Prediction of Reaction Enthalpy and Adiabatic Temperature Rise
Effective management of the heat released from a ch emical process is critical for the safe, successful scale-up of chemical processes. These process heats are used to determine engineering design parameters such as the required jacket temperature(s), the heat load to be handled by heat echangers or condensers, and the adiabatic temperature rise of the desired process. !or an average heat capacity, cp, over the temperature range applicable, one may conveniently calculate the adiabatic temperature rise, deltaTad, as follo"s# delta_Tad = Qrxn/(m.p!
"here m is the reaction mass, and $%rn $% rn is the total heat liberated from the chemical process or reaction. The quantity $%rn is calculated from the enthalpy of reaction delta%r%& and the appropriate quantities of reactants. 'f this adiabatic temperature rise is sufficient to trigger undesired reactions, e.g. a side-reaction or a decomposition, the heat released from these undesired reactions must also be determined. The adiabatic temperature rises for "orst-case scenarios can then be estimated. The accurate and timely determination of the heats of reaction for the desired process and for any plausible undesired processes is clearly a key step in the overall reactive chemicals evaluation process. !urthermore, the process heat (heat of reaction for eample) may very "ell be the most important piece of information needed to accurately assess and engineer the reactive chemicals haard. 't must be emphasised that an eperimental determination of the process heat may be the best approach. This is true, of course, if the means are available (i.e. an analytical technique to determine the etent of reaction and the reaction can be carried out close to the actual conditions of the process). &o"ever there are many cases "here the eperimental approach is not n ot the most feasible. !or eample, there may be safety considerations to contend "ith or eperimental challenges "hich make an eperimental determination unfeasible. The reader is referred to a more detailed discussion discussion of these considerations *it.#reference (+). 't may be desirable to perform both an eperiment and a calculation. 'f both approaches give similar results, then the scale-up of the process can be carried out "ith more confidence in the epected release of energy. The theoretical treatment has its o"n pitfalls one needs to be a"are of. ne limitation of the estimation approach is the lack of accurate data for many important species. nother limitation is the fact that the estimated heats are determined for the chemistry "hich is believed to occur in the reaction vessel. There may be cases "he re one neglects an important source of heat h eat (salt formation for eample) in the calculation but, an eperimental determination "ould probably detect this heat. !inally, the theoretical treatment does not provide information on rate of heat release, a parameter "hich might be critical to the process engineering design. 'n some cases the rate of heat release may be the overriding factor for process control. Thus techniques like /eaction 0alorimetry, "hich carry out the chemical process under conditions identical to the intended, large scale, process, may be the best approach regardless of the (unquantified) etent of
reaction. lso, techniques such as adiabatic calorimetry can yield invaluable information about heat and pressure releases for unkno"n chemical ch emical processes "hich take place at temperatures "ell above those intended in the process due to, for eample, an undesired temper ecursion.
Estimation of Reaction "eats #sin$ %ond Ener$ies
The energy required to break a chemical bond and separate the fragments to infinite distance in the gas phase at ero 1elvin is a common definition of bond energies. en ergies. 2ond energy tables using that definition cannot be used to predict heats of formation but may be used to predict heats of reaction (at ero 1elvin) by summing the bond energies for bonds broken and subtracting the sum of bond energies for bonds formed (being ca reful to account for any additional energy effects such as ring strain - see see discussion in *it.#reference (+) )# equation for delta%r%& n eample of the bond energy approach is given as follo"s# Eample for the Estimation of /eaction &eats 3sing 2ond Energies !or the hypothetical reaction# 4 b 2 --5 c 6 4 d 6 the reaction enthalpy, deltar& deltar& is determined from the standard heats of formation, deltaf&7, of the species in the reaction according to# equation for calculating delta%/%& The 2ond Energy technique is not very accurate (typically 8 +9 kcal:mol). The key to estimation of reaction heats lies in the accurate kno"ledge of the standard standard heats of formation, deltaf&7 of the reactants and products. more more useful approach is that of *it.#2enson (+;) "ho deals "ith partial bond contributions to the gas-phase heat of formation at <=>.+ <=>.+ 1. 2asically the bond contributions are groups "hich can be added together to predict either the heat of formation of a molecule or the heat of reaction directly. ?ore accurate techniques are described belo".
Estimation 3sing a ?ore !ormalised !o rmalised Thermodynamic pproach There are numerous methods and resources for finding or estimating the enthalpy of formation of chemical species. 't is certainly beyond the scop e of this guide to list these. &o"ever, these are summarised and discussed in great detail in reference +. @erhaps the most comprehensive computer program for thermodynamic estimations is the aforementioned 0&ET& code. !or eperimental thermodynamic data, the A'BT Ceb Ceb 2ook, an internet resource is very helpful. !or
inorganic reactions he !innish &B0 0ode *it.#reference (+=) is useful. @ublic literature sources for thermodynamic data are numerous but an ecellent resource for organic data is @edleyDs compilation *it.#reference (+>).
common method is to calculate the reaction heat for a structurally simple reaction, one that is an analogue to the actual chemistry and "here there are available thermodynamic data from the literature for the reaction constituents (reactants and products). This so-called nalog /eaction method is described in more detail in *it.#reference (+).
The eample belo" illustrates the methodology to estimate a reaction heat using the procedure outlined in *it.#reference (+).
&'ERE)T'A* +A))'), -ET"&+
EAE/* !ET3/EB ! 6B0:6T TEBT'A 'n a typical 6B0 or 6T eperiment a fe" milligrams of sample (often in a sealed metal pan) and an inert reference material are heated together at a defined rate of typically +9 degree 0elsius per minute. The temperature of the sample and the reference are monitored and if the temperature of the sample begins to diverge from that of the reference, evidence of thermal activity is inferred. 3sing this data it is possible to determine an onset temperatureF for the thermal event and measure the amount of heat released. 3sing this information the onset temperatureF can be co mpared "ith the materialsG proposed operating or storage temperature. 'f these are similar, it might represent a potential haard. 'n a similar manner the information relating to energy release can be used to estimate the likely temperature and pressure rise if the eothermic reaction did occur. 't is through this type of comparative analysis that 6B0:6T data can be used to estimate the severity of a possible haard. The principle benefit of this type of testing is that it can be conducted in just a fe" hours, requiring limited operator skill. Chen lo" running cost is also taken into accoun t, the "idespread use of 6B0:6T methods can be easily understood. &o"ever, in practice these aspects should be "eighed against the many disadvantages that relate to the possible
reliability of the data and the potential problems that c an ensue from not screening reactions properly.
*'?'TT'AB ! 6'!!E/EAT'* B0AA'A 6T The differential thermal instruments described above are employed for a "ide range of analytical determinations in addition to haard screening. 'n fact most 6B0 and 6T applications do not relate to haard assessment at all and the comments made in this pap er "ill not be concerned "ith these alternative functions. design feature that contributes significantly to the ease of use and lo" running cost of 6B0 and 6T is the small mass of sample required "hich is typically in the milligram scale. !rom a haard evaluation perspective, this e tremely lo" sample mass can be a serious disadvantage due to the increased uncertainty in eperimental reproducibility. Chilst the testing of pure materials presents no problems, taking a representative sample of a miture on such a small scale can be difficult. !or eample, in order to evaluate process intermediates. (and sometimes, even products), it is frequently necessary to dra" a live sample from a reactor. The etraction of such lo" sample masses that are "holly representative of the system under these conditions can be difficult. ne of the most important pieces of information obtained from thermal scanning devices is the so-called Honset temperatureF for eothermic activity. &o"ever, it is important to realise that this HonsetF threshold is not a fundamental property of a reaction and the measured value depends very much on the instrument sensitivity and the procedure by "hich the eperiment is performed. 'n !igure +; , the onset temperature determined by 6B0 is compared "ith that from an adiabatic instrument (in this case the Haccelerating rate calorimeterF, /0) using approimately g of sample . This information "as reported by the 6o" 0hemical 0ompan y from their historical data bank. 't sho"s that in many cases, the Honset temperatureF detected by 6B0 can be as much as 9 o0 higher than that reported from adiabatic instruments. 'n fact in a substantial number of instances the difference bet"een to t"o methods is as much as +99 o0. (There are several cases "here the
6B0 determined Honset temperatureF is lo"er than that reported by the adiabatic testing, but this can be attributed to the study of non-representative 6B0 samples).
&+ s. AR &etection Temperatures
!igure +# comparison of Honset temperaturesF bet"een 6B0 and /0 data. The most significant shortcoming of a 6B0 or 6T type instrument is that they do not provide any information regarding the pressures generated or the rates of pressure rises measured during a screening a test. The authors are a"are that some effort has been made recently to address this problem and produce pressure measuring 6B0 devices, but so far the data reported has been found to be unclear and noisy and such instruments are limited to only a fe" psi. &o"ever, if the 6B0 is to be considered a s an instrument for accurate haard assessment and safe process scale-up this information is crucial, since the etent of thermal haard "ill be directly represented by the pressure generated during the reaction. 't is after allit is the large pressures generated in runa"ay reactions that generally cause product venting and in etreme cases catastrophic damage to plant and equipment.
nother important factor that needs to be considered "hen eamining the haard potential of a material is the aspect of time. 't is vital to remember that it is not only the amount of energy released that is important but also the rate at "hich this energy is released (i.e. the reaction kinetics). lthough the aspect of 6B0 scans do partially reflect the kinetics of the chemical system under study the information that is ob tained is very often indirect and can be applied in a qualitative manner. The time scales determined bear almost no relation to the real life incident and unless considerable effort and rigour is pu t into the kinetic analysis, it is not possible to estimate the rates at "hich events might occur on the plant. Taking into account all the above points, it is clear that choice of 6B0 or 6T, as the primary, or "orse still only, method of thermal screening can be misleading and if used "ithout information from additional adiabatic calorimeters could be potentially dangerous. it is the large pressures generated in runa"ay reactions that generally cause product venting and in etreme cases catastrophic damage to plant and equipment. nother important factor that needs to be considered "hen eamining the haard potential of a material is the aspect of time. 't is vital to remember that it is not only the amount of energy released that is important but also the rate at "hich this energy is released (i.e. the reaction kinetics). lthough the aspect of 6B0 scans do partially reflect the kinetics of the chemical system under study the information that is ob tained is very often indirect and can be applied in a qualitative manner. The time scales determined bear almost no relation to the real life incident and unless considerable effort and rigour is pu t into the kinetic analysis, it is not possible to estimate the rates at "hich events might occur on the plant. Taking into account all the above points, it is clear that choice of 6B0 or 6T, as the primary, or "orse still only, method of thermal screening can be misleading and if used "ithout information from additional adiabatic calorimeters could be potentially dangerous.
2ased on results of the first test, it "as decided that the t"o starting ingredients could be safely pre-mied in drums ready for charging. &o"ever, "hen this "as performed the drums "ere found to rupture after about < hours. Bubsequent testing "ith an adiabatic device (in "hich pressure data "as measured) revealed that "hile the 6B0 had correctly reported a lack of eothermic activity, ho"ever, it had failed to register the fact that, even at room temperature a considerable amount of gas "as being generated albeit at a slo" rate. This highlights the necessity for pressure data "hen scaling up a process.
T&E TB3 B B0/EEA'A T* EBBEAT'* !ET3/EB The TB3 (Thermal Bcreening 3nit) has been developed as an alternative to 6B0 as a primary thermal screening method. schematic diagram and a photograph of the TB3 are sho"n belo" in !igure I.
The sample is contained in a pressure tight metal (or glass) test cell, suspended in the middle of an HovenF. The oven consists of a metal cylinder "ith a heating coil "rapped around the outer surface that is heated at a user-defined rate. n performing a test the user controls the ramp rate(s) of the oven. fter an initial delay due to Hthermal laggingF effects the sample temperature "ill be found to follo" the oven ramp at the same rate "ith a slight HoffsetF ("hich "ill depend on the physical characteristics of the test material such as specific heat). Chen an eothermic or endothermic process is detected the sample temperature "ill be found to deviate from the background-heating rate identifying the Honset temperatureF. The rate of rise in sample temperature (dT:dt) and the maimum va lue reached, T?a before returning to the background-heating rate reflects important characteristics of the thermal event. 'n addition to temperature data, the thermal screening unit is also equipped "ith a pressure transducer that records changes in sample pressure as the reaction proceeds. This provides the operator "ith a second method by "hich sample activity can be identified. This alternative method of sample analysis is particularly useful since it provides a measure of the total pressure generated in the reaction, @?a and the rate of pressure rise (d@:dt). The pressure data also enables very mild eothermic decomposition reactions "hich result in the
production of non-condensable gas to be detected even if the associated temperature rise is too lo" to be reliably detected.
0?@/'BA ! T&E TB3 C'T& 6'2T'0 0*/'?ETE/B diabatic calorimeters such as the /0 or @&'-TE0 ''J are traditionally considered as the most reliable tool for eotherm detection and for genera ting quantitative kinetic and thermodynamic data "ith regards reaction severity. 'n a typical eperiment adiabatic conditions are achieved by taking a sample cell similar to that used in the thermal screening unit and placing it bet"een a set of Hguard heatersF that precisely match the sample temperature. Tests are then performed by heating the chemical in temperature steps, (typically +9 to + 70) holding the sample for a defined period of time and monitoring the sample for evidence of self-heating. 'f an eotherm is indeed detected then the guard heaters "ill follo" and thus maintain an adiabatic environment during the reaction runa"ay. The thermal screening unit possess many of the d esirable characteristics of an adiabatic calorimeter. The system contains a thermocouple "ithin the test cell to enable direct sample temperature measurement and the test cell itself is pressure resistant to over <99 bar and has a thickness of 9.K mm. The screening unit also uses more representative sample sies than either 6B0 or 6T again similar to that employed in an adiabatic system. The thermal screening unit, ho"ever, is considerably smaller than an adiabatic system and also much simpler to set-up and operate, often enabling many runs to be performed in a single "orking day or night. Chen this is coupled "ith the ability to operate multiple screening units (up to four) from a single computer interface and the scales of economy involved bet"een the t"o units it makes the screening unit more suited to rapid screening of multiple compounds 0learly the thermal screening unit has not been designed to act in competition "ith /0 or @&'-TE0 '' type instruments but instead helps to fill the gap in the market for a device that is "holly concerned "ith routine, fast and multiple ha ard screening.
0A0*3B'AB lthough 6B0 and 6T techniques have been applied "ith great success to many areas of chemical testing and process development their application to thermal haard screening has in recent years caused much cause for concern. This is particularly true in the modern chemical manufacturing industry "here the pressures on process scale-up and development organisations are continually increasing but the development-cycle time scales are under a constant squeee. This has resulted in the increasing tendenc y for quick, single test procedures possibly in combination "ith reaction calorimetry for thermal stability assessment, "hich "ithout careful interpretation and eperience cannot p rovide a sufficient guarantee that haardous processes "ill not slip through the net. The thermal screening unit has been developed as tool to address some of these problems associated "ith modern haard screening and provide an alternative instrument to 6B0 and 6T. 'mportantly the unit employs representative sample sies in the range 9. to g. These sample sies not only provide more realistic information for scale up but also enable a "hole range of processes to be eamined including heterogeneous systems, air sensitive materials, starting materials and intermediates. The unit also importantly provides pressure information, "hich enables the study of product storage and provides information on epected reactor pressures if a process "ere to undergo a runa"ay reaction. 2y eamining information such as dT:dt?a and d@:dt?a it is also possible to start to gain information on the rates of energy release from a systems. This latter information is beneficial for the design of vent lines and reactor relief systems and "ill reduce the number of samples that "ill have to be eamined by an adiabatic calorimeter.
Time to maximum rate (TMRad) Time taken for a material to self heat to the maimum rate of decomposition from a specific temperature, under adiabatic conditions.
While much of the emphasis on safety in the pharmaceutical industry relates to ensuring patient safety, as should be the case, often other crucial aspects of safety — in particular process safety — receive less attention than is appropriate. With increasing pressure to deliver faster turnaround times and lower cost, the performance of comprehensive process safety evaluations, which are time consuming and expensive, can be glossed over by contract development and manufacturing organizations (CD!s", most often due to a lac# of awareness and understanding of their importance. $nattention to process safety can, however, lead to devastating conse%uences.
$t is essential that CD!s conduct all pharmaceutical manufacturing processes in a safe manner. &o do so re%uires an effective process safety management system and strategy within a culture that emphasizes safety. 'rior to acceptance, proposed proects must be evaluated to determine whether the capabilities of the CD! are ade%uate, and only those proects for which suitable facilities, e%uipment and s#illed personnel are available should be accepted. !nce a proect is underway, both theoretical and physical analyses must be conducted to determine the thermodynamic and #inetic properties of all materials involved in the process and the process itself, both under no rmal and worst case scenario conditions. !nly with access to this information can the behavior of a process be fully understood with appropriate engineering, safety controls and procedures implemented. &he failure to establish an effective basis of safety can lead to inade%uate process design and protection of operators and, in the most severe cases, the surrounding community and environment.
&he types of processes conducted by CD!s range widely, and conse%uently so do the potential hazards and ris#s they pose. )ome processes, such as those that involve the use of unstable raw materials, exothermic reactions and * or the p roduction of non+condensable gases, are more hazardous than others. $n addition, as reactions are scaled from the lab to the pilot plant and then commercial volumes, the ris# they pose increases.
ot all CD!s are e%uipped to safely manage every possible process re%uired for the production of pharmaceuticals.
&herefore, the first step in establishing an effective process safety strategy is determination of the company-s capabilities — and limitations — with respect to handling process hazards. $n general, a paper
assessment of the hazards presented by a potential new proect, including the potency of the compounds and the potential for highly energetic chemistry * highly hazardous reactivity, should be suitable for determining whether the process presents hazards beyond what the CD! is e%uipped to safely manage. willingness to reect potential proects based on such a safety evaluation is the foundation of an effective safety management strategy.
$f a proect is deemed within the bounds of the CD!-s capabilities, the proposal * %uote submitted to the client should include an outline of all anticipated process safety testing (and associated costs" needed to establish an appropriate basis of safety and potential safety measures.
Comprehensive safety evaluations should begin immediately once a proect is accepted and run parallel to process development activities. &he benefits of this approach are numerous — not only is the identification of any potential ris#s * hazards achieved prior to scale+up/ any necessary changes to the process can be completed prior to process scale+up.
&he comprehensive hazard evaluation should identify both desired and undesired potential material and reaction hazards. &hermal stability testing of the materials and mixtures used in the process is completed using instruments such as a differential scanning calorimeter (D)C", a thermal screening unit (&)u" and * or accelerating rate calorimeter (0C". &he enthalpies of the intended synthetic reactions can be obtained by either estimation techni%ues using available thermodynamic data, or measured with a reaction calorimeter. &he generated data is then used to identify any process hazards and establish a defined basis of safety for each that will minimize the li#elihood of adverse events and, where necessary, provide protection to operators and the environment from a ny potential event that may occur.
!nce a proect is accepted, the CD! should conduct a more thorough paper assessment, considering all of the functional groups of the molecules involved and the process conditions. $f the reaction is sufficiently simple, this phase may include estimation of the heat of reaction using heat of formation data for analogous reactants and products ta#en from the literature. $f no concerns are raised, then reaction 1
calorimetry testing may be deferred until calolater in process d evelopment, so the testing will reflect the process as it will be scaled+up. $f there is any %uestion about the potential stability of the materials in a process, D)C is performed on individual starting materials * reagents * products and * or reaction mixtures. &he sample is heated at a constant rate, and the heat flows to (endothermic change" and from (exothermic change" the sample are recorded as a function of temperature and time. D)C scan provides information about phase changes, decomposition or other self+reactivity behavior of the sample and whether these events occur exothermically or endothermically. 1or reliable results in safety testing, it is crucial that closed pressure rated crucibles be used for these types of D)C experiments.
&here are limitations to D)C methods, however. 1irst, the 2onset3 temperature can vary depending on the instrument sensitivity and the conditions under which the test was conducted. )econd, D)C does not 2
provide any information on changes in pressure, and it is pressure buildup after an energy release due to solvent vaporization or the release of gases that often leads to undesired conse%uences. &herefore, analysis using a &)u or other similar pressure recording screening tool is imperative for evaluating both temperature and pressure responses, which can be studied under either isothermal or ramped temperature conditions. &here are cases, in fact, where only slight exotherms or even endotherms are observed in D)C scans, b ut measurable pressure events are detected during a &)4 analysis (See fgure 1".
Zoo m In
Figure 1 Why Pressure Is Important TSU scan showing no detectable thermal event in the sample temperature profle, but generation o non-condensable gases when comparing PFinal to PInitial Test conditions: !"#$-1%"#$ at 1&#$'min heating rate 5inetic data and the heat of reaction of the desired process chemistry are then obtained using a reaction calorimeter. &raditionally, the ettler &oledo 0C6 has been the industry wor#horse used for reaction calorimetry. disadvantage of this system is that many users have it e%uipped with a 788+m9 or larger reactor, and re%uires substantial %uantities of material. icroreaction calorimeters have been recently developed, however, that utilize 6.7+m9 to :8+m9 vials and re%uire minimal material, ma#ing the test %uic#er and more feasible for regular testing of all processes, regardless of development phase. $t is important to note, however, that while a properly designed and executed microcalorimeter experiment will provide a reliable heat+of+reaction value, it may be difficult to determine the heat+release profile that will be observed during a slow addition that is typically employed on scale+up. $f the microcalorimeter uncovers a large exotherm that will re%uire strict temperature control via addition rate, further testing in larger e%uipment may be appropriate.
&o establish the most appropriate basis of safety therefore re%uires the ability to #now which tests to conduct and how to effectively interpret the obtained data. &he experience and expertise of the CD!-s process safety personnel are thus crucial to successful evaluations. ;oth an understanding of the potential reactivities of molecules based on their structures, and extensive experience conducting safety evaluations of many different processes, are needed to be able to identify the best series of tests that will fully elucidate the reaction behavior for a given process.
1or instance, if a screening test conducted on a reaction mixture uncovers a decomposition event, %uite often the practice is to assign a default safety margin, below which the reaction must be executed.
Zoo m In
&he above D)C scan reveals that the sample melted and then immediately decomposed in the range of 6>8=C+6?8=C. &he actual reaction is run in solvent, however. When in solution, the solid would not undergo melting/ the energetic decomposition could therefore potentially shift to a lower temperature. dditional testing should be conducted to observe the behavior of the compound in the reaction solvent.
Figure 2 DSC Interpretation (S$ scan o solid with melting temp ollowed immediatel) b) decomposition Test conditions: *"#$-+""#$ at 1"#$'min scan rate &he rate at which energy is released, and not ust the amount of energy, should also be considered. $nformation on the time to maximum rate can be gleaned from D)C scans. 1rom a pure %ualitative perspective, a broad decomposition pea# suggests that energy may be released following nth order #inetics, while a sharp pea# indicates a rapid release and possible autocatalytic decomposition. $n the latter case once the initiating event is triggered, there is much less time to correct the situation before a full+blown runaway occurs. $f 0C testing is not readily available, conservative &ime to aximum 0ate
information can often be estimated by conducting D)C a nalyses at multiple isothermal temperatures or scan rates and applying advanced software to develop scalable models of reactions that can be used to predict stability under different heating conditions.
2
$f the results of such an analysis raise any flags, more accurate data can be obtained using an accelerating rate calorimeter. @iven the large instrument footprint and expense, 0Cs are not typically owned by smaller CD!s. )amples must therefore be sent to an external testing laboratory for 0C analysis. &he test provides information about the relationships between time, temperature, pressure and #inetics for exothermic reactions under adiabatic conditions, such as those generally experienced in process e%uipment during loss of cooling.
1inally, it is not sufficient to consider only the desired reaction conditions. &he behavior of the process under various undesirable conditions — worst+case scenarios such as loss of cooling or other e%uipment failure (i.e., stirring, feed pumps, etc." — must also be evaluated in order to determine the most effective basis of safety.
