Hazard Indices: Completion of a Unified Suite of Metrics for the

Jan 14, 2013 - Safety/Hazard Indices: Completion of a Unified Suite of Metrics for the Assessment of “Greenness” for Chemical Reactions and Synthe...
1 downloads 5 Views 2MB Size
Concept Article pubs.acs.org/OPRD

Safety/Hazard Indices: Completion of a Unified Suite of Metrics for the Assessment of “Greenness” for Chemical Reactions and Synthesis Plans John Andraos CareerChem, 504-1129 Don Mills Road, Don Mills, Ontario M3B 2W4, Canada S Supporting Information *

ABSTRACT: An overall Safety/Hazard Index (SHI) is introduced and defined in the same way as the previously described benign index (BI) covering various environmental impacts. Following the same themes and symbolism usage found in the Workplace Hazardous Materials Information System (WHMIS) and National Fire Protection Association (NFPA) 704 code, SHI covers the following safety-hazard potentials: corrosive gas (CGP), corrosive liquid/solid (CLP), flammability (FP), oxygen balance (OBP) applied to combustion reactions and oxidation reactions, hydrogen gas generation (HGP), explosive vapour (XVP), explosive strength (XSP), impact sensitivity (ISP), risk phrase (RPP), occupational exposure limit (OELP), maximum allowable concentration (MACP), dermal absorption (DAP), and skin dose (SDP). In addition, reaction temperature hazard (RTHI) and reaction pressure hazard (RPHI) indices are defined with respect to reference ambient reaction conditions of 25 °C and 1 atm. All three indices vary in value between 0 and 1 to conform to the formalism of BI and other well-known material efficiency green metrics. The methodology is illustrated using single-step and multistep synthesis plans for aniline, phenol, and phenyl isocyanate. Using the best available data, the overall “greenest” routes for these industrially important commodity chemicals are determined with respect to material efficiency, environmental impact, and safety/hazard impact. Results are conveniently presented using radial polygon diagrams and are compared with a modified Edwards−Lawrence inherent safety index formalism.



INTRODUCTION The quest to quantitatively determine overall “greenness”, beyond simple determination of material consumption and waste production, of individual chemical reactions and synthesis plans using green metrics1−10 is an ongoing endeavour. A recent report11 that introduced a benign index (BI) incorporating various environmental impact potentials is the latest contribution to this effort. Although this metric could, in principle, include an unlimited number of parameters encompassing climate change impact, human health impact, and environmental damage as a direct consequence of release of chemicals into the environment (air, water, soil, and sediment), its utility and reliability was shown to be strongly governed by the quality and availability of experimentally and computationally determined fundamental physical data. Furthermore, such source data needed for environmental impact assessments are plentiful only for hydrocarbons (aliphatic and aromatic), halogenated hydrocarbons (CFCs, HFCs, and PCBs), pesticides, simple alcohols and amines, simple inorganics (acids and bases), gases, and famous industrial organic pollutants such as PCBs and dioxins. This meant that applying BI calculations to syntheses of complex pharmaceuticals and fine chemicals was fraught with the problem of dealing with missing data, particularly for all intermediates with progressively more elaborate structures that are isolated along the synthesis paths to these target molecules . These points were thoroughly explored and discussed in the context of an in-depth investigation of syntheses of the following simple, yet industrially important, chemicals: aniline, aspirin, phenol, diphenyl carbonate, and phenyl isocyanate. In particular, phosgene-based and nonphosgene-based chemistries leading to phenyl isocyanate and diphenyl carbonate were compared. The results showed that phosgene reactions were found to be © 2013 American Chemical Society

both more material efficient and environmentally benign than dimethyl carbonate reactions, which were selected as phosgene replacements, within the constraint of data available for environmental impact potentials used. It was suggested that this conclusion could be reversed if a safety/hazard metric were also incorporated in the analysis in addition to benign indices. Generally, it was found that the largest contributing environmental potential associated with environmental damage was toxicity due to direct ingestion of released chemicals by organisms in the environment. On the other hand, climate change potentials such as ozone depletion, smog formation, and global warming were only important if hydrocarbon- and halogen-containing compounds were used as reagents or solvents in chemical reactions. A key point made was that the potential to cause environmental harm is distinguishable from occupational safety/ hazard concerns. Therefore, a safety/hazard metric would have to be defined specifically in the context of occupational health and safety risks associated with workers exposing themselves to chemicals while carrying out chemical reactions in an academic or industrial laboratory environment. This report presents such an index using the same methodology and formalism as for the benign index (BI). In addition, two other indices relating to temperature and pressure reaction conditions are also defined for the hazard assessment of individual chemical reactions with respect to how far those conditions are from ambient temperature (25 °C) and pressure (1 atm). The same synthesis plans of the previously described target molecules are re-evaluated using the new Safety/ Hazard Index (SHI), Reaction Temperature Hazard Index (RTHP), Received: December 7, 2012 Published: January 14, 2013 175

dx.doi.org/10.1021/op300352w | Org. Process Res. Dev. 2013, 17, 175−192

Organic Process Research & Development

Concept Article

weighting factors are used for each chemical species. The inventory score is associated with the tonnage scale of chemicals produced, assuming 8000 h production per year and is not directly linked to the actual mass consumption of materials in a specified chemical reaction. The actual masses of ingredients needed to carry out a synthesis plan may be interpreted as a minimum inventory of chemicals, whereas the 8000 h/year production scale may be interpreted as a maximum inventory. The potential of a chemical to be corrosive is not taken into account. For a given reaction the chemical score is the sum of the maximum values for the four contributing scores taken over all reactants and products. For each reagent and product species the inventory, flammability, explosiveness, and toxicity scores are added up, and then the highest of these sums is selected as the chemical score for that reaction step. This means that only the worst offending reagent or product in a reaction is counted. Reaction solvents and other auxiliary materials employed in a given chemical reaction are not taken into account. The neglect of reaction and auxiliary solvents in particular is a severe liability with the ISI methodology since it is well-known that they contribute most to overall hazard and toxicity due to their greater mass proportion to the total mass of materials used in a chemical reaction. For a reaction sequence, the total ISI for a plan is equal to the sum of ISIs for contributing reactions. Since the ISI is a number, it can be used to rank reactions and plans. Other limitations and controversies associated with its implementation have been documented.17 The shortcomings of all of these methodologies are circumvented by the Safety/Hazard Index presented in this work.

and Reaction Pressure Hazard Index (RPHI), in addition to BI and other material efficiency metrics such as atom economy (AE), reaction yield, and process mass intensity (PMI),12 in order to assess their true global “greenness”. Radial polygon diagrams are used throughout to depict visually the strengths and weaknesses of all reactions and plans. Limitations of the global method are also discussed. Hazard indices have been proposed before, and unlike the indices presented in this work, all are based on penalty scoring systems assigned to various factors, where the reactions or processes that accrue the most penalty points are classified as hazardous and those with the fewest total are classified as safer by comparison. Arguably, the main drawback with such indices defined in this way is that the scoring scales are arbitrarily chosen. Following is a brief description of the main hazard indices documented in the literature arranged in order of sophistication. The Gupta−Babu hazard waste index (HWI)13 follows the National Fire Protection Association (NFPA) scoring scale ranging between 0 and 4 for flammability and reactivity contributors, the Dow Chemical rating for corrosivity, and threshold limit values (TLV) for the toxicity rating. All four components are mass weighted with respect to the mass fraction of the total waste associated with a given chemical. However, the HWI is not treated as a single number derived as a sum or product of contributing factors; rather, it is depicted as an overall code. Since HWI is not a number, it cannot be used for ranking purposes. Furthermore, HWI does not take into account reaction yields, pressure and temperature conditions, and explosiveness potential. The Dow fire and explosion index (FEI)14 is composed of general and specific process hazards factors and a material factor. The material factor covers flammability. The general process hazard factor covers exothermicity, endothermicity, material handling and transfer, accessibility of equipment, process unit design, and drainage and spillage control. The specific process hazard factor covers material toxicity, reaction pressure, quantities of flammable or unstable materials in storage, operation in or near the flammable range of a chemical, reaction temperature, corrosion and erosion, leakage from joints in equipment, fire equipment requirements, use of hot-oil heat exchange systems, and rotating equipment requirements. Although the Dow FEI is comprehensive, as the name suggests, it focuses mainly on flammability and explosiveness hazards associated with chemicals used and equipment requirements and wear and tear. Again, no mass weighting factors are used for each chemical species in a reaction, although the FEI is a real number that can be used for ranking processes. A Turkish group proposed an overall rating value (ORV)15 for hazardous waste that covers several environmental risk factors including bioaccumulation, biodegradability, corrosivity, carcinogenicity, ignitibility, infectiousness, persistence, reactivity, solubility, and toxicity. No mass weighting factors are used for each chemical species. Since ORV is applied to a conglomeration of chemicals found in hazardous waste from multiple processes, parameters such as explosiveness, pressure, temperature, and reaction yield associated with specified chemical reactions are not applicable. Many of the parameters making up ORV are already dealt with in the benign index formalism. Finally, the Edwards−Lawrence inherent safety index (ISI),16,17 conceptually similar to the present investigation and used to evaluate the safety profiles of industrial plans to methyl methacrylate, is composed of a chemical score and a process score. The chemical score covers mass inventory, flammability, toxicity, and explosiveness of a chemical. The process score covers temperature, pressure, and reaction yield. No mass



METHODOLOGY Concept of Safety/Hazard Index. Following the same formalism as benign indices,11 the Safety/Hazard Index (SHI) is defined as shown in eq 1. SHI = 1 − SHZI = 1 −

∑j f j Ωj ∑j Ωj

(1)

where 0 ≤ SHI ≤ 1, SHZI is the Safety/Hazard Impact, j refers to the jth compound, Ωj is the sum of all potentials given by Ωj = (FP)j + (CGP)j + (CLP)j + (OBP)j + (HGP)j + (XVP)j + (XSP)j + (ISP)j + (RPP)j + (OELP)j + (DAP)j + (SDP)j

and f j is the mass fraction of compound j given by fj =

mass of compound j total mass of compounds of type j involved in reaction

The definitions of all abbreviations listed in the sum for Ωj are given at the end of the paper. If SHI = 0 (SHZI = 1), then the Safety/Hazard Impact is maximal, whereas if SHI = 1 (SHZI = 0), then it is minimal and corresponds to a “green” situation with respect to this metric. Each of the potentials, Pxj linked to property x and comprising Ωj, is defined according to eq 2. Pxj =

Xj X ref

(2)

where Xj is the value of property x for compound j and Xref is the value of property x for the reference compound. Just as in the case of the definition of the benign index, SHI according to eq 1 is 176

dx.doi.org/10.1021/op300352w | Org. Process Res. Dev. 2013, 17, 175−192

Organic Process Research & Development

Concept Article

reaction solvents all the way to isolation of the purified product, and the final disposal of all waste materials. In addition to hazards associated with materials used in a chemical reaction two other hazard indices are defined that deal with two fundamental reaction condition variables, namely, temperature and pressure. The reaction temperature and pressure hazard indices (RTHI and RPHI) are given by eqs 4 and 5.

able to support an unlimited number of potentials based on parameters beyond those selected in this work. The only constraints are a choice of a clearly defined and measurable parameter that is directly linked to the hazard in question and that a reference compound is chosen to anchor the definition of its associated potential according to the form of eq 2. The Supporting Information [SI] contains equations of the form of eq 2 for each potential including example calculations and literature references for obtaining necessary data as well as a template HAZARDS algorithm in Excel format to facilitate computation. Table S1 in the SI gives a summary of potential definitions, expressions, and reference compounds associated with safety and hazard. Each of the potentials used may be linked directly to at least one of the hazard symbols used in the Workplace Hazardous Materials Information System (WHMIS)18 or the National Fire Protection Association (NFPA) 704 system19 (see Table of Contents graphic). The only symbol not linked was the one representing biological and infectious hazardous material, which would rarely be encountered in the context of traditional chemical synthesis. Just as in the case of BI, for any chemical reaction SHI may be applied to three cases: all input materials used, all output materials produced, and all waste output materials produced. Input materials include reagents, catalysts and other additives, reaction solvent, workup materials, and purification materials. Output materials include the target product, all side and byproducts, unreacted reagents, reaction solvent, catalysts and other additives, workup materials, and purification materials. Waste materials include all side and byproducts, unreacted reagents, reaction solvent, catalysts and other additives, workup materials, and purification materials. As done before, for the purposes of this work everything except the intended target product is considered waste, and nothing is reclaimed for recycling. The corresponding expressions for the associated SHI parameters are given by eqs 3a−3c. SHI(input) = 1 −

SHI(output) = 1 −

SHI(waste) = 1 −

(3a)

∑j f (output)j Ωj ∑j Ωj

(3b)

(3c)

where f (input)j =

mass of input compound j total mass of input compounds used in reaction

f (output)j =

mass of output compound j total mass of output compounds produced in reaction

(5)

(6)

where rj refers to each of the above metrics ranging in value between 0 and 1. When analyzing synthesis plans to a given target molecule, a common basis scale of 1 ton (907.2 kg) of material was chosen, and all materials required along the synthesis chain were scaled appropriately as previously described.11 In the case of determining overall SHI(input), the masses of all scaled input materials were calculated. In the case of determining overall SHI(waste), the masses of all scaled unreacted reagents, catalysts, reaction solvents, reaction byproducts, and auxiliary materials were calculated. Radial polygons showing overall plan performance were composed of the following six metrics: overall atom economy, overall yield, overall RME (inverse of overall PMI), overall BI(waste), overall SHI(input), and overall SHI(waste). Again, overall plan “greenness” was measured using the corresponding vector magnitude ratio as given by eq 7. Note that the RTHI and

and f (waste)j =

⎡ |Pj − Pambient| ⎤ (RPHI)j = exp⎢ − ⎥ Pambient ⎦ ⎣

⎡ 9 ⎤1/2 VMR = (1/ 9 )⎢∑ (rj)2 ⎥ ⎢⎣ ⎥⎦ j=1

∑j f (waste)j Ωj ∑j Ωj

(4)

where Tj is the temperature for reaction j (K), Tambient = 298 K, Pj is the pressure for reaction j (atm), and Pambient = 1 atm. The exponential function was chosen to guarantee that the numerical value of the index should also fall in the range 0−1. Hence, if Tj = 298 K, then (RTHI)j = 1 which corresponds to a nonhazardous situation implying a benign state. If Tj is far from 298 K, then (RTHI)j = 0 which corresponds to an extreme hazard state. The same logic is used in the case of reaction conditions different from 1 atm. The absolute values of differences of temperature and pressure from ambient conditions imply that conditions equally above and below these reference states would have identical hazard potentials. For example, a reaction carried out 100 °C above room temperature would impart the same hazard potential as one carried out 100 °C below room temperature. However, it must be understood that the manifestation of and potential consequences of accidents occurring due to these two hazards would be different. The SI shows graphs of the exponential functions given in eqs 4 and 5. The use of radial polygon diagrams is again used as a convenient and powerful tool to depict visually the results of all metrics so that strengths and weaknesses may be easily pinpointed. For any given chemical reaction, the following nine metrics are selected: atom economy (AE), reaction yield, excess reagent consumption (1/SF), auxiliary material consumption (MRP), reaction mass efficiency (RME, which is the inverse of the process mass intensity (PMI)), input SHI, RTHI, RPHI, and waste BI. To gauge overall “green” performance over these metrics a vector magnitude ratio is determined according to eq 6.

∑j f (input)j Ωj ∑j Ωj

⎡ |Tj − Tambient| ⎤ (RTHI)j = exp⎢ − ⎥ Tambient ⎦ ⎣

mass of waste compound j total mass of waste compounds produced in reaction

From an occupational hazard point of view, SHI(input) and SHI(waste) are important with respect to all of the input materials a chemical worker would be exposed to in carrying out a given chemical transformation from initial starting reagents and 177

dx.doi.org/10.1021/op300352w | Org. Process Res. Dev. 2013, 17, 175−192

Organic Process Research & Development

Concept Article

formally classified as explosives, XSP and ISP were set to zero. Hydrogen generation potential (HGP) accounts for hydrogen gas evolution from reaction of strong reducing agents such as metal hydrides and alkali metals with hydroxylic solvents. HGP is set to zero since none of the chemicals examined fit this classification. Corrosive gas potential (CGP) is parametrized by LC50 inhalation acute toxicity and corrosive liquid/solid potential (CLP) is parametrized by dermal LD50 acute toxicity (see SI for references). Flammability potential (FP) is estimated from flash point data.28 The oxygen balance29 potential (OBP) covers two kinds of reactions involving oxygen: (1) combustible chemicals are parametrized by the minimum amount of oxygen consumed in the combustion process (negative oxygen balance), and (2) oxidizing chemicals are parametrized by the maximum amount of oxygen available for a complete oxidation process (positive oxygen balance). The SI contains a file in Excel format containing an extensive database of source parameters relevant to the above potentials. Assumptions. The SI contains a listing of all assumptions used in the determination of all hazard potentials for each compound involved in the chemical reactions examined. When multiple values of a given parameter were found in the literature for a given compound, the one selected was the one that gave the highest hazard potential corresponding to worst-case scenarios. For water insoluble compounds having no data, a value of −1 × 10100 was used for log Kow and a value of 1 × 10−100 was used for WS (water solubility). For organic compounds having no available water solubility data, the Mackay equation depending on octanol−water partition coefficients and molecular weights was used.30 Table S2a in SI gives the full list of parameters necessary for hazard impact calculations for all reactions discussed in this work. Table S2b in SI gives a listing of missing parameters and Table S3 in SI gives a summary of all calculated hazard potentials for each chemical. Experimental procedures for reactions were sufficiently described so that all masses or volumes of reagents and other materials and all reaction conditions were specified. In the case when there were missing amounts of catalysts, a 0.1 mol % loading was assumed. As before, any missing data were left blank so as not to introduce any biases. The use of penalty points,31 as was mentioned in the Introduction in connection with previously defined hazard indices, was not adopted because they are necessarily arbitrary. This point is discussed further in the Discussion section in connection with the consequences of implementing a modified Edwards−Lawrence ISI analysis. Evaluations of all reactions and plans were qualified with appropriate commentary showing explicitly what was absent if missing data were encountered. The more missing data there are, the less reliable is the evaluation of the associated hazard indices and, hence, the overall determination of “greenness” of the particular reaction or plan performance.

RPHI indices are not included in the overall VMR since they pertain to temperature and pressure conditions for individual reactions only. ⎡ 6 ⎤1/2 VMR = (1/ 6 )⎢∑ (rj)2 ⎥ ⎢⎣ ⎥⎦ j=1

(7)

Potentials Used. The following hazard potentials were used in this study: CGP (corrosive gas), CLP (corrosive liquid/solid), FP (flammability), OBP (oxygen balance), HGP (hydrogen generation), XVP (explosive vapour), XSP (explosive strength), ISP (impact sensitivity), RPP (risk phrase), OELP (occupational exposure limit), MACP (maximum allowable concentration), and SDP (skin dose). Toluene was chosen as the reference compound for CLP, RPP, OELP, MACP, and SDP in order to maintain consistency with previously reported environmental impact potentials that also used this compound as a reference in their definitions. Dermal absorption potential (DAP) was not selected because few of the chemicals in this work had experimental data available for critical fluxes for dermal absorption. Skin dose potential (SDP) was estimated from water solubility and transdermal permeation coefficient data based on an exposure of a standard 360 cm2 area of skin corresponding to the entire surface area of both hands for 8 h (see SI). The latter parameter was in turn calculated using the Robinson QSAR model equation.20 In the case of OELP (U.S. NOISH usage21) and MACP (European usage22) the lowest figures were selected from both databases to reflect worst-case scenarios. In determining RPP, the Q-factor scaling used by Eissen’s EATOS protocol23 for the European R-phrase DSCL (dangerous substances classification and labelling) system ranging in value between 0 and 5 was multiplied by 10. This was done because in some instances fractional rather than whole numbers were used. Such an operation avoided diminishment of multiplicative products as would result if fractions were multiplied. If a compound had more than one risk associated with it, then its combined risk was taken as the product of the Q-factors rather than the sum to reflect a compounding of risk. This ensured that maximum risk of a given chemical was determined by the R-phrase system over several factors. Also, amplification as a consequence of multiplication has the advantage of separating overall risk between chemicals. For example, a compound with few high value Q-factors would have a comparable overall risk as another compound with many low value Q-factors if sums of Q-factors were compared. If, instead, respective products of Q-factors were compared, the two compounds would be easily distinguishable based on overall risk. A complete listing of Q-factors is given in the SI including five new risk phrases not found in the original list, namely, R69 = stored in pressurized gas cylinder (Q = 40), R70 = strong oxidizer (Q = 40), R71 = asphyxiant (Q = 40), R72 = causes thermal burns due to liquefied cryogenic gas or dry ice (Q = 30), and R73 = no perceived occupational health and safety hazard (Q = 0). Surprisingly, the following descriptors are not mentioned in the European DSCL system: choking agent, cryogenic liquid, lethal, nerve agent, poison, pyrophoric, radioactive, and vesicant. All of these phrases are, however, used in the Emergency Response Guidebook24 applicable to transportation of chemicals in Canada and the United States. XVP, XSP, and ISP cover hazards with respect to explosive risks. These are parametrized by lower explosion limit25 values for vapours, Trauzl lead block test,26 and impact sensitivity,27 respectively. Since none of the chemicals examined are shock sensitive or are



APPLICATION TO SINGLE-STEP CHEMICAL TRANSFORMATIONS Single-step reactions producing phenyl isocyanate,32−35 dimethyl carbonate,36−41 aniline,42,43 and phenol44 were re-examined so that the three new safety/hazard indices were calculated in addition to the previously determined benign indices and material efficiency parameters. Schemes 1−4 show the balanced chemical equations along with reaction conditions. Figures 1−4 show the resultant radial polygons and Tables S4−S7 in the SI summarize the corresponding numerical results of the key parameters along with a listing of missing data for each reaction. The benign VMRs were determined from the three BI indices: 178

dx.doi.org/10.1021/op300352w | Org. Process Res. Dev. 2013, 17, 175−192

Organic Process Research & Development

Concept Article

Scheme 1. Single-step synthesis routes to phenyl isocyanate

Scheme 2. Single-step synthesis routes to dimethyl carbonate

Scheme 4. Single-step synthesis routes to phenol

Scheme 3. Single-step synthesis routes to aniline

For phenyl isocyanate, the phosgene route was the most material efficient having the smallest PMI of 9.8, but also had the second lowest hazard VMR of 0.7962. This was attributed, not to the phosgene reagent, but to the reaction solvent phosphorous oxychloride. Ninety percent of the starting mass of phosgene is converted to product and unreacted phosgene represents only 1% of the total mass of waste materials in the reaction. Phosphorus oxychloride, on the other hand, represents 87% of the total mass of waste materials. This solvent alone contributes 91% of the total mass weighted hazard potential over all 10 hazards examined. The occupational exposure limit potential for phosphorus oxychloride accounts for 98.5% of the total occupational exposure limit potential and 88% of the total mass weighted hazard potential covering all 10 hazards. Within the limitations of available data, the safest route was the diphosgene route followed closely by the triphosgene route. In terms of environmental impact all four routes to phenyl isocyanate had similarly valued benign indices. The dimethyl carbonate route came last in terms of safety and material

BI(waste), BI(input), and BI(output). The hazard VMRs were determined from the five SHI indices: SHI(waste), SHI(input), SHI(output), RTHI, and RPHI. The overall VMRs were determined from the nine metrics according to eq 6. The main missing data were dermal LD50s, which impacted the determination of the CLP potential. The only reaction that did not have missing data was the reaction of methanol with methyl chloroformate to produce dimethyl carbonate. 179

dx.doi.org/10.1021/op300352w | Org. Process Res. Dev. 2013, 17, 175−192

Organic Process Research & Development

Concept Article

Figure 1. Radial polygons for phenyl isocyanate single-step plans showing synthesis performance according to various green metrics covering material efficiency, environmental impact, and Safety/Hazard Impact.

Figure 2. Radial polygons for dimethyl carbonate single-step plans showing synthesis performance according to various green metrics covering material efficiency, environmental impact, and Safety/Hazard Impact.

conditions. The hydrogenation route loses out in terms of safety because of the high temperature and pressure conditions of the reaction. It also has the lowest benign VMR. The reaction of chlorobenzene with ammonium hydroxide has the least environmental impact. Phenol is more efficiently made by decarboxylative oxidation of benzoic acid, but is more safely made with the least environmental impact by direct oxidation of benzene. The cumene oxidation route has the lowest benign

efficiency because of the lowest conversion of starting material to product. Both of these factors combined to make this route have the lowest overall VMR value of 0.5858. Dimethyl carbonate is more efficiently made by reacting methanol with carbon dioxide, but is more safely made with less environmental impact by reacting methanol with methyl chloroformate. Aniline is more efficiently made by hydrogenating nitrobenzene, but is more safely made by reducing nitrobenzene with iron under acidic 180

dx.doi.org/10.1021/op300352w | Org. Process Res. Dev. 2013, 17, 175−192

Organic Process Research & Development

Concept Article

Figure 3. Radial polygons for aniline single-step plans showing synthesis performance according to various green metrics covering material efficiency, environmental impact, and Safety/Hazard Impact.

Figure 4. Radial polygons for phenol single-step plans showing synthesis performance according to various green metrics covering material efficiency, environmental impact, and Safety/Hazard Impact.

parameters for each plan and Figures 6, 8, and 10 show the corresponding radial polygons with associated global VMR values according to eq 7. Comparison of the four routes to phenyl isocyanate beginning from phosgene shows that the three-step diphosgene route is the overall “greenest” with the highest global VMR value of 0.7569 followed closely by the triphosgene route (A) route (A) with a corresponding value of 0.7473 . The triphosgene (A) route has both the least hazards and environmental impact. All four routes produce about the same degree of environmental impact. The one-step phosgene route is clearly the most material efficient route but is the least “green” with the lowest global VMR of 0.6791. When we examine the two routes to phenyl isocyanate from methanol and carbon dioxide, we find that the triphosgene route (B) is all round better than the DMC route (B). This is consistent with results discussed above for the “greener” way of producing dimethyl carbonate. From Figure 5 it can be seen that the phosgene route produces the largest overall hazard impact, primarily because of phosphorus oxychloride solvent usage followed by the DMC (B) route due to the low conversion of

VMR. All of these examples illustrate the compromises that need to be made between material efficiency, environmental impact, and safety/hazard management. It is very difficult to find syntheses that satisfy all criteria optimally.



APPLICATION TO SYNTHESIS PLANS When we examine synthesis plans to the same target molecules but starting from common starting materials we expect fairer comparisons of plan performances since in each plan the pairs of starting reagents and end products are the same. For example, for phenyl isocyanate all four routes can be traced back to the common reagent phosgene. The three aniline routes can be traced back to benzene and four out of five routes to phenol can also be traced back to benzene. Schemes 5 to 7 show all synthesis routes to the three target molecules. Figures 5, 7, and 9 show input and waste safety/hazard profiles for all routes against mass weighted potential. The mass weighted potential corresponds to the numerator of SHZI given in eq 1. Also shown for each route are the fractional contributions of each hazard potential. Tables 1 to 3 summarize the numerical values of key global 181

dx.doi.org/10.1021/op300352w | Org. Process Res. Dev. 2013, 17, 175−192

Organic Process Research & Development

Concept Article

Scheme 5. Multistep synthesis routes to phenyl isocyanate

the waste hazard profiles for the diphosgene, triphosgene (A), and triphosgene (B) routes arising from hydrochloric acid byproduct production. Phenyl Isocyanate. For aniline syntheses, the overall “greenest” plan is the Faith G2 route mainly due to its superior material efficiency performance with a PMI of 2.80. The most hazardous route is the Faith G1 route. By contrast, the least hazardous route is the Faith G3 route that is also the least material efficient. All three routes have similar environmental impacts. The dominant hazard potential for all three routes is the occupational exposure limit. For input materials the next most dominant hazard potential is corrosiveness potential as a liquid/ solid (CLP). For waste materials the skin dose potential is the second largest contributor to hazard potentials after OELP for the Faith G1 and G3 plans arising from hydrochloric acid catalyst and byproduct, respectively. Aniline. For the phenol syntheses, the overall “greenest” route is Faith G5. It is both the most material efficient and the least hazardous. This was the only observation of positive synergy in reaction optimization. The Faith G4 route is the next most material efficient but is the most hazardous and its waste produces the most environmental impact. Both of these factors weigh down its overall performance making it overall the least green of the five routes. The Faith G3 plan is the second best overall in terms of “greenness” but its liabilities are the lower RME and lower SHI(input) values. The lower RME arises because of a large excess of reagents used; whereas, the lower SHI(input) is due to the high skin dose potential from sulphuric acid and the high occupational exposure limit potential from benzene. The input and waste hazard profiles for the phenol plans show the greatest contrast compared to the other two target molecules examined as illustrated in Figure 9. The Faith G3 plan has the highest hazard impacts with respect to input materials; whereas, the Faith G2 and G4 plans have the highest hazard impact with respect to waste materials produced. Close examination of the input hazard profile shows that the skin dose

Scheme 6. Multistep synthesis routes to aniline

Scheme 7. Multistep synthesis routes to phenol

starting material to product. Also, the dominant contributing potential to both input and waste hazard profiles across the board is occupational exposure limit. The skin dose potential figures in 182

dx.doi.org/10.1021/op300352w | Org. Process Res. Dev. 2013, 17, 175−192

Organic Process Research & Development

Concept Article

Figure 5. Input safety/hazard potential profiles for phenyl isocyanate syntheses with respect to mass weighted potential (A1) and percent mass weighted potential; waste safety/hazard potential profiles for phenyl isocyanate syntheses with respect to mass weighted potential (B1) and percent mass weighted potential (B2).

Table 1. Summary of metrics for multistep phenyl isocyanate syntheses plan

number of steps

% overall AE

% overall yield

overall PMI

overall BI (waste)

overall SHI (waste)

overall SHI (input)

4 3 3 1 3 2

39.3 43.9 39.5 52.1 40.2 59.2

56.9 64.1 72.1 89.9 47.6 53.6

20.68 18.90 13.30 9.83 108.15 48.89

0.995 0.974 0.991 0.953 0.990 0.960

0.983 0.975 0.960 0.660 0.975 0.946

0.956 0.942 0.923 0.577 0.972 0.954

a

triphosgene (A) triphosgene (B)b diphosgenec phosgened DMC (A)e DMC (B)f a

Triphosgene route (A): missing OEL for NaCl (step 2); missing LD50 (dermal) for triphosgene (step 4). bTriphosgene route (B): missing Rphrases, SD, and LD50 (dermal) for zirconium oxide (step 1); missing LD50 (dermal) for triphosgene (step 3). cDiphosgene route: missing LD50 (dermal) for 1,4-dioxane, diphosgene, and aniline hydrochloride (step 3); missing LD50 (dermal) for phosgene (step 1). dPhosgene route: missing LD50 (dermal) for phosgene, phosphorous oxychloride, and aniline hydrochloride. eDMC route (A): missing LD50 (dermal) for phosgene (step 1); missing OEL for sodium chloride, missing LD50 (dermal) for chlorobenzene (step 2). fDMC route (B): missing LD50 (dermal), R-phrase, and SD for zirconium oxide (step 1).

Table 2. Summary of metrics for multistep aniline syntheses plan

number of steps

% overall AE

% overall yield

overall PMI

overall BI (waste)

overall SHI (waste)

overall SHI (input)

Faith G1a Faith G2b Faith G3c

2 2 2

32.7 63.3 42.5

92.1 94.9 70.6

5.66 2.80 6.18

0.951 0.948 0.939

0.945 0.959 0.961

0.886 0.923 0.937

a Faith G1: missing LD50 (dermal) for nitric acid and sulphuric acid (step 1); missing LD50 (dermal), SD, and OEL for Fe3O4 (step 2) bFaith G2: missing LD50 (dermal) for nitric acid, sulphuric acid (step 1); missing LC50 and OEL for dihydrogen; missing LD50 (dermal) for copper carbonate (step 2) cFaith G3: missing LD50 (dermal) for copper(I) oxide and chlorobenzene (step 2).

Table 3. Summary of metrics for multistep phenol syntheses plan

number of steps

% overall AE

% overall yield

overall PMI

overall BI (waste)

overall SHI (waste)

overall SHI (input)

Faith G1a Faith G2b Faith G3c Faith G4d Faith G5e

1 2 1 2 2

26.1 35.4 63.3 61.8 60.3

83.0 69.9 75.4 61.1 73.4

7.75 5.26 5.15 2.62 2.34

0.876 0.870 0.977 0.740 0.972

0.954 0.914 0.981 0.833 0.995

0.849 0.835 0.881 0.685 0.997

a

Faith G1: missing LD50 (dermal) for sulphuric acid and sulfur dioxide; missing OEL for sodium sulfite (step 1). bFaith G2: missing LD50 (dermal) for chlorobenzene and sodium hydroxide; missing OEL for sodium chloride (step 2). cFaith G3: missing OEL for dioxygen. dFaith G4: missing OEL for dioxygen (step 2). eFaith G5: missing LD50 (dermal), SD, and R-phrases for cobalt naphthenate (step 1); missing OEL for dioxygen (step 2).

potential dominates for the Faith G3 route due to hydrochloric acid catalyst; whereas, the occupational exposure limit potential

dominates for the Faith G1, G2, G4, and G5 routes, particularly for the Faith G5 route which uses cobalt naphthenate as catalyst 183

dx.doi.org/10.1021/op300352w | Org. Process Res. Dev. 2013, 17, 175−192

Organic Process Research & Development

Concept Article

Figure 6. Summary radial diagrams showing overall green metrics performances for six synthesis plans for phenyl isocyanate (see Scheme 5).

Figure 7. Input safety/hazard potential profiles for aniline syntheses with respect to mass weighted potential (A1) and percent mass weighted potential; waste safety/hazard potential profiles for aniline syntheses with respect to mass weighted potential (B1) and percent mass weighted potential (B2).

Figure 8. Summary radial diagrams showing overall green metrics performances for three synthesis plans for aniline (see Scheme 6).

potential is the second most dominant potential for the Faith G2 and G3 routes arising from hydrochloric acid byproduct and catalyst, respectively. The corrosion potential as a liquid/vapour is important in the Faith G3 and G4 routes due to unreacted benzene.

in the oxidation of toluene to benzoic acid, and benzoic acid in the second decarboxylative oxidation step. The corrosion potential as a liquid/vapour is the next most dominant hazard for the Faith G4 route arising from benzene in the synthesis of cumene. The waste hazard profile shows that the skin dose 184

dx.doi.org/10.1021/op300352w | Org. Process Res. Dev. 2013, 17, 175−192

Organic Process Research & Development

Concept Article

Figure 9. Input safety/hazard potential profiles for phenol syntheses with respect to mass weighted potential (A1) and percent mass weighted potential; waste safety/hazard potential profiles for phenol syntheses with respect to mass weighted potential (B1) and percent mass weighted potential (B2).

Figure 10. Summary radial diagrams showing overall green metrics performances for five synthesis plans for phenol (see Scheme 7).



DISCUSSION The main hazard concerns from an occupational point of view due to exposure to chemicals in the workplace are inhalation toxicity (respiratory and central nervous system damage), skin absorption (irritation to chemical burns of varying degree), explosion potential, and flammability of materials. In all cases the worst-case scenario for adverse effects is death. The results from the few industrially important reactions presented in this work show that the bulk of the hazard risk arises from skin absorption and occupational exposure limit potentials via inhalation. Therefore, inhalation LC50 and dermal LD50 toxicity parameters are more relevant to the workplace environment. Oral LD50s, on the other hand, are relevant to environmental impact since they are linked with organisms in the environment that will be ingesting any released chemical in the air, water, soil, or sediment compartments. The connection with human health is when humans consume organisms that themselves have ingested those chemicals, hence, the problem of upward

propagation of contaminants in the food chain. It is not surprising, then, that ingestion and inhalation toxicity potentials were found to dominate environmental impact profiles.11 The same arguments made before11 on environmental impact potentials with respect to effects of missing data, faulty data, reliance on computational modeling at the expense of avoiding experimental work, the need to significantly augment the database size of industrial chemicals with respect to quantification of hazards, and the overall impact of these problems on the reliability of conclusions drawn about reaction and plan rankings, all apply to this study on hazard potentials. Inconsistencies and sloppiness in MSDS compilations have been well described45 as well as problems associated with computational toxicology.46−56 Following we briefly discuss points about key parameters important for determining hazard potentials and the overall Safety/Hazard Index. Comparison of the U.S.-based OEL and European-based MAC databases, comprising 681 and 426 chemicals, respectively, (see Excel database in SI) reveals that of the 210 common 185

dx.doi.org/10.1021/op300352w | Org. Process Res. Dev. 2013, 17, 175−192

Organic Process Research & Development

Concept Article

points out the following pitfalls of using dermal LD50 data: they refer to acute toxicity with death as the only outcome measured; and they ignore nonlethal, chronic toxicity, accumulation, or repeated dosing effects. Use of dermal LD50s to determine skin notations has also met with controversy.97−99 Nevertheless, despite these shortcomings dermal LD50s and inhalation LC50s are the next most abundant toxicity parameters available after oral LD50s and are currently the best surrogate parameters that can be used to quantify CGP and CLP. Dermal absorption critical fluxes72,99 could be used to determine a dermal absorption potential (DAP), but the limited data available precluded their use for the reactions studied in this work. The hydrogen generation potential is relevant to strong reducing agents, particularly alkali metals and metal hydrides which evolve flammable hydrogen gas upon contact with hydroxylic solvents. This potential will figure prominently in such well-known named organic reactions as summarized in Table 4. None of these reagents were involved in any of the reactions examined in this study.

chemicals found in both databases, 57% have the same exposure limit ratings, 31% have the U.S. threshold limits higher than those of the European, and 12% have the European threshold limits higher than those of the U.S. Since higher threshold limits imply a higher tolerance for exposure to chemicals, it appears that on the whole the European figures are more stringent than the American ones. The skin dose57 potential used in the present study depends on log Kow and water solubility input data. The key parameter to be determined is the transdermal permeation coefficient, Kp, given in cm per hour, which is estimated using QSAR modeling equations.58−68 The modified Robinson model that was used here appears to have been well validated in the literature.20,57 However, there are others that exist with varying degrees of performance depending on the data set of chemicals used. These include the Berner-Cooper,69 Brown-Rossi,70 Cleek-Bunge,71 Fiserova,72 Frasch,73 McKone-Howd,74 Potts-Guy,75 and modified Potts-Guy76 models. Much of the impetus for these studies was the understanding of the mechanisms of skin penetration of topically applied pharmaceuticals with respect to dosing and pharmacokinetics.77−79 The mathematics and performances of these models have been extensively reviewed;58−68 however, a recent paper on experimentally determined Kp values, albeit on a limited number of compounds, indicated that agreement between predictions made using the existing mathematical models with experimental results was disappointingly very poor.80 The doubt cast by that study currently challenges both theoreticians and experimentalists to re-examine the difficult task of reliably measuring and parametrizing absorption of chemicals through the skin, whether they are therapeutic drugs or hazardous contaminants. The same authors questioned the reliability of the Kp parameter as a key measure of skin permeation.81 Arguments have been put forward that in vitro studies of skin permeation are sufficient to replace in vivo ones.82−84 NOISH has an online skin permeation calculator85 using the modified Robinson, PottsGuy, and Frasch models for about a hundred industrial compounds found in the Flynn database.59 It also provides qualitative and semiquantitative skin notation guidelines86−95 for some of the following commodity chemicals listed in the NOISH database: acrylamide, acrylonitrile, ammonia, antimony, asbestos, arsenic, benzene, beryllium, bisphenol A, 2-butoxyethanol, cadmium, carbon disulfide, carbon monoxide, chlorine, chloroform, chromium, cobalt, 1,4-dioxane, dimethylformamide, epichlorohydrin, ethylene glycol, ethylene oxide, fibrous glass, formaldehyde, glutaraldehyde, hexavalent chromium, hydrogen chloride, hydrogen cyanide, hydrogen peroxide, hydrogen sulfide, hydroquinone, hydrazine, isocyanates, lead, manganese, mercury, metal working fluids, methylene chloride, methyl alcohol, methyl ethyl ketone, nickel, nitric acid, nitroglycerin, nitrous oxide, organic solvents, osmium tetroxide, ozone, phenol, phosgene, phosphine, silica, sodium hydroxide, styrene, sulfur dioxide, tetrachloroethylene, toluene, trichloroethylene, vermiculite, and xylene. As noted by the World Health Organization EHC235 Report, corrosive substances are not given a skin notation.96 The reason for this is that skin notation refers specifically to the propensity of a chemical to cross the lipid, protein, and aqueous layers of the skin without damaging the skin tissue. Corrosive substances, on the other hand, cause severe chemical burns by destroying skin tissue, therefore a passive physical mechanism of penetration is not operative. The skin related hazards associated with these substances are best parametrized by the R-phrase Q-factor method and by dermal LD50 data if they are available. The EHC235 WHO report also

Table 4. Summary of named organic reactions that involve reducing agents that can generate hydrogen gas reducing agent

named reduction reaction

Li Li NaBH3CN NaBH4 Na Zn BH3 NaBH4 LiAlH4 n Bu3SnH Ni NaBH3CN

Benkeser Birch Borch borohydride reduction Bouveault−Blanc Clemmensen Corey−Bakshi−Shibata Gribble lithium aluminum hydride reduction radical dehalogenation thioketal desulfurization tosylhydrazone reduction

The oxygen balance potential with respect to oxidation reactions will figure prominently in such well-known named organic reactions as summarized in Table 5. Higher positive oxygen balances correlate with more positive reduction potentials and higher oxygen weight percent content suggesting that stronger oxidizing agents generally have larger OB values (see SI). In this work, compounds having positive OB values included: copper(I) oxide, iron oxide (magnetite), lead(II) oxide, nitric acid, sodium hydroxide, sulphuric acid, and zirconium dioxide. The Q-factor scale applied to R-phrases is, in theme, similar to that of the scaling used in the NFPA diamond system.19 Both scales are arbitrary but are reasonable in terms of relative ranking of chemical risk and are widely adopted. The Globally Harmonized System (GHS)100,101 advanced by the United Nations Economic Commission for Europe (UNECE) is a recent attempt to amalgamate the existing symbolisms used by WHMIS (Canada), NFPA (United States), and ECHA (European Union) into a single set of common icons. This effort is aligned with the European Registration, Evaluation, Authorization, and Restriction of Chemicals (REACH) initiative adopted in 2006.102 Safety, or S-phrases, also exist, but no parallel quantitative Q-factor scaling system is used since they are not specific to risk but to general laboratory husbandry and practice (see SI Excel file for compilation). For practicing synthetic organic chemists the most common compounds encountered in the laboratory that are potentially 186

dx.doi.org/10.1021/op300352w | Org. Process Res. Dev. 2013, 17, 175−192

Organic Process Research & Development

Concept Article

classes of organic compounds, otherwise, the XSP and ISP potentials described in this work cannot be utilized if such compounds are encountered. With such missing data the reliability of the resulting SHI would be severely compromised. This work has shown that rankings of plan performance differ if single-step versus multistep plans is compared. Fairer rankings arise when plans to a given product are compared from common starting materials. Table 6 summarizes the rankings for the best

Table 5. Summary of named organic reactions involving oxidizing agents that result in positive oxygen balances

a

oxidizing agents

named oxidation reaction

CrCl2O2 CrO3 CrO3/2-pyridine Cu(OH)2 H2O2 H2O2/CuSO4 H2O2/FeCl3 HgO HIO4 K2S2O8 K2S2O8 K3[Fe(CN)6]/K2OsO4(OH)2 KHSO5 KMnO4 m-CPBAa m-CPBA m-CPBA m-CPBA NaOCl NaOCl NaOCl O3/H2O2 O3/Me2S O3/PPh3 OsO4 SeO2 t BuOOH

Etard Jones Sarett Fehling Baeyer−Villiger Hooker Fenton Bamford−Stevens Malaprade Boyland−Sims Elbs Sharpless−Jacobsen dihydroxylation Shi asymmetric epoxidation permanganate oxidation of olefins Baeyer−Villiger Fleming Prilezhaev Rubottom Forster Graham Jacobsen epoxidation Harries ozonolysis Harries ozonolysis Harries ozonolysis Lemieux−Johnson Riley Sharpless epoxidation

Table 6. Summary of single-step and multistep plan rankings for best and worst aniline, phenol, and phenyl isocyanate synthesis routes.a best aniline overall greennessb safety/hazardc benign env. impactd material efficiencye phenol overall greenness safety/hazard benign env. impact material efficiency phenyl isocyanate overall greenness safety/hazard benign env. impact material efficiency

m-CPBA = m-chloroperbenzoic acid.

explosive belong to the diazo and azide family. The utility of such reagents in organic synthesis has been well described,103−111 especially in the context of using safe derivatives such as 2-azido3-ethyl-benzothiazol-3-ium tetrafluoroborate,112,113 polymer resin-bound benzenesulfonyl azide,114 N,N-dimethylazidochloromethyleniminium chloride,115,116 4-dodecylbenzenesulfonyl azide,117 trimethylsilyldiazomethane,118 in situ-generated 2-azido1-ethyl-pyridinium tetrafluoroborate,119 azidotris(diethylamino)phosphonium bromide,120 polystyrene-supported benzenesulfonyl azide,121 tosylhydrazone salts,122 replacing dichloromethane for toluene as reaction solvent,123 2-azido-1,3-dimethylimidazolinium salts,124 and imidazole-1-sulfonyl azide hydrochloride.125 It is also well-known that all procedures utilizing diazo and azide reagents carry precautionary warnings about shock sensitivity and explosion potential if heated above room temperature.126 However, there is little experimental work reported on quantitative measures of explosive potential for azides and diazo compounds such as the lead block Trauzl test and the impact sensitivity test. Typically such tests are conducted on compounds that are specifically designed to be explosives for military use.26,27 The only reports that could be found pertained to organic azides,127 ethyl diazoacetate128−130 and, most recently, diazomethane.131 Surprisingly, the data reported on diazomethane came 67 years after Eistert’s and Arndt’s seminal papers on its safe preparation132−134 and 108 years after its discovery by von Pechmann.135,136 Moreover, the thermochemical work done on ethyl diazoacetate suggested that the warnings associated with use of this compound were not as severe as first thought. Clearly, a significant amount of work needs to be done to augment the LBT and IS data available in order to include these important

worst

single-step

multistep

single-step

Faith G1 Faith G1 Faith G3 Faith G2 best

Faith G2 Faith G3 Faith G1 Faith G2

Faith G3 Faith G3 Faith G3 Faith G2 Faith G2 Faith G3 Faith G1 Faith G3 worst

multistep

single-step

multistep

single-step

multistep

Faith G5 Faith G3 Faith G3 Faith G5 best

Faith G5 Faith G5 Faith G3 Faith G5

Faith G2 Faith G4 Faith G4 Faith G4 Faith G4 Faith G4 Faith G1 Faith G1 worst

single-step

multistep

single-step

multistep

diphosgene diphosgene diphosgene

diphosgene DMC (A) triphosgene (A) phosgene

DMC DMC phosgene

phosgene phosgene phosgene

DMC

DMC (A)

phosgene

a

Bolded entries indicate consistent rankings between single-step and multistep plans for each criterion. bBased on VMR calculated using eq 7. cBased on overall SHI(input). dBased on overall BI(waste). eBased on PMI.

and worst aniline, phenol, and phenyl isocyanate plans examined. From these results it is observed that there is more agreement among plan rankings for phenol. No plan scored overall high or overall low in all four categories of material efficiency, benign environmental impact, safety/hazard, and overall “greenness”. However, there were pairwise best rankings for safety and benign environmental impact for the following: phenol − single-step Faith G3 route and phenyl isocyanate − single-step diphosgene route. Conversely, there were the following pairwise worst rankings for the same criteria: phenol − single-step Faith G4 route, phenol − multistep Faith G4 route, and phenyl isocyanate − multistep phosgene route. The only plan scoring high in material efficiency and safety was the phenol − multistep, Faith G5 route, and the only plan scoring low in both categories is the phenyl isocyanate − single-step dimethyl carbonate route. Plans scoring high in terms of overall “greenness” and safety/hazard are the aniline − single-step Faith G1 route, phenol − multistep Faith G5 route, and phenyl isocyanate − single-step diphosgene route. Plans scoring lowest in these criteria are the aniline − single-step Faith G3 route, phenol − multistep Faith G4 route, phenyl isocyanate − single-step DMC route, and phenyl isocyanate − multistep phosgene route. The single-step and multistep Faith G5 plans to phenol score highest in overall “greenness” and material efficiency as does the multistep Faith G2 plan to aniline. 187

dx.doi.org/10.1021/op300352w | Org. Process Res. Dev. 2013, 17, 175−192

Organic Process Research & Development

Concept Article

Figure 11. Histograms showing the ranking of various plans to phenyl isocyanate, aniline, and phenol with respect to safety of waste produced: (Type A) adjusted ISI* using E-factor contribution factors; (Type B) adjusted ISI** using scaling factor; (Type C) SHI(waste). Red bars indicate least safe plans, and green bars indicate safest plans.

following changes were made: (1) a corrosivity scale based on the skin dose was added as a contributor to the chemical score; (2) all chemicals involved in a reaction, including workup and purification, were considered for determining its chemical score with nothing omitted; and (3) the inventory score was replaced with either a scaling factor, defined as the inverse of the product of reaction yields beginning with the last step and working backwards to all chemicals used in a given reaction using a value of unity as the basis scale for the final product in a sequence of steps, or an E-factor contribution factor, defined as the mass ratio of waste due to a given chemical versus a basis mass of target final product collected (1 ton). The scaling factor or E-factor contribution factor was then multiplied by the sum of the flammability, explosiveness, toxicity, and corrosivity scores to obtain the chemical score for a given chemical used in a reaction. This idea is analogous to the Sheldon environmental quotient concept137 where raw E-factors are multiplied by environmental friendliness Q-factors. The calculation is repeated for all chemicals used in a reaction so that an overall chemical score is obtained. This modified summed chemical score was then added to the process score defined as before16 to obtain finally an adjusted ISI for each reaction step in a sequence. The process is again repeated for all reactions in a synthesis plan so that an overall adjusted ISI is obtained for the entire plan. An ISI based on E-factor contribution factors is designated as ISI* and one based on scaling factors is designated as ISI**. The former pertains to a safety assessment of waste materials produced in a synthesis plan and the latter to a safety assessment of input materials. Detailed tables of relevant data are given in the SI (see Tables S8−S35). The key results are shown by the histograms in Figures 11 and 12, which describe the rankings of all plans described in this work according to the safety/hazard scores for waste produced and input materials used, respectively. It is clear that all of the adjusted ISI rankings, except for the best-performing Faith G3 aniline plan and the worst-performing Faith G1 aniline plan (see Figure 12), are not in agreement with the rankings obtained using the SHI presented in this work

There were no plans satisfying both the material efficiency and benign environmental impact criteria at the highest ranking; however, the multistep Faith G3 plan to aniline satisfied these criteria at the lowest ranking. The single-step diphosgene plan to phenyl isocyanate scored highest with respect to overall greenness and benign environmental impact criteria; whereas, the multistep plans to aniline (Faith G3), phenol (Faith G4), and phenyl isocyanate (phosgene route) fulfilled these criteria at the lowest ranking. In terms of overall optimization, if the same plan consistently scores highest in all four criteria then its claim as being truly overall “green” is validated and credible. However, if different plans to a given target molecule score highest in each criterion, then significantly more work is needed to achieve the optimization goal until all criteria are maximized in an orchestrated way toward the same plan. With this philosophy in mind, the most credible claims of best overall “greenness” performance are the phenol (multistep, Faith G5 plan) and the phenyl isocyanate (single-step diphosgene plan) because they score highest in three out of four criteria. However, the strongest claims of worst overall “greenness” performance are the multistep Faith G4 plan for phenol, the multistep Faith G3 plan to aniline, the multistep phosgene route to phenyl isocyanate, and the single-step DMC route to phenyl isocyanate, again because they score lowest in three out of four criteria. It is clear from these findings that reaction optimization with the goal of scoring high in all four categories is a very tough challenge. It is more common to find a plan scoring high or low in two out of the four criteria as discussed above. It is anticipated that the missing dermal LD50 data as indicated in Table S2b in SI will not change the current rankings adversely for the phosgene-based plans because inclusion of those data would only lower further the magnitudes of overall Safety/Hazard Indices which are already at the lowest ranking. Changes in the rankings of the aniline and phenol plans are less certain to anticipate. Finally, for comparison, all of the plans examined in this work were subjected to a modified Edwards−Lawrence ISI analysis addressing the problems pointed out in the Introduction. The 188

dx.doi.org/10.1021/op300352w | Org. Process Res. Dev. 2013, 17, 175−192

Organic Process Research & Development

Concept Article

Figure 12. Histograms showing the ranking of various plans to phenyl isocyanate, aniline, and phenol with respect to safety of input materials used: (Type A) adjusted ISI** using scaling factor; (Type B) SHI(input). Red bars indicate least safe plans, and green bars indicate safest plans.



CONCLUSIONS This work has introduced a new Safety/Hazard Index, covering 10 hazard potentials, defined in the same way as a benign index for environmental impact. In principle, the definition of SHI is able to support an unlimited number of potentials beyond the ones selected, thus making it more versatile than all previously defined safety indices appearing in the literature. The only limitation is the availability of data for each contributing parameter, and the only assumptions that need to be made are the choices of reference compounds to define the potentials. In this regard there are the following additional sources of hazard that could conceivably be added to the list of potentials described in this work: (a) endocrine disruption with respect to occupational exposure to a given chemical; (b) heat of reaction with respect to exothermic and endothermic reactions; (c) rate of reaction with respect to fast and slow reactions; (d) new phase generation with respect to production of gaseous byproducts that can cause overpressurization of reaction vessels, or solids that can precipitate out of solution potentially resulting in mechanical mixing problems; (e) viscosity changes as a given reaction proceeds, either thixotropy (fluid thinning) or rheopexy (fluid thickening) that can cause mixing heterogeneity problems; and (f) unwanted competing side reactions with their own set of reactants, products, stoichiometries, reaction conditions, reaction thermokinetics, environmental impacts, and hazard impact parameters. The utility of the method has in principle been demonstrated for the hazard assessment of industrial plans to aniline, phenol, and

despite attempts to rescue the originally defined ISI method as described above. The most glaring finding is that the phosgenebased route to phenyl isocyanate is ranked simultaneously as the overall “safest” by the ISI method and “most hazardous” by the present SHI method. The two occurrences of agreement between the two indexes observed here for the aniline plans may be interpreted as fortuitous. The underlying kernel problem is that the ISI is fundamentally based on arbitrarily chosen penalty scoring scales for each contributing factor. The SHI, on the other hand, uses ratios of experimentally determined parameters referenced to decidedly well-chosen, but arbitrary, reference compounds as defined by the potential given in eq 2. Though assessments based on penalty points using positive integers are conceptually easy to implement and are attractive for quick “back-of-the-envelope” calculations, they can lead to counterintuitive or completely erroneous conclusions. They may be useful in picking out the overtly good and poor performing plans for a common target molecule but not in ranking the ones that fall in between these extremes. The upshot is that methodologies based on penalty points are completely decoupled from the actual values of parameters on which they are based and are thus unable to capture the true differences between synthesis plan performances. Thus, a penalty score of 5, for example, assigned to two different categories are not equivalent although they are treated as numerically identical in computing an overall penalty score. This is a caveat that should be kept in mind and serves as a cautionary note when synthesis plans are ranked by this approach. 189

dx.doi.org/10.1021/op300352w | Org. Process Res. Dev. 2013, 17, 175−192

Organic Process Research & Development

Concept Article

phenyl isocyanate. The overall “greenness” performances of these plans evaluated according to material efficiency, environmental impact, and hazard resulted in the most reliable rankings found for these procedures, given the constraint of available data for all potentials used. Phosgene routes were found to be the most hazardous, not due to phosgene per se, but because of the noxious phosphorous oxychloride solvent used in the reaction procedures. This result offsets their high material efficiencies making them overall the least “green”. The Edwards−Lawrence inherent safety index (ISI) was found to be wholly inadequate in ranking the safety of plans examined here primarily because of the arbitrary nature of the penalty point scoring system. However, further work in comparing both ISI and SHI methodologies on synthesis plans for other important industrial commodity chemicals are currently being investigated. Past work on material efficiency metrics covering five parameters (AE, reaction yield, excess reagent consumption, auxiliary material consumption, and global reaction mass efficiency (inverse of PMI)), the benign index covering 13 environmental impact potentials, and the present Safety/Hazard Index covering 12 hazard potentials form a complete framework for assessing the global “greenness” of any chemical reaction or synthesis plan. Radial polygon diagrams have been used throughout as a powerful visual tool for chemists to quickly assess the merits and weaknesses of their plans, and hence to direct optimization efforts in a purposeful way. Despite these advances, a precise quantification of uncertainty in the overall “greenness” ranking remains elusive, particularly when dealing with missing data. A reviewer has pointed out the existence of the following semiquantitative hazard assessment tools that complement the present method: ANSI (American National Standards Institute) 355 Greener Chemicals and Processes Information Standard,138 Go Green Chemical Compliance System,139 and Green Screen for Safer Chemicals.140



CLP = corrosiveness potential as a liquid/solid DAP = dermal absorption potential DMC = dimethyl carbonate DSCL = dangerous substances classification and labeling (Europe) EATOS = Environmental Assessment Tool for Organic Synthesis EI = environmental impact ECHA = European Chemicals Agency FEI = Fire and Explosion Index (Dow) FP = flammability potential FLP = flash point (closed cup) GHS = globally harmonized system HFC = hydrofluorocarbons HGP = hydrogen generation potential HWI = hazard waste index (Gupta−Babu) IS = impact sensitivity ISI = inherent safety index ISP = impact sensitivity potential LBT = lead block test (Trauzl) LC50 (inhalation) = lethal concentration to kill 50% of population via inhalation route LD50 (dermal) = lethal dose to kill 50% of population via dermal route LEL = lower explosive limit MAC = maximum allowable concentration (Europe) MACP = maximum allowable concentration potential MRP = material recovery parameter MSDS = material safety data sheets MW = molecular weight (g/mol) NFPA = National Fire Protection Association (U.S.) NOISH = National Institute for Occupational Safety and Health (U.S.) OB = oxygen balance OBP = oxygen balance potential OECD = Organization for Economic Cooperation and Development OEL = occupational exposure limit OELP = occupational exposure limit potential OHG = Occupational Health Guidelines ORV = overall rating value OSHA = Occupational Safety and Health Administration (U.S.) PCB = polychlorinated biphenyl PMI = process mass intensity ppm = parts per million QSAR = quantitative structure−activity relationship RDX = research department explosive (refers to cyclotrimethylenetrinitramine or cyclonite) REACH = Registration Evaluation Authorization and Restriction of Chemicals (Europe) RME = reaction mass efficiency RPHI = reaction pressure hazard index RPP = risk phrase potential RTECS = Registry of Toxic Effects of Chemical Substances RTHI = reaction temperature hazard index SDP = skin dose potential SF = stoichiometric factor SHI = Safety/Hazard Index SHZI = Safety/Hazard Impact TLV = threshold limit value TNT = 2,4,6-trinitrotoluene TWA = time weighted average

ASSOCIATED CONTENT

S Supporting Information *

Database in Excel format (op300352w_si_001.xls) containing source parameters needed for all potentials described in this work; HAZARDS algorithm in Excel format (op300352w_si_002.xls); definitions and equations for all hazard potentials including example calculations; Tables S1, S2a, S2b, S3, S4, S5, S6, and S7; list of assumptions made in calculations; summary of equations for modelling skin permeation; Tables S8 to S35 for the Edwards−Lawrence adjusted ISI analysis. This material is available free of charge via the Internet at http://pubs.acs.org.



AUTHOR INFORMATION

Corresponding Author

E-mail: [email protected] Notes

The authors declare no competing financial interest.

■ ■

ACKNOWLEDGMENTS Dr. Susan van Arnum is thanked for information regarding the National Fire Protection Association (NFPA) 704 diamond system. ABBREVIATIONS USED AE = atom economy BI = benign index CFC = chlorofluorocarbons CGP = corrosiveness potential as a gas 190

dx.doi.org/10.1021/op300352w | Org. Process Res. Dev. 2013, 17, 175−192

Organic Process Research & Development

Concept Article

and Ministry of Communications and Transportation - Mexico). http:// www.tc.gc.ca/media/documents/canutec-eng/ERG2012.pdf, 2012. (accessed October 2012) (25) (a) Lange, N. A. Handbook of Chemistry, 10th ed.; McGraw-Hill Book Co., Inc.: New York, 1961, pp 32−49; 824. (b) Haynes, W. M., Ed. CRC Handbook of Chemistry & Physics, 92nd ed.; CRC Press: Boca Raton, FL, 2011−2012; pp 16−13 and 16−28. (26) Köhler, J.; Meyer, R. Explosives, 4th ed.; VCH: Weinheim, 1993; pp 213−215. (27) Köhler, J.; Meyer, R. Explosives, 4th ed.; VCH: Weinheim, 1993; pp 197−201. (28) (a) Lange, N. A. Handbook of Chemistry, 10th ed.; McGraw-Hill Book Co., Inc.: New York, 1961, pp 32−49. (b) Haynes, W. M., Ed. CRC Handbook of Chemistry & Physics, 92nd ed.; CRC Press: Boca Raton, FL, 2011−2012; pp 15−13 and 15−22. (29) Lothrup, W. C.; Handrick, G. R. Chem. Rev. 1949, 44, 419. (30) Mackay, D. Environ. Sci. Technol. 1982, 16, 224. (31) van Aken, K.; Strekowski, L.; Patiny, L. Beilstein J. Org. Chem. 2006, 2, 3 DOI: 10.1186/1860-5397-2-3. (32) Phosgene route. Burkhardt, T.; Findeisen, K. (Bayer). DE 2329051, 1975. (33) Diphosgene route. Kurita, K.; Matsumura, T.; Iwakura, Y. J. Org. Chem. 1976, 41, 2070. (34) Triphosgene route. Macchioni, A.; Pregosin, P. S.; Ruegger, H.; Koten, G.; van Schaaf, P. A.; van der Abbenhuis, R. A. T. M. Magn. Reson. Chem. 1994, 32, 235. (35) Dimethyl carbonate route . Molzahn, D. C. (Dow). U.S. Patent Appl. US/2008227999, 2008. (36) Tomishige, K.; Ikeda, Y.; Sakaihori, T.; Fujimoto, K. J. Catal. 2000, 192, 355. (37) Schulte-Huermann, W.; Schellmann, E. (BASF). DE 2847484, 1980) (38) Buysch, H. J.; Krimm, H.; Boehm, S. (BASF). EP 21211, 1981. (39) Kurita, K.; Iwakura, Y. Organic Syntheses 1988; Collect. Vol. 6, p 715. (40) Eckert, H.; Forster, B. Angew. Chem., Int. Ed. 1987, 26, 894. (41) Eckert, H. DE 3440141, 1986. (42) Faith, W. L.; Keyes, D. B.; Clark, R. L. Industrial Chemicals, 3rd ed., Wiley: New York, 1966; pp 101, 261, 541. (43) Shreve, R. N. Chemical Process Industries, 3rd ed.; McGraw-Hill: New York, 1967; p 812. (44) Faith, W. L.; Keyes, D. B.; Clark, R. L. Industrial Chemicals, 3rd ed.; Wiley: New York, 1966; pp 23, 140, 152, 261, 472, 583. (45) (a) Ritter, S. K. Chem. Eng. News 2005, 83 (6), 24. (b) Nicol, A. M.; Hurrell, A. C.; Wahyuni, D.; McDowell, W.; Chu, W. Am. J. Ind. Med. 2008, 51, 861. (46) Hogue, C. Chem. Eng. News 2012, 90 (24), 32. (47) Erickson, B. E. Chem. Eng. News 2009, 87 (25), 30. (48) Arnaud, C. H. Chem. Eng. News 2007, 85 (32), 34. (49) Hogue, C. Chem. Eng. News 2005, 83 (51), 43. (50) Hogue, C. Chem. Eng. News 2003, 81 (23), 19. (51) Hogue, C. Chem. Eng. News 2003, 81 (6), 21. (52) Hogue, C. Chem. Eng. News 2001, 79 (37), 27. (53) Rusyn, I.; Daston, G. P. Environ. Health Perspect. 2010, 118, 1047. (54) Russom, C. L.; Breton, R. L.; Walker, J. D.; Bradbury, S. P. Environ. Toxicol. Chem. 2003, 22, 1810. (55) Valeiro, L. G., Jr. Toxicol. Appl. Pharmacol. 2009, 241, 356. (56) Kavlock, R. J.; Ankley, G.; Blancato, J.; Breen, M.; Conolly, R.; Dix, D.; Houck, K.; Hubal, E.; Judson, R.; Rabinowitz, J.; Richard, A.; Setzer, R. W.; Shah, I.; Villeneuve, D.; Weber, E. Toxicol. Sci. 2008, 103, 14. (57) Chen, C. P.; Ahlers, H. W.; Dotson, G. S.; Lin, Y. C.; Chang, W. C.; Maier, A.; Gadagbui, B. Reg. Toxicol. Pharmacol. 2011, 61, 63. (58) Guy, R. H.; Hadgraft, J.; Maibach, H. I. Toxicol. Appl. Pharmacol. 1985, 78, 123. (59) Flynn, G. L. In Principles of Route-to-Route Extrapolation for Risk Assessment; Garrity, T. R., Henry, C. J., Eds.; Elsevier: New York, 1990; pp 93−127. (60) Barratt, M. D. QSAR Toxicol. In Vitro 1995, 9, 27.

UNECE = United Nations Economic Commission for Europe US EPA = United States Environmental Protection Agency VMR = vector magnitude ratio WHMIS = Workplace Hazardous Material Information System (Canada) WS = water solubility (mg/mL) XSP = explosive strength potential XVP = explosive vapour potential (applies to gases and vapours)



REFERENCES

(1) Lapkin, A., Constable, D. C., Eds. Green Chemistry Metrics: Measuring and Monitoring Sustainable Processes; Blackwell Scientific: Oxford, 2008. (2) Andraos, J. The Algebra of Organic Synthesis: Green Metrics, Design Strategy, Route Selection, and Optimization; CRC Press: Boca Raton, 2012. (3) Calvo-Flores, F. G. ChemSusChem 2009, 2, 905. (4) Andraos, J. Pure Appl. Chem. 2012, 84, 827. (5) Andraos, J. Pure Appl. Chem. 2011, 83, 1361. (6) Andraos, J. Org. Process Res. Dev. 2009, 13, 161. (7) Andraos, J.; Sayed, M. J. Chem. Educ. 2007, 84, 1004. (8) Andraos, J. Org. Process Res. Dev. 2005, 9, 149. (9) Mercer, S. M.; Andraos, J.; Jessop, P. G. J. Chem. Educ. 2012, 89, 215. (10) Werner, L.; Machara, A.; Sullivan, B.; Carrera, I.; Moser, M.; Adams, D. R.; Hudlicky, T.; Andraos, J. J. Org. Chem. 2011, 76, 10050. (11) Andraos, J. Org. Process Res. Dev. 2012, 16, 1482. (12) Jiménez-González, C.; Ponder, C. S.; Broxterman, Q. B.; Manley, J. B. Org. Process Res. Dev. 2011, 15, 912. (13) Gupta, J. P.; Babu, B. S. J. Haz. Mater. 1999, A67, 1. (14) Dow’s Fire and Explosion Index Hazard Classification Guide, 7th ed.; American Institute of Chemical Engineers: New York, 1994. (15) Talinli, I.; Yamantürk, R.; Aydin, E.; Basakcilardan-Kabakci, S. J. Haz. Mater. 2005, 126, 23. (16) Edwards, D. W.; Lawrence, D. Proc. Safety Environ. Protect. 1993, 71, 252. (17) Gupta, J. P.; D.W. Edwards, D. W. J. Haz. Mater. 2003, 104, 15. (18) (a) Workplace Hazardous Materials Information System (WHMIS) (http://www.health.gc.ca/whmis). (accessed October 2012) (b) WorkSafeBC (Workers’ Compensation Board of British Columbia) WHMIS Core Material: A resource manual for the application and implementation of WHMIS; Crown Publications: Victoria, BC, 2007; ISBN 0-7726-4468-3, http://www.worksafebc.com/publications/health_and_safety/by_ topics/assets/pdf/whmis.pdf. (accessed October 2012) (c) Workplace Hazardous Materials Information System (WHMIS): A guide to the legislation; Ministry of Labour: Toronto, ON, 2008; ISBN 978-1-42496997-5, http://www.labour.gov.on.ca/english/hs/pdf/whmis.pdf). (accessed October 2012) (19) NFPA 704: Standard System for the Identification of the Hazards of Materials for Emergency Response; National Fire Prevention Association: Quincy, MA, 2007; http://law.resource.org/pub/us/cfr/ibr/004/nfpa. 704.2007.pdf. (accessed October 2012) (20) Wilschut, A.; ten Berge, W. F.; Robinson, P. J.; McKone, T. E. Chemosphere 1995, 30, 1275. (21) NIOSH Pocket Guide to Chemical Hazards; National Institute for Occupational Safety and Health, Centers for Disease Control and Prevention: Atlanta, GA; September Publication No. 2005-149, 2007; http://www.cdc.gov/niosh). (accessed October 2012) (22) Czerczak, S.; Kupczewska, M. Appl. Occup. Envirron. Hyg. 2002, 17, 187. (23) Eissen, M. Bewertung der Umweltverträglichkeit organischchemischer Synthesen (Assessment of Environmental Impact of Organic Chemical Syntheses). PhD Thesis. University of Oldenburg: Oldenburg, Germany, 2001. (24) Emergency Response Guidebook: A guidebook for first responders during the initial phase of a dangerous goods/hazardous materials transportation incident. Transport Canada; U.S. Department of Transportation (Pipeline and Hazardous Materials Safety Administration); 191

dx.doi.org/10.1021/op300352w | Org. Process Res. Dev. 2013, 17, 175−192

Organic Process Research & Development

Concept Article

(99) Kupczewska-Dobecka, M.; Jakubowski, M.; Czerczak, S. Environ. Toxicol. Pharmacol. 2010, 30, 95. (100) Globally Harmonized System of Classification and Labelling of Chemicals (GHS), 3rd ed.; United Nations, United Nations Economic Commission for Europe: New York and Geneva, 2009 (ISBN: 978-921-117006-1), http://www.unece.org/trans/danger/publi/ghs/ghs_ rev03/03files_e.html. (accessed October 2012) (101) Winder, C.; Azzi, R.; Wagner, D. J. Hazard. Mater. 2005, 125, 29. (102) Williams, E. S.; Panko, J.; Paustenbach, D. J. Crit. Rev. Toxicol. 2009, 39, 553. (103) Regitz, M. Chem. Ber. 1966, 99, 3128. (104) Regitz, M. Angew. Chem., Int. Ed. 1967, 6, 733. (105) Regitz, M. Synthesis 1972, 351. (106) Black, T. H. Aldrichimica Acta 1983, 16 (1), 3. (107) Regitz, M. Diazo Compounds: Properties and Synthesis, Academic Press: Orlando, 1986. (108) Padwa, A.; Austin, D. J. Angew. Chem., Int. Ed. 1994, 33, 1797. (109) Ye, T.; McKervey, M. A. Chem. Rev. 1994, 94, 1091. (110) Zhang, Z.; Wang, J. Tetrahedron 2008, 64, 6577. (111) Maas, G. Angew. Chem., Int. Ed. 2009, 48, 8186. (112) Balli, H.; Kersting, F. Ann. Chem. 1961, 647, 1. (113) Balli, H.; Kersting, F. Ann. Chem. 1961, 647, 11. (114) Roush, W. R.; Feitler, D.; Rebek, J. Tetrahedron Lett. 1974, 1391. (115) Kokel, B.; Viehe, H. G. Angew. Chem., Int. Ed. 1980, 19, 716. (116) Kokel, B.; Boussouira, N. J. Heterocycl. Chem. 1987, 24, 1493. (117) Hazen, G. G.; Weinstock, L. M.; Connell, R.; Bollinger, F. W. Synth. Commun. 1981, 11, 947. (118) Aoyama, T.; Shioiri, T. Chem. Pharm. Bull. 1981, 29, 3249. (119) Monteiro, H. J. Synth. Commun. 1987, 17, 983. (120) McGuiness, M.; Shechter, H. Tetrahedron Lett. 1990, 31, 4987. (121) Green, G. M.; Peet, N. P.; Metz, W. A. J. Org. Chem. 2001, 66, 2509. (122) Fulton, J. R.; Aggarwal, V. K.; de Vicente, J. Eur. J. Org. Chem. 2005, 1479. (123) Titz, A.; Radic, Z.; Schwardt, O.; Ernst, B. Tetrahedron Lett. 2006, 47, 2383. (124) Kiramura, M.; Tashiro, N.; Miyagawa, S.; Okauchi, T. Synthesis 2011, 1037, 113. (125) Goddard-Borger, E. D.; Stick, R. V. Org. Lett. 2007, 9, 3797. (126) Urban, P. G., Ed. Bretherick’s Handbook of Reactive Chemical Hazards, 3rd ed.; Butterworth-Heinemann Ltd.: Oxford, 1995; Vol. 1. (127) Tuma, L. D. Thermochim. Acta 1994, 243, 161. (128) Clark, J. D.; Shah, A. S.; Peterson, J. C. Thermochim. Acta 2002, 392−393, 177. (129) Clark, J. D.; Shah, A. S.; Peterson, J. C.; Patelis, L.; Kersten, R. J. A.; Heemskerk, A. H. Thermochim. Acta 2002, 386, 73. (130) Clark, J. D.; Shah, A. S.; Peterson, J. C.; Patelis, L.; Kersten, R. J. A.; Heemskerk, A. H.; Grogan, M.; Camden, S. Thermochim. Acta 2002, 386, 65. (131) Proctor, L. D.; Warr, A. J. Org. Process Res. Dev. 2002, 6, 884. (132) Arndt, F. Org. Synth. 1935, 15, 3. (133) Eistert, B. Angew. Chem. 1941, 54, 124. (134) Arndt, F. Organic Syntheses; Wiley and Sons: New York, 1943; Collect. Vol. 2, p 165. (135) von Pechmann, H. Chem. Ber. 1894, 27, 1888. (136) von Pechmann, H. Chem. Ber. 1895, 28, 855. (137) Sheldon, R. A. ChemTech 1994, 24 (3), 38. (138) ANSI (American National Standards Institute) and Process Information Standard (NSF International, The Public Health and Safety Company). 355 Greener Chemicals; http://www.nsf.org/business/ sustainability/product_greener_chemicals.asp?program=Sustainability; http://www.nsf.org/business/sustainability/index.asp?program= Sustainability. (accessed January 2013) (139) Go Green. Chemical Compliance System. http://www.chemply. com/gogreen.htm. (accessed January 2013) (140) Green Screen for Safer Chemicals. Clean Production Action. http://www.cleanproduction.org/Greenscreen.php. (accessed January 2013)

(61) Sartorelli, P.; Aprea, C.; Cenni, A.; Novelli, M. T.; Orsi, D.; Palmi, S.; Matteucci, G. Ann. Occup. Hyg. 1998, 42, 267. (62) Patel, H.; ten Berge, W.; Cronin, M. T. D. Chemosphere 2002, 48, 603. (63) Moss, G. P.; Dearden, J. C.; Patel, H.; Cronin, M. T. D. Toxicol. In Vitro 2002, 16, 299. (64) Abraham, M. H.; Martins, F. J. Pharm. Sci. 2004, 93, 1508. (65) Mitragotri, S.; Anissimov, Y. G.; Bunge, A. L.; Frasch, H. F.; Guy, R. H.; Hadgraft, J.; McDougal, J. N.; Boeniger, M. F. Crit. Rev. Toxicol. 2002, 32, 291. (66) Krüse, J.; Golden, D.; Wilkinson, S.; Williams, F.; Kezic, S.; Corish, J. J. Pharm. Sci. 2007, 96, 682. (67) Kasting, G. B.; Lane, M. E.; Roberts, M. S. Int. J. Pharm. 2011, 418, 115. (68) Rodford, R.; Patlewicz, G.; Walker, J. D.; Payne, M. P. Environ. Toxicol. Chem. 2003, 22, 1855. (69) Berner, B.; Cooper, E. R. In Transdermal Delivery of Drugs; Kydonieus, A. F., Berner, B. Eds.; CRC Press: Boca Raton, 1987; Vol. 11, pp 41−55. (70) Brown, S. L.; Rossi, J. E. Chemosphere 1989, 19, 1989. (71) Cleek, R. L.; Bunge, A. L. Pharm. Res. 1993, 10, 497. (72) Fiserova-Bergerova, V.; Pierce, T.; Droz, P. O. Am. J. Ind. Med. 1990, 17, 617. (73) Frasch, H. F. Risk Anal. 2002, 22, 265. (74) McKone, T. E.; Howd, R. A. Risk Anal. 1992, 12, 543. (75) Potts, R. O.; Guy, R. H. Pharm. Res. 1992, 9, 663. (76) Guy, R. H.; Potts, R. O. Am. J. Ind. Med. 1993, 23, 711. (77) Scheuplein, R. J.; Blank, I. H. Physiol. Rev. 1971, 51, 702. (78) Michaels, A. S.; Chandrasekaran, S. K.; Shaw, J. E. AIChE J. 1975, 21, 985. (79) Kubota, K.; Ishizaki, T. Comput. Biol. Med. 1986, 16, 7. (80) Korinth, G.; Schaller, K. H.; Bader, M.; Bartsch, R.; Göen, T.; Rossbach, B.; Drexler, H. Arch. Toxicol. 2012, 86, 423. (81) Korinth, G.; Schaller, K. H.; Drexler, H. Arch. Toxicol. 2005, 79, 155. (82) Franz, T. J. J. Invest. Dermatol. 1975, 64, 190. (83) Williams, F. M. Env. Toxicol. Pharmacol. 2006, 21, 199. (84) Fasano, W. J.; McDougal, J. N. Reg. Toxicol. Pharmacol. 2008, 51, 181. (85) NOISH skin permeation on-line calculator (http://www.cdc. gov/niosh/topics/skin/skinPermCalc.html). (accessed October 2012) (86) Grandjean, P.; Berlin, A.; Gilbert, M.; Penning, W. Am. J. Ind. Med. 1988, 14, 97. (87) De Cock, J.; Heederik, D.; Kromhaut, H.; Boleij, J. S. M. Ann. Occup. Hyg. 1996, 40, 611. (88) Hemmen, J. J. Occup. Environ. Med. 1998, 55, 795. (89) Bos, P. M.; Brouwer, D. H.; Stevenson, H.; Boogaard, P. J.; de Kort, W. L.; van Hemmen, J. J. Occup. Environ. Med. 1998, 55, 795. (90) Sartorelli, P. Occup. Med. 2002, 52, 151. (91) Semple, S. Occup. Environ. Med. 2004, 61, 376. (92) Nielsen, J. B.; Granjean, P. Am. J. Ind. Med. 2004, 45, 275. (93) Sartorelli, P.; Ahlers, H. W.; Alanko, K.; Chen, C. P.; Cherrie, J. W.; Drexler, H.; Kezic, S.; Johanson, G.; Filon, F. L.; Maina, G.; Montomoli, L.; Nielsen, J. B. Reg. Toxicol. Pharmacol. 2007, 49, 301. (94) Lavoué, J.; Milon, A.; Droz, P. O. Ann. Occup. Hyg. 2008, 52, 747. (95) NOISH study. Sartorelli, P.; Ahlers, H. W.; Alanko, K.; Chen, C. P.; Cherrie, J. W.; Drexler, H.; Kezic, Dotson, G. S.; Chen, C. P.; Gadagbui, B.; Maier, A.; Ahlers, H. W.; Lentz, T. J. Reg. Toxicol. Pharmacol. 2011, 61, 53. (96) Kielhorn, J.; Melching-Kollmuß, S.; Mangelsdorf, I. Environmental Health Criteria 235: Dermal Absorption, joint sponsorship of the United Nations Environment Programme, the International Labour Organization, and the World Health Organization: Geneva, 2006; Section 12.8, pp 117−121 (ISBN 978-92-4-157235-4), http://www.inchem. org/documents/ehc/ehc/ehc235.pdf. (accessed October 2012) (97) Chen, C. P.; Boeniger, M. F.; Ahlers, H. W. Appl. Occup. Env. Hyg. 2003, 18, 154. (98) Czerczak, S.; Kupczewska, M. Appl. Occup. Env. Hyg. 2003, 18, 156. 192

dx.doi.org/10.1021/op300352w | Org. Process Res. Dev. 2013, 17, 175−192