the potentials to surface charge by the means or possibly varying dielectric constant close to the sur- of Eq. (6). Note that in this equation a concentration face. However, these non-DLVO attractions seem to be of free ions and not nominal concentration of salt has present, when ions strongly interact with the surface [9]. to be used. The resulting average surface charge densi- The presence of these forces in alcohol solutions seems ties for TBAB and LiCl solutions are −0.50 ± 0.10 and to confirm this observation, since due to lower dielectric −0.40 ± 0.04 mC/m2, respectively. The slightly lower constant the electrostatic interaction between the ions magnitude of the surface charge density for LiCl solu- and the surface is enhanced. In water, the interaction tions is possibly connected to the stronger association of between ions and surface is strong enough only in the the Li+ ion with the negatively charged silanol groups. case of multivalent counterions, or if ions interact through This association is probably less prominent for TBA+, other strong non-electrostatic interactions, for example and therefore, this ion is less effective in neutralizing the hydrophobic force. surface charge. In addition to non-DLVO attraction, a short-range Finally, let us look at the non-DLVO interactions ob- non-DLVO repulsion is also observed in alcohols, see served in alcohol solutions. Similarly, the non-DLVO Fig. 7b. These forces have been observed in alcohols 7

60 TBAB (a) 60 LiCl (b) The Kirkwood cross-over was also observed in aqueous 40 40 solutions of simple ions and solutions of ionic liquids, al- 50 mM 20 20 50 mM beit one has to increase the concentrations beyond few 0 0 molar in these systems [23, 31]. In the present alcohol -20 -20 case this cross-over occurs at concentrations of few tens 5 mM -40 -40 of mM, which is at two orders of magnitude lower con- -60 van der Waals -60 centrations. These low concentrations provide a larger

Force/Effective Radius (µN/m) Force/Effective

Force/Effective Radius (µN/m) Force/Effective 0 1 2 3 4 5 0 1 2 3 4 5 window for exploration of these effects in future. Separation Distance (nm) Separation Distance (nm) In order to extract the wavelength of the oscillations, λ and decay of exponential damping ξ, we model these FIG. 8. Oscillatory forces observed at 50 mM electrolytes forces with Eq. (14). The fitted wavelengths, λ, for (a) TBAB and (b) LiCl. The full lines represent the fit to TBAB and LiCl are equal to 1.0 ± 0.1 nm and 0.9 ± 0.1, Eq. (14), while dashed lines show van der Waals interaction. respectively. The decay of the oscillations, ξ =0.65±0.1, Note that in the case of LiCl van der Waals force is added to is the same for both salts. The observed wavelength is the damped oscillatory force. slightly lower for LiCl as compared to TBAB, but in both cases the wavelength is about two times larger than before [26, 29] and are caused by structuring of alcohol the diameter of the ion pair. The corresponding wave molecules close to the solid surface. When sharp AFM tip length in 2 M NaCl aqueous solution was reported to be is used for the measurements the resulting profile is oscil- ∼ 0.5 nm, which is about the size of an ion pair [31]. Sim- latory with a period of about 0.95 nm in 1-propanol [29]. ilarly, the wavelength in ionic liquid-solvent mixtures was In our case, the oscillations are probably smeared out due measured to be about the ion pair size [23]. While cur- to surface roughness of the colloidal probe, however the rently we have no explanation, why in the present alcohol steps with the period of about 1 nm can be clearly ob- solutions, the wavelength is about two times larger than served in the sample without added salt. Upon addition diameter of the ion pair, this observation might be again of salt the structuring of alcohol molecules close to the connected with strong electrostatic coupling in alcohol surface seem to be disturbed and the steps become less solutions. Further theoretical studies would be needed clear. This disturbance of the alcohol layering is probably to understand Kirkwood cross-over and oscillatory forces caused by adsorption of ions to the surfaces. in these strongly electrostatically coupled systems. At increased salt concentration another type of interactions becomes evident. One can observe a transition between monotonic interaction and damped oscilla- IV. CONCLUSIONS tory force. In Fig. 8 this transition is shown for TBAB salt. At 5 mM, monotonic van der Waals interaction Forces between negatively charged silica particles were is present, while oscillations in the force profile become measured in monovalent salt solutions in isopropanol. clearly evident at 50 mM. Similarly, an oscillatory pro- An extremely rich behavior of these systems is observed; file on top of van der Waals attraction is observed in this includes, van der Waals forces, double-layer forces, 50 mM LiCl shown in Fig. 8b. We suspect that the tran- attractive non-DLVO forces, repulsive solvation forces, sition from monotonic to the oscillatory behavior is the and damped oscillatory interactions at increased concen- Kirkwood cross-over [60, 61]. Recently, different theoret- trations. The richness of these systems is connected to ical and simulation approaches were used to study this strong electrostatic coupling, which is due to low dielec- transition [32, 35–38, 60, 62]. These approaches predict tric constant of isopropanol. the transitions at salt concentrations corresponding to The interactions between silica surfaces are repulsive κd ∼ 1 − 2, where d is the mean ion diameter. In the at low salt levels and become progressively more attrac- present case we observe the transition below κd ∼ 0.5. tive with increasing salt concentration. This behavior is This shift to lower concentrations, might be connected consistent with DLVO theory. However, the decay of the to the strong electrostatic coupling, which is present in double-layer repulsion is much longer than expected from the current alcohol system. However, dressed-ion theory Debye lengths calculated from nominal salt concentra- predicts the shift to larger κd values for the 2:2 elec- tions. This observation can be explained by ion pairing trolytes in water, where the electrostatic coupling is also and quantified by Bjerrum theory. The association con- stronger [63]. The origin of the observed shift of the stants for the 1:1 electrolyte in alcohol are comparable to monotonic to oscillatory transition in alcohol solutions is the association constants in the 2:2 and 3:3 aqueous elec- therefore not clear and could be also caused by solvent trolytes, since Bjerrum lenghts in isopropanol are about molecules [32]. 4-5 times larger as compared to water solutions. 8

At distances below ∼ 10 nm the experimental force studies and found in aqueous systems. Furthermore, the proﬁles deviate from DLVO theory, and additional non- wavelength of the oscillations is longer than what is ex- DLVO forces are observed. The additional attractive pected from the diameter of the ion pair. These dif- forces are possibly caused by surface charge hetero- ferences might be connected to strong electrostatic cou- geneities, which are induced by strong ion adsorption. pling in alcohol systems. We hope that the present ex- This strong adsorption is driven by strong electrostatic perimental data will enable testing of recent theoretical interaction between ions in solution and charged surface approaches and spark the interest for more detailed ex- groups. ploration of phenomena found in these systems.

At distances below 2 nm repulsion due to structuring of isopropanol molecules close to the surface is present. ACKNOWLEDGMENTS This structuring is disturbed with increasing salt concentration, due to adsorption of counterions. This research was supported by the Swiss National Science Foundation through grant 162420 and the Uni- Finally, the transition from monotonic to oscillatory versity of Geneva. The authors are thankful to Michal forces is observed at increased concentrations. We be- Borkovec for insightful discussions and for providing lieve that this observation is the consequence of the Kirk- access to the instruments in his laboratory and to wood cross-over, albeit the concentration of this cross- Plinio Maroni for the help with AFM measurements and over seems to be lower than predicted by theoretical Poisson-Boltzmann solution software.

[1] J. J. Valle-Delgado, J. A. Molina-Bol´ıvar, F. Galisteo- 088001 (2016). González, M. J. Gálvez-Ruiz, A. Feiler, and M. W. Rut- [16] T. Rojac, A. Bencan, B. Malic, G. Tutuncu, J. L. Jones, land, J. Phys. Chem. B 108, 5365 (2004). J. E. Daniels, and D. Damjanovic, J. Am. Ceram. Soc. [2] R. Govrin, I. Schlesinger, S. Tcherner, and U. Sivan, J. 97, 1993 (2014). Am. Chem. Soc. 139, 15013 (2017). [17] G. Wang, P. Sarkar, and P. S. Nicholson, J. Am. Ceram. [3] B. Bolto and J. Gregory, Water Res. 41, 2301 (2007). Soc. 80, 965 (1997). [4] M. Cerbelaud, A. Videcoq, P. Abelard, C. Pagnoux, [18] D. Kokkinis, M. Schaffner, and A. R. Studart, Nat. Com- F. Rossignol, and R. Ferrando, Soft Matter 6, 370 (2010). mun. 6, 8643 (2015). [5] D. Kuscer, G. Stavber, G. Trefalt, and M. Kosec, J. Am. [19] K. Sun, T.-S. Wei, B. Y. Ahn, J. Y. Seo, S. J. Dillon, Ceram. Soc. 95, 487 (2012). and J. A. Lewis, Adv. Mater. 25, 4539 (2013). [6] M. Zanini, C. Marschelke, S. E. Anachkov, E. Marini, [20] H. K. Christenson and R. G. Horn, Chemical Physics A. Synytska, and L. Isa, Nat. Commun. 8, 15701 (2017). Letters 98, 45 (1983). [7] J. Israelachvili, Intermolecular and Surface Forces, 3rd [21] H. Christenson, J. Chem. Soc. Faraday Trans. 1 Phys. ed. (Academic Press, London, 2011). Chem. Condens. Phases 80, 1933 (1984). [8] G. Trefalt, T. Palberg, and M. Borkovec, Curr. Opin. [22] H. K. Christenson and R. G. Horn, Journal of Colloid Colloid Interface Sci. 27, 9 (2017). and Interface Science 103, 50 (1985). [9] A. M. Smith, M. Borkovec, and G. Trefalt, Adv. Colloid [23] A. M. Smith, A. A. Lee, and S. Perkin, Phys. Rev. Lett. Interface Sci. 275, 102078 (2020). 118, 096002 (2017). [10] O. Zohar, I. Leizerson, and U. Sivan, Phys. Rev. Lett. [24] D. T. Atkins and B. W. Ninham, Colloids and Surfaces 96, 177802 (2006). A: Physicochemical and Engineering Aspects Prof. B. W. [11] K. Besteman, M. A. G. Zevenbergen, H. A. Heering, and Ninham, 129-130, 23 (1997). S. G. Lemay, Phys. Rev. Lett. 93, 170802 (2004). [25] J. L. Parker and P. M. Claesson, Langmuir 8, 757 (1992). [12] M. Moazzami-Gudarzi, G. Trefalt, I. Szilagyi, P. Maroni, [26] Y. Kanda, T. Nakamura, and K. Higashitani, Colloids and M. Borkovec, Phys. Chem. Chem. Phys. 18, 8739 and Surfaces A: Physicochemical and Engineering As- (2016). pects 139, 55 (1998). [13] Y. Zeng, S. Grandner, C. L. P. Oliveira, [27] Y. Kanda, S. Iwasaki, and K. Higashitani, Journal of A. F. Thünemann, O. Paris, J. S. Pedersen, S. H. Colloid and Interface Science 216, 394 (1999). L. Klapp, and R. von Klitzing, Soft Matter 7, 10899 [28] C. E. McNamee, M. Matsumoto, P. G. Hartley, P. Mul- (2011). vaney, Y. Tsujii, and M. Nakahara, Langmuir 17, 6220 [14] S. H. L. Klapp, D. Qu, and R. von Klitzing, J. Phys. (2001). Chem. B 111, 1296 (2007). [29] V. Franz and H.-J. Butt, J. Phys. Chem. B 106, 1703 [15] M. Moazzami-Gudarzi, T. Kremer, V. Valmacco, P. Ma- (2002). roni, M. Borkovec, and G. Trefalt, Phys. Rev. Lett. 117, 9

[30] M. A. Gebbie, M. Valtiner, X. Banquy, E. T. Fox, W. A. [46] E. Thormann, Current Opinion in Colloid & Interface Henderson, and J. N. Israelachvili, Proc. Natl. Acad. Sci. Science 27, 18 (2017). U. S. A. 110, 9674 (2013). [47] V. Valmacco, M. Elzbieciak-Wodka, C. Besnard, P. Ma- [31] A. M. Smith, A. A. Lee, and S. Perkin, J. Phys. Chem. roni, G. Trefalt, and M. Borkovec, Nanoscale Horiz. 1, Lett. 7, 2157 (2016). 325 (2016). [32] F. Coupette, A. A. Lee, and A. Härtel, Phys. Rev. Lett. [48] A. M. Smith, P. Maroni, G. Trefalt, and M. Borkovec, 121, 075501 (2018). J. Phys. Chem. B 123, 1733 (2019). [33] A. A. Lee, C. S. Perez-Martinez, A. M. Smith, and [49] A. P. dos Santos, A. Diehl, and Y. Levin, J. Chem. Phys. S. Perkin, Phys. Rev. Lett. 119, 026002 (2017). 132, 104105 (2010). [34] R. Kjellander, J. Chem. Phys. 148, 193701 (2018). [50] R. M. Fuoss and L. Onsager, J. Phys. Chem. 61, 668 [35] R. Kjellander, Soft Matter 15, 5866 (2019). (1957). [36] R. M. Adar, S. A. Safran, H. Diamant, and D. Andel- [51] R. M. Fuoss, J. Am. Chem. Soc. 80, 3163 (1958). man, Phys. Rev. E 100, 042615 (2019). [52] H. Sadek and R. M. Fuoss, J. Am. Chem. Soc. 81, 4507 [37] Y. Avni, R. M. Adar, and D. Andelman, Phys. Rev. E (1959). 101, 010601 (2020). [53] C. Valeriani, P. J. Camp, J. W. Zwanikken, R. van Roij, [38] S. W. Coles, C. Park, R. Nikam, M. Kanduˇc, J. Dzubiella, and M. Dijkstra, Soft Matter 6, 2793 (2010). and B. Rotenberg, J. Phys. Chem. B 124, 1778 (2020). [54] N. Bjerrum, Kgl Dan. Vidensk. Selsk. 7, 1 (1926). [39] J. E. Sader, J. W. M. Chon, and P. Mulvaney, Rev. Sci. [55] Y. Marcus, Chem. Rev. 88, 1475 (1988). Instrum. 70, 3967 (1999). [56] Y. Marcus and G. Hefter, Chem. Rev. 106, 4585 (2006). [40] W. B. Russel, D. A. Saville, and W. R. Schowalter, Col- [57] A. M. Smith, P. Maroni, M. Borkovec, and G. Trefalt, loidal Dispersions (Cambridge University Press, Cam- Colloids Interfaces 2, 65 (2018). bridge, 1989). [58] A. M. Smith, P. Maroni, and M. Borkovec, Phys Chem [41] M. Kanduˇc, M. Moazzami-Gudarzi, V. Valmacco, Chem Phys 20, 158 (2018). R. Podgornik, and G. Trefalt, Phys. Chem. Chem. Phys. [59] N. Adˇzićand R. Podgornik, Phys. Rev. E 91, 022715 19, 10069 (2017). (2015). [42] B. Stojimirović, M. Vis, R. Tuinier, A. P. Philipse, and [60] R. J. F. L. de Carvalho and R. Evans, Mol. Phys. 83, 619 G. Trefalt, Langmuir 36, 47 (2020). (1994). [43] B. Uzelac, V. Valmacco, and G. Trefalt, Soft Matter 13, [61] J. G. Kirkwood, J. Chem. Phys. 7, 919 (1939). 5741 (2017). [62] P. Keblinski, J. Eggebrecht, D. Wolf, and S. R. Phillpot, [44] V. Valmacco, M. Elzbieciak-Wodka, D. Herman, G. Tre- J. Chem. Phys. 113, 282 (2000). falt, P. Maroni, and M. Borkovec, J. Colloid Interface [63] J. Ennis, R. Kjellander, and D. J. Mitchell, J. Chem. Sci. 472, 108 (2016). Phys. 102, 975 (1995). [45] M. Dishon, O. Zohar, and U. Sivan, Langmuir 25, 2831 (2009).