You are on page 1of 20

This is an open access article published under an ACS AuthorChoice License, which permits

copying and redistribution of the article or any adaptations for non-commercial purposes.

pubs.acs.org/OPRD Article

Process Safety in the Pharmaceutical IndustryPart I: Thermal and


Reaction Hazard Evaluation Processes and Techniques
Ayman D. Allian,* Nisha P. Shah, Antonio C. Ferretti, Derek B. Brown, Stanley P. Kolis,
and Jeffrey B. Sperry

Cite This: https://dx.doi.org/10.1021/acs.oprd.0c00226 Read Online

ACCESS
See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

Metrics & More Article Recommendations

ABSTRACT: Process safety groups in the pharmaceutical industry are important components of active pharmaceutical ingredient
(API) development through its life cycle from discovery to commercial scale. The pharmaceutical process safety laboratory staff
Downloaded via 212.119.46.179 on October 24, 2020 at 14:33:56 (UTC).

conduct a series of tests to identify chemically unstable reagents, intermediates and solvents, and mixtures to ensure that the
proposed operating conditions provide a sufficient safety margin from the onset of undesired and potentially catastrophic thermal
decomposition. Across several pharmaceutical companies, the methods used for these assessments and how results and conclusions
are made are widespread (vide infra). A working group was created with members from several pharmaceutical companies within the
International Consortium for Innovation and Quality in Pharmaceutical Development (IQ), with the goal of precompetitive
collaboration and to understand each of the participating companies’ procedures and assessment regarding process safety. Each
company was invited to provide input using a blind survey format. This was done in the interest of making this knowledge accessible
for the participating companies and the wider community of other pharma and chemical companies and even academic institutions
in the US and throughout the world. This article provides the results of this in-depth survey of the members of the IQ Consortium
thermal hazard working group. General issues around different tools used to assess thermal hazard risk and questions regarding
staffing and tech transfer of process safety data/information from development to manufacturing were addressed. A snapshot of how
various assessment strategies are employed as a function of stage of development (early, mid, and late) is also presented.
KEYWORDS: process safety, thermal stability, thermal hazard, pharmaceutical, calorimetry, safety

1. INTRODUCTION to enable their respective companies to run potentially


hazardous chemistry such as ozonolysis.7−19
Most large pharmaceutical companies have a process safety
Drug development is an iterative process that requires many
laboratory (PSL) group that is accountable for the under-
years to complete.20 During the journey of a drug from the
standing of the thermal hazards associated with reagents, bench to the market, different phases of development require
chemicals, and reactions at hand.1,2 Their work focuses on increasingly larger amounts of material for clinical assessment,
minimizing risks associated with handling hazardous chemicals, and the scales at which manufacturing processes are executed
waste streams, and running hazardous reactions. A lack of may increase from milligrams of the limiting reagent in a
understanding of chemical hazards can potentially expose these round-bottom flask to metric tons of the limiting reagent in
reagents and/or mixtures to conditions, temperatures, and/or 2000 gallon (US) equipment. The hazards of operating with
mechanical stress that can trigger unsafe conditions such as flammable solvents and energetic materials/reactions drasti-
fire, deflagrations, and explosions, which will lead to significant cally increase with the increase in scale. Therefore, most
material loss, delay, supply chain disruptions, injuries, and/or pharmaceutical companies have a PSL. PSL staff are tasked
loss of life. Process safety groups conduct a series of tests to with the assessment of thermal and reaction hazards in order to
identify chemically unstable intermediates and reaction ensure the safety of the personnel handling these materials in
mixtures and ensure that the proposed operating conditions the lab and plant operators who will be executing the process.
provide a sufficient safety margin from the onset of undesired The most seasoned practitioners of process chemistry and
and potentially catastrophic thermal decompositions.3 There engineering would agree that safety is of paramount
are several illustrations available in the literature where importance in running on scale and cannot take a back seat
pharmaceutical process safety groups successfully managed to
avert catastrophic incidents by stopping dangerous chemistry Received: May 13, 2020
from being scaled-up.4,5 For example, potential safety hazards
associated with the violent thermal decomposition of dimethyl
sulfoxide in the presence of an acid were identified and
mitigated prior to scale-up.6 Other examples include the use of
continuous operation that was championed by their PSL staff

© XXXX American Chemical Society https://dx.doi.org/10.1021/acs.oprd.0c00226


A Org. Process Res. Dev. XXXX, XXX, XXX−XXX
Organic Process Research & Development pubs.acs.org/OPRD Article

Figure 1. Pharmaceutical research and development process overlaid with the phases of drug substance process development used for the survey.
Reproduced from the PhRMA website with permission.

to any other process characteristics (yield, cycle time, cost, cantly based on the scales of operation involved, which parallel
etc.),21 as any incidents can lead to significant capital loss and different stages of drug development. Figure 1 shows a
human loss and can be very damaging to a company’s brand. snapshot of the typical research and development dynamics
Outsourced processes also need appropriate attention, as an which occur when transforming a molecule into a drug. During
incident at a strategic contract manufacturing organization the early stages, a very large number of molecules are screened,
(CMO) can lead to significant business interruption, supply and their structure is subsequently modified and optimized
chain delays, injury to staff, and damage to equipment through a diverse array of chemical transformations carried out
executing the chemistry on the innovator company’s behalf. at a small scale (mg). Many molecules do not progress to early
Over the past decade, the pharmaceutical industry has faced development, and therefore, very limited resources are devoted
many new challenges from both a cost and timing to optimizing and understanding the synthetic chemistries
perspective.22 Patient needs and market pressures have involved. Process safety experiments may be evaluated for
increased the need to run at speed, and as a result, cycle reactions that are known to be hazardous, but in general
times for development have decreased. In addition, many large reactions are not screened or evaluated by the process safety
pharmaceutical companies have outsourced chemistry and group (vide inf ra). As a molecule transitions to early
typically one CMO works with multiple pharma companies at development and is introduced in the clinic, larger amounts
the same time. Given each may have a different approach to of active pharmaceutical ingredient (API) (g to kg) are
process safety assessment and risk tolerance, this may cause required to support clinical and product development studies.
confusion and potentially lead to decreased safety perform- At this stage, process chemistry groups become involved and
ance. Collaboration in the field of process safety can bring new chemical routes and scalable processes are invented and
tremendous benefit to the pharma companies; however, there developed. At this stage, it is atypical to execute at this scale
have been limited venues and appropriate platforms to allow without an appropriate process safety assessment. At this stage,
such collaboration. It is in this vein that the IQ thermal hazard processes may be fit-for-purpose to supply the clinical program.
working group set out on a survey of member companies As clinical trials progress, the attrition rate is still high and only
designed to answer some fundamental questions. Specifically, a small fraction of programs advance to late-stage develop-
questions arose around the common “best practices” at each of ment. The chemical processes at this stage need to be further
the member companies with respect to assessment of thermal developed to be commercially viable, that is, highly optimized,
hazards at all stages of development. Finally, a dissemination of well characterized, robust, sustainable, with established proven
the pharmaceutical industry practices could have the effect of
and acceptable ranges, efficient, and cost effective. At the late
harmonizing language for the industry and creating less
stage, the process safety aspects must be deeply characterized
confusion for our partners.
and understood.
The survey was constructed with input from all 15 of the
2. METHODOLOGY AND SURVEY QUESTIONNAIRE working group member companies.1 During the design phase
CRAFTING AND DESIGN of the survey, we elected to create three subsections of the
In order to achieve our goal of understanding how different survey that roughly paralleled the traditional stages of
pharma companies approach thermal hazard evaluation, a pharmaceutical drug substance development (Figure 1).
significant effort was initiated to draft a comprehensive set of From the perspective of operational definition and guidance,
questions to be distributed among the participants. It was the following definitions were used to demarcate the stages for
quickly recognized that process safety practices vary signifi- the survey:
B https://dx.doi.org/10.1021/acs.oprd.0c00226
Org. Process Res. Dev. XXXX, XXX, XXX−XXX
Organic Process Research & Development pubs.acs.org/OPRD Article

Figure 2. Key differentiators considered in the hazard evaluation of chemical processes.

• Early stage: programs conducted in a laboratory the high attrition rate of clinical programs (see Figure 1).
environment, at the discovery/medicinal chemistry scale. However, this makes the PSL role more critical as potentially
• •Mid-stage: programs that either progressed from dangerous chemistries can move forward from discovery
discovery to process research and development, first- groups with little to no attention to thermal hazards.
in-human and/or are in process characterization in the The PSLs are commonly a part of chemical development
development lab, kilo lab, and/or pilot plant. organizations and composed of highly trained dedicated staff
• Late stage: programs that either progressed from the with a background in either chemistry or engineering.
mid-stage to tech transfer, engineering runs, validation, Pharmaceutical companies also outsource various safety tests
and commercial delivery. especially when advanced equipment or expertise are not
In addition to these three sections, a general overview available internally. Companies were asked if staff outside the
section was used to capture items that span all three stages of process safety group, but within the same company, also carry
development. The final survey that was sent to the working out dedicated safety experiments. Six of the 15 companies
group members consisted of 91 questions distributed indicated that process development chemists or engineers may
(unevenly) among the four different sections and consisted run initial process safety screening by differential scanning
of free response questions, menu drop down choices, and calorimetry (DSC). In these cases, the PSL staff is still
option selection types of questions. The survey was sent to 15 responsible for the overall process safety evaluation (PSE),
companies and participation rate was 100%. In this paper, interpreting the results generated by partner groups (either
results and analysis of the survey are discussed and described. internal or external), providing recommendations, and
2.1. Process Safety General Approach. It is generally deploying more advanced thermal hazard assessment techni-
accepted that a chemical process cannot be executed at large ques as needed.
scales without a proper process safety assessment.23 PSL staff As a clinical program moves toward late-stage development
are accountable for evaluating the thermal and pressure safety and commercialization, the scale at which the manufacturing
of chemical processes throughout all unit operations. The process needs to be executed may increase significantly. As a
outcome of an experimental process safety study may enable a result, the level of experimental depth of the process safety
process to be scaled as proposed, which lead to recom- study also increases. In addition to the scale, (vide infra), there
mendations on process changes and/or necessary specialized are several “triggers” which may initiate a PSE or increase the
equipment, reducing the scale of the process, or determining level of experimental scrutiny. Companies were probed for
that the process is not scalable and needs to be redeveloped. In which factors, excluding the scale, can alert the PSL staff when
this section of the survey, we focused on general information evaluating a new process: the most cited (top 4 out of 8)
such as (a) when PSL staff are typically engaged, (b) general criteria, as shown in Figure 2, were the presence of highly
criteria used to assess risk, (c) how PSL staff are involved in energetic materials, energetic reactions, gas evolution, and the
the tech transfer process, and (d) literature references utilized use of highly corrosive and highly hazardous materials (e.g.,
to identify hazardous reagents and reactions. toxic or low flash point)
First, the participating companies were asked to choose The presence of highly energetic functional groups or
between two options when the PSLs are engaged:
energetic reactions are key factors mentioned by all 15
• For known hazardous chemistry, PSL staff are involved companies. These are identified using a combination of
in the early phase and potentially contribute to route literature and internal knowledge on hazardous chemistries.
selection Among the widely used literature resources are Bretherick’s
• Only after the process chemistry is “locked” Handbook of Reactive Chemical Hazards24 and Stoessel’s
Only 2 of the 15 companies surveyed indicated that their Thermal Safety of Chemical Processes.25 Also, the list of
PSL has an active role in route selection, providing data to recommendations on transport of dangerous goods by the
inform the decision on which route to pursue. This can be due United Nations26 along with other publications27,28 have been
to the fact that resources are limited during the early phase and referenced.
C https://dx.doi.org/10.1021/acs.oprd.0c00226
Org. Process Res. Dev. XXXX, XXX, XXX−XXX
Organic Process Research & Development pubs.acs.org/OPRD Article

Figure 3. Triggers for repeating a PSE on a process.

Gas evolution is another highly cited factor because of Figure 3 shows that changes in the amount or type of
potential pressure hazards. Companies were asked whether reagent and/or solvent, the addition sequence, or conditions
they have a written list of reactions which generate gas. The trigger an additional PSE because the latter changes can
choices available were: drastically alter the thermal hazard profile and thus require
further testing. In terms of handling minor changes in the
A No
process that are deemed too small to have an impact, 5 out of
B Yes, we have standard list (please share) the 15 companies at the commercial stage would still repeat a
full experimental hazard evaluation. At the small scale, 7 out of
C Chemist is asked for a balanced equation
the 15 companies indicated that an experimental re-evaluation
According to the survey, option B was not selected, might not be necessary and only a paper exercise could suffice.
indicating that none of the companies possess comprehensive The PSLs are typically equipped with an array of calorimetry
literature lists to flag reactions that either generate gas as a side tools to measure the heat generated by (a) the desired reaction
product of the desired chemistry or as an undesired at the process temperature, (b) runaway reactions, and/or (c)
decomposition product that may be generated under off-target thermal decompositions that might occur at higher temper-
conditions. The potential for gas generation is initially assessed atures. Criteria to classify the severity of the latter thermal
by analyzing the balanced chemical equations (11/15, option events, based on adiabatic temperature rise (ΔTad), were
C) relative to the process, which is provided by the proposed by Stoessel, as shown in Table 1.34 ΔTad is a key
development scientist. Only one company screens every
process for unexpected pressure events. Table 1. Severity of an Exothermic Event Based on Heat
The presence of corrosive or highly hazardous reagents was Released and the Corresponding Adiabatic Temperature
cited by 13 out of 15 companies as a significant hazard Risea
concern. In this context, it is worth noting that the choice of
adiabatic temperature rise (°C) energy released J/g
equipment in terms of the material of construction also has
process safety implications, as side reactions can either be high >200 >400
initiated or have lower onset temperatures because of medium 50−200 100−400
interactions between equipment and the process streams. low <50 <100
a
Furthermore, incompatibility of reagents with equipment can Adopted from Stoessel, F., Thermal Safety of Chemical Processes:
lead to significant corrosion and damage. Incompatibilities can Risk Assessment and Process Design. Wiley-VCH: Weinheim, 2008.
also compromise product quality by introducing metal
contamination into the product. Nine pharmaceutical
quantity used in the field of process safety to evaluate the
companies indicated that they have references that are used
exothermic behavior of a process and can be calculated
to the screen materials of construction compatibilities. The
through the following equation
main references are the Cole Parmer chemical compatibility
database,29 Schweitzer Tables,30 Corrosion Data Survey,31 Q rxn
ASM Handbook,32 and The Pilot Plant Real Book.33 ΔTad =
m × Cp (1)
Typically, a PSE is only valid for the exact process that has
been tested, and changes to the reaction (temperature, where Qrxn is the total heat evolved by the chemical reaction
concentration, reagents, solvents, order of addition) can (Joules), m is the mass of the reaction (g), and Cp is the heat
cause a dramatic change in the outcome of the PSE. Because capacity of the reaction mixture (J/g·K). The adiabatic
the PSL staff operate in a space where processes are constantly temperature rise is a conservative estimation of the severity
being improved and optimized, there are questions on how to of the hazard; it assumes that all heat generated by the reaction
address process changes and whether a full PSE, both is retained in the reaction system to increase the temperature
experimental and paper, is needed because of a single process of the reaction mass. As such, the ΔTad is the largest
change. Figure 3 shows companies’ responses on what might temperature increase one could expect to see if the reactor
trigger a new PSE evaluation. cooling was to malfunction.
D https://dx.doi.org/10.1021/acs.oprd.0c00226
Org. Process Res. Dev. XXXX, XXX, XXX−XXX
Organic Process Research & Development pubs.acs.org/OPRD Article

Figure 4. (a) Classification of exotherms in terms of adiabatic temperature rise as low (green), medium (yellow), high (red). Each line corresponds
to the criteria given by a single company. (b) Stoessel criticality classes, used by 12 out of 15 companies.

For example, if a reaction exhibits a ΔTad of 200 °C and is The criticality class is assigned only based on temperatures
operating at room temperature, 22 °C, it can potentially that are descriptive of the process. It starts by identifying the
increase the temperature of the reactor contents to 222 °C if conditions of the intended process, namely, the temperature of
cooling failed. Clearly, this can be catastrophic as the the process (Tp). Then, typically the heat of the desired
temperature increase could lead to further speed up of the reaction is measured by reaction calorimetry and the ΔTad is
reaction rate, solvent evaporation, and initiate other thermal calculated using eq 1. The maximum temperature of the
decompositions. Based on the Stoessel classification criteria synthesis reaction (MTSR) is defined as shown in eq 2
(Table 1), a thermal event that exhibits a ΔTad of 200 °C or
MTSR = Tp + ΔTad (2)
higher is considered a high severity event. On the other hand, a
ΔTad of 50 °C or less is considered low severity, while a ΔTad
of 50−200 °C is considered medium. In addition to the MTSR, two other figures of merit are
Surveyed companies were asked to express severity in terms considered important for the assignment of the Stoessel class.
of the energy associated with an exotherm and/or in terms of First, the maximum technical temperature (MTT) is typically
defined as the solvent boiling point for an open system. For a
the adiabatic temperature rise (ΔTad) associated with that
closed system, it is the temperature that corresponds to the
exotherm. Figure 4a shows what is considered a low exotherm
pressure to which the safety valve is set. Second, the TD24 (vide
(green), a medium exotherm (yellow), or a high exotherm supra) is defined as the temperature at which the time to the
(red) in the 10 companies that answered this question. Figure maximum rate of a runaway reaction is 24 h under adiabatic
4a indicates that most companies consider an adiabatic conditions. With these temperatures in hand, the Stoessel
temperature rise below 50 °C as a low severity exotherm, criticality class can be determined.
while a ΔTad higher than 200 °C would be classified as a high For a Stoessel class 1 reaction, the heat of reaction is
severity exotherm which reflects the classification proposed by insufficient to cause the reactor content temperature to reach
Stoessel. However, three companies were on the conservative the solvent boiling point (MTT) and therefore reaching the
side and considered ΔTad higher than 50 °C as high severity. TD24 is very unlikely. However, for a class 5 reaction, the heat
The probability of reaching a thermal decomposition is of reaction is capable of causing the temperature to reach the
measured using the Stoessel Criticality classes25 that is widely TD24, which is below the MTT. As such, the boiling of the
used in the pharmaceutical industry, as was mentioned by 12 of solvent cannot act as a barrier to prevent the temperature from
the 15 companies surveyed. These criticality classes rank a reaching the TD24, and the probability of triggering a thermal
process relative to the probability that a runaway reaction may decomposition is very high.
be initiated in the worst-case scenario of the loss of cooling to a Across industries, the discipline of process safety seeks to
reactor. The classification goes from class 1, the safest, to class quantify the risk inherent to executing a process. A Stoessel
5, the most dangerous where the probability of triggering a criticality class can be used as a guideline for the probability of
runaway reaction is very high. a thermal hazard event occurring, as shown in Figure 4b.
E https://dx.doi.org/10.1021/acs.oprd.0c00226
Org. Process Res. Dev. XXXX, XXX, XXX−XXX
Organic Process Research & Development pubs.acs.org/OPRD Article

Figure 5. Process volumes that trigger thermal hazard evaluation in the early stage.

Pharmaceutical companies tend to be conservative. For Because pharmaceutical companies often acquire external
example, if the probability of an exothermic event is high, programs that have already advanced to the mid- to late-stage
(Stoessel class 4 and 5), there is a consensus that these development, there might be cases where process safety data
processes would not be allowed to move forward. There is the have already been generated. In such cases, 7/14 companies do
constant objective to design processes that are inherently safe not adopt the data and repeat the PSE, while the remaining 7/
(i.e., Stoessel class 1 or 2), where the probability of triggering 14 might use the available data on a case-by-case basis.
an uncontrolled exothermic event is low regardless of the When assessing the overall safety of a chemical process,
severity of the event. there might be the need to investigate concepts beyond
A series of questions were asked regarding how the PSLs thermal hazards. For example, there might be issues regarding
engage with internal or external manufacturing plants. When a the toxicity of certain reagents, side products, or byproduct,
process is scaled-up within the same organization, it needs to and in order to produce a comprehensive safety assessment,
be determined what group is ultimately accountable for there might be the option to involve experts from an
process safety. For 7 out of 15 companies, the central PSL is environmental, health, and safety group (EHS). Surprisingly,
fully accountable. However, 3 out of the 15 companies the majority (8 out of 15) indicated that EHS is not consulted,
indicated that the receiving site is accountable while the while 4 out 15 companies mentioned that EHS is involved in
remaining 5 indicated that it is a partnership between the two the PSE and/or participate in writing the process safety report;
sites. the remaining 3 companies responded that EHS is somewhat
For a process that is transferred to an external CMO, there involved.
are often questions relative to how to share safety data 2.2. Process Safety Approach at the Early Stage. This
generated at the innovator’s thermal hazard lab. A majority of section will focus on PSL staff’s approach to thermal hazard
the surveyed companies (10/15) have responded that the analysis in early development. Knowing that several labs do not
available process safety data are shared without providing evaluate thermal hazards at this stage, we asked the participants
interpretation or recommendations. It is then the responsibility if their safety labs support the medicinal chemistry group. Five
of the CMO to determine whether the process is safe for their companies indicated that either their organization does not
facility and how the process needs to be run based on their have medicinal chemistry, or they have it but strategically
own internal risk tolerance. Indeed, 5/15 companies elected not to support it; these participants were exempted
responded that they require the CMO to also generate their from the remaining questions in this section.
own thermal hazard data. Only three companies indicated that At the early stage, as discussed in the introduction, medicinal
in addition to sharing data, they provide recommendations on chemistry is typically carrying out a large number of small-scale
how to translate the data into process design or equipment reactions. Therefore, companies are taking a risk-based
setup. It is worth mentioning that four companies selected two approach to minimize thermal hazard evaluation at this stage.
or three of these choices, indicating that some companies do The 10 companies that do support early stage have minimal
not have a unified policy in terms of sharing data and their involvement and experimental work. Two companies indicated
approach can be CMO or program specific. that the medicinal chemistry staff are equipped with a DSC
Instances where the central PSL and the CMO have and test their own samples, rather than relying on the PSL.
disagreements on whether a process is safe to scale are rare but However, it is worth mentioning that, as stated in Section 2.1,
do occur. In some instances, there might be a conflict between while the staff outside the PSL are permitted to gather their
the way the PSL interprets the data and SOP of the CMO. An thermal hazard data, the PSL staff remains accountable for the
example that was mentioned was a case where the CMO used overall PSE. Furthermore, to keep the experimental work to a
the 100 °C rule35 to determine safety margins as a part of their minimum at this early stage, in 7 out of the 10 companies, the
SOPs, while the innovator company did not. The rule requires PSL staff rely on desk screening performed by the medicinal
100 °C between the MTSR and the onset temperature of a chemistry staff. The medicinal chemistry staff then engages or
significant exotherm as detected by DSC or a similar consults with the PSL staff only when dealing with energetic
technique.25 functional groups and/or potentially hazardous reactions. To
F https://dx.doi.org/10.1021/acs.oprd.0c00226
Org. Process Res. Dev. XXXX, XXX, XXX−XXX
Organic Process Research & Development pubs.acs.org/OPRD Article

Figure 6. Materials tested by DSC in the early stage.

understand if there is a scale that prompts PSL staff by DSC. The goal of using the ARC is to get more quantitative
engagement, participants were asked if there is a threshold measurement of the heat generated and more accurate onset
above which the PSL staff involvement is required. Most temperatures. Also, ARC allows pressure measurement, a
companies did not report a scale threshold to trigger PSL capability not available in the DSC. Four companies use DSC
involvement and simply react to medicinal chemistry requests exotherm energy as a trigger for ARC. Specifically, two
irrespective of scale. The response from the few companies that companies mentioned that the presence of an exothermic
reported a scale threshold is shown in Figure 5. It is worth event by DSC in the amount of 100 and 800 J/g, respectively,
mentioning that companies that reported these thresholds will trigger an ARC run.
indicated that there are cases at a small scale, less than the In addition to triggering ARC testing, seven of the ten
threshold, where testing may be done if energetic material is respondents use DSC to prompt impact sensitivity and
present or a significant amount of gas is expected. For example, explosion propagation testing, using Yoshida correlations.38
two companies specified that they still would perform thermal The Yoshida correlations are an empirical method of
hazard testing before scaling to ≥1 g if flagged as high energy predicting a material’s hazard potential when subjected to
or potentially hazardous. low energy stresses based on onset temperature and
In cases where testing is deemed necessary, we asked the decomposition energy measured by DSC. These correlations,
participating companies the type of thermal tests that are or a modified form thereof,5 are used by five companies to
deployed at this stage. By far, the primary method of testing (6 trigger additional testing. One company uses the Yoshida
out of 9) is DSC.3 DSC provides fast thermal stability data for correlations but also carries out impact sensitivity and
components of reactions, while requiring only milligram explosion testing for materials that did not trigger Yoshida
quantities of samples. This is important at this early stage correlations if the DSC has an exotherm energy of ≥800 J/g.
where the material is very valuable and scarce. One company Two companies do not use the Yoshida correlations; rather,
uses thermal screening unit (TSu)36 and reaction calorimetry further testing at one is triggered by the exotherm onset and
(RC) in certain scenarios. The remaining two companies who energy (<500 °C and >500 J/g, respectively) and at the other
responded to this question do not do any thermal testing at solely by exotherm energy (>800 J/g). We asked the
this early stage. At this point, the PSL staff’s hazard evaluation participating companies to specify the type of test used if
is mainly focused on isolated reagents, starting materials, and Yoshida correlations were positive; two companies specified
products, as shown in Figure 6. Waste stream and distillation Fallhammer,5 one of them in conjunction with high rate Carius
concentrates are less likely to be evaluated. This is probably Tube,a as the next test to perform. Unfortunately, the
because processes, solvents, and unit operations at this stage remaining six companies did not specify the impact or friction
are not fixed and still being investigated. One company only sensitivity tests of choice.
tests streams that contain known explosive functionality. Heat of reaction is an important parameter to measure, as a
After gathering DSC data, most companies do not set a limit part of the overall assessment of the thermal hazard (see
on an acceptable window between the operating temperature Section 2.1). However, experimental heat of reaction measure-
and the onset of a significant exothermic event because the ment can require amounts of materials that might not be
associated risk is low at this small scale. However, three of ten available at this early phase. Not surprisingly, seven
respondents restrict the operating temperature to at least 100 respondents do not measure the heat of reaction at this
°C below the decomposition onset by DSC. Another company stage and two will if the DSC indicates energetic decom-
went further and dictated a full calorimetric evaluation and position and/or a large exotherm is observed in lab
requiring a 100 °C window between the MTSR and the experiments. With the experimental heat of reaction not
decomposition onset by DSC. Of course, the latter is not just widely used at this stage, we asked the participants the type of
conservative but also requires additional testing as MTSR is desktop methods commonly used to estimate heats of reaction
typically determined by reaction calorimetry. in lieu of testing. Participants indicated using estimates based
DSC screening is the main platform for thermal stability on similar chemistry, CHETAH,39 and quantum methods.40
investigation at the early stage. However, accelerating rate One participant has developed their own proprietary heat
calorimetry (ARC)37 is consistently used for further inves- estimation tool. Because most respondents do not evaluate
tigation when red flags such as large exotherms are determined heats of reaction experimentally at this stage, there are no
G https://dx.doi.org/10.1021/acs.oprd.0c00226
Org. Process Res. Dev. XXXX, XXX, XXX−XXX
Organic Process Research & Development pubs.acs.org/OPRD Article

Figure 7. Materials tested for pressure hazards in early stage.

Figure 8. Scale that triggers thermal hazard testing in the mid-stage.

triggers for further advanced calorimetric testing. We will show remaining two conduct pressure screening if DSC indicates a
later in the article that advanced heat of reaction testing is used severe exotherm, and then they move to more advanced
in the later phases of drug development. Two companies may techniques if the pressure screening indicates uncontrolled off-
perform further measurements if desk screening indicates high gassing. Next, participants were asked if pressure increase or
heats of reaction. rate of gas evolution is examined. In agreement with the
Next, participants were asked questions related to gas discussion earlier around the desired gas evolution, most
evolution inherent to desired chemistry. The participants were companies (9 out of 10) do not investigate these parameters at
given the following options. this early stage. One company looks into gas hazards only if
A No, no gas measurement is done at this stage reaction is known to generate gas, or if the scale is large
B No, we just assume that the theoretical amount is enough to warrant testing. It is worth mentioning that we
liberated asked the participants to indicate if red flags from the pressure
C Yes, but only for reactions that are known to generate screening test will trigger more advance testing; as expected, 8
gas (please describe how this is measured) companies said “no” as most companies at this stage do not
D Yes, only if the screening tool gives red flags (please even conduct the initial screen, let alone deploying an
describe) advanced gas hazard test. Figure 7 shows the types of samples
E Yes, all reactions are screened for pressure (please tested if pressure hazards are deemed worthy of investigating at
describe how it is measured) this stage. Most of the companies select samples based on
None of the responders picked choice E, as that would be hazard potential, with 4 companies testing any streams
experimentally exhaustive. Eight companies do not exper- expected to be highly hazardous. Waste streams and distillation
imentally measure total gas generated, option A and B. It is concentrates are rarely considered at this stage.
either not considered at this stage or they assume the Finally, survey participants were asked if any results or
theoretical amount is liberated. The remaining two companies outcomes would prompt the process safety lab to require
will investigate if gas evolution appears severe, uncontrolled, or modifications to the chemistry prior to moving forward.
hazardous in lab experiments. In general, rates of gas evolution Although open-ended, the responses were relatively consistent.
and maximum pressures are not examined at this stage. Four of the ten companies listed a positive impact or friction
For undesired/unintended reactions, eight respondents do sensitivity as a justification to require process modifications.
not investigate gas generation in early development. The Additionally, six of the ten will stop reactions from moving
H https://dx.doi.org/10.1021/acs.oprd.0c00226
Org. Process Res. Dev. XXXX, XXX, XXX−XXX
Organic Process Research & Development pubs.acs.org/OPRD Article

Figure 9. Material subject to thermal hazard testing in the mid-stage.

forward if DSC and/or ARC data indicate high energy (see based on the responses from the survey, the scale varied from 2
Section 2.1 for definition of high energy) and low onset L to pilot scale, see Figure 8. The data suggest that certain
exothermic activity that is close to the operating temperature. pharmaceutical companies are leaning more conservative and
The responses were also consistent with our previous running thermal hazard testing even at a scale less than or
discussion on the appropriate safety window, with companies equal to 2 L. On the other hand, other companies appear to be
indicating that they will not allow chemistry to move forward if more risk tolerant and will not do testing until the process is
a significant exotherm is within 100 °C of the operating scaled-up in reactors above 100 L. It is worth mentioning that
temperature measured by DSC. Another interesting response 3 companies indicated that even though they are committed to
referenced the increase of flow chemistry to enable chemistry a scale at which a level of testing is needed, some testing will be
that is hazardous in cases where an exothermic event is very done even below that threshold if the reaction is deemed risky
close to the operating temperature.41 due to the presence of high-energy functional groups. A similar
2.3. Process Safety Approach at the Mid-Stage. This approach by the pharmaceutical companies was discussed in
section is aimed at understanding the IQ members’ thermal the early stage as well, indicating that these thresholds serve as
hazard testing in the mid-stage. The participants were asked guidelines but are not set in stone.
some similar questions to those in the early stage in order to Next, we asked participants to describe the thermal tests
examine the increase in complexity of process safety testing used and materials being tested. The participants were giving
with the increase in scale. the following options:
As in the early stage, the participants were asked two A None
questions. First, is there a scale where little to no testing is
done? Second, is there a scale above which a level of thermal B Quick screening tools, (please specify).
hazard evaluation/testing must be done? Then, we asked the C Full safety evaluation with calorimetry, (please specify).
participants to select one of these options: Not a single company picked choice (a) that no work is
1) No, all reactions are tested in a similar fashion done at this stage. This is understandable, as at this stage, the
2) Yes (please specify) volume is large enough to pose more risk and some form of
testing is needed. Indeed, some testing was deployed at the
3) Yes, but other factors are also considered (please early stage, see Section 2.2. It is worth reminding the reader
specify) that in the early phase, apart from one response, testing was
Only two companies selected option (1), which indicated largely limited to screening tools like the DSC. At this stage, 13
that most companies (13 of 15) are using the scale to guide out of the 15 picked choice (c), meaning that companies did
when the PSL staff need to conduct thermal hazard testing, as not stop at screening tools and deployed other advanced
shown in Figure 8. The use of the scale as guidance appears to calorimetric tools as needed. However, 2 companies indicated
help PSL labs prioritize their work even when the programs that they still rely strictly on quick screenings. These results
transition from the early stage, especially when the scale is still provide solid evidence for a sharp increase in the complexity of
small. In other words, the majority of the companies are taking thermal hazard tools utilized once the program progress from
a risk-based approach to reduce the testing burden on the PSL the early to mid-stage.
personnel, especially when the scale is small, even at early- and At the early stage, as shown in Figure 6, the types of
mid-stages. In the early stage, Section 2.2, limited involvement materials subjected to thermal hazard screening showed that
was observed from the PSL staff and their engagement was the focus was limited to the isolated material and product.
dictated by the medicinal chemistry staff needs. However, When the same question was posed to the participants on their
clearly at the mid-stage, the PSL staff now drive the decision approach in the mid-stage, the response is starkly different, as
on their involvement in thermal hazard evaluation. Similar to shown in Figure 9. At this stage, IQ members did not just test
the early stage, ass shown in Figure 5, the scale that would isolated reagents and isolated products, but workflows evolved
trigger thermal hazard testing in the mid-stage also varied to include other streams. Now, the majority (13 out of 15) will
significantly. For reactions free of energetic functional groups, test the initial and final reaction mixtures and streams that were
I https://dx.doi.org/10.1021/acs.oprd.0c00226
Org. Process Res. Dev. XXXX, XXX, XXX−XXX
Organic Process Research & Development pubs.acs.org/OPRD Article

Figure 10. Acceptable safety window for operating temperature of the reaction.

less likely to be tested in the early stage. Similarly, distillation sensitivity and explosion propagation testing. We posed the
concentrates and waste streams rarely tested in the early stage same question for the mid-stage, namely, what kind of results
now are being evaluated by most participants. One company would trigger impact sensitivity and explosion propagation
goes further and tests the wet cake for thermal hazard. testing and the following options were given:
Once reagents and reaction streams were tested, the A Yes, please describe.
participants were asked what constitutes an acceptable safety
window from the operating temperature, as shown in Figure B No, not at this phase
10. Based on the DSC exotherm onset data, the 100 °C degree The majority of the companies (12 out of 15) selected
rule was the most common approach, with 7 companies using option A. However, three companies indicated that it is still
this approach. The use of the 100 °C degree rule is also too early to deploy these tests at this stage. The authors
popular in the early stage, see Section 2.2. Interestingly, in the assume that the shipment and handling of large quantities of
mid-stage two companies diverted slightly from the 100 °C solids are not expected and that explains why these companies
rule, using a slightly more relaxed approach, where the 70 °C elected to delay testing until assets are commercialized. The
degree rule or 2/3 of the onset was used. With the use of twelve positive responses can be divided into two groups. The
advanced calorimetric techniques in the mid-stage, compared first group, 5 companies, relied on the process safety
to the early stage, PSL has access to more data to analyze and personnel’s prior experience to trigger these tests such as the
extrapolate. The use of time to maximum rate (TD24) along presence of certain functional groups. The other group, the
with Stoessel criticality classification is gaining momentum to remaining 7 companies, was using quantitative triggers to
determine the safety window, as four companies are using this initiate impact sensitivity and explosion propagation testing,
approach, see Section 2.2 for more details. Companies that use either using DSC data alone (3 companies) or DSC data with
Stoessel criticality classification indicated that class 4 and 5 Yoshida correlations (4 companies). The companies that use
reactions will not move forward, and rarely class 3 will be DSC data alone relied solely on the decomposition energies
allowed to be scaled-up (Section 2.2). For ARC data, two observed. These energy thresholds were 400, 500, and 800 J/g,
companies indicated they use the 50 °C rule, which is because respectively, above which warrant further testing. The
of the increased accuracy of the ARC versus the DSC. remaining 4 companies utilize the Yoshida correlation to
However, it should be noted that ARC testing requires more examine if further testing is needed. It is worth mentioning that
material and is much slower than DSC. It is worth mentioning impact sensitivity was deployed when necessary by 7
that in the early stage, companies did not define a safety companies in the early stage. At this mid-stage, impact
margin; however, at this stage, they now have an internally sensitivity is more widely used with clear internal guidelines
well-defined safety window that they enforce. on when to deploy it.
Next, IQ companies were asked what kind of results would Heat of reaction is a key component of the thermal hazard
trigger further testing beyond the screening tests mentioned assessment. In the early stage, see Section 2.2, the heat of
above. It appears that 8 companies do not have a quantitative reaction measurement was not done by most companies and
measure for deployment of further tests and rely on process instead they relied on desktop methods to estimate it. Again,
safety expertise to trigger it. However, 7 companies indicated we posed the same question here to see if the increase in scale
that if the operating temperature is within 100 °C of the onset affected thermal hazard and impact sensitivity measurement. In
temperature measured by DSC, further testing, most likely in that spirit, participants were asked if heat of reaction is
the form of ARC and/or DSC isothermal hold, will be measured and were asked to pick from these options:
deployed. Other techniques such as the use of modeling A No, no heat of reaction is measured at this stage
(example AKTS) were also mentioned when the operating B Yes, always. Please specify instrument
temperature is close the onset measured by DSC.
In the early stage, see Section 2.2, the survey showed that C Yes, only if heat is expected or previous screening tests
pharmaceutical companies are already deploying impact indicate that this is necessary (please describe)

J https://dx.doi.org/10.1021/acs.oprd.0c00226
Org. Process Res. Dev. XXXX, XXX, XXX−XXX
Organic Process Research & Development pubs.acs.org/OPRD Article

Figure 11. Material tested for pressure hazard in the mid-stage.

Figure 12. Will companies deploy additional testing to justify long hold times of reaction mixtures.

None of the IQ companies picked option A, indicating that preventing scale-up when maximum heat flow is above 25−
at this stage of development, heat of reaction is a topic of 30 W/L (i.e., typical value for heat removal capability of
concern. Indeed, 10 out of 15 will always measure heat of reaction vessel at the scale), otherwise the process needs to be
reaction, while 5 companies indicated that the heat of reaction modified. Four companies indicated using commercially
is only measured if necessary, that is, based on prior available advanced modeling software like DynoChem to
knowledge. Again, these responses are strikingly different understand a vessel’s suitability to handle the reaction at hand.
than those obtained from the early stage, where most Surprisingly, 9 out of 15 companies appear to have no criteria
companies said that the heat of reaction is not measured. or consideration of this parameter at this stage.
Companies’ responses indicate that there is a plethora of tools The use of desktop methods, as in the early stage, is still very
that can be used for this measurement, namely, RC1, EasyMax popular at this stage with 9 out of the 15 companies using heat
HF, Omnical, ARSST,b Setaram, and THT microcalorimeter. estimating techniques, literature, and/or internal archived data
It is worth mentioning that the measurement of the heat of of similar reactions to estimate the heat of reaction. However,
reaction, Qrxn, is required to be able to calculate ΔTad and the remaining six companies will only do experimental work to
MTSR using eqs 1 and 2, respectively. These values are used to eliminate risks. Next, we asked the participating companies an
assign a criticality class to the reaction using Stoessel’s open question, namely, in the mid-stage, can red flags from the
methodology (see Section 2.1). With more companies using heat of reaction screening trigger further calorimetric measure-
this approach, one can see why the measurement of heat of ments? The authors’ intent was to check if the companies that
reaction is more prevalent. Next, the participating companies measure the heat of reaction using screening tools will
were asked if the maximum heat of reaction is a parameter that eventually go back and use more advanced calorimetric
is examined to make sure that the heat generated from the techniques if a significant exotherm was observed during the
reaction at any point during the course of the reaction is within screening. Several companies indicated that if the adiabatic
the vessel’s ability to remove heat, preventing undesired temperature rise is above 50 °C from screening tools then
heating of the reactor contents. Four companies said that this advanced techniques like RC1 will be deployed to quantify the
parameter is not examined, and nine companies do examine it. heat of reaction more accurately. Other companies strictly use
Two of the nine companies that indicated looking into this RC1 calorimetry and do not utilize the fast screening tools to
important parameter said that they have guidelines for start with.
K https://dx.doi.org/10.1021/acs.oprd.0c00226
Org. Process Res. Dev. XXXX, XXX, XXX−XXX
Organic Process Research & Development pubs.acs.org/OPRD Article

Questions in this section thus far focused on thermal the same question in the early stage most companies, 9 out of
hazards; the survey at this point pivots and focuses on pressure 10, do not check this. However, in the mid-stage, 13 out of the
hazard screening as it did in the early stage. We asked 14 companies that responded to this question indicated that
participating companies, in the mid-stage, is gas generation of these parameters, gas evolution and pressure increase, are
the desired/intended reaction measured? being carefully examined. Most companies are now concerned
A No, no gas measurement is done at this stage with pressure limitation in the vessel and its ability to vent the
B No, we just assume that theoretical amount is liberated generated gas. One interesting response was a defined
C Yes, but only for reactions that are known to generate threshold of 2 ft3/min for gas generation above which the
gas (please describe how this is measured) process will be deemed unacceptable.
D Yes, only if screening tool give red flags (please describe) For the question whether more advanced gas measurement
E Yes, all reactions are screened for pressure (please techniques are used whenever pressure screening tests find any
describe how this is measured) red flags, 8 companies answering yes and 7 answering no at this
stage. The 8 companies with the positive answers indicated
At the mid-stage, most companies’ PSL are striving to that they typically deploy TSu or ARC for quick screening, and
understand gas generation. Of the positive responses, option C if a significant pressure event was observed, they follow it up
was dominant with 10 companies. This means that while the with advanced testing like Vent Sizing Package (VSP). While
pressure hazard is a concern, only samples that are known to the latter techniques focused on the volume of gas generated, it
generate gas will be tested, probably to reduce the workload, is worth mentioning that authors are aware that certain
that is, taking a risk-based approach. Four companies will test pharmaceutical companies have implement on-line technique
gas evolution for all reactions in any case, and the remaining to identify the composition of the generated gas.
company does not measure but assumes that the theoretical Gas generation during reactions poses another hazard
amount of gas is liberated (option B). This is in contrast to the besides over pressurization, namely, fire risk when the
early stage where 80% of companies picked option A and B, generated gas is flammable. Pharmaceutical companies are
that is, do not measure gas evolution because theoretical conservative, their first safety layer is making sure that there is
amount is assumed, or it is not considered. Interestingly, an absence of ignition sources. The second safety layer
option D was also not selected which indicates that the use of deployed is inerting the reaction system. In order to achieve
pressure screening tools are not widely used. For companies this goal, companies target bringing the flammable gas
that measure gas generation, the majority (8) use a mass flow
concentration below its lower explosive limit (LEL) or
meter coupled with their calorimetric setup. Other companies
reducing the oxygen concentration below the limiting oxygen
indicate using ARSST and RITTER clocks for gas measure-
concentration (LOC) which then renders the headspace
ment.
inflammable even when an ignition source is present. Because
Besides the desired reaction, participants were asked if
of potential inaccuracies in the quantitative measurement of
individual process streams are also tested for pressure hazards
LEL and LOC, many companies develop their own safety
in the mid-stage. As shown in Figure 11, the response was
window between the measured or known LOC/LEL level and
widespread with at least 4 companies testing all streams, and at
what they require in practice. We have asked each company
least one company indicated that no pressure hazard testing is
what their acceptable level is, that is, the typical threshold they
done unless there are specific data highlighting a potential
target to ensure safety. Only 10 companies responded to this
pressure hazard. The majority of companies (11) subject
question. Four companies target dilution to 25% of the LEL,
reaction mixtures to testing. Other streams like waste streams,
two companies target 50% of the LEL. One company took the
distillation concentrates, and end-of-reaction mixtures are also
conservative approach of reaching 10% of the LEL. One
of concerns and tested by 4−7 companies for pressure hazard.
company targets LOC and another two will use enough inert
In addition, gas generation can also be due to an upset
gas to keep O2 below 5%.
scenario where loss of cooling can trigger a decomposition that
Several questions were asked regarding extended hold times
releases gas, or the heat generated can drive solvent
for reactions, waste streams, and storage of starting materials
vaporization. The survey is attempting to probe best practices
and intermediates. First, is additional thermal testing required
to identify these potentially very hazardous scenarios. In this
for reaction mixtures to justify long hold times? Three
regard, the survey asked the participants if the rate of undesired
companies responded that testing for stability is not done at
gas evolution is measured and the following options were
given. this stage. On the other hand, three companies always do
additional testing to justify long hold time, while the majority
A No, not at this stage (9) only do such testing if there are red flags in the process that
B Yes, but only if DSC indicates significant exotherm warrant it. One of these red flags is if the hold temperature is
(please describe) within 50 °C of a significant exotherm. Some companies listed
C Yes, all undesired chemistry is screened for gas what specifically the additional tests were, which included a
generation (please describe) variety of isothermal calorimetry tests, (e.g., TSu, ARC, and
Similar to the question on the desired reaction gas evolution, DSC) and one company does kinetic modeling.
three companies do not consider pressure hazards at this stage Next, the same question was asked to justify long hold times
(option A). Among the remaining 12 companies, 5 companies of waste streams. Six companies do not do additional testing in
do so only if a significant exotherm is found in the process this case, where 9 do it if there are red flags in the process that
(option B), and 7 companies measure gas generation of warrant it. These red flags include the waste generating gas or
undesired chemistry (option C) regardless. Next, and as in the showing a significant exotherm from screening tests. Next,
early stage, participants were asked if values of the gas companies were asked whether tests are done to justify storage
evolution rate or maximum pressure increase can be used to conditions of starting materials, intermediates, or products and
designate a process safe to run. The reader is reminded that for what would trigger these tests. Approximately half, 8, said that
L https://dx.doi.org/10.1021/acs.oprd.0c00226
Org. Process Res. Dev. XXXX, XXX, XXX−XXX
Organic Process Research & Development pubs.acs.org/OPRD Article

Figure 13. (A): Number of companies who responded that they consider MOC compatibility. (B): Further breakdown of the “yes” responses.

they do no such testing, while 7 do. Further testing is usually Answers to this question were across the board but could be
triggered by the thermal stability data collected in routine collectively summarized by the following themes:
screening. Examples of scenarios where testing would be • Stoessel classification of 4 or 5
needed include (a) low onset temperatures and/or (b) high • Decomposition detected close to the operating temper-
decomposition energies. ature
Figure 13 shows that the majority of companies (14) • Large exotherm (i.e., adiabatic temperature rise >50 °C)
consider material of construction compatibility with chemicals • Uncontrolled gas generation
or reaction mixtures. The 14 participants with positive • Potential for explosivity
responses were asked to describe their approach. Eleven
The fact that reactions with a classification of 4 or 5 are not
companies responded and their responses were fairly evenly allowed to move forward is consistent with the companies’
divided between using online databases and literature data response in Section 2.1, where there is a consensus among
versus actual tests, where coupon corrosion tests are pharmaceutical companies that these reactions are too
performed. Five companies indicated deploying both desktop hazardous and will not be permitted to move forward.
and experimental methods. Considering whether a formal process hazards analysis
Many of the ubiquitous solvents used in the pharmaceutical (PHA) is performed when scaling-up to a kilo lab or pilot
industry are nonconductive and thus there is a risk of building plant, five companies responded yes and eight responded no.
static especially when solution is mechanically agitated or from Out of those who responded yes, one company performed
friction because of flowing in a pipe. This electrostatic PHA if the process is very high risk and another company does
discharge can lead to a spark which can start a fire or/and so if the process is over 200 L scale.
cause damage to the reactor.41 The survey shows that slightly With more pharmaceutical companies these days doing
less than half of companies (7) consider stream conductivity as chemistry in flow,42 the question was asked whether further
a potential for electrostatic hazards. A closer look at the thermal hazard tests are conducted if a reaction is done in
positive responses shows that companies either use literature continuous mode. Surprisingly, ten companies responded no
values for conductivity or experimentally measured values or N/A, with only five companies responding yes. This shows
when data are not available. The remaining 8 companies did that the PSL approach to evaluating thermal hazard with flow
not consider conductivity at this stage. chemistry is still evolving.
For shipment of hazardous materials, the self-accelerating The use of modeling in thermal hazard safety evaluation was
decomposition temperature (SADT) is a value that is widely another topic of interest that the survey aimed to probe. When
used to determine whether the material must be transported failure modes are likely, participants were asked whether they
under temperature-controlled conditions. The SADT is experimentally evaluate them or rely on models to understand
defined as the lowest environmental temperature at which that failure. Only four companies answered no, while the
the center of the material within the package heats to a majority 11 companies answered yes, which indicates that the
use of modeling is becoming a very integral part of thermal
temperature 6 °C greater than the environmental temperature
hazard evaluation. Situations that are likely to be modeled are
after a seven-day period or less.26 This period is measured from
reactor and condenser cooling failure scenarios, heat of
the time when the center of the package reaches 2 °C below
reaction at upset conditions, and reactions with high adiabatic
the environmental temperature. Six out of the 15 companies temperature rise.
responded that they consider SADT for the shipment of Finally, participants were asked if thermal hazard data were
material. Those who use SADT mentioned that they typically utilized for kinetic modeling. Three companies indicated
deploy this measurement when there is an onset temperature modeling reactions using the heat flow data using DynoChem
below 125 °C or to ensure compliance with DOT and UN and also using AKTS to study kinetic of decomposition. On
transportation standards. The method of determining the the other hand, ten companies indicated that they do not use
SADT can vary; usually this is done with isothermal screening data for kinetic modeling at this stage, which is unfortunate
tests, or by using DSC data and modeling methods to predict because heat flow data can be used to better understand
the kinetics of decomposition. reaction kinetics and mechanisms.43,44
Companies were asked what results or outcomes would 2.4. Process Safety Approach at the Late Stage. This
prompt them to stop a reaction from moving forward and section will focus on the last portion of the drug journey (the
propose changing the chemistry to mitigate the observed risks. late stage). It is worth mentioning that several pharmaceutical
M https://dx.doi.org/10.1021/acs.oprd.0c00226
Org. Process Res. Dev. XXXX, XXX, XXX−XXX
Organic Process Research & Development pubs.acs.org/OPRD Article

Figure 14. Material subject to thermal hazard testing in the late stage.

companies indicated that much of the thermal hazard quick screening tools, but now at this late stage, this is not an
evaluation has been done before arriving at this stage. Most option which is expected.
pharmaceutical companies partner with external CMOs for The next question asked each company to identify the
manufacturing both the drug substance intermediates and drug specific reagents, materials, and/or streams that were tested at
substance. This partnership between the two organizations the late stage. The 8 companies who responded to the question
becomes increasingly critical with respect to the process safety indicated that they perform testing on isolated reagents and
assessment of chemical processes. Each IQ member company starting materials, isolated products, initial and final reaction
was asked to answer a series of questions pertaining to their mixtures, as well as the distillation end point. Seven of the eight
process safety testing practices for commercial scale companies perform testing on waste streams (Figure 14). The
production. Five out of the 15 companies who took part in response is very similar to the mid-stage, the only noticeable
the survey noted that all commercial manufacturing is difference is waste stream evaluation, where 8 of the 15
externalized, and therefore, only minimal process safety testing companies in the mid-stage do it, but at the late stage it is 7 out
is performed in-house. Because the CMO partner is ultimately of 8.
responsible for the safety of their plant and people, the CMO The next question polled the participants about acceptable
will gather any additional process safety data not provided by safety windows for operating temperatures based on available
the pharmaceutical company with which they are working. thermal hazard data. Of the ten companies that do in-house
These five companies skipped this section and the remaining testing for late-stage processes, two said that only Stoessel class
analysis is done on the remaining ten companies who support 1, 2, or 3 reactions will be performed. Two companies
thermal hazard testing at the late stage. mentioned that there are site-specific rules regarding safe
The ten remaining IQ members were asked a series of operating temperatures, and one company said they apply 2/3
questions that are similar to those asked for the mid-stage to of the DSC onset temperature. Finally, three companies noted
capture the increased complexity as the program progresses to that they use 100 °C below the DSC onset temperature and/or
the commercial and manufacturing phase. The first question is 75 °C off the TSu onset temperature as these are the common
“which thermal tests are employed for assessment of late-stage margin of safety adjustments for these instruments. The
processes?” and the following options were given to the response is in good agreement to the question asked earlier in
participants: the mid-stage (Figure 10). This suggests that these safety
A None margins are set and are consistent across stages.
B Quick screening tools (please specify) After thermal hazard testing is performed, oftentimes the
data lead to additional testing. The ten member companies
C Full safety evaluation with calorimetry (please specify) who did internal late-stage manufacture were polled: “can red
D Evaluation is similar to what is done for the mid-stage flags lead to additional testing?”. Four companies said the
Not surprisingly, half of the companies indicated that the approach is the same as their mid-stage evaluation. One
evaluation is similar to what is done in the mid-stage, option D, company noted that if the exotherm is >100 J/g and within the
as typically the chemical processes do not change significantly MTSR then additional testing would be required. The
from the mid-stage to commercial stage. In addition, the remaining four companies noted that internal guidance
process safety tests do not necessarily change from the mid to governs what additional testing may be required, for example,
late stage. This answer shows that process safety tests are testing of upset scenarios like mischarge or thermal stability.
widely applicable to the assessment of early stage to late stage This could include ARC testing and vent sizing calculations. It
chemical processes. The remaining half of the participants do is worth mentioning that in the mid-stage, about half of the
full evaluation with calorimetry, option C. The open responses participants did not have a quantitative trigger for deploying
indicate that companies will do additional testing related to advanced testing, but that is not the case here at the late stage.
changes in the process, take a closer look at the waste streams, The participants were asked if red flags trigger impact
and also analyze authentic samples from the commercial sites. sensitivity and explosive propagation testing, and the following
In the mid stage, two out of 15 companies were still using options were given to the participants.
N https://dx.doi.org/10.1021/acs.oprd.0c00226
Org. Process Res. Dev. XXXX, XXX, XXX−XXX
Organic Process Research & Development pubs.acs.org/OPRD Article

A Yes, please describe these criteria and any further tests stage do not use screening calorimetric techniques and by
that are done default exclusively use advanced calorimetry. Other companies
B No, not at this phase indicated that the need for advanced calorimetric techniques
C Evaluation is similar to what is done for mid-stage would have been addressed earlier in the mid-stage.
The next set of questions addressed the measurement of off-
Majority (seven out ten) responses were very similar to what
gas for reactions performed at a commercial scale. For gas
is done in the mid-stage, namely, option C. Only one company
generation from the desired reaction, the following options
noted that this would not be done presumably because the
were given:
company simply outsources this testing, and participants
meant that this test is not done internally. Finally, two A No, no gas measurement is done at this stage
companies noted that a full testing package (impact, friction, B No, we just assume that the theoretical amount is
and dust explosivity) would be performed for all late-stage liberated
commercial manufacturing processes regardless of whether or C Yes, but only for reactions known to generate gas, please
not the tests were triggered by a screen. Among the triggers describe how this is measured.
discussed in the response were DSC decomposition >500 J/g, D Yes, only if screening tool give red flags (please describe)
>800 J/g and peak shape. The response is similar to what has E Yes, all reactions are screened for pressure (please
been discussed in the mid-stage and the early stage. describe how is measured)
The next four questions addressed gathering calorimetric F Evaluation is similar to what is done for the mid-tage
data for reactions performed at a commercial scale. First, the
participants were asked if the heat of reaction is measured and Six of the ten companies indicated using the same strategy
the following options were provided. used at the mid-stage. Two companies indicated that all
reactions are tested at this stage regardless of the screening
A No, no heat of reaction is measured at this stage
results. Two companies noted that the measurement of off-gas
B Yes, always. Please specify instrument used (example: is only determined for reactions that are known to specifically
Omnical, RC1, THT). generate pressure during the reaction. This is drastically
C Yes, only if heat is expected or previous screening tests different than the response from the mid-stage, where 11 out of
indicate that this is necessary (please describe) 15 companies indicated the measurement of off-gas only for
D Evaluation is similar to what is done for the mid-stage reactions known to generate gas. This suggests that at the late
All companies answered yes (option B and D) which is not stage companies conduct gas measurement even for reactions
surprising as most companies measure the heat of reaction that are not known to generate gas. It is worth mentioning that
even at the mid-stage. The 10 positive responses were equally the tools described in this section for the measurement of gas
split between option B and D, that is, half of the companies do evolution are the same as that described in the mid-stage.
this testing in the mid-stage and if it was not done, they will The member companies were asked to specify which streams
definitely do it at this stage. The participants were asked to undergo analysis for pressure generation as we did in the mid-
provide the type of instrument used to measure calorimetry, stage, as shown in Figure 11. Seven of the companies
and the following platforms RC1 and C80 were mentioned. responded that the same streams undergo testing at this
Again, the response is similar to the mid-stage but excluding stage as for pressure generation as for mid-stage. The
some screening tools like microcalorimetry. Not surprisingly, remaining 3 companies indicated testing every stream on the
option (A) was not selected, that is, all companies that perform list, namely, isolated reagents, starting materials, isolated
internal commercial manufacture note that calorimetry is (or product, reaction mixtures, end of the reaction, distillation
has been) performed at this scale. Even option C was not concentration, and waste streams. In the open response,
selected, indicating that the pharmaceutical companies at this companies were encouraged to list other streams that they
stage will measure heat of reaction irrespective if a screening might look into, and we want to highlight that the companies
triggers it or not. focused on reactions with potential gas generation like
Next, the participants were asked if the maximum heat flow intermediates with a tert-butyloxycarbonyl protecting group
was determined. Nearly all participants (nine out of ten) (BOC) group. Intermediates with a BOC moiety are known to
indicated using the same approach as the mid-stage, where the release isobutylene and CO2 gas.
participants have a rule of thumb of not scaling up a reaction In terms of gas generation measurement for unintended
where the maximum heat flow is >30 W/L, see Section 2.3. reactions or decompositions, companies were given the
Several companies indicated relying on their chemical following options:
engineering group to ensure that the heat generated from A No, not at this stage
the reaction can be removed. B Yes, but only if DSC indicates significant exotherm
For heat of reaction measurements, the majority (7 out of (please describe)
10) will rely on experimental results which is understandable
because of the increased risk at this scale. Only three of the C Yes, all undesired chemistry is screened for gas
participants noted that desktop methods for screening heats of generation (please describe)
reaction were used at a commercial scale but indicated that D Evaluation is similar to what is done for mid-stage
only if the comparable reactions are sufficiently similar to the The answers suggest that the approach here for the
one performed at scale. The use of desktop methods was more undesired gas generation is similar to what is done in the
popular in the mid-stage where 9 out of the 15 companies are mid-stage, where most companies (seven out of 10) picked
using CHETAH or literature to estimate the heat of reaction. option D. Two companies indicated being conservative where
Next, companies were polled if heats of reaction screen all undesired chemistry will be screened. One company picked
measurements could trigger additional calorimetric measure- option B, namely, testing for unintended gas generation will be
ments. The responses indicated that companies at this late done only if DSC data indicate a significant exotherm. Next,
O https://dx.doi.org/10.1021/acs.oprd.0c00226
Org. Process Res. Dev. XXXX, XXX, XXX−XXX
Organic Process Research & Development pubs.acs.org/OPRD Article

participants were asked if gas evolution and/or maximum extended storage times could be justified with DSC and ARC
pressure are examined, the following options were given. testing, one company used ARC testing alone to cover
A Yes, please describe extended storage times, and another one used isothermal TSu
B No experiments.
C Evaluation is similar to what is done for the mid-stage For raw materials, all companies except for one said that
storage conditions are examined for late-stage development.
All ten companies noted that the rates and total pressure are Among the eight companies that responded positively, three
a concern and examined, option A and C. The fact that companies performed storage evaluations at earlier stages and
pressure and gas evolution are examined by all companies is five companies said that the evaluation would be performed
not surprising at this stage, as these parameters are already based on the available thermal stability data. This response is
examined by 14 of the 15 companies even in the mid-stage. much higher than that observed in the mid-stage, where 8
Looking closer at the open responses to see if there are criteria companies responded negatively to the question, that is, no
on what is an acceptable pressure rise and/or gas generation testing was done (Section 2.3).
threshold, the respondents did not provide a quantitative Participants were also asked to describe the type of tests
answer but indicated that there are concerns around not deployed to justify storage conditions and what triggers them.
exceeding the maximum allowable pressure of the vessel and Companies used a combination of DSC, TSu, and ARC data
the need to do vent calculations to ensure safe operating for thermal stability. Beyond thermal stability, one company
pressures. indicated conducting dust and explosivity testing on their
When asked if red flags that arise from the pressure stored materials. For raw materials, intermediates, and
screening tests could warrant additional, more advanced products, companies were asked again, as in the mid-stage, if
testing, and the following options were given: SADT is used to assess risks for decompositions during
A No shipping. Of the ten companies that do internal late-stage
B Yes (please describe) manufacturing, eight use SADT at this stage while two
C Evaluation is similar to what is done for mid-stage companies said no. Again, in the mid-stage, 9 out of the 15
It is worth mentioning that 7 of the 15 companies at the participants did not use SADT which suggest that the use of
mid-stage responded yes to this question. However, at this late SADT is more predominant in the late stage. Looking closer at
stage all companies selected options B and C, that is, additional the open answers for the companies that answered positively,
testing would be deployed. Three companies picked option B three companies determined SADTs at earlier stages. Two
and the remaining seven companies, option C, meaning that companies determined the SADTs of materials if deemed
the red flags would have been picked up and additional tests appropriate by EHS during the HazOp. One company used
performed at earlier stages. Taking a closer look at the open SADT if a material exhibited an exothermic decomposition
answers, one company noted that ARC testing would be below 125 °C. Two of the eight companies used advanced
performed and one said that vent sizing calculations would be software packages (model-free kinetics or AKTS) to determine
done at this stage. a material’s SADT, which is again similar to the approach in
With respect to flammable gas generation during the course the mid-stage.
of a process, if the flammable gas is diluted with nitrogen, four The participants were asked whether or not MOC
companies noted that the margin of safety is 25% of the LEL of compatibility was considered. All responses were positive
the mixture. One company targets 10% of the LEL. The other that testing is done which is not surprising as virtually all
companies that responded noted that the margin of safety is companies (14 out 15) are doing this testing in the mid-stage,
site specific. Overall, the responses are very similar to the as shown in Figure 13. The positive responses include two
responses from the mid-stage (Section 2.3). companies that use coupon testing. Two companies also stated
The next series of questions covered protracted hold times. that if incompatibilities are known, then additional testing
This included reaction mixtures, waste streams, and the storage would be performed. One company said that the receiving site
and transportation of materials. All companies noted that the is responsible for all MOC considerations. Finally, one
testing needed to cover protracted reaction hold times, which company stated that the process safety labs study how MOC
would have already been completed at the mid-stage. affects thermal hazard testing results. Again, the responses are
Interestingly, at this late stage, participants were given the very similar to those seen in the mid-stage.
option of “testing only if there is a red flag”; however, that Next, we asked if stream conductivity is considered for
option was not selected which suggests that these tests are potential electrostatic hazards, and all responses were positive.
deployed regardless. It is worth mentioning that this option of Five companies noted that this data would have been gathered
testing only when there are red flags was the predominant at previous stages. One company said that conductivity testing
choice, 9 out of 15, in the mid-stage, as shown in Figure 12. is done for all processes. One company said that the data are
For waste stream hold, the same question was asked and the pulled from literature sources. Finally, one company said that
response was again all positive. Seven companies said this the data are gathered at an external vendor. Overall the
would have been performed in the mid-stage, and as with the response suggests an increase in scrutiny of the electrostatic
reaction hold, none of the companies said that the red flags hazard at this stage as 8 out of 15 companies did not even
would trigger these tests, which suggests that at this stage these consider the streams’ conductivity in the mid-stage.
streams are tested regardless. The response is significantly For process safety testing requirements for continuous flow
different than the approach in the mid-stage, where 6 processes, four companies said that the evaluation of flow
companies said these streams are not tested and 9 said these processes would have already been completed at an earlier
waste streams will only be tested if there are red flags in the stage. Two companies said the evaluation of a flow process is
screening process. Taking a close look at the open responses identical to batch processes and no additional testing is done.
for companies that perform the testing, one company said that Two companies also said that no late-stage processes are
P https://dx.doi.org/10.1021/acs.oprd.0c00226
Org. Process Res. Dev. XXXX, XXX, XXX−XXX
Organic Process Research & Development pubs.acs.org/OPRD Article

Figure 15. Survey results about specific techniques used for PHA.

performed in continuous mode. Overall the response is similar data would have been gathered at earlier stages of develop-
to what is observed in the mid-stage. ment.
Companies were next asked “what results/outcome would Process development of the unit operations in the
prompt your lab to stop a reaction from moving forward and pharmaceutical industry presents a design space where
propose changing the chemistry to mitigate the observed risk”. variability in the process parameter and input variables (like
As in the mid-stage answers varied and here are the key temperature, reagent equivalent) are permitted as it would still
responses. provide assurance of quality. These variations within the design
• One company said a reaction’s Stoessel classification space are typically not evaluated during thermal hazard
would be used to give a go/no go decision. evaluation, and safety tests are done at the intended center
point. In the next question of the survey, companies were
• One company noted that if adequate protection asked how they evaluate safety when operating with the design
measures could not be implemented to contain the space, although experimental evaluation and results might be
consequences of the event then the reaction would limited to the center point. The five companies that perform
require further development. commercial-scale manufacture noted that they would not carry
• The remaining six companies said that all issues that out further testing beyond the center point. The companies
would prevent a reaction from being executed at a that considered the safety of the process over the design space
commercial scale would have been resolved at earlier mentioned doing the testing at the worst-case scenario and
stages. some companies use a paper exercise based on FMEA.
When asked about assessing the mixing of waste streams,
All companies were asked if a formal PHA is required. At only one company that performs commercial manufacturing
this stage, all companies responded positively that PHA is stated that they did not assess the mixing of waste streams.
required with the exception of one who said that it depends on This could be due to the location specific requirements for
the CMO. This is a significant increase compared to only 5 out waste handling (i.e., waste is shipped off site for incineration).
of 15 companies in the mid-stage that deploy a PHA. The other seven companies perform some type of waste mixing
Companies stated that multiple PHA methodologies could assessments. Some companies will perform waste neutraliza-
be implemented within the same company. Six perform Hazard tion if required by the site.
Operability (HAZOP) methodology. Four use failure mode The final question asked the participating companies how
and effects analysis (FMEA). One company used layers of they keep track of changes made to processes post validation.
protection analysis. Four companies did not disclose the Two companies said they do not track changes to processes
specific PHA techniques they employed (see Figure 15). post validation. In this instance, the manufacturer is
If a failure mode is identified, the companies were asked how responsible for assessing the safety of any proposed changes.
these modes were evaluated, with experimental data and/or Two companies track changes made to processes through
modeling. Seven companies use experimental data to evaluate written reports that are archived within the respective
failure modes and one company did not gather any additional company. Four companies rely on their external sourcing
data. There was no indication that modeling was used to study groups to reach out when changes to processes are proposed.
these failure modes. Modeling was slightly more popular in the
mid-stage (11 out 15); it appears that late-stage companies are
more cautious in terms of adopting modeling which is 3. CONCLUSIONS
understandable. As for general kinetic modeling use, three The IQ thermal hazards and process safety working group
companies said they will not use modeling for studying upset completed a survey of its membership to assess approaches to
scenarios. Three companies also stated that they will only use process safety development. Participation rates among the
modeling for low-risk activities. Two companies said that these member companies was excellent, and the information
Q https://dx.doi.org/10.1021/acs.oprd.0c00226
Org. Process Res. Dev. XXXX, XXX, XXX−XXX
Organic Process Research & Development pubs.acs.org/OPRD Article

gathered covered all phases of development and manufactur- Derek B. Brown − Process Development, One Amgen Center
ing. While many commonalities exist, it should not come as a Drive, Amgen Inc., Thousand Oaks, California 91320, United
surprise that each company has a different approach for States; orcid.org/0000-0002-7909-0867
gathering and developing process safety information, and these Stanley P. Kolis − Small Molecule Design and Development, Eli
approaches are likely reflective of the individual member Lilly and Company, Indianapolis, Indiana 46285, United
companies’ cultures and risk tolerance. In addition, there were States; orcid.org/0000-0001-6268-9470
very few commonalities noted among the way member Jeffrey B. Sperry − Vertex Pharmaceuticals, Boston,
companies interact with CMOs and share their thermal hazard Massachusetts 02210, United States; orcid.org/0000-0003-
results. 0365-5646
The key findings from the survey in terms of engaging PSL Complete contact information is available at:
throughout the drug development lifecycle is the following. At https://pubs.acs.org/10.1021/acs.oprd.0c00226
the early phase, some companies reported not having
medicinal chemistry groups to support, whereas others are Author Contributions
either supporting the early phase in a scaled-back manner or The manuscript was written through contributions of all the
not at all. Most companies surveyed have a threshold that authors. All the authors have given approval to the final version
triggers an evaluation by PSL; however, that threshold varies of the manuscript.
significantly from company to company. The “mid-stage” of
Notes
development (roughly corresponding to the execution of drug
The authors declare no competing financial interest.


substance campaigns in the kilo-lab or pilot plant) is where the
majority of companies surveyed are executing process safety
ACKNOWLEDGMENTS
work. Almost half of the companies’ PSL are not providing
process support in the late phase, which likely reflects a desire The authors are grateful to the entire IQ thermal hazard team
to have most process safety risks understood and discharged that worked together to craft and design the questionnaire and
prior to transfer to manufacturing. subsequently took the time to answer them. That includes
Among the common themes among the companies, they all Nelson Landmesser (TEVA), Brian D. Phenix (Vertex), Simon
indicated that exothermic and pressure generating reactions are Leung (Bristol-Myers Squibb), John D. Weaver III (Pfizer),
their prime area of focus regardless of the stage of Frank Dixon Jr. (GlaxoSmithKline), Lady Mae Alabanza
development. Also, most member companies will closely (Genentech), Megan Roth (Merck), Chris Mitchell (Takeda),
examine any process changes and most likely any change will Charles Papageorgiou (Takeda), Zhe Wang (Abbvie), Steve
trigger an experimental thermal hazard examination of the Richter (Abbvie), Max Sarvestani (Boehringer Ingelheim),
change on process safety. Across the member companies, the Shasha Zhang (Bristol-Myers Squibb Company), Mark Hoyle
Stoessel criticality classification system is becoming a common (AstraZeneca), and Christopher Tickner (Bristol-Myers
language for assessment of thermal risk. On the other hand, Squibb Company). The team appreciates the support of the
when it comes to establishing a “margin of safety”, variation International Consortium for Innovation and Quality in
was noted among companies that participated in the survey. Pharmaceutical Development (IQ, www.iqconsortium.org)
Finally, the authors’ desire is that this article will be used by throughout the process. We would especially like to thank
the pharmaceutical companies and other institutions that Maja Leah Marshall for her support of the project during the
handle chemicals including and not limited to CROs, rollout of the Survey, and her expertise in SurveyMonkey was
crucial to the success of this effort.


universities, and other chemical industries to evaluate their
current PSE programs. However, the reader should be
reminded that approaches discussed in this paper might not NOMENCLATURE
be the best practices and are subject to change as they ΔHrxn (a) the desired reaction at the process temperature,
represent the current state of how surveyed companies conduct (b) by runaway reactions and/or (c) thermal
their hazard evaluation. Changes in the current state can come decompositions J/mol
from learning from incidents and/or evolution of testing ΔTad adiabatic temperature rise associated with that
methodologies. Indeed, we plan to conduct a follow-up survey exotherm, °C (K)
in the future to assess how surveyed companies change their Qrxn thermal Energy of the reaction J
safety testing methodologies. AKTS advanced kinetics and technology solutions


ARC accelerating rate calorimetry
BOC tert-butyloxycarbonyl protecting group
AUTHOR INFORMATION
CMO contract manufacturing organizations
Corresponding Author Cp heat capacity of reaction mixture J/g °C
Ayman D. Allian − Process Development, One Amgen Center CRO contract research organization
Drive, Amgen Inc., Thousand Oaks, California 91320, United DSC differential scanning calorimetry
States; orcid.org/0000-0002-1604-6738; Email: aallian@ LEL lower explosive limit
amgen.com LOC limiting oxygen concentration
m mass g
Authors MTSR maximum temperature of the synthesis reaction °C
Nisha P. Shah − Process Development, Gilead Sciences, Inc., MTT maximum temperature for technical reasons °C
Foster City, California 94404, United States; orcid.org/ Phi (ϕ) correction factor used to account for the thermal
0000-0001-7524-5959 mass of a sample bomb. ϕ = 1 + (mbCpb/msCps),
Antonio C. Ferretti − Chemical Process Development, Bristol where m and Cp are the mass and heat capacity of
Myers Squibb, Summit, New Jersey 07901, United States the bomb and the sample. Unitless
R https://dx.doi.org/10.1021/acs.oprd.0c00226
Org. Process Res. Dev. XXXX, XXX, XXX−XXX
Organic Process Research & Development pubs.acs.org/OPRD Article

PSL process safety laboratories (15) Likhite, N.; Lakshminarasimhan, T.; Rao, M. H. V. R.;
RC reaction calorimetry Shekarappa, V.; Sidar, S.; Subramanian, V.; Fraunhoffer, K. J.; Leung,
Tp temperature of the process °C S.; Vaidyanathan, R. A Scalable Synthesis of 2-(1,2,4-Oxadiazol-3-
TSu thermal screening unit yl)propan-2-amine Hydrobromide Using a Process Safety-Driven
TD24 temperature at which the time to maximum rate for a Protecting Group Strategy. Org. Process Res. Dev. 2016, 20, 1328−
1335.
decomposition reaction is 24 h °C (16) Veedhi, S.; Babu, S. R. Process Safety Evaluation To Identify
VSP vent sizing package


the Inherent Hazards of a Highly Exothermic Ritter Reaction Using
Adiabatic and Isothermal Calorimeters. Org. Process Res. Dev. 2013,
ADDITIONAL NOTES 17, 1597−1602.
a
The Carius Tube test is used to determine the onset (17) Zhu, H.-T.; Arosio, L.; Villa, R.; Nebuloni, M.; Xu, H. Process
Safety Assessment of the Iron-Catalyzed Direct Olefin Diazidation for
temperature for decomposition of reagents and the rate of gas the Expedient Synthesis of Vicinal Primary Diamines. Org. Process Res.
evolution from such decomposition. Dev. 2017, 21, 2068−2072.
b
Advanced reactive system screening tool. (18) Shilcrat, S. Process Safety Evaluation of a Tungsten-Catalyzed

■ REFERENCES
(1) Grawunder, S.; Crockford, C.; Kalan, A. K.; Clay, Z.; Stoessel,
Hydrogen Peroxide Epoxidation Resulting In a Runaway Laboratory
Reaction. Org. Process Res. Dev. 2011, 15, 1464−1469.
(19) Tang, W.; Sarvestani, M.; Wei, X.; Nummy, L. J.; Patel, N.;
Narayanan, B.; Byrne, D.; Lee, H.; Yee, N. K.; Senanayake, C. H.
A.; Hohmann, G. Response to Garcia and Dunn. Curr. Biol. 2019, 29,
R734−R735. Formation of 2-Trifluoromethylphenyl Grignard Reagent via
(2) Dermaut, W. Process Safety and Reaction Hazard Assessment. In Magnesium−Halogen Exchange: Process Safety Evaluation and
Chemical Engineering in the Pharmaceutical Industry: R & D to Concentration Effect. Org. Process Res. Dev. 2009, 13, 1426−1430.
Manufacturing; Am Ende, D. J., Ed.; John Wiley & Sons, Inc., 2011; (20) Administration, Under Floor Air Distribution https://www.fda.
pp 155−182. gov/patients/learn-about-drug-and-device-approvals/drug-
(3) Frurip, D. J.; Elwell, T. Effective use of differential scanning development-process (accessed 16 Sept, 2019).web
calorimetry in reactive chemicals hazard evaluation. Process Saf. Prog. (21) Butters, M.; Catterick, D.; Craig, A.; Curzons, A.; Dale, D.;
2007, 26, 51−58. Gillmore, A.; Green, S. P.; Marziano, I.; Sherlock, J.-P.; White, W.
(4) Hoare, J.; Duddu, R.; Damavarapu, R. A Safe Scalable Process Critical Assessment of Pharmaceutical ProcessesA Rationale for
for Synthesis of 4,6-Bis(nitroimino)-1,3,5-triazinan-2-one (DNAM). Changing the Synthetic Route. Chem. Rev. 2006, 106, 3002−3027.
Org. Process Res. Dev. 2016, 20, 683−686. (22) Taylor, D. The Pharmaceutical Industry and the Future of Drug
(5) Sperry, J. B.; Minteer, C. J.; Tao, J.; Johnson, R.; Duzguner, R.; Development. In Pharmaceuticals in the Environment; The Royal
Hawksworth, M.; Oke, S.; Richardson, P. F.; Barnhart, R.; Bill, D. R.; Society of Chemistry, 2016; pp 1-33.
Giusto, R. A.; Weaver, J. D. Thermal Stability Assessment of Peptide (23) Laird, T. Chemical and Process Safety. Org. Process Res. Dev.
Coupling Reagents Commonly Used in Pharmaceutical Manufactur- 2012, 16, 1979.
ing. Org. Process Res. Dev. 2018, 22, 1262−1275. (24) Seabury, R. W.; Stoessel, A. M.; Steele, J. M. Vancomycin
(6) Wang, Z.; Richter, S. M.; Bellettini, J. R.; Pu, Y.-M.; Hill, D. R. Trough Concentration Poorly Characterizes AUC: Is It Time to
Safe Scale-Up of Pharmaceutical Manufacturing Processes with Transition to AUC-Based Vancomycin Monitoring? Ann. Pharmac-
Dimethyl Sulfoxide as the Solvent and a Reactant or a Byproduct. other. 2017, 51, 926−927.
Org. Process Res. Dev. 2014, 18, 1836−1842. (25) Stoessel, F. Thermal Safety of Chemical Processes: Risk Assessment
(7) Allian, A. D.; Richter, S. M.; Kallemeyn, J. M.; Robbins, T. A.; and Process Design; Wiley-VCH: Weinheim, 2008.
Kishore, V. The Development of Continuous Process for Alkene (26) UNECE UN Model Regulations: UN Recommendations on
Ozonolysis Based on Combined in Situ FTIR, Calorimetry, and the Transport of Dangerous Goods. https://www.unece.org/?id=
Computational Chemistry. Org. Process Res. Dev. 2011, 15, 91−97. 3598 (accessed 16 Sept, 2019).web
(8) Martinot, T. A.; Ardolino, M.; Chen, L.; Lam, Y.-h.; Li, C.; (27) Dermaut, W. Process Safety and Reaction Hazard Assessment.
Maddess, M. L.; Muzzio, D.; Qi, J.; Saurí, J.; Song, Z. J.; Tan, L.; In Chemical Engineering in the Pharmaceutical Industry; Am Ende, D. J.,
Vickery, T.; Yin, J.; Zhao, R. Process Safety Considerations for the Ed., 2010.
Supply of a High-Energy Oxadiazole IDO1-Selective Inhibitor. Org. (28) Pasman, H. Foreword. In Lees’ Process Safety Essentials;
Process Res. Dev. 2019, 23, 1178−1190. Mannan, S., Ed.; Butterworth-Heinemann: Oxford, 2014; p xxi.
(9) Monteiro, A. M.; Flanagan, R. C. Process Safety Considerations (29) Cole-Parmer Chemical Compatability Database. https://www.
for the Use of 1 M Borane Tetrahydrofuran Complex Under General coleparmer.com/chemical-resistance (accessed 16 Sept, 2019).web
Purpose Plant Conditions. Org. Process Res. Dev. 2017, 21, 241−246. (30) Schweitzer, P. Corrosion Resistance Tables, 4th ed.; Marcel
(10) Agosti, A.; Bertolini, G.; Bruno, G.; Lautz, C.; Glarner, T.; Dekker: New York, NY (United States), 1995; p Medium: X; Size:
Deichtmann, W. Handling Hydrogen Peroxide Oxidations on a Large [600] p.
Scale: Synthesis of 5-Bromo-2-nitropyridine. Org. Process Res. Dev. (31) Metals Section. In Corrosion Data Survey, 6th ed.; Graver, D. L.,
2017, 21, 451−459. Ed.; Nace International, 1985.
(11) Emerson, K.; Muzzio, D.; Fisher, E. Identification of Significant (32) Craig, B.; Anderson, D. Handbook of Corrosion Data, 2nd
Process Safety Risks in the Preparation of Methyl-N-cyanocarbamate. edition; ASM International: Materials Park, OH (United States),
Org. Process Res. Dev. 2019, 23, 1352−1358. 1995; p Medium: X; Size: 1000 p.
(12) Reeves, J. T.; Sarvestani, M.; Song, J. J.; Tan, Z.; Nummy, L. J.; (33) McConville, F. X. The Pilot Plant Real Book: A Unique
Lee, H.; Yee, N. K.; Senanayake, C. H. Process Safety Evaluation of a Handbook for the Chemical Process Industry, 2nd ed.; FMX Engineering
Magnesium−Iodine Exchange Reaction. Org. Process Res. Dev. 2006, and Design: Worcester, MA, 2007.
10, 1258−1262. (34) Stoessel, F. Fundamentals of Thermal Process Safety. Thermal
(13) Smith, I. H.; Alorati, A.; Frampton, G.; Jones, S.; O’Rourk, J.; Safety of Chemical Processes; Wiley-VCH: Weinheim, 2008; pp 37−38.
Woods, M. W. Rapid Scale-Up of the Matrix Metalloproteinase (35) Mannan, S. Reactive Chemicals. In Lees’ Process Safety
Inhibitor CH5902: Process Safety and Route Development Essentials; Mannan, S., Ed.; Butterworth-Heinemann: Oxford, 2014,
Considerations. Org. Process Res. Dev. 2003, 7, 1034−1039. Chapter 22, pp 409−424.
(14) Kryk, H.; Hessel, G.; Schmitt, W. Improvement of Process (36) Singh, J.; Simms, C. The Thermal Screening Unit (TSu) as a
Safety and Efficiency of Grignard Reactions by Real-Time Tool for Reactive Chemical Screening. Inst. Chem. Eng. Symp. Ser.
Monitoring. Org. Process Res. Dev. 2007, 11, 1135−1140. 2001, 148, 67−80.

S https://dx.doi.org/10.1021/acs.oprd.0c00226
Org. Process Res. Dev. XXXX, XXX, XXX−XXX
Organic Process Research & Development pubs.acs.org/OPRD Article

(37) Townsend, D. I.; Tou, J. C. Thermal hazard evaluation by an


accelerating rate calorimeter. Thermochim. Acta 1980, 37, 1−30.
(38) Yoshida, T.; Yoshizawa, F.; Itoh, M.; Matsunaga, T.; Watanabe,
M.; Tamura, M. Prediction of Fire and Explosion Hazards of Reactive
Chemicals. I. Estimation of Explosive Properties of Self-Reactive
Chemicals from SC-DSC Data. Kogyo Kayaku 1987, 48, 311−316.
(39) Harrison, B. K. CHETAH 10.0, the ASTM Computer Program
for the Prevention of Reactive and Flammability Hazards. J. Test. Eval.
2016, 44, 1035−1049.
(40) van Speybroeck, V.; Gani, R.; Meier, R. J. The calculation of
thermodynamic properties of molecules. Chem. Soc. Rev. 2010, 39,
1764−1779.
(41) Hafner, A.; Mancino, V.; Meisenbach, M.; Schenkel, B.;
Sedelmeier, J. Dichloromethyllithium: Synthesis and Application in
Continuous Flow Mode. Org. Lett. 2017, 19, 786−789.
(42) McWilliams, J. C.; Allian, A. D.; Opalka, S. M.; May, S. A.;
Journet, M.; Braden, T. M. The Evolving State of Continuous
Processing in Pharmaceutical API Manufacturing: A Survey of
Pharmaceutical Companies and Contract Manufacturing Organiza-
tions. Org. Process Res. Dev. 2018, 22, 1143−1166.
(43) Mathew, J. S.; Klussmann, M.; Iwamura, H.; Valera, F.; Futran,
A.; Emanuelsson, E. A. C.; Blackmond, D. G. Investigations of Pd-
Catalyzed ArX Coupling Reactions Informed by Reaction Progress
Kinetic Analysis. J. Org. Chem. 2006, 71, 4711−4722.
(44) Ferretti, A. C.; Mathew, J. S.; Blackmond, D. G. Reaction
Calorimetry as a Tool for Understanding Reaction Mechanisms:
Application to Pd-Catalyzed Reactions. Ind. Eng. Chem. Res. 2007, 46,
8584−8589.

T https://dx.doi.org/10.1021/acs.oprd.0c00226
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

You might also like