Downloaded by guest on September 28, 2021 chocolate powder–liquid of class ubiquitous a choco- mixing. that exemplifies suggest the conching and shift, discuss thus late this We system, of content. origins the microscopic solid consid- of possible flowable a fraction maximum volume effect the jamming to increasing the combine in surfactants energy shift mechanical of erable of addition input sim- staged the a that and of find rheology we the formulation, Studying plified particulates, dispersants. of and mixture oil, inhomogeneous triglyceride an suspen- we from flowing prepared homogeneous, dispersions,” is a sion the which “granular incorporation: in recent solid powder chocolate, such of such on possible conching of of example Building highest paradigmatic rheology a operation. the study the industrial into with in particles common dispersion primary advances a a 50-µm is to form 10- 2019) content to 4, of February liquid review powder for (received a a 2019 of 8, April mixing approved and The MA, Cambridge, University, Harvard Weitz, A. David by Edited and 10003; Kingdom; NY York, 08801; New NJ University, York Annandale, Company, New Netherlands; Physics, The of Utrecht, Department CC Research, 3584 University, Utrecht Science, Nanomaterials a Cates E. Michael www.pnas.org/cgi/doi/10.1073/pnas.1901858116 “φ Such transition. liquid to disper- solid of a raises i.e., amount further pension, small and a reducing friction of interparticle turn addition the the reduces conche, sant the of thus part aggregates, particulate apart breaks increasing conche the of part first fraction, volume jamming the from solids, of stress, fraction yield the viscosity, manufacturing, high-shear chocolate param- rheological in key the eters of two Specifically, jamming. and flow and ceramics the e.g., in, (1). analogs sectors has flowing pharmaceuticals that a seek process into and a together mixture effect, suspension, dispersants inhomogeneous this and nonflowing, on a action focus transform addition mechanical we how staged paper, understand the this to and In action dispersants. triglyceride mechanical of (a prolonged choco- butter (liquid by suspension cocoa late) flowing and homogeneous, a solids) into cocoa mixture) and sugar, to solids, (including is particulates milk function of physical mixture major inhomogeneous its an turn but by development, achieved flavor is for tant (3)] fat lower [= content solid “conching.” high manufactur- where chocolate is ing, example Another (2). strength higher material where ceramic examples, and a important replacement are is bone bodies” content or “green building solid for maximizing Cements Often, goal. (1). key industries many in fraction tant volume solid with suspension flowing T Blanco Elena content solid maximal suspension with flowable to solid from jammed transition frictionally prototypical a is chocolate Conching colo hsc n srnm,TeUiest fEibrh dnug H F,Uie Kingdom; United 3FD, EH9 Edinburgh Edinburgh, of University The Astronomy, and Physics of School efidta h e hsclpoessaefriction-dominated are processes physical key the that find We impor- is 1879, in Lindt Rodolphe by invented (4), Conching il iei h rnlrrne(∼ range par- granular primary the with in powder size dry ticle into liquid of incorporation he | rheology φ J g asCooaeU t. luhS14X ntdKingdom United 4JX, SL1 Slough Ltd., UK Chocolate Mars a,1 eaiet h xdms rcin ntesecond the In fraction. mass fixed the to relative σ ailJ .Hodgson M. J. Daniel , y a,f | and ftecooaefruaini situated is formulation chocolate the of φ, jamming sblaVnDamme Van Isabella , η 2 r otoldb o a h volume the far how by controlled are , η 2 eutn nfliiaino h sus- the of fluidization in resulting , | f eateto ple ahmtc n hoeia hsc,Uiest fCmrde abig B W,United 0WA, CB3 Cambridge Cambridge, of University Physics, Theoretical and Mathematics Applied of Department incorporation φ J 10 edmntaeta the that demonstrate We . o50 to µm a,1,2 g ihe Hermes Michiel , n isnC .Poon K. C. Wilson and , φ & J )t oma form to µm) 50% engineering” σ φ y n the and , improves simpor- is φ J in , a,b,1 e oprt taei eerh xoMblRsac n Engineering and Research ExxonMobil Research, Strategic Corporate c u Besseling Rut , nrcs-S,54 BOs h Netherlands; The Oss, AB 5349 InProcess-LSP, tfinite At φ η and exceeds observed, it coefficient, when friction tact interparticle the stress, applied both of tion φ ulse nieMy7 2019. 7, May online Published Edinburgh from downloaded 2 be can work this 1 in plotted (https://datashare.is.ed.ac.uk/handle/10283/3281).y data DataShare All Deposition: Data λ where at plateau tonian oeiaie ies . C BY-NC-ND) (CC 4.0 NoDerivatives License distributed under is article access Ltd.y open This UK Submission.y Chocolate Direct Mars PNAS by a G.L.H. funded is part, and article in This is, R.B., work This M.H., statement: interest D.J.M.H., of paper.y Conflict E.B., the wrote W.C.K.P. research; and D.J.M.H. performed E.B., and research; data; G.L.H. analyzed designed W.C.K.P. and and R.B., I.V.D., M.E.C., D.J.M.H., P.M.C., M.H., contributions: Author applied the when value Newtonian stress, low-stress high-φ a a suspension from of granular increases viscosity in The advances (5–14). recent rheology briefly, review, first We Suspensions Thickening Shear of production the on rely that industries dispersions. high-solid-content diverse to common is o-adhg-tesviscosities, high-stress and low- owo orsodnesol eadesd mi:[email protected] Email: addressed. be should correspondence work.y whom this To to equally contributed M.H. and D.J.M.H., E.B., 2 rcp m µ→∞ h oua tn f“unn ncornstarch.” to on related “running of is stunt describe dispersant. popular we the a are physics of underlying conching addition the reduction of the Intriguingly, the through roles friction and main interparticle aggregates the of of involved. of breakdown two mechanisms mechanical that the physical show the we Here, of been has there understanding now, until process, poor a the of into antiquity transformed the choco- and of is late popularity solids the Despite cocoa liquid. flowing and homogeneous sugar, fat, inhomogeneous of an which mixture in process the is conching Chocolate Significance ' h ihsrs icst,nwdvre tsome at diverges now viscosity, high-stress the (φ), a 2 o ooiprehr pee Fg 1A) (Fig. spheres hard monodisperse For . o pee 1,1) h amn point, jamming The 16). (15, spheres for σ ≈ η r xed oeoststress, onset some exceeds , h o-tesviscosity low-stress the µ, 0.54 = η 1,2 PNAS ”i rcie means practice, in “∞,” (where a,c η σ /η r h atrbgn opesprilsit con- into particles press to begins latter The . ayL Hunter L. Gary , 0 iegsa admcoepacking, close random at diverges σ b | η with otCnesdMte,DbeIsiuefor Institute Debye Matter, Condensed Soft  r a 1 2019 21, May = σ σ ? A ? η With . h upninsertikn.The thickens. shear suspension The : 0  . stesletviscosity, solvent the as 1 y − raieCmosAttribution-NonCommercial- Commons Creative η 1 φ µ | η J µ d,e and → φ 1 (σ o.116 vol. (φ) alM Chaikin M. Paul , σ d osertiknn is thickening shear no 0, ) ? etrfrSf Matter Soft for Center η  ecigahge New- higher a reaching , 2 tl iegsat diverges still −λ ieg as diverge , | , rnlrsuspension granular o 21 no. φ µ rcp φ & J | safunc- a is , n the and µ, ≈ A 8 17). (8, 1) φ 10303–10308 φ J d 0.64 ' φ J , = rcp = and 1, φ φ but , m µ and rcp [1] < .

PHYSICS superscript, which, again, we will drop unless needed), and A µ diverges at φm. Competition between friction and adhesion gives rise to a ? range of rheologies (19). If σ /σa  1, the suspension shear thins at σ > σy to the frictional viscosity, η2. The state diagram of such a system is shown schematically in Fig. 1B; Fig. 1B, Inset shows a typical flow curve. However, a suspension with ? σ /σa  1 first shear-thins at σ > σy, and then shear-thickens as σ exceeds σ?. Modifying system additives (e.g., removing poly- ? meric depletants or adding surfactants) can increase σ /σa and change the first type of behavior to the second type (21, 22). Conching Phenomenology We worked with a simplified chocolate formulation of “crumb powder” dispersed in sunflower oil with lecithin (23). For one experiment, we also added a second surfactant, polyglycerol polyricinoleate (PGPR). Crumb is manufactured by drying a water-based mixture of sucrose crystals, milk, and cocoa mass followed by milling (24). To perform a laboratory-scale conche, we used a planetary mixer (Fig. 2) to prepare 500-g batches. The total lecithin added was 0.83 wt%. In the first step, the B “dry conche,” we mixed the solids with the oil and 0.166 wt% of lecithin (20% of the total) in the planetary mixer at ∼ 100 rpm until the material smears around the bowl right after it has cohered into a single lump around the blade. At φ = 0.55, this took ∼ 40 min. Then, for the “wet conche,” the remaining lecithin was added and mixed for a further 20 min. Conched samples were prepared with solid concentrations in the range 0.4 . φ . 0.6, with φ calculated using measured den- sities (see Materials and Methods), so that a weight fraction of 74% converts to φ = 0.55 (assuming all of the fat contained in the crumb melts during conching). In each case, the flow curves of the as-conched sample as well as that of samples succes- sively diluted with pure oil were measured using parallel-plate rheometry (see Materials and Methods). Fig. 3 stages A through H show the phenomenology of conching a mixture with solid volume fraction φ0 = 0.55 (or 74 wt.%) to which, initially, 20% of the final total of 0.83 wt.% of lecithin has been added; the accompanying plots show the power consumption of the planetary mixer as well as mea- sured densities of the sample as conching proceeds. We define Fig. 1. (A) The high-shear viscosity of suspensions of granular hard spheres the solid volume fraction as the ratio of solid volume to total normalized by the solvent viscosity, ηr, plotted against the volume fraction solid plus liquid volume, discounting any air that may be present; φ, with friction coefficient increasing from µ = 0 (red), diverging at φrcp, to this differs from the granulation literature, where the air is typ- µ → ∞ (blue), diverging at φµ→∞.( ) The jamming volume fraction, m Inset ically taken into account. Almost immediately after addition of φJ, where ηr diverges, as a function of the coefficient of static friction µ (replotted from ref. 17). (B) The jamming state diagram of a frictional gran- the sunflower oil to the crumb powder (t = 0 min), all of the liq- ular suspension with interparticle adhesion. The adhesive strength is set by uid appeared to have been absorbed. The sample then proceeded ? σa  σ . Shaded region is jammed. (Inset) The flow curve of a suspension to granulate, with the granule size increasing with time. The first with volume fraction ϕ. It has a yield stress σy(ϕ). granules were visually matt and dry (Fig. 3, Bottom, stages A–C) and did not stick to each other during mixing.

µ (Below, we drop the “µ” in φm unless it is needed.) A granular suspension at φ > φm cannot flow at high stress either steadily or homogeneously (12): It shear-jams (7). Instead, theory (7) and experiments (18) suggest that it granulates. The onset stress, σ?, correlates with the force to over- come an interparticle repulsive barrier; typically, σ∗ ≈ d −ν with ν . 2, where d is the particle diameter (9). For granular suspensions, σ? is far below stresses encountered in liquid– powder mixing processes, so that they always flow with viscosity η2(φ, µ), which diverges at φm < φrcp. To formulate a flow- able granular suspension with maximum solid content is there- fore a matter of maximizing φm, e.g., by lowering µ (Fig. 1A, Inset). Interparticle adhesion introduces another stress scale, σa, characterizing the strength of adhesive interactions (19). A Fig. 2. Schematic of a planetary mixer. A blade (bold) rotates inside a bowl, µ µ yield stress, σy, emerges above some φa < φm that is depen- full circle, which counterrotates. Shearing occurs in the gap between the dent on both adhesion and friction (19, 20) (hence the µ blade and the bowl.

10304 | www.pnas.org/cgi/doi/10.1073/pnas.1901858116 Blanco et al. Downloaded by guest on September 28, 2021 Downloaded by guest on September 28, 2021 iin otewtcnh.Tesml ail udzdit a into fluidized rapidly sample The conche. tran- conche wet dry the the and to added, 3, is 3, Fig. sitions lecithin in (Fig. of line to easily dashed 80% flow (red remaining adhering not this the after did ball Soon that G). a paste stage a Bottom, into became formed material decreased, the rapidly had and consumption power material at the Thereafter, peaked the blade. the sharply of consumption all power when The F. stages and by and finished E slightly, coalesced, had granules Compactification decreased constant: as remained density incorporation then air envelope to due The presumably (25). 3, granulation manufactured powder (Fig. by washing resemble structures somewhat that “raspberry-like” D) stage larger into lesced 3, (Fig. slowly first volume, increased line). skeletal the black the consumption macroscopic to power over converged its and The increased compacted, by density. granules pores, the divided as liquid-filled min sample 15 and a air- of including mass envelope The the components. as liquid average and system 3, solid the (Fig. density the to of rapidly during density converging decreased bulk minutes, sharply volume few density first the skeletal the The by con- pores. externally divided excludes air therefore contains condensed nected and phases, mesoscopic it these of liquids) by occupied mass and the (solids is phases Methods), and Materials Letter conche. in (Images the system. of the stages various at stage in samples correspond mixer stages of planetary of appearance the the labels Visual and of (Bottom) added conche. is out lecithin wet taken Red of to density. shot dry envelope second from which the transition at is time circle the denotes box, dashed line blue black the dashed In the density. inside skeletal the density is average cluster shaded gray the of density lnoe al. et Blanco typical a for time mixing of function a with as formulation chocolate line) model (blue density envelope and 3. Fig. A EFGH After 3, (Fig. material a of density skeletal The A ostage to oe osmto bakln) kltldniy(e line), (red density skeletal line), (black consumption Power (Top) ≈ 15 i h rnlsbcm iil os n coa- and moist visibly became granules the min E lebxand box blue Top, ystage by ; BCD otm A–H Bottom, h rnl iehsdvre otesz of size the to diverged has size granule the F, Top φ and 0 r 0m wide.) mm 40 are = ,defined Methods ), and Materials rnl ieicessfrom increases size Granule Bottom. 5(≡ 0.55 4w.) nterdbx the box, red the In wt.%). 74 e o and box red Top, ≈ Bottom, 30 Top), Top, min, n hsvle h ufc ftesml rasu,adi sno is it an and at up, occurs breaks This tools. sample approximately rheometer the the of between contained surface longer the value, this at ing plateau Newtonian a ht flcti ttebgnigo h e oce different a conche, wet to dots): (here, red the 4, of (Fig. beginning obtained is the rheology at lecithin of shot” (28). interface suspension–air free the of out poke tension ∼ σ with curve, flow mixture the crumb shows dots) (black 4 Fig. Rheology Chocolate on through Conching uncover of under- Effect to be seek may we which granulation flowing physics, rheology. of a generic form stages by to suggest pinned various powder liquid sectors into via fixed diverse liquid at suspension of across incorporation mixing similarities espe- the concrete note, that Such time-lapsed for We the (27). curves H). in content similarity stage (26), power visual at F) and striking case, stage a images amount our is our there to in the that (equivalent (as cially, typi- as at curve later suspension, would seen power than flowing the rather system is a of into the H, peak turn the However, i.e., through “overwet,” increases. A become cally liquid stages powder to added 3, events dry agro- of of Fig. a sequence from in to similar ranging A that added applications pharmaceuticals. gradually for to chemicals is granules liquid manufacture to where (25), tion powder of time batch the same highly the and is provided used. consumed time peak, is power the of reach total function to the a required both as power in fluidization. consumed reproducible, this the accompanied that consumption Note power in 3, decrease (Fig. suspension pourable glossy, eihnol,sertiknn smse 1,2,2)by 22) 21, (19, masked is thickening shear only, lecithin of stress t GRcnet.Dse ie ud h y ohg-ha viscosity high-shear to eye intermedi- the for are guide curves lines Intermediate Dashed PGPR. at contents. 1.2% PGPR with ate ( but lecithin before, as 0.83% with conched i.4. Fig. y σ > σ 0.1Σ/a neetnl,i GRi de oehrwt h “second the with together added is PGPR if Interestingly, granula- wet in reported widely is phenomenology Similar ≈ 40 oe hclt o uvs lc osdenote dots Black curves. flow chocolate Model frac Pa, Σ σ σ . ? (∼ y ≈ 0 & ≈ η 30 300 2 Above ∞. → 10 φ a hssget ht ntesml oce with conched sample the in that, suggests This Pa. mN· 0 −2 -independent afrorsse with system our for Pa PNAS a,rvaigsertiknn iha onset an with thickening shear revealing Pa), m φ −1 0 0 = ,weeprilsmyb xetdto expected be may particles where ), | . a 1 2019 21, May 5 .54 φ σ 0 Pa y = h apesertistoward thins shear sample the , σ 7 t%.Blwayedstress, yield a Below wt.%). (73 · 5.Rddt denote dots Red 0.55). frac .Hwvr utbfr reach- before just However, s. ≈ tg ) sharp A H). stage Bottom, σ 400 | y η o.116 vol. a sdaaial lowered dramatically is (σ a ls otestress, the to close Pa, ≈ faflyconched fully a of ), 10 µ σ | ? n surface and m /σ o 21 no. a  σ ? σ ,crumb 1, /σ y | σ > a 10305  ? 1; ,

PHYSICS but the high-shear viscosity is nevertheless the shear-thickened, A frictional contacts-dominated η2. [φ0] This high-shear viscosity, ηfc , of model chocolates fully- conched (fc) with lecithin at nine different solid fractions, φ0, is plotted in Fig. 5A (red open circles). In four cases, we successively diluted the conched samples with sunflower oil and measured the high-shear viscosity along each dilution [φ0] series, ηfc (φ). Each dataset can be fitted to Eq. 1 with A = 1 (Fig. 5A, solid lines), confirming what is already obvious from inspection, namely, that these datasets diverge at different points: gold, φm = 0.639 (λ = 1.88) for φ0 = 0.596; green, φm = 0.627 (λ = 1.78) for φ0 = 0.586; blue, φm = 0.612 (λ = 1.72) for φ0 = 0.576; and purple, φm = 0.562 (λ = 1.53) for φ0 = 0.536. In each remaining case, we estimate φm without generating a full dilution series at each φ0, by fitting Eq. 1 through each of the five [φ0] ηfc data points using the averaged exponent from the four full dilution series λ = λ¯ = 1.73 (Fig. 5A, thin red curves). Plotting all available pairs of (φ0, φm), Fig. 5A, Inset con- firms that φm increases as we conche the mixtures at higher solid fraction φ0. That is to say, conching at a higher solid fraction gives rise to a higher jamming volume fraction. There is, however, an upper bound to such φm optimization, which B we can estimate by noting that, empirically, φm(φ0) is approxi- mately linear in our range of φ0. A linear extrapolation shows max that φm = φ0 at φ0 ≈ 0.8. This is likely an overestimate: The approximately linear relation shown in Fig. 5A, Inset probably becomes sublinear and perhaps saturates at higher φ0. Never- max theless, the existence of some φ0 < 0.8 beyond which conch- ing will not increase φm seems to be a reasonable inference from our data.

Conching as Jamming Engineering

Conching Reduces Aggregate Size and Increases φm. Interestingly, our observations may be interpreted in terms of an “inverse conching” experiment performed some 50 y ago. Lewis and Nielsen (29) measured the viscosity of 30- to 40-µm glass spheres suspended in Aroclor (a viscous Newtonian organic liquid) as a function of volume fraction (Fig. 5B, dataset ℵ) and repeated the measurements with glass beads that were increasingly aggre- gated by sintering before dispersal in Aroclor (Fig. 5B), giving average number of primary particles in an aggregate N of 1.8 (A), 5 (C), 8 (D), 12 (E), and 200 to 300 (K). Neutral silica Fig. 5. (A) Relative high shear viscosity of chocolate suspensions vs. solid ? in an apolar solvent likely has a vanishingly small σ , so that [φ0] volume fraction. Red open circles denote ηfc , suspensions fully conched in Lewis and Nielsen were measuring η2(φ), the shear-thickened the planetary mixer. Filled circles denote diluted fully conched suspensions: viscosity, which diverges at φm. Thus, φm is clearly lowered by gold, diluted from φ0 = 0.596; green, diluted from φ0 = 0.586; blue, diluted aggregation. from φ0 = 0.576; and purple, diluted from φ0 = 0.536. Matching-color ver- We interpret our data (Fig. 5A) as “Lewis and Nielsen in tical dotted lines denote φm from fitting Eq. 1 to these four datasets. Thin reverse.” The primary particles in raw crumb are aggregated in red curves denote Eq. 1 with λ = 1.73, consistent with single open red cir- storage due to moisture, etc. (30). Conching reduces aggregation cle data points. (Inset) Frictional jamming φm of chocolate suspensions as a function of the conched volume fraction φ0. Symbols are as A. (B) Replotted and therefore increases φm, with the effect being progressively µ φ data of Lewis and Nielson (29) for 30- to 40- m glass spheres suspended in more marked as the material is being conched at higher 0. The Aroclor, with each dataset fitted to Eq. 1; aggregate size increases from ℵ latter effect is probably because the same external stress gener- and then A to K. ates higher particle pressure at higher φ (12, 31), which breaks up aggregates more effectively. The linear relation in Fig. 5A, Inset extrapolates to a finite are needed to distinguish N -mers with small N , even when the intercept at (0, 0.11 ± 0.02), suggesting that the viscosity of primary particles are quasi-monodisperse. [See the instructive unconched crumb powder would diverge at φm = 0.11. This value study of monomers and dimers by Johnson et al. (32).] Simi- is perhaps unrealistically low: The real φm(φ0) dependence prob- larly, light scattering will be, at best, a crude tool for studying ably becomes sublinear at low φ0 and saturates at some value that conching. In fact, measuring changes in φm by rheology is likely is > 0.11. Nevertheless, there seems little doubt that unconched, a good, albeit indirect, method to detect changes in low N -mer aggregated crumb suspensions jam at volume fractions consid- aggregation. erably below those used for real chocolate formulations (φ0 & 0.55). From Granules to Flowing Suspension. We can now describe the The change in aggregation during the liquid–powder mixing whole conching process in terms of suspension rheology. Con- process is often monitored by laser light scattering. This method sider a crumb–oil–lecithin mixture at φ0 = 0.576, somewhat would not have separated Lewis and Nielsen’s samples ℵ, A, higher than in Fig. 3. The jamming point of the initial mixture µ C, D, and E. Highly accurate data at very low scattering angles at time t = 0, φm(init), is substantially lower (our lower-bound

10306 | www.pnas.org/cgi/doi/10.1073/pnas.1901858116 Blanco et al. Downloaded by guest on September 28, 2021 Downloaded by guest on September 28, 2021 revlm ntesse,s htboth that so system, the increasing the until up, increase, in broken 33, being volume (25, are cluster–cluster clusters free aggregates the of parallel, of kinetics In properties 34). the mechanical the by and gran- controlled collisions The is adhesion. process interparticle are a ulation and maintaining and there tension packing system, surface of particle of entire result jammed combination the a a by saturate as together to form held liquid granules insufficient These 3, D). being (Fig. through (7) granulates B (Fig. it stages region agitation, jammed mechanical the under inside where, deep 6A), is diagram system state starting In the homogeneously. terms, flow cannot therefore mixture the estimate, φ dropping down, esi aysses h oe sosre opa utas just peak i.e., (26), to overwet observed jargon) is granulation jammed power (in hap- becomes shear what The material indeed, a systems: the is, longer many This no maintain. in is to pens tension there surface with since for albeit viscosity, suspension, state flowable high a very into turn a to system the expect to to coefficient friction interparticle the suspen- unconched both increasing the and of aggregates at up point breaking formulation jamming The the sion, (A) than concentrated 1B). more Fig. (compare diagram 6. Fig. lnoe al. et Blanco sus- the conche, dry the of end flow, the because principle, At in right. could, the pension to boundary jamming C B A m µ σ tti on,tesse eoe ul auae,adw may we and saturated, fully becomes system the point, this At 0 wt,tesse snwaflwn suspension. flowing a now is system the (wet), a 0 φ σ < m µ tgauae ne ehnclaiain simultaneously agitation, mechanical under granulates It (B) (init). h ocigpoesrpeetda hfsi h amn state jamming the in shifts as represented process conching The σ a frac hrb hfigtejmigbudr ate otergtand right the to farther boundary jamming the shifting thereby , σ < y φ y σ necp fFg 5A, Fig. of intercept dy.(C (dry). y m µ wt below (wet) utexceeds just diino h eodso flcti reduces lecithin of shot second the of Addition ) σ φ frac 0 φ φ < Since . 0 µ . 0 m µ en .1.Ti initial This 0.11). being Inset, µ < dy,bt nfc,cno oso, do cannot fact, in but, (dry), φ φ 0 n h deieinteraction adhesive the and m µ snwcnieal below considerably now is and φ m µ φ a µ and φ hssitn the shifting thus , 0 sconsiderably is φ a µ steadily h r oce h il stress, yield the conche, dry the 4). not homogeneously, Fig. flow does with because to still yield probably can is suspension the it i.e., This the before at matt. fractures 3), case, visually sample Fig. our the appears and In in easily surface. F flow shiny (stage a peak with power suspension flowing a seta dai htcnhn,ad oegnrly e milling, wet The generally, more 6. and, Fig. conching, in that summarized is granular is idea essential in picture jamming resulting emerg- and The an thickening as dispersions. shear well of as understanding existing We literature ing granulation using chocolate. the measurements model from flowing and knowledge crumb We a observations of sectors. into our conching the industrial oil interpreted detail, sunflower many in and across process powder such goal one solid generic studied maximal have a of is dispersions content solid-in-liquid flowable Creating Conclusions and Summary lowering φ causes by lecithin effect that conclude this therefore to in may failed We This oils paste. correspond- (1.4%). the concentration the oil liquefy lecithin the of to maximum volume the repeated due to equivalent we ing not an fraction, was added volume and this sample experiment that the check lowering To lecithin 7). (Fig. viscosity, high-shear tration The liquefied. into a which mixed produced η paste, were again this lecithin which of of omitted, aliquots conche amounts lecithin dry Various of a paste. shot nonflowing prepared first we the role, with this for evidence experimental φ lowers shot second the in Lubricant. a as Lecithin yield chocolate. the liquid of drops lowering below and dramatic ably right resulting the to The stress to 6C). shifts (Fig. this abruptly that down boundary suggest We lowers jamming lecithin effect. The additional dramatic the a because has is lecithin of shot ond stress, fracture i.7. Fig. ihu eihna ucino usqetyaddlecithin. added subsequently of function a as lecithin without 2 J a µ ftersligssesosdcesdwt eihnconcen- lecithin with decreased suspensions resulting the of , ute ocigcniust increase to continues conching Further = yrlaigcntanso h ytm(9.T rvd direct provide To (19). system the on constraints releasing by σ φ y m µ ihservsoiyo oe hclt at chocolate model of viscosity High-shear σ > σ Fg 1A, (Fig. y frac φ (wet) m (wet) cmaeteerirdsuso of discussion earlier the (compare σ σ < frac Inset). PNAS meitl rdcsaflwn suspension, flowing a produces immediately frac .Hr,teadto ftesec- the of addition the Here, 6B). (Fig. ehv ugse httelcti added lecithin the that suggested have We µ nasse hr,now, where, system a in | µ n oicesn h amn point, jamming the increasing so and a 1 2019 21, May n hrfr nrae both increases therefore and σ y nyjs xed the exceeds just only (dry), | µ o.116 vol. φ and m φ 0 ni,a h n of end the at until, = σ 5 dry-conched 0.557 σ | a φ frac o 21 no. 0 to sconsider- is µ associated 0 and φ | m µ 10307 and σ a 0 .

PHYSICS µ µ is about “jamming engineering”—manipulating φm and φa by UK). The latter consists of a mixture of phospholipids (∼ 60%) with some changing the state of aggregation, and “tuning” the interparticle residual soya oil. Our sunflower oil was Newtonian at 20 ◦C, with viscos- friction coefficient µ and the strength of interparticle adhesion, ity η0 = 54 mPa·s. Its density was measured by an Anton Paar DMA density meter to be 0.917 g·cm−3. Using sunflower oil rather than cocoa butter σa. Importantly, many additives ostensibly acting as dispersants obviates the need for heating during rheological measurements. The rhe- to reduce interparticle attraction and so lower σa may, in fact, µ ology of the chocolate suspension obtained from conching our mixture function primarily as lubrications to lower and so increase resembles that of fresh liquid chocolate made using cocoa butter (23). Our φm. Our scheme (Fig. 6), with appropriate shifts in σfrac, can be PGPR was also supplied by Mars Chocolate UK. used to understand liquid incorporation into powders in many Conching was performed using a Kenwood kMix planetary mixer with different specific applications. a K-blade attachment, adding the lecithin in two successive batches as Our proposed picture for conching/wet milling poses many described in Conching Phenomenology. We measured the skeletal density questions. For example, the rheology of a suspension at the end at various stages of conching by performing helium pycnometry (Quan- of dry conche, in which flow is, in principle, possible (φ0 < φm) tachrome Ultrapyc). The envelope density was measured using a 2.00 ± but, in practice, ruled out by surface fracture occurring before 0.02-mL Pyrex microvolumetric flask, with sunflower oil as the liquid phase. bulk yielding, has not yet been studied in any detail. Neither is Rheometric measurements were performed using a stress-controlled rheometer (DHR-2; TA Instruments) in a cross-hatched plate–plate geome- the role of changing φm during the granulation process under- try (diameter 40 mm, 1 × 1 × 0.5 mm serrated grid of truncated pyramids) stood. Our results therefore constitute only a first step toward to minimize wall slip, at a gap height of 1 mm and a temperature of 20 ◦C. a unified description of liquid incorporation, wet milling, and All data plotted in this work can be downloaded from https://datashare. granulation. is.ed.ac.uk/handle/10283/3281.

Materials and Methods ACKNOWLEDGMENTS. We thank Ben Guy, John Royer, and Jin Sun Our crumb powder (supplied by Mars Chocolate UK) consists mostly of (Edinburgh) and Will Taylor (Mars Chocolate) for illuminating discussions. We thank Mars Chocolate UK Ltd. for initiating and funding part of this faceted particles with mean radius a ≈ 10 m (polydispersity of 150%) µ & work. Other funding came from Engineering and Physical Sciences Research according to laser diffraction (LS-13 320; Beckman-Coulter). It has a den- −3 Council Grants EP/J007404/1 and EP/N025318/1. Research at New York Uni- sity of 1.453 g·cm and a specific (Brunauer–Emmett–Teller) surface area versity was supported partially by the Materials Research Science and 2 −1 of 2.02 m ·g (data provided by Mars Chocolate UK). We used sunflower Engineering Centers Program of the National Science Foundation under oil as purchased (Flora) and soy lecithin as supplied (by Mars Chocolate Award DMR-1420073.

1. Kaye B (1997) Powder Mixing (Springer, Dordrecht, The Netherlands). 18. Hodgson DJM, Hermes M, Poon WCK (2015) Jamming and the onset of granulation 2. Carneim TJ, Green DJ (2004) Mechanical properties of dry-pressed alumina green in a model particle system. arXiv:1507.08098. bodies. J Am Ceram Soc 84:1405–1410. 19. Guy BM, Richards JA, Hodgson DJM, Blanco E, Poon WCK (2018) Constraint-based 3. Tao R, Tang H, Tawhid-Al-Islam K, Du E, Kim J (2016) Electrorheology leads to approach to granular dispersion rheology. Phys Rev Lett 121:128001. healthier and tastier chocolate. Proc Natl Acad Sci USA 113:7399–7402. 20. Liu W, Jin Y, Chen S, Makse HA, Li S (2017) Equation of state for random sphere 4. Gutierrez´ TJ (2017) State-of-the-art chocolate manufacture: A review. Comp Rev Food packings with arbitrary adhesion and friction. Soft Matter 13:421–427. Saf 16:1313–1344. 21. Gopalakrishnan V, Zukoski C (2004) Effect of attractions on shear thickening in dense 5. Seto R, Mari R, Morris JF, Denn MM (2013) Discontinuous shear thickening of suspensions. J Rheology 48:1321–1344. frictional hard-sphere suspensions. Phys Rev Lett 111:218301. 22. Brown E, et al. (2010) Generality of shear thickening in dense suspensions. Nat Mater 6. Wyart M, Cates ME (2014) Discontinuous shear thickening without inertia in dense 9:220–224. non-Brownian suspensions. Phys Rev Lett 112:098302. 23. Taylor JE, Van Damme I, Johns ML, Routh AF, Wilson DI (2009) Shear rheology of 7. Cates ME, Wyart M (2014) Granulation and bistability in non-Brownian suspensions. molten crumb chocolate. J Food Sci 74:55–61. Rheol Acta 53:755–764. 24. Beckett ST (2003) Is the taste of British milk chocolate different. Int J Dairy Tech 8. Mari R, Seto R, Morris JF, Denn MM (2014) Shear thickening, frictionless and frictional 56:139–142. rheologies in non-Brownian suspensions. J Rheol 58:1693–1724. 25. Salman AD, Hounslow M, Seville JP, eds (2006) Granulation, Handbook of Powder 9. Guy BM, Hermes M, Poon WCK (2015) Towards a unified description of the rheology Technology (Elsevier, New York), vol. 11. of hard-particle suspensions. Phys Rev Lett 115:088304. 26. Betz G, Burgin¨ PJ, Leuenberger H (2003) Power consumption profile analysis and 10. Lin NY, et al. (2015) Hydrodynamic and contact contributions to continuous shear tensile strength measurements during moist agglomeration. Int J Pharm 252:11–25. thickening in colloidal suspensions. Phys Rev Lett 115:228304. 27. Cazacliu B, Roquet N (2009) Concrete mixing kinetics by means of power 11. Royer JR, Blair DL, Hudson SD (2016) Rheological signature of frictional interactions measurement. Cem Concr Res 39:182–194. in shear thickening suspensions. Phys Rev Lett 116:188301. 28. Brown E, Jaeger HM (2014) Shear thickening in concentrated suspensions: Phe- 12. Hermes M, et al. (2016) Unsteady flow and particle migration in dense, non-Brownian nomenology, mechanisms and relations to jamming. Rep Prog Phys 77:046602. suspensions. J Rheol 60:905–916. 29. Lewis TB, Nielsen LE (1968) Viscosity of dispersed and aggregated suspensions of 13. Comtet J, et al. (2017) Pairwise frictional profile between particles determines dis- spheres. Trans Soc Rheol 12:421–443. continuous shear thickening transition in non-colloidal suspensions. Nat Commun 30. Pietsch W (2004) Agglomeration in Industry: Occurrence and Applications (John 8:15633. Wiley, New York). 14. Clavaud C, Berut´ A, Metzger B, Forterre Y (2017) Revealing the frictional transition in 31. Boyer F, Guazzelli E,´ Pouliquen O (2011) Unifying suspension and granular rheology. shear-thickening suspensions. Proc Natl Acad Sci USA 114:5147–5152. Phys Rev Lett 107:188301. 15. Maron SH, Pierce PE (1956) Application of Ree-Eyring generalized flow theory to 32. Johnson PM, van Kats CM, van Blaaderen A (2005) Synthesis of colloidal silica suspensions of spherical particles. J Colloid Sci 11:80–95. dumbbells. Langmuir 21:11510–11517. 16. Krieger IM, Dougherty TJ (1959) A mechanism for non-Newtonian flow in suspensions 33. Iveson SM, Litster JD (1998) Growth regime map for liquid-bound granules. AIChE J of rigid spheres. Trans Soc Rheol 3:137–152. 44:1510–1518. 17. Silbert LE (2010) Jamming of frictional spheres and random loose packing. Soft Matter 34. Iveson S, et al. (2001) Growth regime map for liquid-bound granules: Further 6:2918–2924. development and experimental validation. Powder Technol 117:83–97.

10308 | www.pnas.org/cgi/doi/10.1073/pnas.1901858116 Blanco et al. Downloaded by guest on September 28, 2021