Dynamics of fronts in chemical and bacterial media: If you’ve seen one front, you’ve seen them all Paul Ronney Department of Aerospace & Mechanical Engineering Univ. of Southern California, Los Angeles, CA, 90089 University of Southern California Established 125 years ago this week! …jointly by a Catholic, a Protestant and a Jew - USC has always been a multi-ethnic, multi-cultural, coeducational university Today: 32,000 students, 3000 faculty 2 main campuses: University Park and Health Sciences USC Trojans football team ranked #1 in USA last 2 years USC Viterbi School of Engineering Naming gift by Andrew & Erma Viterbi Andrew Viterbi: co-founder of Qualcomm, co-inventor of CDMA 1900 undergraduates, 3300 graduate students, 165 faculty, 30 degree options $135 million external research funding Distance Education Network (DEN): 900 students in 28 M.S. degree programs; 171 MS degrees awarded in 2005 More info: http://viterbi.usc.edu Paul Ronney B.S. Mechanical Engineering, UC Berkeley M.S. Aeronautics, Caltech Ph.D. in Aeronautics & Astronautics, MIT Postdocs: NASA Glenn, Cleveland; US Naval Research Lab, Washington DC Assistant Professor, Princeton University Associate/Full Professor, USC Research interests Microscale combustion and power generation (10/4, INER; 10/5 NCKU) Microgravity combustion and fluid mechanics (10/4, NCU) Turbulent combustion (10/7, NTHU) Internal combustion engines Ignition, flammability, extinction limits of flames (10/3, NCU) Flame spread over solid fuel beds Biophysics and biofilms (10/6, NCKU) Paul Ronney Motivation Propagating fronts are ubiquitous in nature Flames » (Fuel & Oxidant) + Heat More heat Solid rocket propellant fuels » (Fuel & Oxidant) + Heat More heat Self-propagating high-temperature synthesis (SHS) - reaction of metal with metal oxide or nitride, e.g. Fe2O3(s) + 2Al(s) Al2O3(s) + 2Fe(l) » (Fuel & Oxidant) + Heat More heat Frontal polymerization » Monomer + initiator + heat polymer + more heat Autocatalytic chemical reactions (non-thermal front) » Reactants + H+ Products + more H+ Bacterial front (non-thermal front) » Nutrient + bugs more bugs All of these might be construed as “reaction-diffusion systems” Today’s topic: what is similar and what is different about these different types of fronts? Reaction-diffusion systems Two essential ingredients Reactive medium (e.g. fuel-air mixture) Autocatalyst - product of reaction that also accelerates the reaction (e.g. thermal energy) Self-propagation occurs when the autocatalyst diffuses into the reactive medium, initiating reaction and creating more autocatalyst, e.g. A + nB (n+1)B Enables reaction-diffusion fronts to propagate at steady rates far from any initiation site Premixed flame (SHS, solid propellant similar) Reaction zone 2000K Product concentration Direction of propagation Speed relative to unburned gas = SL Temperature Reactant concentration 300K Distance from reaction zone Convection-diffusion zone - /SL = 0.3 - 6 mm Reaction-diffusion systems - characteristics After initial transient, fronts typically propagate at a steady rate Propagation speed (SL) ~ (D)1/2 » D = diffusivity of autocatalyst or reactant » = characteristic reaction rate = (reaction time)-1 D depends on “sound speed” (c) & “mean free path” () » D ~ c Propagation rate generally faster in turbulent media due to wrinkling (increased surface area) of front Thermal fronts require high Zeldovich number (Ze) so that products >> reactants, otherwise reaction starts spontaneously! T T E Tad T ad Ze (T ) T T RT Tad T Tad ad ad ad Flammability or extinction limits when loss rate of autocatalyst ≈ production rate of autocatalyst Instability mechanisms Instability mechanisms may preclude steady flat front Turing instability - when ratio of reactant to autocatalyst diffusivity differs significantly from 1 Thermal fronts: Dautocatalyst/Dreactant = Lewis number Low Le: additional thermal enthalpy loss in curved region is less than additional chemical enthalpy gain, thus local flame temperature in curved region is higher, thus reaction rate increases drastically, thus “blip” grows High Le: pulsating or travelling wave instabilities Hydrodynamics - thermal expansion, buoyancy, Saffman-Taylor Direction of propagation Fuel diffusion Unburned gas Heat diffusion Heat diffusion Fuel diffusion Flame front Burned gas Polymerization fronts First demonstrated by Chechilo and Enikolopyan (1972); reviewed by Pojman et al. (1996), Epstein & Pojman (1998) Decomposition of the initiator (I) to form free radicals (Ri*): I R1* + R2* - highest activation energy step e.g. (NH4)2S2O8 2NH4SO4* Followed by addition of a radical to a monomer (M): M + Ri* RiM* - initiates polymer chain, grows by addition: RiMn* + M RiMn+1* Most of heat release occurs through addition step Note not chain-branching like flames Chain growth eventually terminated by radical-radical reactions: RiMn* + RjMm* RiMn+mRj Chain length can be controlled by chain transfer agents affects properties of final product Polymerization front Reaction zone 500K Polymer concentration 10 cm2/s 1.2 Polymerization front Monomer concentration Temperature 0.001 cm2/s Viscosity (log scale) 0.01 cm2/s 300K Distance from reaction zone Density relative to reactants 0.96 5 mm Polymerization fronts Potential applications Rapid curing of polymers without external heating Uniform curing of thick samples Solventless preparation of some polymers Filling/sealing of structures having cavities of arbitrary shape without having to heat the structure externally Limitations / unknowns Thermally driven system - need significant T between reactants and products to haveproducts >> reactants Previous studies: use very high pressures or high boiling point solvent (e.g. DMSO) to avoid boiling since mixtures with Tad < 100˚C won’t propagate …but water at ambient pressure is the solvent required for many practical applications Idea: use a very reactive monomer (acrylic acid) highly diluted with water to minimize peak temperature, and control heat losses to avoid extinction …but nothing is known about the extinction mechanisms! Polymerization fronts - approach Simple apparatus – round tubes Need bubble-free model polymerization systems 2-hydroxyethyl methacrylate (HEMA) monomer in DMSO solvent Acrylic acid (AA) monomer in water solvent Both systems: ammonium persulfate (AP) initiator, Cab-o-sil (fumed silica powder) viscosity enhancer Control thermal boundary conditions & assess heat loss Varying tube diameter Water bath, ambient air or insulated tube to control external temperature Polymerization front Typical speeds 0.01 cm/s, SL ≈ ()1/2 -1 ≈ 14 s From plot of ln(SL) vs. 1/Tad can infer E ≈ 13.5 kcal/mole, Ze ≈ 20 Extinction at Pe ≈ (0.004 cm/s)(1.6 cm)/(0.0014 cm2/s) ≈ 4.6 - close to classical flame theory predictions Plot of SL vs. “fuel” concentration approaches vertical at extinction limit as theory predicts With insulation, limiting SL and %AA much lower 0.016 16 mm tube Uninsulated 16 mm tube 10% AP 0.014 0.01 0.008 Front speed (cm/s) Front speed (cm/s) 0.03 Mass % AP 0.006 5% 8% 10% 12% 15% 0.004 0.012 0.01 0.008 0.006 0.004 Insulated Uninsulated 0.002 0.002 0 15 20 25 30 35 Mass percent AA 40 45 10 15 20 25 Mass percent AA 30 35 Polymerization fronts - thermal properties Far from limit Peak T same with or without insulation, speed and slope of T profile same, uninsulated case shows thermal decay in products Close to limit Uninsulated case shows substantial thermal decay in products; ratio (peak + slope)/(peak - slope) ≈ 12 Insulated case much slower, thicker flame, little or no thermal decay, limit not well defined 80 Temperature (ыC) Temperature (ыC) 100 80 Slope = 0.056ыC/s 60 27.5% AA / 10% AP 16 mm tube 40 Insulated Uninsulated Adiabatic 20 14.7% AA, insulated 22.2% AA, uninsulated 70 60 50 40 30 10% AP 16 mm tube 20 0 100 200 300 400 Time (seconds) 500 600 700 0 500 1000 1500 2000 Time (seconds) 2500 3000 Polymerization front High Lewis number - spiral & travelling-wave instabilities like flames (middle and right videos, viscosity-enhancing agent added to suppress buoyant instabilities) Quick Time™a nd a YUV4 20 cod ec dec ompr esso r ar e need ed to see this pictur e. Quick Time™a nd a YUV4 20 cod ec dec ompr esso r ar e need ed to see this pictur e. Quick Time™a nd a YUV4 20 cod ec dec ompr esso r ar e need ed to see this pictur e. Movies courtesy Prof. J. Pojman, University of Southern Mississippi Lean C4H10-O2-He mixtures; Pearlman and Ronney, 1994 Autocatalytic aqueous reactions - motivation Models of premixed turbulent combustion don’t agree with experiments nor each other! Pope & Anand 1987 (zero heat release) (large heat release) 30 Turbulent Burning Velocity (S T /S L ) Bray 1990 (zero heat release) (large heat release, = 7) Sivashinsky 1990 25 Yakhot 1988 Bychov 2000 =7 20 Experiment (Bradley, 1992) x (Re L=1,000) 15 10 Gouldin 1987 (Re L 5 =1,000) (Where Re is not reported, predictions are independent of Re L 0 0 10 20 30 Turbulence Intensity (u'/S 40 L ) L ) 50 Modeling of premixed turbulent flames Most model employ assumptions not satisfied by real flames, e.g. Adiabatic (sometimes ok) Homogeneous, isotropic turbulence over many LI (never ok) Low Ka or high Da (thin fronts) (sometimes ok) Lewis number = 1 (sometimes ok, e.g. CH4-air) Constant transport properties (never ok, ≈ 25x increase in and across front!) u’ doesn’t change across front (never ok, thermal expansion across flame generates turbulence) (but viscosity increases across front, decreases turbulence, sometimes almost cancels out) Constant density (never ok!) Autocatalytic front (bacterial fronts similar) Reaction zone Aqueous chemical front Reactant concentration Product concentration Viscosity (log scale) 0.01 cm2/s 0.01 cm2/s 303K Temperature 300K 0.9994 Density relative to reactants Distance from reaction zone 0.01 mm “Liquid flame” idea Use propagating acidity fronts in aqueous solution Studied by chemists for 100 years Recent book: Epstein and Pojman, 1998 Generic form A + nB (n+1)B - autocatalytic / << 1 - no self-generated turbulence T ≈ 3 K - no change in transport properties Zeldovich number ≈ 0.05 vs. 10 in gas flames Aqueous fronts not affected by heat loss!!! Large Schmidt number [= /D ≈ 500 (liquid flames) vs. ≈ 1 (gases)] - front stays "thin” even at high Re 2 u' /LT LI u' D 1/ 2 u' 1 Ka 2 ~ ~ Re Sc L SL /D u' LI LT SL2 S L 2 Approach - chemistry Iodate-hydrosulfite system IO3- + 6 H+ + 6e- I- + 3 H2O -2 + 4 H O 6 e- + 8 H+ + 2 SO -2 S O 2 4 2 4 _________________________________________________ IO3- + S2O4-2 + H2O I- + 2 SO4-2+ 2 H+ Comparison assumptions with turbulent combustion Adiabatic Homogeneous, isotropic turbulence over many LI Low Ka or high Da (thin fronts) due to high Schmidt # Constant transport properties u’ doesn’t change across front Constant density Conclusion: liquid flames better for testing models! model Taylor-Couette apparatus Motor Rotation Ar +-ion Laser Sheet Product (Not fluorescing) Inner cylinder s ST Mirror Reactant (Fluorescing) LDV Probe Cylindrical Lens Outer Cylinder Beam Splitter Rotation 3-D Traversing System Fiber-Optic Transmitte r + Ar Laser Fiber Motor Photomultiplier FFT Signal Analyzer Computer Capillary-wave apparatus Ar +-ion Laser Sheet Mirror Vibration Product (Not fluorescing) Reactant (Fluorescing) s LDV Probe Cylindrical Lens Vibrating Platform Beam Splitter Optical Fiber Loudspeaker Fiber-Optic Transmitter + Ar Laser Photomultiplier FFT Signal Analyzer 3-D Traversing System Computer Results - liquid flames QuickTi me™ a nd a MPEG-4 Vid eo d eco mpres sor a re ne eded to see this picture. Results Thin "sharp" fronts at low Ka (< 5) Thick "fuzzy" fronts at high Ka (> 10) No global quenching observed, even at Ka > 2500 !!! High Da - ST/SL in 4 different flows consistent with Yakhot model u' S 2 ST exp L 2 SL S S T L Low Da - ST/SL lower than at high Da - consistent with Damköhler model over 1000x range of Ka! Rising, buoyantly-unstable fronts in Hele-Shaw flow show unexpected wrinkling - subject of separate investigation Propagation rate (S T /S L ) Liquid flames - comparison to Yahkot (1988) Hele-Shaw Capillary wave T aylor-Couette Vibrating grid (Shy et al. ) T heory (Yakhot) Power law fit to expts. 100 10 Power law fit (u'/S L > 2): S T /S L = 1.61 (u'/S L ).7 42 1 0.1 1 10 100 "Turbulence" intensity (u'/S L ) 1000 Results - liquid flames - propagation rates Capillary wave experiments Taylor-Couette experiments 1 0.1 Flamelet Distributed T L S /S (experiment) / S T L /S (theory, Yakhot) Data on ST/SL in flamelet regime (low Ka) consistent with Yakhot model - no adjustable parameters Transition flamelet to distributed at Ka ≈ 5 0.01 0.1 1 10 100 Karlovitz number (Ka) 1000 10 4 Results - liquid flames - propagation rates 3 1 0.8 Flamelet 0.6 Distributed Experiments (Taylor-Couette) Experiments (capillary wave) 0.4 T L S /S (experiment) / S T L /S (theory, Damköhler) Data on ST/SL in distributed combustion regime (high Ka) consistent with Damköhler’s model - no adjustable parameters 0.1 1 10 100 Karlovitz number (Ka) 1000 10 4 Front propagation in one-scale flow ST exp u' SL 1 exp u' SL S S SL S S T L T L Front propagation rate (s/c) Turbulent combustion models not valid when energy concentrated at one spatial/temporal scale Experiment - Taylor-Couette flow in “Taylor vortex” regime (one-scale) Result - ST/SL lower in TV flow than in turbulent flow but consistent with model for one-scale flow probably due to "island" formation & reduction in flame surface (Joulin & Sivashinsky, 1991) 250 Th eory (Eq. 1) Th eory (1-s cal e) CW e xp erime nt TC e xp erime nt 1-s cal e expe rimen t 200 150 100 50 0 0 100 200 300 400 Turbulence intensity (q/c) 500 600 Fractal analysis in CW flow Fractal-like behavior exhibited D ≈ 1.35 ( 2.35 in 3-d) independent of u'/SL Same as gaseous flame front, passive scalar in CW flow Theory (Kerstein & others): D = 7/3 for 3-d Kolmogorov spectrum (not CW flow) Same as passive scalar (Sreenivasan et al, 1986) Problem - why is d seemingly independent of Propagating front vs. passively diffusing scalar Velocity spectrum Constant or varying density Constant or varying transport properties 2-d object or planar slice of 3-d object Fractal analysis in CW flow 10 5 1.5 1.4 u'/S L = 220 Slope = 0.732 d = 1.268 Fractal dimension Area (number of pixels) u'/S L = 77 Slope = 0.776 d = 1.224 1.3 1.2 Al l da ta a t u'/S 1.1 L > 60: Mean = 1 .31, RMS deviation 0.06 10 4 1 1 10 Measurement scale (number of pixels) 0 50 100 150 Disturbance intensity (u'/S 200 L 250 ) Bacterial fronts Many bacteria (e.g. E. coli) are motile - swim to find favorable environments - diffusion-like process - and multiply (react with nutrients) Two modes: run (swim in straight line) & tumble (change direction) - like random walk Longer run times if favorable nutrient gradient Suggests possiblity of “flames” Motile bacteria Bacteria swim by spinning flagella - drag on rod is about twice as large in crossflow compared to axial flow (G. I. Taylor showed this enables propulsion even though Re ≈ 10-4) (If you had flagella, you could swim in quicksand or molasses) Flagella rotate as a group to propel, spread out and rotate individually to tumble QuickTi me™ and a Sor enson Video decompr essor ar e needed to see this picture. http://www.rowland.org/bacteria/movies.html Analogy with flames Flame or molecular property Temperature Fuel Heat diffusivity Е c Fuel diffusivity Sound speed (c) Mean fr ee path () Reaction timescale Heat loss Extinguishment Microbiological equivalent Concentration of b acteria Nutrients Diffusivity of bacteria Diffusivity of nutr ient Swimming speed of bacterium in "run" mode c multipled by av erage time to switch f rom run mode to tumble mode and back Reproduction time Death (of individual bacterium) Death (of all bacteria) Reaction-diffusion behavior of bacteria Bacterial strains: E.coli K-12 strain W3110 derivatives, either motile or non-motile Standard condition: LB agar plates (agar concentration of 0.1 - 0.4%) Variable nutrient condition: Tryptone/NaCl plates (agar concentration of 0.1, 0.3%) All experiments incubated at 37˚C 60 0.2% agar 0.3% agar Front diameter (mm) 50 40 30 20 10 0 0 1 2 3 4 5 6 7 8 Time (hours) Fronts show a steady propagation rate after an initial transient Propagation rates of motile bacteria fronts As agar concentration increases, motility of bacteria (in particular “sound speed” (c)) decreases, decreases effective diffusivity (D) and thus propagation speed (s) decreases substantially No effect of depth of medium Above 0.4% agar, bacteria grow along the surface only Recently: very similar results for Bacillius subtilis - very different organism - E. coli & B. subtilis evolutionary paths separated 2 billion years ago Front speed (mm/hr) 10 8 6 4 2 0 0.050 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 Agar concentration, % Effect of nutrient concentration Increasing tryptone nutrient concentration increases propagation speed (either due to increased swimming speed or increased division rate) but slightly decreases propagation rate beyond a certain concentration - typically motility decreases for high nutrient concentrations (detectors saturated?) Front Speed (mm/h) 4.5 4 3.5 3 2.5 2 1.5 1 0 0.2 0.4 0.6 0.8 Tryptone Concentration (%) 1 Quenching limit of bacteria fronts Quenching limit: min. or max. value of some parameter (e.g. reactant concentration or channel width) for which steady front can exist Quenching “channels” made using filter paper infused with antibiotic bacteria killed near the wall, mimics heat loss to a cold wall in flames Bacteria can propagate through a wide channel but not the narrow channel, indicating a quenching limit Quenching described in terms of a minimum Peclet Number: Pe = sw/D (w = channel width) For the test case shown s ≈ 1.75 x 10-4 cm/s, D = 3.7 x 10-5 cm2/s, w at quenching limit ≈ 2.1 cm Pe ≈ 9.8 - similar to flames and polymer fronts 6 mm wide channel 35 mm wide channel E. coli, 0.1% agar, 100 µl of kanamycin per side, 6.5 hours after inoculation Comparison of fronts in Mot+ and Mot- bacteria Some mutated strains are non-motile but D due to Brownian motion ≈ 104 smaller Fronts of Mot- bacteria also propagate, but more slowly than Mot+ bacteria 1.2 Front speed (mm/hr) 1 0.8 0.6 0.4 0.2 0 0 0.05 0.1 0.15 Agar % 0.2 0.25 Quantitative analysis Bacteria D as estimated from measured front speeds SL for Mot+ ≈ 5.3 x 10-5 cm/s for 0.3% agar Reproduction time scale () of E.coli ≈ 20 min D ≈ s2 ≈ (5.3 x 10-5 cm/s)2(1200s) ≈ 3.3 x 10-6 cm2/s Similarly, D ≈ 3.7 x 10-5 cm2/s in 0.1% agar Bacteria diffusivity estimated from molecular theory “Mean free path” () estimated as the “sound speed” (c) multiplied by the time (t) bacteria swim without changing direction c ≈ 21 µm/s, t ≈ 1.4 s ≈ 3.0 x 10-3 cm, D ≈ 6.3 x 10-6 cm2/s, similar to value inferred from propagation speed Diffusivity of Mot- E. coli due to Brownian motion (0.75 µm radius particles in water at 37˚C) ≈ 2.9 x 10-9 cm2/s, ≈ 1700x smaller than Mot+ bacteria Fronts should be (1700)1/2 ≈ 40x slower in Mot- bacteria Consistent with experiments (e.g. 8 mm/hr vs. 0.2 mm/hr at 0.1% agar) Comparison of fronts in Mot+ and Mot- bacteria Dnutrient (≈ 10-5 cm2/s) close to Dbacteria, so “Lewis number” ≈ 1 Do bacteria choose their run-tumble cycle time to produce D required for Le ≈ 1 and avoid instabilies??? Switching from Mot+ to Mot- bacteria decreases the bacteria diffusivity (Dautocatalyst) by ≈ 1700x but nutrient diffusivity (Dreactant) is unchanged decreases the effective “Lewis number” Mot- fronts “cellular” but Mot+ fronts smooth - consistent with “Lewis number” analogy Mot+ 5 hr 30 min after inoculation Mot- 50 hr after inoculation 0.1% Agar dyed with a 5% Xylene Cyanol solution (Petri dish 9 cm diameter) Biofilms Until recently, most studies of bacteria conducted in planktonic (free swimming) state, but most bacteria in nature occur in biofilms attached to surfaces Recently many studies of biofilms have been conducted, but the effects of flow of the nutrient media have not been systematically assessed No flow: no replenishment of consumed nutrients - little or no growth Very fast flow: attachment and upstream spread difficult Most flow studies have reported only volumetric flow rate or flow velocity - not a useful parameter - why should it matter what the flow is far from the surface when the biofilm is attached to the surface? Biofilms can spread upstream - is spread rate ~ shear as with upstream flame spread on a solid fuel bed? Fluid mechanics tells us the shear rate at the surface is the key Our approach: use flow in tubes (shear not separated from mean flow rate) and Taylor-Couette cells (shear and mean flow independently controlled) Biofilm experiments - laminar flow in tubes No biofilm above interface Dense biofilm at air-medium interface Low-density biofilm below interface LB + E. coli (1) 12 hr incubation (2) Biofilm initiated Peristaltic pump LB only Waste (3) Flow testing (4) Sectioning & analysis Biofilms - images Control (incubated but not placed in flow apparatus) 0.3 ml/min, 3.5 hr 0.6 ml/min, 3.5 hr 0.6 ml/min, 3.5 hr 0.6 ml/min, 12 hr 2 ml/min, 3.5 hr 0.6 ml/min, 24 hr 16 ml/min, 3.5 hr Experiments show an effect of flow velocity or shear rate on growth rate and upstream spread Biofilms Abs or banc e ( ar b. uni t s) 2.5 Flow = 0.6 ml/min 1/8" diameter tube u = 0.13 cm/s 2 m um/d = 3.2/s Control 10 min 28 min 76 min 210 min 12 hr 24 hr 1.5 1 0.5 0 -6 -4 -2 0 2 4 6 Distance from inoculation point (cm) Fixed flow / varying elapsed time: more time, more growth, but maybe some sloughing at long times Biofilms Abs or banc e ( ar b. uni t s) 3 2.5 Elapsed time 3.5 hr 1/8" diameter tube u = 0.21 cm/s for 1 ml/min m 2 0 ml/min 0.3 ml/min 0.6 ml/min 2.0 ml/min 7.8 ml/min 16 ml/min 1.5 1 0.5 0 -4 -2 0 2 4 Distance from inoculation point (cm) Fixed elapsed time / varying flow: optimal flow/shear rate that maximizes growth Biofilms - Taylor Couette cell concept Motor Waste outlet Rotation Fluid region Inner cylinder Fluid region LDV Probe Optical fiber bundle Outer Cylinder + Ar Laser 3-D Trav ersing System Rotation Motor Fiber-Optic Transmitter Photomultiplier Media inlet FFT Signal Analyzer Computer Conclusions Broad analogies can be drawn between different types of reaction-diffusion fronts in disparate types of physical / chemical / biological systems Steady propagation rates Effects of reactant and product diffusivities Instabilities (i.e. pattern formation) Quenching behavior Applications to Combustion engines Solid propellant rockets Synthesis of ceramics Polymer synthesis Assessment of turbulent combustion models Colonization of new environments by swarms of bacteria Biofilms - bacteria growing on surfaces - far more resistant to antibiotics & other stresses than “planktonic” (free-swimming) bacteria Thanks to… National Cheng-Kung University Prof. Y. C. Chao, Prof. Shenqyang Shy Combustion Institute (Bernard Lewis Lectureship)