A Systematic Evaluation of the Resource Consumption of Active

ARTICLE pubs.acs.org/est

A Systematic Evaluation of the Resource Consumption of Active Pharmaceutical Ingredient Production at Three Different Levels Geert Van der Vorst,† Jo Dewulf,*,† Wim Aelterman,‡ Bruno De Witte,‡ and Herman Van Langenhove† † ‡

Research Group ENVOC, Ghent University, Coupure Links 653, Ghent, B-9000, Belgium, Johnson & Johnson PRD, Janssen Pharmaceutica nv, Turnhoutseweg 30, Beerse, 2340, Belgium

bS Supporting Information ABSTRACT: In this paper, the development and the advantages of a methodology which allows the systematic assessment of the environmental impact on the resource side of specific pharmaceutical production processes with limited data entry is presented. The quantification of the process-specific mass and energy balances over three different system boundaries (process, gate-to-gate, and cradleto-gate) is based on the methodology explained in Van der Vorst et al. (Ind. Eng. Chem. Res. 2009, 48(11), 5344
5350). These mass and energy balances are now coupled with the thermodynamic term exergy allowing the quantification of the resource efficiency at the process and gate-to-gate level and the environmental impact at the cradle-to-gate level. The advantages of such a calculation tool for the resource evaluation are illustrated with five consecutive pharmaceutical production steps which are part of the galantamine (anti-Alzheimer medication) pathway. It is shown that such a quantitative and systematic evaluation tool allows a detailed and relatively fast evaluation of the resource efficiency of active pharmaceutical ingredient (API) production processes at the three different levels. Combining thermodynamics and the systematic data inventory methodology for the quantification of the resource efficiency first allows results to be merged into a single impact value (exergy loss/mol API or CEENE/mol API) for fast benchmarking and evaluation of different API production processes. Second, it also allows results to be divided over different categories depending on the users' interest and make thorough analysis of processes in order to pinpoint process improvements and quantitatively justify the introduction of second generation production processes or production techniques.

’ INTRODUCTION Developing sustainable production processes for active pharmaceutical ingredients (API) involves working from an early clinical phase of development with economically and ecologically sustainable processes.1 Chemical developers have to select as early as possible the best ecological and economical process for further development. For this selection procedure, the potential and impact of new processes and technologies are to be quantified, requiring the development of appropriate metrics and indicators. Different groups of metrics and indicators can be distinguished.2 The tools used by pharmaceutical companies as Johnson & Johnson, AstraZeneca,3 GlaxoSmithKline,4 and Pfizer,5 etc. are mainly dealing with emission and toxicity type indicators and indicators dealing with mass efficiency. These metrics and indicators, however, show three shortcomings. First, the energy requirements for the production of one mole API are barely included. In the best case, the follow up of the energy consumption currently occurs at the building level because allocation of the energy consumption to one specific batch production process is difficult due to complex networks of mass and energy supply in multipurpose plants.6
8 A second limitation is splitting up mass and energy inputs: kg vs kJ. When looking to the overall production chain, numerous resources can fulfill both functions. For the r 2011 American Chemical Society

evaluation of two alternative processes A and B, it is possible that based on mass requirements, A scores better, while for energy requirements, B is favorable. Third, there is the frequently too narrow system boundary approach. It is very well-known in industry what mass goes in and out of specific equipment to perform one process. These data can also be found quite easily in the “recipes” of a specific production process, being the batch production reports (BPRs).9 By enlarging the system boundaries to the gate-to-gate level, one considers also on-site resource requirements for the supply of utilities, recuperation of solvents, and treatment of waste streams. Eventually by enlarging the system boundaries to the cradle-to-gate level, all off-site resource requirements for the supply of all industrial products and services to sustain the process are taken into account. However, taking into account more data means more data have to be inventoried. Data inventory is the main drawback for making an integral resource consumption evaluation of pharmaceutical production processes.10
12 Life Cycle Assessment (LCA) in

Received: May 11, 2010 Accepted: February 28, 2011 Revised: February 24, 2011 Published: March 10, 2011 3040

dx.doi.org/10.1021/es1015907 | Environ. Sci. Technol. 2011, 45, 3040–3046

Environmental Science & Technology general is therefore barely used for the evaluation of pharmaceutical products and production processes. In this article, the methodology as presented in Van der Vorst et al.1 for the quantification of mass and energy balances over three system boundaries (process, gate-to-gate, and cradle-togate) is combined with the environmental impact assessment methodology using thermodynamics.13
16 Both methodologies are combined into an MS Excel-based resource evaluation tool for specific pharmaceutical production processes. This calculation tool offers the answer to all three above-mentioned limitations: both mass and energy are taken into account and quantified in one single unit (joule exergy) and this is done for the process, plant, and overall industrial network level, all with a very limited data inventory.

’ MATERIALS AND METHODS The methods and materials can be split into two parts. In the first part the methods and materials required for the setup of the combined methodology for the resource evaluation tool are presented. The second part includes the materials required for the illustration of the possibilities of the resource evaluation tool by five consecutive production steps. Data of these five production steps are from Janssen Pharmaceutica Belgium, part of the Johnson & Johnson group. A. Setup of the Resource Evaluation Tool for Specific Pharmaceutical Production Processes. For setting up the mass

and energy balances, this MS Excel tool is based on the data inventory methodology as presented in Van der Vorst et al.1 This methodology for the calculation of mass and energy balances for specific pharmaceutical production processes with reduced data inventory is summarized in Figure S1 in gray in the Supporting Information. The basic principle of this methodology is the quantification of the mass and energy balances for basic operations (BOs) (j) occurring in typical multipurpose equipment (i) over three system boundaries (plant, gate-to-gate, and cradle-to-gate). These levels are similar to the system boundaries R, β, and γ as defined in Van der Vorst et al.1 The R system boundary is the boundary around the process level, the β system boundary is around the gate-to-gate level, and the γ system boundary is around the cradle-to-gate level. A pharmaceutical production process or production step performed in a multipurpose production plant is a combination of different BOs (j) in different equipment (i). For the calculation of the mass and energy balances over three system boundaries of a complete production process, only the balances of all the BOs occurring in the production step under consideration have to be summated. All equipment and BOs per equipment for which data have to be available in the MS Excel-based tool are presented in Table S1 in the Supporting Information. These data allow the calculation of the mass and energy balances over the three previously defined system boundaries per BO. BPRs and the chemical reaction pathway are the only inputs for the calculation of the mass and energy balances of basic operations over three different levels. The functional unit for the evaluation is one mol API or intermediate produced. The focus of the evaluation methodology as presented in this article is on the resource side of the API production processes. This involves transforming the mass and energy balances into a resource-oriented environmental impact assessment unit. In this methodology, the resource requirements will be expressed as exergy losses occurring at the process level and the gate-to-gate level. At both levels, the exergy losses are quantified instead of the

ARTICLE

exergy input. Hereby the effect of considering valorizable waste streams as exergy loss or not, can be taken into account. This is not the case when only quantifying the exergy input at the process and gate-to-gate level. For the calculation of the exergy values, data available from literature, data obtained by means of the group contribution method,17 and Gibbs formation energy data are employed. In the γ system boundary (cradle-to-gate level), the CEENE methodology is used to express the integral resource consumption of the specific API products and production processes. The method for the calculation of the cumulative exergy extracted from the natural environment CEENE is retrieved from Dewulf et al.13 Next to using thermodynamics for the evaluation of the resource requirements, this methodology can also be extended to express the environmental impact by other life cycle impact assessment methods. This is also illustrated for the five cases. The mass and energy balance at the cradle-to-gate level are retrieved using the ecoinvent database (output 3 in Figure S1). Using the ecoinvent database allows an easy coupling of other life cycle impact assessment methods from ecoinvent.18 In this manuscript, only resource-oriented LCIA methods are combined with the results from OP3 as presented in Figure S1. The methodologies used for this tool can, however, be used to make a similar tool for the evaluation of the environmental impact of emissions over the three system boundaries for specific pharmaceutical production processes. This is, however, outside the scope of this article. B. Five Consecutive Production Steps for the Production of a Galantamine Precursor. In this article, the possibilities and advantages of both methodologies combined in a calculation tool are presented for five different cases. These five production steps are steps in the production of a galantamine (anti-Alzheimer medication) precursor. The important streams of the five production steps are illustrated in Figure 1, which shows the yields of all five steps, resulting in 1 mol intermediate (product E). The toluene recycling between step 4 and step 5 is shown because of its impact on the end result. A more detailed illustration of the five consecutive production steps including a description of the chemistry is provided in the Supporting Information (Figure S2). Names of the chemical structures cannot be given due to confidentiality issues. For these five production steps, the chemical reaction pathway is known and the BPRs are available. From each of the five BPRs, all used equipment is selected in the tool, followed by the selection of all BOs per equipment, and finally entering the required process-dependent parameters. For these 5 production steps in total 27 pieces of equipment and 267 BOs are used. For the calculation of the resource consumption this involves the selection of 27 pieces of equipment and the selection of 267 BOs in the MS Excel calculation tool. For each of the 267 BOs a limited amount of parameters are to be retrieved from the BPRs and entered into the MS Excel calculation tool. In Supporting Information Figure S3 a print screen of the data entry sheet for a reactor R1 in step 2 is presented. In this way all specific processdependent information is utilized and combined with the process-independent data in order to deliver the results as presented in the following parts.

’ RESULTS Two results can be distinguished. The first result is the setup of the combined methodology for the quantitative evaluation of the resource consumption of specific pharmaceutical production 3041

dx.doi.org/10.1021/es1015907 |Environ. Sci. Technol. 2011, 45, 3040–3046

Environmental Science & Technology

ARTICLE

Figure 1. Flow sheet of the production pathway of 1 mol E (precursor of API galantamine).

processes. This combined methodology is put into a MS Excelbased calculation tool. The second result is the illustration of the capabilities and advantages of the use of this MS Excel-based resource evaluation tool by five consecutive production steps of a galantamine intermediate. A. Setup of the Systematic Resource Evaluation Methodology and Related Excel Tool. The input data for this resource evaluation methodology are the data from the BPRs and the knowledge of the chemical reaction of the process to be evaluated. The setup of this methodology and also tool is visualized in Figure S1, available in the Supporting Information. The methodology to calculate the mass and energy balances of API production processes as presented in Van der Vorst et al. 1 and put in Figure S1 in gray is the basis for this new methodology and tool. The modifications allowing environmental impact assessment are presented in Figure S1 in green. The mass and energy balances (in Figure S1 presented as output 1, 2, and 3), are used as input for three new operators (OP4, OP5, and OP6). In these new OPs the mass and energy balances are combined with exergy values or CEENE values in order to calculate the exergy losses at the R system boundary (process level) and the β system boundary (gate-to-gate level) (output 4 and 5) and finally the CEENE values at the γ system boundary (cradle-to-gate level) (output 6). The OPs 4, 5, and 6 contain simple search functions and formulas for multiplying mass and energy values from outputs 1, 2, and 3 with the corresponding exergy values and finally categorizing and summating these exergy values. The meanings of these categories are explained in Table S2 in the Supporting Information. The main advantage of combining OPs 4, 5, and 6 with the methodology of Van der Vorst et al 1 is the possibility to aggregate the huge amount of output data available from outputs 1, 2, and 3. The problem of this huge amount of data is also mentioned in Van der Vorst et al. 1 and is the reason of the use of life cycle impact assessment (LCIA) methods in general. Thermodynamics and more in specific exergy analysis allows a scientifically correct aggregation of resources.13
17 Aggregation is used for a facilitated interpretation of large data sets but does not mean all resource are substitutable by each other. Two new process-independent databases are required for the use of exergy. An exergy database containing exergy values per kg product or per MJ energy stream is added as illustrated in Figure S1 enabling the calculation of the exergy losses occurring at the process and the gate-to-gate level. For the calculations at the cradle-to-gate level, a CEENE database is added as presented in Figure S1. In figure S4 in the Supporting Information, the

calculations performed by a calculation tool based on the methodology from Figure S1 are illustrated for one BO: “Heating or cooling reactor (closed system)”. A 4000-L stainless steel (SS) reactor is assumed containing 1000 L water at 25 °C. The resource requirements for the heating up to 50 °C are calculated and expressed in thermodynamic units. For this specific BO, the only input required from the BPR is the end temperature (x1: T end = 50 °C) and the time span for heating (x2: dT = 1 h = 3600 s). All the other data are retrieved from the previous BOs in the same equipment (e.g., addition of water 1000 L) and from the databases R1, R2, R3, β1, γ1, the exergy database, and the CEENE databases. All data are combined by simple calculations as illustrated in Figure S4. By only inserting into the resource evaluation tool the few inputs being the equipment, the BO, and parameters x1 and x2, the resource requirements over the process, plant, and overall system boundary can be calculated as presented in “output 4, 5, and 6” in Figure S4. B. Illustrative Case. The results and possibilities of such a calculation tool are illustrated by the evaluation of the resource requirements of five consecutive production steps as previously presented. In a first part, the processes are evaluated independently from each other. The results are presented for the five processes producing 1 mol product A, B, C, D, and E individually. In a second part, the cumulative results for 1 mol product E are presented. In this second part, the resource requirements of the previous production steps are taken into account. B.1. Results Per Process. The five processes are evaluated separately and results are presented in Figure 2. At the process level (Figure 2a and b), the exergy losses occurring in the equipment as such are presented. These results include irreversibility due to the electricity consumption for stirring and pumping, the heating and cooling requirements, cleaning, and the chemical reactions. The exergy content of waste streams sent to waste treatment are not taken into account as exergy losses in this process level. These streams when leaving the process boundary still can be reused and thus the exergy content is not yet lost. Once sent to the treatment, the exergetic content of these streams is lost and thus taken into account for the gate-togate level. From Figure 2a, it can be concluded that steps 1 and 2 have low exergy losses per mol compared to steps 3, 4, and 5. In step 1 and 2, 30% and 50%, respectively, of the exergy losses are due to the dG of the chemical reactions. The other exergy losses are mainly due to the use of electricity for stirring and pumping. Steps 1 and 2 do not require excessive heating and the equipment occupancy is for both steps approximately 1 min per mol 3042

dx.doi.org/10.1021/es1015907 |Environ. Sci. Technol. 2011, 45, 3040–3046

Environmental Science & Technology

ARTICLE

Figure 2. Environmental impact at the resource side of separate production processes, producing 1 mol A, B, C, D, and E.

produced. The equipment occupancy is defined as the time (in minutes) all equipment (reactors, filters, centrifuges, dryers, and storage tanks) is in use for the specific production step, divided by the total amount of moles produced in the process step under consideration. Step 3 has a higher irreversibility, due to the consumption of electricity and an equipment occupancy of 6.5 min/mol produced. Steps 4 and 5, which are interlinked due to the toluene recycling, have high irreversibilities at the process level because of the electricity consumption, heating, and cooling. The impact of heating and cooling on the exergy losses are a result of the reflux required in step 4, and the evaporation of 96% of the toluene solvent in step 5. The equipment occupancy is also high: 15.1 min/mol (step 4) and 7.6 min/mol (step 5). The reason

for this high equipment occupancy in step 4 is the product concentration in the toluene solvent which is almost 10 times lower than in step 1, 2, and 3. Step 5 has a high equipment occupancy because different batches from step 4 with a low product concentration are combined in step 5, where 96% of the toluene from 6 batches of step 4 has to be evaporated. This is very time-consuming, but finally results in a higher product concentration in the reactor. At the plant level (Figure 2c and d), the exergy losses over the whole production plant are quantified. This involves the exergy loss of waste streams induced by the process under consideration. Also the production and supply of the utilities required for the process under consideration is taken into account at the plant level. From the increased impact of exergy loss of chemicals 3043

dx.doi.org/10.1021/es1015907 |Environ. Sci. Technol. 2011, 45, 3040–3046

Environmental Science & Technology (Figure 2d), it is clear that at the plant level, the waste streams which cannot be recuperated are taken into account as exergy losses. For steps 1, 4, and 5, the exergy losses at the gate-to-gate level are 7.5
9.5 times the exergy losses at the process level. The reasons are the losses of organic solvents. In step 1 the solvent is sent to the wastewater treatment. In steps 4 and 5, even when taking into account 96% direct toluene reuse, the exergy loss from the 4% toluene which cannot be recuperated has a large impact. Next to the tremendous increase of chemical exergy losses, also the impact of heating and cooling increases because of their production efficiencies. For steps 2 and 3, the increase in exergy loss between the process and plant level is limited to 3
4%. For step 2 this can be explained by the use of an aqueous solvent. The low increase of exergy losses for step 3 can be explained by the electricity consumption. The electricity consumed at the process level comes from the national grid and is equal to the electricity loss at the plant level. This is in high contrast with, e.g., the exergy loss of heating. If at the process level 1 MJ heat exergy from heating oil (175 °C) is required, then at the plant level, approximately 4 MJ exergy is lost in order to produce steam, which is used to heat up the heating oil (see Figure S4). At last, results are presented at the cradle-to-gate level. Here the CEENE values representing the total amount of resources extracted from the natural environment for the production of 1 mol are shown. First the CEENE values are presented according to their use in the plant level (Figure 2e and f). Going from the plant level to the cradle-to-gate level means shifting from exergy losses to the total exergy requirements over the entire industrial network in order to produce one mol A, B, C, D, or E. At this level, the resource requirements for the production of inert gas are added. In the process and gate-to-gate level the exergetic content of inert gas was negligible for all process steps. In the cradle-to-gate level similar trends are visible compared to the process and plant level results except for the production process of 1 mol E. The CEENE for the production of 1 mol E remains low compared to the results at the plant level. The reason is the toluene recycling from step 5 back to step 4. The exergy losses as defined at the plant level are mainly due to the loss of the 4% non recuperated toluene. This toluene has, however, entered step 5 as the output from step 4. Therefore the CEENE value related to the fresh toluene is all allocated to step 4, while the toluene exergy losses are allocated to both step 4 and step 5. Again the integral resource consumption of the previous production steps is not included in the results of Figure 2. This means that the CEENE of 1.27 mol D in toluene, which is required for the

ARTICLE

production of 1 mol E, is put equal to zero. Second, in Figure 2g and h the CEENE values are presented according their source. The CEENE values are divided in eight categories as presented in Dewulf et al.13 The share of fossil resources has overall a large impact, except for step 2 (production of 1 mol B) due to the use of an aqueous solvent. This second part in the results over the gamma system boundary allows getting an idea of the “cradle” of the integral resource requirements next to the “sinks” at the plant level as shown in Figure 2e and f. Overall, it can be concluded that the CEENE value of step 4 is the highest. Most resources are to be extracted from the natural environment for the production process of 1 mol D, even not taking into account the previous production steps. The main reason is the low product concentration in toluene during step 4. This results in a high CEENE input of toluene, even taking into account 96% toluene reuse. Based on this, step 4 can be identified as the production step to be improved. B.2. Cumulative Results for the Production of 1 mol Intermediate E. The cumulative results, taking into account the yields over different production steps and resource requirements of previous production steps, are presented in Figure S5 in the Supporting Information. Except for the accumulation and taking into account resource requirements of the previous steps, the meaning of the process, plant, and cradle-to-gate levels as well as the exergy categorization remain similar. In Figure S5a and b, it is visible how the impact of electricity consumption on the irreversibilities at the process level accumulates after the third production step. The yield of 32% of production step 4 is also very clear in Figure S5a. The low yield, combined with the low product concentration, as discussed in the previous part, makes step 4 the ultimate candidate for a second generation process. Going to the plant level (Figure S5c and d), the exergy losses related to the nonrecoverable solvents and chemicals dominate the picture. Producing 1 mol intermediate E involves 75% exergy losses due to chemicals, 10% from electromechanical losses, and 15% by the other groups. For the production of 1 mol intermediate E, the exergy losses at the plant level are 5.6 times higher than at the process level. The biggest reasons, however, are the nonrecoverable organic waste solvent streams which are sent to waste treatment with negligible exergy recovery. Also at the plant level, the low yield of step 4 is clearly visible. The results at the cradle-to-gate level are presented in Figure S5e up to h. Similar conclusions as in the previous system boundaries can be drawn. The low yield of step 4 has the largest impact on the final CEENE value of 1 mol intermediate E. In Figure 3, the total CEENE flow over the five consecutive

Figure 3. CEENE flow of consecutive production steps for the production of 1 mol precursor E. 3044

dx.doi.org/10.1021/es1015907 |Environ. Sci. Technol. 2011, 45, 3040–3046

Environmental Science & Technology

ARTICLE

Table 1. Comparison Based on Different Impact Assessment Methods of a Five-Step Production Process to 1 kg Intermediate E and a Ten-Step Production Process of 1 kg “Substance A” as Defined in Wernet et al.8 1 kg “substance A” Wernet et al. 8

functional unit

1 kg E

number of synthesis steps

(5 steps)

CED [MJ-eq/kg product]

4.89  10

3

Eco-indicator 99 (H/A) resources [points/kg product]

1.02  10

1

IMPACT2002þ resources [points/kg product]

2.91  10
2

production steps is visualized showing the impact of the yields. The low yield of step 4 results in a high requirement of previous products A, B, and C. It can be concluded that for the production of 1 mol precursor E (MW = 387 g/mol and an exergy content of 22.24 MJ/kg), 2.57 GJ exergy is extracted from the natural environment. This means only approximately 0.5% of the required resources (CEENE) are captured in the final intermediate E. Such low resource efficiency is typical for the pharmaceutical industry. The E-factor of a pharmaceutical is around 25
100 while for bulk chemical this is between 1 and 5.19 As illustration, 2.57 GJ of exergy can be put equal to the exergy value of approximately 75 L (60 kg) of gasoline. When comparing Figure S5 c and e, it can also be concluded that approximately 2/3 of the exergy losses for the production of 1 mol intermediate E occur in the overall industrial network and 1/3 in the production plant (gate-togate). Making API processes more resource efficient is thus not only a commitment of the API producer itself, but also of the suppliers of utilities and chemicals. This systematic resource calculation tool allows evaluating complete API production process chains in a consistent way and with limited data inventory which results in the identification of production processes with the highest environmental impact. Also the potential improvement can be evaluated by combining this methodology with scenario analysis and estimations. This tool can be used to build a large data set with resource requirements for a large amount of API production processes. This eventually can be used for the predictive modeling of the resource requirements based on a limited amount of input parameters of certain API production processes. Predictive modeling of the environmental impact in early phases of API product development allows better and faster synthesis route selection and avoids the introduction of second generation production processes in a later phase. B.3. Other Resource Impact Assessment Methods. The resource evaluation tool as presented in this article is an extension of the methodology to set up the mass and energy balances of specific pharmaceutical production processes.1 Similar to the CEENE database, which is added to the process independent data as presented in Figure S1, data of other LCIA methods can be added to the process-independent database. In Table 1, three environmental impact results for 1 kg precursor E, as presented in this article, are compared with life cycle results as presented in Wernet et al.8 Although the impacts of the processes described in this article are higher, the results are in the same order of magnitude. It can be seen that the impact for the five-step synthesis of 1 kg E, on average is 3 times higher than the impact of the ten-step synthesis as discussed in Wernet et al.8 More thorough evaluation and comparison of both production processes is not given because the final APIs are different. Also the type of chemistry, reactions,

(10 steps) 1.43  103 2.73 9.40  10
3

and the production scale can have an important impact on the final results of life cycle assessments. This calculation methodology not only allows a fast and detailed quantification and evaluation of the resource requirements at three different levels (process, gate-to-gate, and cradleto-gate) using thermodynamics, but can also be used for the environmental impact assessment of pharmaceutical production processes, using other life cycle assessment impact assessment methods.

’ ASSOCIATED CONTENT

bS

Supporting Information. Figures and tables explaining and visualizing more into detail the setup of this resource evaluation methodology and the illustrative case results from such a tool; finally a list with abbreviations and symbols. This material is available free of charge via the Internet at http://pubs. acs.org.

’ AUTHOR INFORMATION Corresponding Author

*E-mail: [email protected]; tel: þþ32 9 264 59 49; fax: þþ32 9 264 62 43.

’ ACKNOWLEDGMENT We acknowledge the financial support of the Institute for the Promotion of Innovation through Science and Technology in Flanders (IWT-Vlaanderen) ’ REFERENCES (1) Van der Vorst, G.; Dewulf, J.; Aelterman, W.; De Witte, B.; Van Langenhove, H. Assessment of the Integral Resource Consumption of Individual Chemical Production Processes in a Multipurpose Pharmaceutical Production Plant: A Complex Task. Ind. Eng. Chem. Res. 2009, 48 (11), 5344–5350. (2) Dewulf, J.; Van der Vorst, G.; Aelterman, W.; De Witte, B.; Vanbaelen, H.; Van Langenhove, H. Integral resource management by exergy analysis for the selection of a separation process in the pharmaceutical industry. Green Chem. 2007, 9 (7), 785–791. (3) Hargreaves, C. R.; Manley, J. B. Collaboration to Deliver a Solvent Selection Guide for the Pharmaceutical Industry. The AIChE Fall National Meeting in Philadelphia, 2008. (4) Curzons, A. D.; Jimenez-Gonzalez, C.; Duncan, A. L.; Constable, D. J.; Cunningham, V. L. Fast life cycle assessment of synthetic chemistry (FLASC (TM)) tool. Int. J. LCA 2007, 12 (4), 272–280. (5) Alfonsi, K.; Colberg, J.; Dunn, P. J.; Fevig, T.; Jennings, S.; Johnson, T. A.; Kleine, H. P.; Knight, C.; Nagy, M. A.; Perry, D. A.; Stefaniak, M. Green chemistry tools to influence a medicinal chemistry and research chemistry based organisation. Green Chem. 2008, 10 (1), 31–36. 3045

dx.doi.org/10.1021/es1015907 |Environ. Sci. Technol. 2011, 45, 3040–3046

Environmental Science & Technology

ARTICLE

(6) Geisler, G.; Hofstetter, T. B.; Hungerbuhler, K. Production of fine and speciality chemicals: Procedure for the estimation of LCIs. Int. J. LCA 2004, 9 (2), 101–113. (7) Lampret, M.; Bukovec, V.; Paternost, A.; Krizman, S.; Lojk, V.; Golobic, I. Industrial energy-flow management. Appl. Energy 2007, 84 (7
8), 781–794. (8) Wernet, G.; Conradt, C.; Isenring, H. P.; Jimenez-Gonzalez, C.; Hungerbuhler, K. Life cycle assessment of fine chemical production: A case study of pharmaceutical synthesis. Int. J. LCA 2010, 15 (3), 294–303. (9) Bieler, P. S.; Fischer, U.; Hungerbuhler, K. Modeling the energy consumption of chemical batch plants: Bottom-up approach. Ind. Eng. Chem. Res. 2004, 43 (24), 7785–7795. (10) Jimenez-Gonzalez, C. PhD. Dissertation, North Carolina State University, 2000. (11) Jimenez-Gonzalez, C.; Curzons, A. D.; Constable, D. J.; Cunningham, V. L. Cradle-to-gate life cycle inventory and assessment of pharmaceutical compounds. Int. J. LCA 2004, 9 (2), 114–121. (12) Wernet, G.; Hellweg, S.; Fischer, U.; Papadokonstantakis, S.; Hungerbuhler, K. Molecular-structure-based models of chemical inventories using neural networks. Environ. Sci. Technol. 2008, 42 (17), 6717–6722. (13) Dewulf, J.; Bosch, M. E.; De Meester, B.; Van der Vorst, G.; Van Langenhove, H.; Hellweg, S.; Huijbregts, M. Cumulative exergy extraction from the natural environment (CEENE): A comprehensive life cycle impact assessment method for resource accounting. Environ. Sci. Technol. 2007, 41 (24), 8477–8483. (14) Dewulf, J.; Van Langenhove, H.; Muys, B.; Bruers, S.; Bakshi, B. R.; Grubb, G. F.; Paulus, D. M.; Sciubba, E. Exergy: Its potential and limitations in environmental science and technology. Environ. Sci. Technol. 2008, 42 (7), 2221–2232. (15) Cornelissen, R. L.; Hirs, G. G. The value of the exergetic life cycle assessment besides the LCA. Energy Convers. Manage. 2002, 43 (9
12), 1417–1424. (16) Sciubba, E.; Wall, G. A brief Commented History of Exergy From the Beginnings to 2004. Int. J. Therm. 2007, 10 (1), 1–26. (17) Szargut, J. Exergy method technical and ecological applications. In International Series on Developments in Heat Transfer; Sunden, B., Ed.; Witpress: Southampton, 2005; p 164. (18) Frischknecht, R.; Rebitzer, G. The ecoinvent database system: a comprehensive web-based LCA database. J. Cleaner Prod. 2005, 13 (13
14), 1337–1343. (19) Henderson, R. K.; Kindervater, J.; Manley, J. B. Lessons learned through measuring green chemistry performance
The pharmaceutical experience. 2009; http://portal.acs.org/portal/PublicWebSite/greenchemistry/industriainnovation/roundtable/CTP_005585 (accessed May 1, 2010).

3046

dx.doi.org/10.1021/es1015907 |Environ. Sci. Technol. 2011, 45, 3040–3046