Case Studies in Process Safety: Ensuring the Basis of Safety Throughout the Development Process Susan Shilcrat Process Safety WW GlaxoSmithKline Pharmaceuticals Organic Process Research & Development 21 March 2012; San Francisco Process Safety Adjusting to New Environment How can we use the safety test to obtain the maximum amount of information with the least amount of effort to improve our success in scale-up? Shorter timelines Same standard for the basis of safety Less resource of time and manpower Consideration of scale Consideration of goals “Nice-to-haves” later in development Case Studies from the Trenches Investigation of an industrial incident N-Halo imide chemistry – achieving a basis of safety despite inherent incompatibilities Investigation of a laboratory runaway Hazards of DMSO chemistry Investigation of an Industrial Process Safety Incident Iodination Incident in Manufacturing I COOH H 3C X Unsaturated Acid + I2 Y Iodine H 3C COOH X I Y Halogenated Component Iodination of unsaturated fatty acid in manufacturing facility Relatively small scale process run only several times/year (160 kg) Dedicated vessel • Acid is preheated to 60 °C using jacket steam • Open manway charge of iodine • Questionable physical state of iodine • Exothermic reaction on iodine dissolution • When the operator finished adding the iodine and tried to close manway port → • Violent ejection of material, ‘like a volcano’ ensued • Iodine vapors were all over the manufacturing and packaging area. Reactor After Incident • • • • • No personnel were injured Plant was evacuated Production halted until investigation was complete Current stockpile of material ~ 2-3 months • Investigation showed mis-calibration of a key thermocouple • Operator added the iodine when temperature indicator was at 60 °C However the real temperature was 77 °C Original Process Safety Hypothesis By adding the iodine to the unsaturated acid at an elevated temperature (77 ºC vs. 60-65 ºC), the reaction mixture reached an unusually high process temperature. – At this elevated temperature, decomposition of the iodine/fatty acid mixture initiated, generating more heat and gaseous by-products. – Unable to close manway due to pressurization. Uncontrolled temperature rise + rapid gas evolution led to a thermal runaway reaction, with material violently ejected from the vessel. Possibility that the process was always on the borderline of process safety incident Characterize typical reaction and model possible runaway scenarios. . Initial Investigation Authentic manufacturing samples were not quickly available – Worked with available chemical supplies Unprocessed halogenated material was not available; only diluted, formulated commercial product. Unfortunately, an authentic sample of the material ejected from the vessel had not been saved. DSC of reaction components showed no areas of concern. Isothermal Reaction Calorimetry via RC1 Rate of Heat Output (W/kg Acid) 120 100 Heat Output if Run at 72 oC 80 60 40 Began with isothermal experiments to simulate actual and mal-op processes: 93-103 kJ/kg acid patr 45-47 °C No gas Mild process without obvious concerns 20 Heat Output if Run at 60 oC 0 Iodine Addition -20 0 1 2 3 Time (h) 4 5 6 Thermal Stability on Material from Isothermal Experiments (ARC) 275 7 250 6 225 5 175 150 4 125 100 3 75 50 2 25 0 1 0 400 800 1200 Time (min) Mild exotherm initiates ~ 205 – 210 °C Moderate pressure rise above 175 °C 1600 2000 Pressure (bar) Temperature (oC) 200 Iodine + Unsaturated Acid: Thermal Stability A mixture of iodine and unsaturated acid was tested for thermal stability under 10 adiabatic conditions. 275 250 9 225 200 7 175 6 150 5 Pressure 125 4 100 Pressure (bara) Temperature (oC) 8 Temperature 3 75 2 50 25 1 0 5 10 15 20 25 30 35 40 45 50 55 Time (h) No significant pressure generation until temperatures > 140 ºC. Several small exothermic events with low self-heating rates each with a temperature rise of 5-10 ºC Results are independent of material source (Aldrich versus Manufacturing site). . 275 10 250 9 225 8 200 7 175 6 Temp Hast. 150 5 125 4 Pressure (bar) Temperature (oC) Possible Metal Contaminants Adiabatic experiments on the acid + iodine using hastelloy and stainless steel test bombs. Pressure Hast 100 3 Temp SS 75 2 Pressure SS 50 1 25 0 0 5 10 15 20 25 30 35 40 45 50 55 60 Time (h) The onset of exothermic activity appears somewhat earlier in stainless steel bombs (90 ºC versus 105 ºC). A rise in pressure also appears somewhat earlier and more rapidly in stainless steel over hastelloy. However, the rate of temperature and pressure rise (max. 0.037 ºC/min and 0.06 bar/min) are not such to cause a rapid ejection of material as seen on site. Iodine + Acid: Reaction Calorimetry on Simulated Incident 150 120 Temp 100 Conv 110 80 90 70 60 50 Heat 30 40 Iodine Addition 10 20 -10 0 Reaction Temperature (oC) or Thermal Conversion (%) Rate of Heat Output (W/kg Oleic Acid) 130 • Iodine was added to the acid at 77 ºC in a single portion. • After 20 min, the jacket temperature was rapidly raised to 110 ºC to simulate the ingress of steam -30 -50 -20 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 Time (h) The addition of iodine initiated an immediate heat maximum (132 W/kg), tapering down over 30 min. Raising the jacket temperature accelerated the rate of heat output, reaching a second maximum at 105 ºC (142 W/kg). – Heat output diminished to zero over 1 h. The overall heat of this operation was 124 kJ/kg starting acid with the capacity to raise the reaction temperature rise 49 ºC. No gas output was observed. Water Contamination? • Process is run in a dedicated vessel cleaned with City water. • ArSST test data of unsaturated acid, iodine and water to examine whether the possible presence of cleaning water catalyzed a decomposition reaction. Observations Under several experimental regimes, no evidence of significant or rapid gas generation was observed until process temperatures > 150 ºC. – The difficulty in closing the manway is not readily explained by pressurization caused by gas generation. The total enthalpy of this process is in the range 93-124 kJ/kg unsaturated acid with the potential to raise the process temperature 45-49 ºC. – Therefore even operating at a higher starting temperature, the maximum achievable reaction temperature should be no greater than 121 ºC. This is below the temperature that pressurization was observed. Mixtures of unsaturated acid + iodine prepared according to the reaction conditions do not show early onset (< 150 ºC) of decomposition. Heating unsaturated acid + iodine under adiabatic conditions shows low rates of exothermic activity and pressure increase which are not consistent with a runaway chemical reaction. Conclusions Following our initial investigations, the loss of containment on site does not appear to be due to a runaway chemical decomposition caused by the inherent thermal instability of the reaction mixture near the operating temperatures. – Process is safe under ordinary operating conditions. – Production can continue on schedule. Other Possible Explanations for Incident: The presence of an unknown contaminant(s) may have catalyzed a decomposition process at a lower temperature, leading to a runaway. – Must consider possible contaminants. Starting temperature was actually > 77 ºC. – Gross over-heating scenarios were not investigated. N-Halo Imide Chemistry – Achieving a Basis Of Safety Despite Inherent Incompatibilities NCS Chlorination: Initial Campaign Halogenation via an N-halo imide (NBS or NCS) can have safety concerns: Control of a batch reaction Possible intense heat spikes due to rapid reaction Examination of this chemistry showed several significant issues: Additional NCS (2x) was needed to push the reaction to completion Heat output showed induction period and appeared autocatalytic Total heat output could bring the reaction to vigorous reflux Intense heat spikes were observed which would not be adequately controlled on scale, leading to possible: Condenser flooding Reactor pressurization Loss of containment Further Process Development 3000 70 65 2500 Temperature 2250 60 2000 Heat: Impure Rxn Mixture 55 1750 1500 50 1250 45 1000 750 40 500 Heat: Clean Rxn Mix 35 250 0 Reaction Temperature (oC) Rate of Heat Output (W/kg) 2750 On development, the unfavourable safety profile was attributed to byproducts from the previous stage By-products not fully characterized / quantitated by existing analytical methods 30 25 35 45 55 65 75 85 95 Time (min) By-products chlorinated preferentially to starting material Secondary reaction contributed to large ΔHrxn Autocatalytic behaviour possibly due to impurity mixture No basis of safety without use-test of authentic input material Total Heat Maximum Heat Rate Adiabatic Temperature Rise Impure Starting Material Use-Test Starting Material 1470 kJ/kg 366 kJ/kg 2965 W/kg 175 W/kg 200 W/L 18 W/L 60 ⁰C 22 ⁰C Next Campaign – Process Changes Chemistry developments in previous stage Changes in order of transformations – Fewer by-products transferred to chlorination R’ changes slightly Changes in solubility necessitate solvent change Tetrahydrofuran instead of isopropyl acetate Excess NCS Safety Test of NCS Chlorination in THF 120 Heat Heat-up & 3 h hold at 60 °C – thermoneutral; IPM very little reaction → Heat to 65 °C Once the reaction mixture was slightly above 60° C The reaction initiated and an intense heat spike was observed. patr = 66 ºC Possibility of loss of containment Heat output (W/kg) 2000 1500 100 80 Conv 60 Temp 1000 40 500 20 0 0 0 1 2 3 Time (hr) 4 5 Thermal Conversion (%) or Reactor Temp (C) 2500 Examination of Heat Profile Reaction had initiated before heat-up Autocatalytic profile No Basis of Safety NCS + THF: Adiabatic Testing 130 125 120 115 110 105 100 95 90 85 80 75 70 65 60 55 50 45 40 35 30 25 20 15 10 3.5 3.3 3.1 Temp Pressure 2.7 2.5 2.3 Immediate interaction upon reaching ambient temp → Search for new solvent system by synthetic chemist & safety scientist Safety & quality concern 2.1 1.9 1.7 0 50 100 150 200 Time (min) 250 300 350 1.5 400 Pressure (bar) Temperature (oC) 2.9 Solvent Screening for Safety & Performance Chemistry ran better than ever in dioxane-water – Reduced impurities Worked to establish basis of safety NCS Interaction with 50-50 Dioxane-Water With SM present, reaction kinetics are favored over NCS-solvent interaction (and gives off much less heat) atr = 15 & 35 ⁰C No permanent gas evolution Mass spectroscopy data on NCS/dioxane interaction shows ring-opening and chlorination of the dioxane moiety Reaction Calorimetry on Desired & Secondary Reactions Comparative RC1 data showing NCS interaction with 50-50 dioxane water • Data plotted from two RC1 experiments, both with 1.4 eq NCS, 10 vol dioxane/water, 80 °C • NCS – dioxane – water • NCS – dioxane – water – SM Kinetic basis of safety established due to system preference for desired reaction to undesired solvent interaction. • Recommendation to plant – ensure all materials (NCS and starting material) have been charged prior to heat-up to avoid undesired interaction Another example of N-Halosuccimide Interaction First Scaleup Campaign: NBS showed an initial incompatibility with the solvent of choice – Switched to chlorinated solvent Aprotic solvent necessary for second transformation A cumbersome solvent swap was performed between the two transformations Scale-up required adjustments and simplifications to two-solvent system NBS + DMA 100 90 Temperature deg C 80 Chemistry preference for dimethylacetamide as aprotic solvent Experiment 1 70 Experiment 2 60 50 40 Onset PATRa 40 °C 35 °C 23 °C 25 °C 30 20 0 100 200 300 400 500 600 700 800 Time (min) No significant improvement in the safety profile using recrystallized NBS The maximum rate of self-heating was moderate. No significant pressurization was observed; – Decomposition did not produced permanent gaseous by-products. Reaction Chemistry 120 1800 Rxn 1: NBS solid Addition 1600 Heat Rate of Heat Output (W/kg) 1400 Conv 80 1200 Addn 1000 60 800 40 600 400 20 200 NBS Addition (%) or Thermal Conversion (%) 100 Rxn 2: Substrate Soln Addition 0 120 Heat 180 -200 Addn -20 0 5 10 15 20 25 30 35 40 45 100 160 50 Time (min) Conv 80 Rate of Heat Output (W/kg) 140 120 60 100 40 80 60 20 40 0 20 0 -20 0 patr 20 ºC for both transformations 5 10 15 20 25 30 Time (min) 35 40 45 50 55 Reagent Addition (%) or Thermal Conversion (%) 200 0 Sub-Ambient Calorimetry at Two Concentrations NBS-DMA Experiment 7.3 wt % NBS in DMA 13.0 wt % NBS in DMA Observed Onset Temp 32. 47 ºC 34.03 ºC Self-Heat Rate at To 0.023 ºC/min 0.031 ºC/min Temp at Max. Rate 52.03 ºC 74.10 ºC Maximum Self-Heat Rate 0.343 ºC/min 2.43 ºC/min Final Adiabatic Temp 67.20 ºC 121.20 ºC Adiabatic Temp Rise 34.80 ºC 80.30 ºC Adiabatic Temp Rise (Φ corrected) 60.55 ºC 130.09 ºC Final Adiabatic Temp 93.02 ºC 164.12 ºC Enthalpy of Reaction 400 kJ/mol 400 kJ/mol Time to Maximum Rate @ 20 ºC 10 hours 5 hours Time to Maximum Rate @ 0 ºC 300 hours; 12.5 days 66.7 hours; 2.77 days The onset of exothermic activity not concentration dependent Adiabatic temperature rise, maximum self-heating rate, moderate, concentration dependent Worse-case analysis of the unintended reaction of NBS and dimethylacetamide must be dependent on process concentration RC1 Experiments on NBS/DMA Stability NBS into DMA @ 20 °C: Color change – Clear to orange between 2-3 h Heat spike; 250 kJ/kg; patr =10 °C; maximum 67 W/kg. – Orange solution no longer reactive with starting material NBS in DMA at ambient temperature has limited stability of a few hours. NBS into DMA a @ 0 °C and held for 24 hours. A color change to bright orange by the end of the experiment – Thermal activity was below the level of instrument noise – Use test unsuccessful Solution NBS in DMA at 0 ºC also has limited stability of less than 24 hours. Basis of Safety NBS in dimethylacetamide = unstable, self-heating mixture, but This combination safe in the proposed process – NBS will react preferentially with starting material over the solvent DMA. Worse-case scenario of total decomposition of NBS in DMA: – Heat generation will not bring the reaction mixture into a critical, runaway situation Loss of cooling = loss of batch Caveats: Changes in reaction concentration and/or processing temperature must be reevaluated with respect to the stability of the NBS + DMA interaction NBS must not be premixed with dimethylacetamide. – Any proposed process change involving a solution charge of NBS must be thoroughly evaluated for safety implications. Another Incompatibility Example A request for process safety examination on an older process Feedstock of pyrophosphoryl chloride in THF – Concentrated feedstock – No competing reactions to minimize interaction More potentially hazardous than previous process – Possibility of many maloperations to initiate interaction – Feedstock prepared too soon, forgotten, remainder sent to inappropriate waste, etc. Warm climate at processing facility Cl Cl O O P P O Cl Cl Initial ARC Experiment on Feedstock 110 Exp terminated 100 90 Temperature (oC) 80 70 60 50 40 30 20 0 25 50 75 100 125 150 Time (min) Initial exotherm on mixing proceeds to runaway Onset at ambient temperatures 20-30 ºC 175 200 Cryogenic ARC Experiment Cryogenic ARC Experiment Approximately 13 °C, an exotherm is detected At ~65 °C, the rate of self-heating increased Once the sample reaches 170 °C an exothermic runaway occurs with a huge pressure increase Significant permanent pressure is generated due to by-product, noncondensable gas(es) of unknown identity A very large adiabatic temperature rise (> 450 °C) is achievable Onset Temperature Rate at 12.77 °C = 0.21 °C/min The temperature rate at 13 °C is very close to the sensitivity threshold Some self-heating can be detected as low as 1 °C Time to Maximum Rate At 30 °C, the maximum reaction rate will be reached in approximately 460 min or 7.7 h. (TMR @ 13 °C = 16.9 h) < 8 h to detect the mal-operation, take corrective action, and mitigate the mal-operation before a catastrophic situation is reached. – A safety critical temperature in a real system will be achieved earlier An excess of pyrophosphoryl chloride is used in this process. – Post-addition reaction mixture contains a potentially self-heating interaction of this reagent with THF – Low operating temperature (-10 °C) – Higher dilution (5.8 wt%) – Mal-operations prior to the aqueous quench can lead to a loss of control Loss of cooling, overcharge of reagent solution, prolonged hold Protective Basis of Safety Feedstock not admixed until actual reaction vessel charged and at operating temperature Feed tank vented with temperature monitoring Emergency Quench tank prepared Defined temperature of emergency response Immediate post-reaction feed tank rinse to quench tank Investigation of a Laboratory Runaway Hazards of a Hydrogen Peroxide Epoxidation Runaway Laboratory Oxidation 5 g Epoxidation run – No incident – Mediocre yield and quality 30 g Epoxidation setup to run overnight at ambient temp – Chemist returns in morning to find mess Another 30 g experiment setup – Short induction period – Violent ejection from flask – Flask hot to touch; gas evolution uncertain Reaction Chemistry Literature Reference: OH + H2O2 H2WO4 NaOAc O OH MeOH 2, 4-pentadien-1-ol Actual Reaction: OH + H2O2 t rans, t rans-2, 4-hexadien-1-ol H2WO4 NaOAc O OH MeOH target epoxide OMe MeO OH ring-opened by-product Reaction Conditions OH + H2O2 t rans, t rans-2, 4-hexadien-1-ol H2WO4, NaOAc MeOH Batch reaction Allylic alcohol added last in one portion – 1.5 eq 30% Hydrogen peroxide – 4 Vol methanol – 0.1 eq Tungstic acid – 0.3 eq Sodium acetate Catalyst loading 10X literature Target epoxide – Only one literature reference – No preparative procedure O OH target epoxide Preliminary Analysis: Theoretical Calculations Olefin + H2O2 → Epoxide + water – CHETAH & Examples of ∆Hf from the literature ΔHrxn = -235 kJ/mol; ΔTad = 154 ºC Heat sufficient to bring reaction to reflux; vaporize 56% of total methanol Transformation of target epoxide to ring-open by-product; Low enthalpy process: ≈ -20 kJ/mol Theoretical Models of the Enthalpy of Epoxidation Olefin OH + H2O2 OH O OH + H2O ΔHf (kJ/mol) -172 4.3 Epoxide O OH O 53 O ΔHf (kJ/mol) ΔHrxn (kJ/mol) -298 -224 -131 -233 -98 -249 Reaction Calorimetry 110 650 H2 O2 Addn Thermal Conv. Total Heat Output: – -257 kJ/mol Adiabatic temp rise: – 155 ºC Maximum rate: – 588 W/kg Rate of Heat Output (W/kg) 550 90 80 450 70 Heat 60 350 50 250 40 30 150 20 10 50 0 -50 0 1 2 3 4 5 Reaction Time (h) No gas evolution 5% heat during addition Final product: – 9% starting material & 91% ring-opened by-product Autocatalytic profile Vessel with good heat transfer runaway is avoided 6 7 8 -10 H2O2 Addition or Thermal Conversion (%) 100 Heat Profile: 2 Hour Addition Time Longer addition time: Still poor control Thermal Stability Studies: ArSST ARSST Experiments on Reaction Mixtures Entry Experimental Conditions Self-Heat Rate (°C/min) Self-Pressurization Rate (psi/min) Maximum Rate Initiation Temperature Maximum rate Initiation Temperature Pressure Gain (psi) 1 0.01 eq H2WO4; no 1 82 47 ºC 2.4 42 ºC 1.7 2 0.1 eq H2WO4; no 1 71 40 ºC 11.5 33 °C 11.6 3 0.1 eq Na2WO4; no 1 8300 45 ºC 618 40 ºC 13.7 4 0.01 eq H2WO4; with 1 4970 40 ºC 70 35 ºC 0.3 5 0.1 eq H2WO4; with 1 6760 42 ºC 480 35 ºC 0.3 6 No H2WO4 catalyst; with 1 314 65 ºC 309 55 ºC 4.9 7 No H2WO4 catalyst; no 1 15 105 °C 7.9 95 °C 12 8 0.1 eq H2WO4; no 1; no methanol 3730 45 °C 7510 35 °C 30 Blank Experiment 1: 0.1 eq H2WO4 Self-Heat & Pressurization Rates o C/min 10 psi/min Pressure vs Temperature 1 315 310 0.1 20 25 30 35 40 45 50 55 60 65 70 75 Temperature (oC) 80 85 90 95 100 105 110 115 Pressure (psig) Self-Heat Rate (oC/min) & Self-Pressurization Rate (psi/min) 100 305 306 psig 300 295 294.4 psig 290 20 30 40 50 60 70 80 Temperature (oC) 90 100 110 120 Blank Reactions: Hydrogen Peroxide decomposition at low temperatures – Surprisingly low onset (33-47 ºC) for common synthetic methodology – Pressure more sensitive than heat – Tungsten/H2O2: Not specifically reported in literature Gaseous decomposition – Permanent non-condensable gas – ~1/3 of hydrogen peroxide loading Severity of runaway function of catalyst loading Sodium tungstate far more severe than tungstic acid/sodium acetate – Buffered pH = stabilizing effect Self-Heat Rate (oC/min) & Self-Pressurization Rate (psi/min) 100 o C/min 10 psi/min 1 0.1 20 25 30 35 40 45 50 55 60 65 70 75 Temperature (oC) 80 85 90 95 100 105 110 115 Decomposition in the Presence of Hexadien-1-ol Self-Heat & Pressurization Rates – 0.01 eq W 10000 100 o 1000 10 100 10 1 1 0.1 0.1 Self-Heat & Pressurization Rates – 0.1 eq W 100000 1000 psi/min 0.01 20 40 60 80 100 120 140 160 10000 180 o C/min Temperature (oC) 100 1000 10 100 1 10 0.1 1 0.01 0.1 0.01 0.001 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 Temperature (oC) Self-Pressurization Rate (psi/min) 0.01 Self-Heating Rate (oC/min) Self-heat Rate (oC/min) psi/min Self-Pressurization Rate (psi/min) C/min Hexadien-1-ol reactions Extreme self-heating scenarios Little or no non-condensable gas formation Hydrogen peroxide not available for decomposition Target epoxide thermally unstable at low onset temperatures (~40 ºC) 100000 10000 10000 Self-Heat Rate (oC/min) 1000 100 100 10 10 Heat & Pressure: Blank 1 1 0.1 0.1 0.01 0.01 20 30 40 50 60 70 80 Reaction Temperature (oC) 90 100 110 Self-Pressurization Rate (psi/min) 1000 Heat & Pressure: with Sorbitol Root Causes Moderately exothermic batch reaction with high patr, high catalyst loading, and autocatalytic nature Tungsten catalyzed decomposition of hydrogen peroxide at fairly low temperatures – Possible non-condensable gas formation Thermal and chemical instability of target molecule Poor cooling capacity of vessel configuration in lab; TNR exceeded = Classic runaway scenario! Competing secondary decomposition processes Metal catalyzed decomposition of H2O2 Decomposition of target epoxide Recommendations for Laboratory Risk Assessments Chemistry unsuitable for scale-up – Target epoxide probably not a suitable synthetic intermediate No procedural changes necessary Incident caused by combination of factors which due diligence and desk screening by laboratory chemist would not detect. – Some complacency in following literature procedure for closely related molecule Ordinary laboratory safety procedures mitigated consequence to cleanup job Existing procedures adequately addressed consequences of unknown chemical interaction. However, Although tungsten catalyst + H2O2 is a green reagent, the development chemist must be aware of potential hazards and barriers to scale-up Heterogeneous Batch Reactions in DMSO at Elevated Temperatures Or ‘Almost Fooled’ Reaction & Proposed Mechanism X + ArOH Br X Cs2CO3, Cu2O Ar N,N-dimethylglycine DMSO 3 vol. > 120 oC + CsHCO3 + CsBr O + CuO + OH N O Heterogeneous batch reaction heated to 120 oC for 5-10 hrs Cu(I) O Br Ar R Cu(III) Br Reductive elimination -Cu(I) Oxidative addition (2e- are added to Br-Ar bond to make electrophilic species) Ar O-Cs+ Cu(III) O R O Ar +CsBr + HCsCO3 O N Cu(I) N,N-Dimethylglycine is added to make Cu more nucleophilic DMSO is Not an Inert Solvent DMSO reacts violently or explosively with: – Acetyl chloride, benzenesulfonyl chloride, cyanuric chloride, phosphorus trichloride, phosphoryl chloride, tetrachlorosilane, sulfur dichloride, disulfur dichloride, sulfuryl chloride or thionyl chloride. Violent reactions are explained in terms of exothermic polymerisation of formaldehyde produced under a variety of conditions by interaction of the sulfoxide with reactive halides, strong acidic or basic reagents It is believed that HBr catalyses DMSO decomposition at elevated temperatures, which is then accelerated by its decomposition product, dimethylsulfide, resulting in rapid temperature and pressure rises. The autocatalytic nature of the DMSO decomposition in presence of cesium bromide is also described Thermal Stability of DMSO by ARC 280 60.0 DMSO 260 240 50.0 220 Temperature (°C) 180 160 140 40.0 30.0 120 100 20.0 80 60 Temperature 40 20 Pressure Recommended maximum processing temp ~130 °C Various 'additives' can dramatically reduce this figure 0 0 200 400 600 Time (min.) 800 1000 10.0 0.0 1200 Pressure (bara) oC Exothermic decomposition from 162 Exotherm detected from 162°C Significant gas generation Runaway near the boiling point 200 Thermal stability by DSC No significant issues in the thermal stability of individual components Real issue in stability of DMSO at elevated temperatures in the reaction matrix Isoage @ 120 oC (Tj) for 24 hrs, then HWS 160oC 135oC Exotherm detected from 120 oC, came out of exotherm after ~13.5 hrs ∴ Lower the process temperature to 100 °C RC1: Batch Process @ 100 oC 105 120 95 Temperature 100 80 75 65 60 55 40 45 20 Heat Flow 35 25 0 0 20 40 60 80 100 120 140 160 180 200 Time (mins) • Heat-up over 30 min: ∆Ta d= 12 oC, ∆H = 120 kJ/kg • Remaining rxn: ∆ Tad = 18 oC, ∆H = 180 kJ/kg • No gas generated • Reaction was complete by HPLC after 6 hrs @ 100 oC Temperature [kg] Heat Flow [W/kg] 85 DMSO + Additives DMSO + Cs2CO3 HWS No safety problems observed near reaction temperature of 100 oC → Reaction appeared fully characterized for safety DMSO + Cs2CO3 + Cu2O + dimethylglycine Last Set of Experiments: VSP – Stirred System 600 1600 1400 1200 400 1000 300 800 600 200 400 100 0 Pressure (psia) Temperature (oC) 500 200 0 200 400 Time (mins) 600 800 0 Batch @ 110 °C overnight: Slow exotherm After ~10 h; rapid exotherm, rapid pressure; – > 1000 psi/min – Bursting of test container – Probably autocatalytic Onset problematic, dependent on reaction conditions Cooling failure on scale could lead to violent loss of containment In a system with agitation, previously undetected exotherm is revealed Assumed to be due to better contact between DMSO & CsBr VSP2, Cut-off @ 110 oC, Runaway after~ 10 hours 600 1600 1400 1200 400 1000 300 800 600 200 400 100 0 200 0 200 400 Time (mins) 600 800 Pressure (psia) Temperature (oC) 500 0 • No Basis of Safety • Move away from DMSO & select alternative aprotic solvent • Chemist quickly reworks process Modified chemistry (NMP) VSP2, cut-off @ 130oC X + ArOH N,N-dimethylglycine NMP 3 vol. > 120 oC Ar + CsHCO3 + CsBr O + CuO 160 140 140 manual cool-dow n initiated 120 Temperature ( o C) 120 100 100 80 80 60 60 40 40 20 20 0 0 200 400 600 800 Time (min) 1000 1200 1400 + OH N O 0 1600 Pressure (psia) Br X Cs2CO3, Cu2O Conclusions DSC & ARC data may not be sufficient for assessment of heterogeneous batch reactions – Combination of techniques to establish basis of safety Heterogeneous batch reaction to be investigated under adiabatic conditions with agitation Discourage the use of DMSO at elevated temperatures in presence of reactive halides Early Process Safety Incident: 1386: Geoffrey Chaucer, Canterbury Tales: THE CANON’S YEOMAN’S TALE It happens, like as not, There’s an explosion and good-bye to the pot! These metals are so violent when they split Our very walls can scarce standup to it…… Some said the way the fire was made was wrong; Others said, “No - the bellows. Blown too strong.”…… I’ve no idea why the thing went wrong; Recriminations through the air were hot and strong. “Well,” said my lord, “there’s nothing more to do. I’ll note these dangers for another brew;” Thanks to the WW Process Safety Team at GSK