A multidisciplinary investigation of the technical and environmental performances of TAML / peroxide elimination of Bisphenol A compounds from water † ‡

Designing technologies that mitigate the low-dose adverse effects of exposures to large-volume, everyday-everywhere chemicals such as bisphenol A (BPA, 1a) requires an understanding of the scope of the exposures and the nature of the adverse effects. Therefore, we review the literature of, (i) the occurrences of 1a in humans, waters and products and the effectiveness of widely deployed mitigation methods in 1a stewardship and, (ii) the adverse effects of 1a exposures on human cells and fish. Within this broad context, we present and evaluate experimental results on TAML/H2O2 purification of 1a contaminated waters. TAML/H2O2 catalysis readily oxidizes BPA (1a) and the ring-tetramethyl (1b), tetrachloro (1c), and tetrabromo (1d)-substituted derivatives. At pH 8.5, TAML/H2O2 induces controllable, oxidative oligomerisation of 1a (2-, 3-, 4-, and 5-unit species were identified) with precipitation, establishing a green synthetic pathway to these substances for biological safety characterisation and an easy method for near quantitative removal of 1a from water. TAML/H2O2 (24 nM/4 mM) treatment of 1a (10 000 μg L ) in pH 8.5 (0.01 M, carbonate) lab water effects a >99% reduction (to <100 μg L 1a) within 30 min. Yeast oestrogen screens (YES) of the pH 8.5, TAML/H2O2 treated, catalase quenched, and filtered oxidation solutions show elimination of 1a oestrogenicity. Zebrafish developmental assays of TAML/H2O2 treated, unfiltered, agitated pH 7, 1a solutions showed no significant incidences of abnormality among any of 22 endpoints—treated samples showed an insignificant increase in mortality. At pH 11, the TAML/H2O2 oxidations of 1a–d are fast with second order rate constants for the substrate oxidation process (kII values) of (0.57–8) × 10 M s. The 1a oxidation gives CO and CO2 (∼78%), acetone (∼25%) and formate (∼1%). In striking contrast with pH 8.5 treatment, no oligomers were detected. TAML/H2O2 (150 nM/7.5 mM) treatment of 1a (34 244 μg L) in pH 11 (0.01 M, phosphate) lab water effected a >99.9% reduction (to <23 μg L 1a) within 15 min. The pH dependent behaviour of 1a was examined as a possible origin of the differing outcomes. The 1 and 2 pKa values of 1a were estimated by fitting the pH dependence of the UV-vis spectra (pKa1 = 9.4 ± 0.3; pKa2 = 10.37 ± 0.07). At pH 8.5, coupling of the radical produced on initial oxidation evidently outcompetes further oxidation. A linear free energy relationship between the logarithm of the pH 11, kII values and the redox potentials of 1a–d as determined by differential pulse voltammetry in CH3CN is consistent with rate-limiting, electron transfer from the dianionic


Introduction
Green Chemistry has much to offer at every stage of the lifecycle of chemical products and processes. Both during and at the end of commercial utility, many chemical products become water contaminants. 1,2 These may or may not be persistent. Those that negatively impact flora or fauna at low concentrations (ng L −1 -μg L −1 ) are called "micropollutants" (MPs). Some MPs are endocrine disruptors (EDs). An endocrine disruptor is, "an exogenous chemical, or mixture of chemicals, that interferes with any aspect of hormone action". 3 Minimizing exposures to EDs is a significant sustainability challenge. High volume EDs that sufficiently elude water treatment processes to threaten the environment and human health are among the most difficult MPs to manage. 4,5 For example, phenolic compounds are not completely removed by the combined physical, biological and chemical processes in water treatment plants resulting in contamination of released effluent streams. 6,7 Bisphenol A (2,2-bis(4-hydroxyphenyl)propane, BPA, 1a, Fig. 1) is one such continually emitted, anthropogenic, xenooestrogenic, high volume, commodity, phenolic ED found in multiple products. [8][9][10][11] Although 1a is often regarded as weakly oestrogenic, its capacity to impact biological processes is modulated by several factors including interactions with plasma oestrogen-binding proteins, 12,13 varying potential for metabolism, [13][14][15][16][17][18][19] and the types and quantity of oestrogen receptors (ERs) present, including those bound to membranes. 20,21 In some cases, 1a has been shown to have the same effect as and be as potent as the endogenous, primary, female sex hormone oestradiol (E2, Fig. 1), 3,13,20,22,23 which can alter the functioning of cellular proteins at sub-picomolar to nanomolar concentrations. 13,24 Oestrogens regulate development [25][26][27] and actions in bone, brain, cardiovascular, liver, and reproductive tissues. 28 ERs are targeted by some endogenous hormones and pharmaceuticals. Such drugs include oestrogens, anti-oestrogens and selective oestrogen receptor modulators like 1a, 13 which lack the steroid ring structure of oestrogens (see E2, Fig. 1), but retain structural elements necessary to bind to ERs. 28,30 Drugs that target ERs include fertility enhancers and contraceptives, as well as menopausal hormone, 31 breast 32 and prostate cancer 33,34 therapeutics. In addition to acting as an oestrogen, 1a can act as a thyroid hormone 35,36 and androgen. 37 Inappropriate adjustment of oestrogen, thyroid hormone and androgen regulated processes, such that which can result from exposure to xeno-oestrogens like 1a, can have negative effects. 20,[25][26][27]31,[38][39][40][41][42][43][44] Consequently, in 2016 the European union voted to recognize 1a as a presumed human reproduc-tive toxicant. 45 In 2017, it voted to also add 1a to the list of substances of very high concern for adverse effects on human mammary gland development, cognitive functions and metabolism, identifying it as a general disruptor of the human endocrine system. 46 Approximately 15 billion pounds of 1a are produced annually. 1 This figure is expected to increase with a compound annual growth rate (CAGR) of almost 6% by 2020. 47 About 95% of the 1a produced is incorporated into BPA polycarbonate plastic and bisphenol A diglycidyl ether (BADGE) epoxy resins 48 (Fig. 1), including those used to line food and water containers such as metal cans, 49,50 metal and concrete drinking water pipes 51,52 and residential water storage tanks. 52,53 BPA is also added to phenoxy, 54 polyacrylate, 55,56 polyarylate, 57,58 polyetherimide, 58 polyester, 55,56,58,59 polyesterstyrene, 55 and polysulfone 54,56,58,59 plastic and resins, 54 and rubber, polyethylene tetraphthalate, 13,54,55,59 and polyvinyl chloride (PVC) products where it can function as a stabilizer, 54,60,61 antioxidant 62,63 or inhibitor of end polymerization. 64 Many of these products leach 1a to contribute to ubiquitous environmental contamination. 50,53,65 For example, 7.8 μg L −1 Fig. 1 Bisphenol A (BPA, 1a), its derivatives (1b-d) and the TAML activator (2) used in this work, as well as compounds referred to in this work. TAML is a registered trademark of Carnegie Mellon University covering tetra-organic-amido-N macrocyclic ligand complexes. 29 of 1a was detected in the water of an epoxy resin lined drinking water tank, 66,67 the contents of canned goods have been estimated to contain 4-23 μg of 1a as a result of leaching from the epoxy resin can linings, 50,68 and, at room temperature, polycarbonate food containers have been estimated to be capable of leaching ca. 6.5 μg of 1a into each gram of food stored inside. 68,69 Aqueous 1a concentrations of 1.98-139 μg L −1 have been found to result from the room temperature soaking of each of one gram of PVC products and synthetic leather for two weeks in the dark. 60,67 Room temperature, 24-hour exposure of pure water to PVC hoses resulted in leaching of 4-1730 μg L −1 of 1a. 67,70 Contamination also occurred upon passage through the tube and increased from 8.7-558 μg L −1 as the water residence time increased from 0-24 hours. 70 At 20°C, polycarbonate tubes leach 1a into lab, river, and sea waters at rates of 0.15, 0.2, and 1.6 μg L −1 day −1 , respectivelythe amount of 1a leached is less than that from PVC hoses. 67,70,71 Herein we examine the consequences of widespread BPA use and present an enticing laboratory study of a potential method for BPA stewardship. The impacts of BPA use are presented in two parts, a 'Mini-Review of the occurrence of BPA' and a 'Mini-Review of BPA toxicity'. These draw extensively upon reviews both from within and outside of traditional chemistry. 1,10,11,13,25,28,48,49,68, The BPA occurrence minireview encompasses the industrial synthesis and associated releases, the diverse product space, the associated contamination of air, water, soils, food crops, and recycled products, and measured human body burdens. The immense scope of BPA use and contamination is detailed to highlight the complexity of the stewardship challenge. Thus, the design and implementation of sustainable, chemical processes for removing BPA from waters is an important goal for green chemists. Therefore, the technical performances of currently deployed BPA water treatment technologies are surveyed. The BPA toxicity mini-review outlines the health and ecological implications of exposures to BPA at the concentrations currently observed in humans and environmental waters. With this understanding, we evaluate the technical and environmental performances of TAML/peroxide (2/H 2 O 2 , Fig. 1) oxidation of heavily BPA-contaminated lab water at pHs 8.5 and >11 at similar concentrations to certain processing streams and landfill runoff. The TAML/peroxide processes are found to be remarkably simple and effective warranting further studies in a variety of real-world scenarios.

Mini-review of BPA occurrences
This mini-review demonstrates the panoptic contamination of the ecosphere by BPA (1a) and 1a removal by the water treatment strategies currently deployed. We have largely focused on the sources and surface water occurrences of 1a. However, it is also important to recognize the significance of the reported contamination of oceans [110][111][112] and sediments. 11,67,113 The presence of polycarbonate in oceans is of special concern because leaching of 1a from polycarbonate occurs more rapidly in sea than in fresh waters, 74,114 1a oxidation by radical oxygen species is slower in seawater than in control water, 71 and marine bacteria strains have been found to degrade 1a much more slowly than freshwater strains. 107,115 Furthermore, 1a has been detected in marine organisms including phytoplankton, zooplankton, mussels, herring, flounder, and cod, and bioamplification between phytoplankton and zooplankton has been documented. 110,111,116 The reported sediment contamination levels exceed those of both sea and fresh water. 67 This assessment is consistent with the observation of higher concentrations of 1a in the lower layer of water columns than in the upper layers and a positive correlation between the low layer and sediment concentrations. 117 Since 1a degradation is slower under anaerobic than aerobic conditions, 67,109,117,118 and agents in the environment accumulate in the accessible reservoir in which they have the longest half-life and that reservoir then becomes a secondary source of emissions, 119 the potential for rerelease of sediment-bound 1a is concerning. A report on the fates of plastics and the need for reforms has recently been released. 113 Biomonitoring studies have revealed that 1a is also a pervasive human contaminant-unconjugated 1a has been detected in foetal, child and adult fluids and tissues (central tendency: 0.3-4.4 μg L −1 or 1.3-19 nM). 120 Total (unconjugated + bioconjugated) 1a urinary concentrations have been detected in 92.6% of 2517 Americans older than 6 years (mean: 5.2 μg L −1 or 23 nM, range: 0.3-149 μg L −1 or 1.3-653 nM). 121 Concentrations have been found in human colostrum (total 1a, all samples, 1-7 μg L −1 , mean of 3.41 μg L −1 ) 122 and breast milk (total 1a, 90% of samples, <0.3-6.3 μg L −1 , mean of 1.3 μg L −1 ). 123 Breast milk concentrations were found to rise from 6.2 μg L −1 to ca. 30 μg L −1 one hour after consumption of a canned coffee drink containing 37.4 μg, 124 indicating that exposure of mothers correlates with that of nursing children. An overview of the available science on the impacts of human exposures to these concentrations is presented later in the 'Mini-Review of BPA toxicity'.
The concentrations found in humans, colostrum, and breastmilk are comparable to some of the higher surface water concentrations of 1a that have been reported throughout the globe (US: median of 0.14 and maximum of 12 μg L −1 ; 125 Atibaia River watershed of São Paulo, Brazil: weighted average of 4.6 and maximum of 13 μg L −1 ; 126 Dongguan watershed of China: average of 6.5 and maximum of 56 μg L −1 ; 127 Nagara River of Japan: average of 4.8 and maximum of 22.2 μg L −1 ; 117 and Portugal: 0.07-4 μg L −1 ). 128 An overview of the available science on the impacts of fish exposures to these concentrations is also presented in the 'Mini-Review of BPA Toxicity'.
Canned goods are perhaps the most well-studied source of dietary 1a. Globally, contamination with 1a has been detected in canned beers, 154 decaffeinated and non-decaffeinated coffees, 162,164 soft drinks 165,166 including diet and regular ginger ales, diet and regular root beers, 165 diet and regular colas, 154,165 energy drinks, 154,165 orange and lemon soft drinks, 154 flavoured and unflavoured soda waters, 165 teas 162,165 and tonic waters, 165 infant 167-169 and follow up formulas, 170 the liquid phases of canned artichokes, green beans, corn, mushrooms, peas, and mixed vegetables, 50 the homogenized liquid and solid contents of canned fruit products including coconut cream, 154,171 coconut milk, 156 lychees, mangoes, 154 olives, 171 peaches, light pineapples, 154 tomatoes, 63,155,156,171 tomato juice 156 and tomato paste, 172 and fruit pieces and cocktails, 63 the homogenized liquid and solid contents of canned vegetable products including asparagus, 156 baked beans in tomato sauce, green beans, 63 beetroot, 171 carrots, 63 corn, 63,171,172 mount elephants, mushrooms, 156 peas, 63,171 jalapeño peppers, 153 potatoes, 63 and goulash, 154 the homogenized liquid and solid contents of canned soups 171,172 including cream of chicken, chicken and white wine, 63 potato, 154 tomato, 63 and Tom Kha, 154 the homogenized liquid and solid contents of canned sauces 171 of many varieties including demi-glace, fond de volaille, gratin, meat, tomato, and white, 156 the homogenized liquid and solid contents of canned pastas in tomato sauce, 63 the homogenized liquid and solid contents of canned seafood including Japanese sand lance, mackerel, 156 pilchards in tomato sauce, 63 salmon, 63,156,171 sardine, 156 sardine in tomato sauce 63 shrimp, squid, 156 tuna, 156,171,172 and fish and vegetable mixtures, 156 the homogenized liquid and solid contents of canned meats 171 including chicken, corned beef, fish balls, ham, hot dogs, and pork, 63 the homogenized liquid and solid contents of canned quail eggs, 156 the homogenized liquid and solid contents of desserts including evaporated milk 63,173 and creamed rice, 63 the solids of canned crushed tomatoes, young peas, 154 corn, 154,174 haricot beans, red kidney beans, lentils, 154 mushrooms, 174 tuna in oil, and sardines in oil, 154 mackerel filet in tomato sauce, and canned dinners. 166 Contamination of vegetable solids has been found to be greater than that of the liquids. 174 Alone, the estimated child and adult dietary intakes (0.008-14.7 μg kg −1 day −1 ) 82 are unlikely to account for the concentrations observed in human bloodstreams. 79 The human oral dosage necessary to maintain the average unconjugated 1a blood concentrations observed in healthy adults, adults having diseases, pregnant women, and foetuses (0.5-10 μg L −1 , average ca. 1-3 μg L −1 , greater than that found in most surface waters) has been estimated at >500 μg kg −1 day −1 . 81 This estimate agrees with a prediction of an oral dose of 1.43 mg kg −1 day −1 for 10 days as necessary to effect human steady state blood levels of 0.9-1.6 μg L −1 . 175 Therefore, alternate routes of exposure which obviate the first pass metabolism of the liver that greatly reduces blood concentrations of unconjugated 1a, 79 including dermal uptake, which increases when skin is wet and/or greasy, 176,177 and sublingual absorption 78,79 are thought to contribute to human blood concentrations of 1a. For example, the handling of thermal receipt papers, which are coated in amounts of unbound 1a that have been found to be at least one thousand times greater than that of food can epoxy linings, 178 is the most well-known route of dermal exposure. 140,177,[179][180][181][182] Given the myriad of products into which 1a is incorporated or contaminates, and the known sublingual, oral and dermal routes of exposure, other less-discussed sources likely also contribute to human blood concentrations of 1a. While not directly proven to raise blood concentrations of 1a, it is reasonable to expect additional exposures from, inter alia, indoor and outdoor air, [183][184][185][186] indoor dust, 183-185,187-189 5 gallon water carboys, 157 polyethylene tetraphthalate 13,132,190 and polycarbonate water bottles, 129 drinking water generators 52 and pipes, 51 water main 55 and tap 131 filters, fungicides, 54,55,59 randomly selected fresh foods including fresh cherries, courgettes, eggplants, medlars, oranges, peaches, peppers, and tomatoes, 191 white clams, crabs, blood cockles, fish, prawn, and squid, 111 buns, flour, hard cheese, minced meat, sausages, hamburgers, sliced salami and turkey, and frozen pizza in plastic packaging, bread in plastic or paper packaging, liver paté in plastic packaging with metal foil, fish pudding in plastic or paper packaging, caviar spread in a metal tube, jam in glass jars with metal or plastic screw caps, whole eggs packaged in cardboard, 166 honey packaged in glass or plastic that was imported in epoxy-lined metal drums, 192 baby food products in glass 193 and high-density polyethylene plastic 168 jars 52 with metal lids, solid 185 and liquid foods served at day care centers, 183,184 baby teethers, 194 breastpumps, 52 children's books and toys, 52,195,196 cosmetics, 197 training cups, 157 dishwasher and laundry detergents, 198 protective and general sporting equipment, 52,79 inhaler housings, musical instrument mouthpieces, 52 plastic plates and utensils 52,79,133 including those used at elementary schools, 162 nail polish, 55,198 pillow protectors, 198 outdoor play area soil, 183 artificial teeth, 55 hand, hard floor surface and food preparation surface wipes, 184 personal care-hygiene pro-ducts 52,60,209,210 and electronic waste, 60,91,209 distributor boxes, plug connectors, large advertising displays, fuses, lamp globes, inductors, lamp holders, lamps, electrical meters, switch modules, solar panels, sockets, streetlights, battery power stations, switches, transformers, kitchen tools and appliances including front panels for electric cookers, 52 refrigerator crisper drawers, 55 electrical kettles, coffee makers, microwaves, mixers, 52 and food processors, 55 and PVC products 48,60 including synthetic leather [211][212][213] in, for example, seating materials and clothing, 214 shower curtains, 198 and cord coverings. 60,211 It is important to note that the concentrations found can be high. For example, a Nigerian study recorded in the occurrences above reported 1a contamination ranges in consumer products and food/drink samples of 915.00-1415.50 μg L −1 and 163.00-2785.00 μg L −1 , respectively. 133 When used, discarded or recycled, 1a-containing goods and the contaminated solutions generated in the manufacture and processing of 1a and 1a-containing goods, become the largest sources of 1a in the environment. 60,211,215,216 In 2008, the government of Canada concluded that "bisphenol A is entering or may be entering the environment in a quantity or concentration or under conditions that have or may have an immediate or long-term harmful effect on the environment or its biological diversity or constitute or may constitute a danger in Canada to human life or health". 217 Consequently, in 2009, it recommended 1a releases be reduced to the lowest level technically and economically feasible and proposed an upper limit of 1.75 μg L −1 in emissions. 218 This value was chosen because it is an order of magnitude greater than the 2009 partial no effect concentration (PNEC) of 1a 218 -it has been argued that the PNEC for 1a in surface waters should be lowered to 0.06 μg L −1 , 219 and a maximum concentration of 0.03 μg L −1 has been proposed to safeguard 95% of exposed species from the chronic toxicity of 1a. 74 Therefore, easy to implement, safe technologies like the potential process documented in this contribution are critical to achieving emissions standards that, like the proposed Canadian standard of 1.75 μg L −1 , safeguard human health and the environment.
In 2014, the 258 million tons (mt) of municipal solid waste generated in the US before recycling, composting or combustion was composed of 69 mt of paper and paperboard (28% of which was landfilled, 65% was recycled and 7% was combusted), 33 mt of plastics (76% landfilled, 10% recycled and 15% combusted), 16 mt of textiles (65% landfilled, 16% recycled and 19% combusted), and 8 mt of rubber and leather (51% landfilled, 18% recycled and 32% combusted). 220 Landfilled postconsumer goods release 1a. 211 Consequently, water that has contacted waste, known as landfill leachate 221,222 (mostly pH 6.5-8.5), 61,89 has been reported to be a major source of 1a in the environment 202 227 These mixed pollutant streams containing 1a may or may not be treated based on local judgments of the need for such. 54 Treatments often include recycling through the landfill, on-site biological treatment, 90 coagulation, filtration, and/or activated carbon adsorption, 54,89 prior to discharge to surface waters or a municipal WWTP.
For surface water discharges, a higher treatment requirement often applies. One such treatment process entails pumping collected leachate containing ca. 100 μg L −1 1a into an adjustment tank where it is combined with leachate of unreported 1a concentration from four other landfill blocks and from neighbouring landfill sites and aerated. This reduced the concentration of 1a to ca. 0.15 μg L −1 . This combined influent was further treated by phosphate addition and passage through three biological contactors, coagulation with FeCl 3 and NaOH, mixing, sedimentation, and sterilization prior to discharge into a river. 61 When operated at ca. 25°C, the stages of the treatment process that followed combination with leachate of undisclosed composition and aeration removed ca. 80% of the influent ca. 0.15 μg L −1 1a and produced effluents of ca. 0.03 μg L −1 . Biological treatment and coagulation followed by sedimentation each account for roughly half of the amount of 1a removed. When the process was operated at ca. 20°C, the effluents were found to contain higher concentrations of 1a than the post-aeration influents. Both biological treatment and coagulation produce contaminated secondary wastes that must be landfilled, which risks rereleases of 1a, or incinerated.
Another treatment entailed 1a removal through the use of a membrane bioreactor (MBR) consisting of three activated sludge tanks. The first two tanks engaged nitrifying bacterial cultures (NH 3 → NO 3 − ). The third engaged a denitrifying culture (NO 3 − → N 2 ). This was followed by ultrafiltration (UF). 224 The influent ca. 500 μg L −1 1a was reduced to ca. 300 μg L −1 by the biological treatment of the MBR and the subsequent UF of the MBR effluent reduced this to ca. 2 μg L −1 . Biological treatment and UF produce secondary wastes in the forms of contaminated sludge and retentate, respectively, that must be landfilled, which risks rerelease of 1a, or dried and incinerated. Further treatment by nanofiltration (NF) increased the MBR effluent concentration of 1a to ca. 3 μg L −1 . Ozonation (0.2-0.5 kg O 3 per kg COD) of the NF retentate, which contained ca. 4 μg L −1 1a, removed ca. 83% to give effluent containing ca. 0.7 μg L −1 . For discharge to municipal WWTPs, a lower treatment requirement often applies. One such treatment entailed collection of the raw leachate in a regulating reservoir for pumping through a sequence of adjustment and aeration, mixing and sedimentation tanks before discharge into a sewer. 61 About 30% of the influent ca. 90 μg L −1 1a was removed giving effluents of ca. 65 μg L −1 . Another treatment entailed passage through a MBR which is similar to that discussed previously. 224 The biological treatment of the MBR reduced the highly contaminated, ca. 5000 μg L −1 1a-containing leachate to ca. 1300 μg L −1 . This was further reduced to ca. 70 μg L −1 by UF. Further treatment by passage through a granular activated charcoal (GAC) column removed ca. 50% of the remaining 1a, resulting in final discharges of ca. 35 μg L −1 . GAC treatment produces contaminated GAC which, when removal performance diminishes due to saturation, must be landfilled with risk of 1a rerelease and replaced at cost, or thermally regenerated.
In addition to landfill leachate, other process and waste streams containing 1a in excess of 1.75 μg L −1 have been reported worldwide. Some are treated before release and some are not. For example, 1a has been detected in untreated paper production and recycling solutions (Spain 2004: 3.0-142 μg L −1 , mean of 53.6 μg L −1 ). 228 Primary and secondary treatment at the plant reduced these 1a concentrations to 1.6-27 μg L −1 (mean of 13.8 μg L −1 ). 228 231 industries. Unfortunately, we were unable to locate data on the concentrations of 1a in various other industrial solutions and effluents including washing residue and wastewater generated in the production of 1a itself 91 and the sink-float/heavy media separation slurries, chemical etchants, detinning, and chemical delacquering solutions employed in the recycling of, for example, cans and bulk metals. 102 As with landfill leachate treatment plant effluent, industrial effluents may or may not be treated before release to the environment or wastewater treatment plants (WWTPs). 216 When directed to WWTPs, treated or untreated landfill leachate and industrial effluents may join with domestic wastewater which often also contains 1a. For example, the use of toilet tissue made from recycled paper has been estimated to contribute ca. 36 000 pounds of 1a to wastewater per year. 202,232 Consequently, at various times, 1a concentrations in WWTP influent have been reported to be high (Austria 2000: 10-37 μg L −1 , mean of 21 μg L −1 ; 229 Canada 2004: 0.16-28.1 μg L −1 ; 233 Germany 2008: <0.02-12.2 μg L −1 , weighted average of 3.67 μg L −1 ). 234 Though the removals of EDs including 1a have been observed to vary with the operating conditions, 216,235-237 conventional wastewater ( pH 7-8) treatment processes have been reported to remove 82 ± 12% of influent 1a. 87,233 Primarily, this removal occurs through sorption to primary sludge and sorption to activated sludge which may or may not be followed by biological degradation. 87,237,238 The final effluents are often discharged to surface waters. 233 Reports indicate that ca. 2% of the influent 1a is not degraded by and, as a result, remains in the activated sludge, along with a mixture of other EDs, the mass-normalized concentrations of which are significantly greater than those typically found in effluents and surface waters. 5,95,98,216,[239][240][241][242][243] As a consequence of lesser degradation in anaerobic than aerobic biological treatment, adsorption of several EDs, including 1a, to sludge occurs to a greater extent in anaerobic treatment. 239 Therefore while the use of nitrifying cultures can provide cleaner effluents, 244 these can come at the expense of increased sludge contamination. 202 The contaminated sludge generated in biological treatment can also become a source of 1a. Sludge concentrations of 0.10-3.2 × 10 7 μg kg −1 of dry weight have been reported. 11,[245][246][247][248] While incineration is practicable, the high costs, technological demands, low to no energetic gain, concentration of hazardous metal content in the ash produced which is usually landfilled, air emissions, and negative public perception can hinder its deployment. 249 Consequently, the contaminated sludge is often disposed of by landfilling or application to agricultural lands and composting. 99 For example, in 2004 the US produced 7.18 × 10 6 dry US tons of sewage sludge solids, 55% (3.95 × 10 6 tons) of which was land applied, 250 and in 2005, the EU-27 countries produced ca. 10.96 × 10 6 dry tons of sewage sludge solids, 41% (4.49 tons) of which was directed towards agricultural use. 99 These percentages are consistent with the 50% land application and 50% incineration reported in 2004 for the Canadian Ashbridges Bay, Humber and North Toronto WWTPs. 233 While the practice recycles nutrients and, thus, can enhance the sustainability of societies, land application of sewage sludge can increase the risk of ED rerelease to the environment 233 and contributes to ED contamination of soils. 5,238,[251][252][253][254][255][256] This has led to a considerable literature of the effects on animals and their offspring that graze on pastureland amended with sludge containing multiple EDs. 95,[257][258][259][260][261][262][263][264][265][266][267] Since these studies mostly concern multi-ED exposures, we consider this area outside the scope of the current mini-reviews. Sludge is the primary source of 1a soil contamination. 238 If practiced for an extended period of time, 239,255,256 land application to agricultural soils may contribute to the aforementioned levels of 1a detected in fresh produce. 191 Soil 1a can partition into soil water 83 which, in turn, can result in contamination of leafy vegetables, 238,268 root crops 238 and cereal grains. 238 Unfortunately, while never applying sewage sludge or only applying decontaminated sewage sludge would limit soil concentrations of 1a, this alone cannot ensure 1a-free soil.
Ongoing global deposition from ubiquitous contamination 186 of the atmosphere 238 and irrigation with treated and untreated contaminated water also contribute to soil concentrations of 1a. 70,268 Sources of atmospheric aerosol concentrations of 1a include volatilisation during processing, handling and transportation of 1a (at least 109 metric tons per year) 91 the open burning of municipal wastes (US: estimated to be >75 000 kg per year) 269 and plastics, 186 the thermal degradation of polycarbonate, [270][271][272] waste electrical and electronic equipment recycling, disposal and burning, 186,210 waste sorting facilities including metal shredders, 226 and landfills. 226 For example, 1a is a major product of the depolymerisation of polycarbonate observed on heating to 475°C in air, 272 and flash pyrolysis gas chromatography of polycarbonate at 500-850°C shows the release of H 2 O, CO 2 , 1a, phenol, isopropenyl phenol, and diphenyl carbonate in addition to higher molar mass compounds. [273][274][275] Resuspension of contaminated soil also contributes to atmospheric concentrations. 186 In addition to contributing to dietary exposures, cured, epoxy resins made from the 1a-containing prepolymer Bisphenol A diglycidyl ether (BADGE, Fig. 1), such as those used in the metal coatings of canned goods, may contribute to atmospheric concentrations of 1a when thermally degraded. In studies of heating the purified, cured resins made from BADGE prepolymers, stepwise alterations occur as the temperature increases. From ca. 250-350°C, evaporation occurs of the compounds encaged in the cured polymer, including unreacted modifiers such as diols, and residual prepolymer mono-and bis(hydroxyether)s of 1a. 276,277 From 300-340°C, thermal degradation begins and 1a is a major product released along with smaller amounts of isopropenylphenol, isopropylphenol and phenol. 278 Up to 500°C, there is very little decomposition of the 1a moiety. 278 From 500-600°C, arylalkyl ether bonds are further cleaved and 1a remains a major product released. [276][277][278][279] At >600°C, the release of high boiling pyrolyzates with epoxide end groups decreases markedly, 279 presumably giving a more 1a-rich product mixture.
The heating of cured resins made from BADGE prepolymers occurs in the recycling of metals. In, for example, aluminium recycling, thermal decoating or delacquering of the shredded metal is commonly performed prior to melting. 102,103 Approximately 18 metric tons of scrap pass through each delacquering machine per hour. 103 There are two major thermal delacquering approaches for Used Beverage Cans (UBCs). 103 The first is based on conveying crushed and shredded UBCs through zones of increasing temperature, which may be fixed at 520°C 103 or may progressively rise to ca. 540°C. 280 The furnace temperatures must be maintained below ca. 566°C to prevent ignition of the surface contaminants. 280 The second approach relies upon roasting the scrap in a rotary kiln through various temperature stages, the last of which occurs at near 615°C, with recirculation of the produced combustion gases. 103 Consequently, the gases produced in the delacquering process can be rich in 1a. These then may or may not be treated with activated charcoal 281 and/or lime or calcium carbonate 282 before or after passage through a baghouse filter followed by release via a high stack. 283 These purifications create contaminated materials which must be disposed of by, for example, landfilling or incineration. 282 Alternatively, the delacquering gases may be combusted. 282 Activated charcoal may, or may not, be added to the resulting flue gases 281 which are then purified by passage through a baghouse filter. 283 This creates contaminated activated charcoal. In either case, any compounds not sequestered or destroyed are released.
Despite the activated sludge removals, WWTP effluent concentrations of 1a in excess of 1.  285 While these alone are cause for concern, of perhaps greater concern are the by-products that can be formed at landfill leachate 135 and wastewater 286 treatment facilities if effluents are disinfected by chlorination 233 prior to release.
These disinfection by-products can include mono-and poly-chlorinated forms of 1a including the tetra-chlorinated 1c. 134,135 When product mixtures resulting from chlorination of 1a in water were tested in a binding assay employing a recombinant form of the endogenous human classical oestrogen receptor, 287 ERα, 24-fold greater activity than the untreated 1a solution was observed. 134 The mixtures also induced β-galactosidase activity in a yeast two-hybrid system employing human ERα. 134 In addition, 2-ClBPA, 2,2′-diClBPA, 2,6-diClBPA, 2,2′,6-triClBPA and 1c exhibited oestrogenic activity as measured by oestrogen-induced green fluorescent protein expression in transgenic human breast carcinoma MCF7 cells (ERE-GFP-MCF7 cells). 288 In addition to being oestrogenic, the multiply chlorinated forms of 1a are more resistant to activated sludge treatment than 1a. 203 Notably, 2,2′-diClBPA only undergoes very slow degradation while 2,2′,6-triClBPA and 1c are not degraded. Thus, in addition to emissions resulting from the use of 1c in flame retardants and as an epoxy intermediate, 243 chlorination of 1a-containing waters may contribute to the 1c concentrations which have been detected in sewage sludge, 243,289-291 river water 289 and sediment. 289 Further treatment of WWTP effluent with powdered activated carbon (PAC) or ozone has been advanced to enhance removal of MPs from wastewater. 292 The energy and resource demands of PAC treatment followed by sand filtration (PAC-SF) to retain the PAC and ozone are significant and similar. 293 PAC has been observed to remove 89.5-99% and 33-98.5% of 1a from pH 8.3 lab and pH 8.3-8.4 raw drinking waters, respectively, and these results are expected to transfer to waters containing ≥500 ng l −1 1a. 294 However, difficulties arise from handling of the finely powdered PAC and retention of the PAC which can necessitate employment of ultrafiltration membranes at significantly greater cost than either PAC-SF or ozone. 293 Additional problems derive from interference from dissolved organic carbon, a decrease in the performance over time due to saturation and management of the spent PAC, which is expensive to produce for replacement or to regenerate at elevated temperature. 293 BPA is readily degraded by ozone. [295][296][297][298][299] For example, treatment of 10 045 μg L −1 1a in pH 6.0 lab water with ca. 0.1 mg L −1 O 3 (4.05 mg O 3 per min) effects a >98% removal in 10 minutes with ca. 20% mineralization. 295 The technical performance of ozone decreases as the water matrix becomes more complex 296 and despite the extensive deployment of ozone in some countries, it is not widely used globally-the reasons vary by country. 88,293 Where it is deployed, staff training, safety measures 293 and additional infrastructure are necessary.
Significant amounts of 1a have also been detected in recycled cellulose fiber (RCF) plant process solutions, effluents and products, the last of which comprise roughly 50% of the furnish for worldwide paper and board production. 140,202,300 This can result in the previously noted contamination of virgin and RCF products. Thermal receipt papers typically containing low mg g −1 quantities of 1a contribute significantly to RCF contamination. 140 Up to 10% of the thermal paper produced is never used and directly enters recycling streams along with an additional 30% of that which is used. 284 A phase-out of 1a would remove it from the paper cycle, however a lag period of 10-30 years has been estimated before 1a concentrations would reach the limit of nondetection. 301 The contamination of RCF with EDs of any description is a wrench in the sustainability machinery required to move the civilization progressively toward renewable feedstocks. For example, incineration of highly 1a-contaminated waste at state-of-the-art, low-emission facilities has been advanced over recycling as a method for avoiding further contamination of recycled materials. 226 However, incineration precludes the energetic gains and decreased production emissions of material reuse, generates emissions including carbon dioxide 302 and can also generate PCDDs and PCDFs. 104,303,304 Another recently suggested option for dealing with 1a contamination of feedstocks may involve establishing acceptable levels for 1a in recycled products. 215 However, any level of 1a in goods exposes the consumer and compromises material streams. Therefore, improving decontamination methods for removing 1a from recycling process solutions and waste streams is an important sustainability research trajectory advanced by the empirical results detailed in this study.
Opportunities for 1a removal vary with whether or not the process includes a deinking stage. In the deinking of paper products, the addition of NaOH, sodium silicate, and surfactants extracts 95% of the contaminating 1a into the aqueous solution ( pH 9.5-11) and sludge. 284,305 The deinking sludge is then separated from the pulp slurry and dewatered. The process water may be clarified prior to reuse resulting in contaminated sludge and the enrichment of 1a in the process water. 306 Effluents from the pulp thickening process have been reported to contain 196-10 300 μg L −1 1a. 230 Washing of the pulp also generates contaminated solutions that are reused and may be bleached. 305,306 If alkaline paper recycling plant pulping solutions are chlorine-bleached, chlorinated forms of 1a, including 1c, can be generated, 203 as noted above. Primary treatment of these process waters transfers 95.9% of the influent 1a to the primary sludge. 284 The water is then sent to a WWTP for secondary treatment, often with activated sludge. In a study of 40 Korean WWTPs that receive influent comprised of varied proportions of industrial and domestic effluent, the concentrations of 1a detected in sludge at plants receiving primarily industrial effluent (I-WWTPs, >70% of inflow rate from industrial wastewater) were an order of magnitude greater than those receiving primarily domestic effluents (D-WWTPs, 0-3% of inflow rate from industrial wastewater). 216 Of the I-WWTPs, the highest sludge concentrations of 1a were found at plants receiving wastewater from the paper industry.
If deinking is not necessary, such as in the production of corrugated packing materials, the majority of 1a remains in the finished paper products limiting opportunities for removal and creating sources of environmental contamination. However, 10% of the influent 1a is transferred to water from the pulping process which is sent to primary treatment where 50% is removed with 18% incorporated into the primary sludge and 32% unaccounted for. 284 Final effluents have been reported to contain significant quantities of 1a (Japan 2002: 0.2-370 μg L −1 , average of 59 μg L −1 ). 230 In most plants, this effluent is subjected to secondary treatment with activated sludge. 306 The deinking, primary, and secondary sludges are dewatered and dried. These are then incinerated with negative to low net energy production, used in biogas production, landfilled, or applied to agricultural land where allowed by law. Thus, the massive global use of BPA in myriad products and processes further burdens an overstrained water treatment infrastructure where upgrading and maintenance require great public expense. Low cost, high efficiency BPA removal approaches, such as the lab experimental results herein promise, can improve both the technical and cost performances of dealing with this contamination.
TAML/peroxide removal of BPA from water One green science strategy for reducing exposures to compounds such as 1a is to pursue safer, more efficient and flexible stewardship processes. To this end we have been developing TAML activators (2 is the prototype, Fig. 1 315,331 suggests that 2 should easily oxidise electron rich 1a and BPAlike compounds. In fact, well over a decade ago at Carnegie Mellon, 1a was one of the first compounds studied in this context. At that time, 2/H 2 O 2 was found to readily catalyse the oxidative elimination of BPA from water. 332 However, further studies revealed a pH-dependence of the product distribution. As detailed herein, at pH > 10 2/H 2 O 2 oxidizes 1a rapidly and deeply giving a potential solution for BPA water contamination. At or near neutral pH (optimal for most water treatment processes), 2/H 2 O 2 induces oligomerisation of 1a to form precipitates as the principal products. While this could also represent a solution, the toxicity properties of the aqueous products were unknown, and the intervening years of this long study were primarily focused on developing confidence that no new toxicities were being introduced.
This concern is well-grounded in literature precedent. The unanticipated formation of phenolic oligomers found in the demethylation of anethole (Fig. 1) resulted in the discovery of hexestrol (Fig. 1). 94 Hexestrol is a mono-hydrogenated form of diethylstilbestrol (DES, Fig. 1), a compound structurally similar to 1 that, at the time, was one of the most potent known oestrogens with ca. one hundred thousand-fold greater oestrogenic activity than 1a as indicated by the relative minimum total weights of a substance required to induce a full oestrus response in ovariectomized female rats when administered by six injections over three days of a solution in sesame oil. 333,334 Additionally, the oxidation of 1a by fungal manganese peroxidase 335 has been observed to give a product mixture containing hexestrol. Therefore, in all cases, evidence was clearly needed that the treated solutions do not contain compounds that are also MPs across the domain of endocrine endpoints and beyond. Addressing these concerns necessitated the development of methodologies for analysing the catalysts and post-treatment solutions. In 2008, through the leadership of J. P. Myers and Advancing Green Chemistry, a coalition of environmental health scientists and green chemists formed to develop and eventually publish in this journal the Tiered Protocol for Endocrine Disruption (TiPED). 336 The TiPED is an organized suite of mammalian, fish and amphibian, cellular and computational assays designed to detect low dose, adverse effects as a pre-commercial guide to the chemical enterprise for avoiding EDs and MPs. The development of this protocol and the various resulting collaborations, 4,307,308,336 as extended herein, have allowed TAML activators and TAML-treated BPA media to be scrutinized for low dose toxicity.
Herein, we present a study of 2/H 2 O 2 treatment of 1a in lab water at near neutral pH and at pH 11 to determine the potential for improved decontamination of 1a-containing waters. This work demonstrates the many levels of complexity that accompany the oxidative degradation of BPA and its derivatives in water treatment. Here we report (i) that 1a-d are all readily decomposed by 2/H 2 O 2 at pH 11 and substantially mineralized and effectively eliminated from water, (ii) that 2/H 2 O 2 treatment of high concentrations of 1a at near neutral pH leads to a green procedure for oligomerising BPA, (iii) on the acid-base and redox properties of 1, (iv) on an improved synthesis of 12 (Scheme 4), a product of enzymatic degradation of BPA, (v) on a kinetic and mechanistic study of the oxidation of 1a-d and, (vi) on the toxicity of 1a samples before and after 2/H 2 O 2 treatment at pHs 8 and 11 via bacterial, oestrogenicity, and zebrafish developmental assays. Given that 1a and 1d are deployed commercially in large quantities, the work also highlights the requirement for further investigation of the degradation profiles of 1b-d and points to the need for expanded studies on the environmental safety of the 1a-d oligomers.

Materials
Bisphenol A (1a, Sigma Aldrich, GC grade >99%) was purified by re-crystallization using a mixture of hot ethanol and water and 2 was obtained from GreenOx Catalysts, Inc. Fresh stock solutions of H 2 O 2 were prepared daily from reagent grade H 2 O 2 (30% w/w, Fluka) and standardized by measuring the absorbance at 230 nm (ε = 72.8 M −1 cm −1 ). 337 All other chemicals and solvents obtained from Sigma-Aldrich or Fischer Chemicals were of ACS reagent grade quality or higher and were used as received. Water was either Fisher HPLC grade or deionized Milli-Q water (Millipore). Regenerated Cellulose (RC) 15 mm syringe filters (0.2 µm pore diameter) were supplied by Phenomenex.

Synthesis of 4-(4-(4-hydroxyphenyl)-2-methylpent-4-en-2-yl)phenol, a precursor of 12 (Scheme 4)
Bisphenol A (15.2 g, 0.07 mol) was dissolved in concentrated H 2 SO 4 (50 mL) and stirred (20 min) at 22°C. The reaction mixture was quenched by transfer into deionized water (900 mL) in an Erlenmeyer flask with vigorous stirring. The solids were filtered on a medium porosity frit and the filtrate was extracted 3 times with diethyl ether (3 × 200 mL). The organic extracts were washed with aqueous NaHCO 3 (5%, 100 mL) and combined with the solids. The resulting solution was dried over MgSO 4 , filtered, and the solvent was removed by rotary evaporation. The residue was purified by flash chromatography on silica gel with 2 : 1 hexane : ethyl acetate to yield a white solid (3.5 g, 0.01 mol, 15%). 1

Synthesis of 4-( prop-1-en-2-yl)phenol (12) (Scheme 4)
4-(4-(4-Hydroxyphenyl)-2-methylpent-4-en-2-yl)phenol (3.5 g, 0.01 mol) and NaOH (9 mg) were placed in a vacuum distillation apparatus to which a water aspirator was connected. The product was distilled under vacuum at 200°C using a silicone oil bath and a fraction boiling in the range of 75-135°C was collected. The light-yellow solid was dissolved in diethyl ether (5 mL) and was added to deionized water (25 mL). The mixture was stirred rapidly and sparged with nitrogen until a white precipitate was observed. This was isolated by suction filtration on a fine porosity glass frit (2.3 g, 0.020 mol, 86%). An analytical sample was recrystallized by slow evaporation of diethyl ether from heptane. 1

Methods
UV-vis spectra and kinetic data were obtained using a Hewlett Packard 8453 Diode Array spectrophotometer equipped with a thermostatted cell holder and an 8-cell sample positioner. The temperature was maintained at 25°C using a Thermo digital temperature controller RTE17 with a precision of ±1°C. Stock solutions of 1 were prepared by dissolving solid (7.50 × 10 −5 mol) in water made basic with KOH (5.0 mL). Stock solutions were then diluted with phosphate buffer (usually pH 11). For experiments at pH 8.5, aliquots of BPA stock solutions (10 000 ppm in CH 3 OH) were added to the required volumes of 0.01 M buffer (sodium carbonate/bicarbonate) to give solutions with final concentrations of 10 ppm BPA. Aliquots of a 2 stock solution (40 µM in deionised H 2 O) were added to the 1 containing buffer solutions to give the required final concentration (4-40 nM). Differential pulse voltammetry was performed with an Autolab PGSTAT100 potentiostat and GPES 4.9 software. The working electrode was a glassy carbon disk, with a saturated calomel reference electrode and platinum wire counter electrode. HPLC measurements were performed using a Waters® 600 system with 717 autosampler and 2996 photodiode array detector. Separations were carried out on a Varian Microsorb-MV 100-5 C18 (250 × 4.6 mm internal diameter, particle size 5 mm) column. The system at 40°C was run in isocratic mode with an acetonitrile/water (3/1) mobile phase. HPLC measurements for the determination of k II at pH 11 were performed using a Shimadzu HPLC system with a Shimadzu CMB-20A controller, LC-20AB pump, DGU-20A3 degasser, SPD-M20A diode array detector, RF-20A XS fluorimeter detector, CTO-20A column oven, and SIL-20A HT auto sampler. Separations were performed on a Phenomenex EVO C18 column at 40°C with a mobile phase of 50% methanol : 50% water. After H 2 O 2 addition to initiate reactions, aliquots (1 mL) were quenched by addition to an HPLC vial containing a catalase solution-12 000 units of bovine liver catalase or 60 times the concentration capable of destroying 2.0 mL of H 2 O 2 (4.0 × 10 −3 M) in 1 min with shaking (5 min). 1 H NMR spectra (500 MHz Bruker Avance 500) of the reaction mixtures were recorded for reaction mixtures containing 10% D 2 O and 1a (1.5 × 10 −4 M), 2 (1.5 × 10 −7 M) and H 2 O 2 (7 × 10 −3 M). The Watergate water suppression technique was applied. Ion chromatography (IC) studies were conducted on a Dionex DX500 chromatography system with a GP50 gradient pump, an AS40 automated sampler, an ED40 electrochemical detector, a LC25 chromatography oven, and an ASRS® 300 (P/N 064554) self-regenerating suppressor. Chromatographic data were analysed using Chromeleon chromatography software (Version 6.70 Build 1820, S/N 50398). IonPac® AS9-HC (4 × 250 mm) analytical and IonPac® AG9-HC (4 × 50 mm) guard columns were obtained from Dionex. The IC analysis was performed under isocratic conditions with an aqueous Na 2 CO 3 (9 × 10 −3 M) mobile phase at a flow rate of 1 mL min −1 with the oven temperature set at 35°C and the SRS current set at 100 mA. The injection volume for all IC samples was 100 μL. The IC mobile phase was prepared with water from a Barnstead Nanopure system. Total organic carbon (TOC) analysis was performed by Analytical Laboratory Services, Inc. Middletown, PA-samples consisted of 1a treated with 2/H 2 O 2 for a 15 min at pH 11. Reaction solutions at pH 8.5 were also subjected to solid phase extraction (SPE) using 500 mg hydrophilic-lipophilic balance (HLB) cartridges from Waters Corp. The cartridges were preconditioned with methanol (2 mL) followed by Milli-Q water (2 mL) and the sample was passed through the cartridge at a flow rate of 10.0 mL min −1 . The SPE cartridges were dried under high vacuum and then eluted with methanol (5.0 mL) at a flow rate of 3 mL min −1 . The eluent was collected and dried under a gentle stream of nitrogen gas. The residue was dissolved in methanol (2.0 mL) for analysis by high resolution (HR) mass spectrometry (Bruker micro-ToF-QII, Bruker Daltonics, Germany) coupled with a Dionex Ultimate 3000 HPLC with autosampler (Dionex, Germany) following a previously described procedure. 330 Samples were scanned within the range m/z 50-1500 in both the positive and negative ESI modes. Pure samples of 1a and 2 were similarly analyzed. Nitrogen dried methanol eluents were also derivatised for GC-MS analysis by treatment with BSTFA + TMCS (150 µL, 2 h, 60°C). The samples were then dissolved in benzene (350 µL) and injected (1 µL) onto a column (Restek RXi-5 ms, 30 m long, 0.25 mm ID, 0.25 µm) for GC-MS analysis using an Agilent GC 7890A gas chromatogram equipped with an Agilent 5975C inert XL MSD mass spectrometer with a Triple-Axis detector. 1 H NMR analysis (Bruker AVIII-400 MHz spectrometer) was also carried out using a sample obtained by SPE extraction of a BPA oxidation reaction after the methanol eluent had been evaporated to dryness under a gentle stream of nitrogen and the products redissolved in CD 3 OD.

Kinetic measurements
Monitoring of pH 11 processes was performed at the wavelength of maximal absorbance (λ max ) of the corresponding phenol. Initial rates were calculated using the independently measured pH 11 extinction coefficients [ε/M −1 cm −1 (λ max , nm)]; 1a Yeast oestrogen screens (YES). For oestrogen screening, samples (2.0 mL) were withdrawn at set intervals and assessed after catalase treatment by the four-hour yeast oestrogen screen (YES) bioassay-the procedure for preparing the yeast culture, growth medium and the testing regime has been detailed elsewhere. 338 This method allowed for the screening of multiple samples in a short amount of time without the need for sample preparation. 339 All sample solutions and blanks (without substrate) were evaluated by measuring the hormone-induced chemiluminescent signal on a Xenogen IVIS-200 optical in vivo imaging system. The chemiluminescent signals from solutions collected prior to the addition of 2 or H 2 O 2 were normalized to 100% oestrogenicity, which served as a reference point for subsequent screening tests during the reaction.
Zebrafish development. In order to further test the effects of 2/H 2 O 2 treatment of BPA on aquatic organisms, we assessed changes in 22 morphological endpoints over the first 5 days of zebrafish development. 340 All appropriate reaction controls were conducted.
The oxidation of 1a was studied using very low concentrations of 2 to achieve gentle oxidising conditions as a way of gaining insight into the formation of insoluble material by slowing its rates of formation and potential further oxidation. Following 2/H 2 O 2 (4 × 10 −9 M/4 × 10 −6 M) treatment of 1a (4.4 × 10 −5 M) at pH 8.5 (0.01 M carbonate buffer) for 180 min, HR-ESI-MS (negative ion mode) of the SPE collected reaction media (ESI, Fig. S1a ‡) showed two major ions. One was attribu- To better characterize the dimer, the crude material was TMS-silylated and analysed by GC-MS. Two major ions were observed. The first, at 51.4 min with m/z 742 [M] + , is consistent with a tetrasilylated dimer (ESI, Fig. S2a ‡). This suggests the formation of the C-C coupled product 3 (Fig. 3). The second, at 53.79 min with m/z 670 [M] + , is consistent with a trisilylated dimer (ESI, Fig. S2b ‡) suggesting the formation of the C-O coupled product 4 (Fig. 3). The 3 : 4 ratio of the relative integrals was ca. 14 : 1 i.e. the C-C coupling is a dominant pathway. Since authentic samples of the silylated dimers were not available, standard response curves were not generated. 1 H NMR analysis of small amounts of the crude reaction product before silylation gave spectra with poor signal to noise ratios consistent with a similar 3 : 4 ratio. Signals assigned to the proposed C-C coupled product 3 were clearly visible at δ 6. 78 Fig. S3 ‡). C-O coupled isomer signals could not be clearly assigned.
The influence of the 2 concentration on 3 and 4 formation was investigated (ESI, Fig. S4 ‡). The reactions were monitored at set time intervals by low resolution electrospray ionization mass spectrometry (LR-ESI-MS, negative ion, selected ionmonitoring mode (SIM), m/z 453) without SPE. With 4 nM 2, the yields of 3 and 4 reached a maximum after 60 min and then declined slightly over the next 120 min. This 2/H 2 O 2 oxidative procedure provides a green chemical synthesis for the oligomers. At 8 and 16 nM 2, approximately the same amounts of 3 and 4 formed more rapidly, but then declined more quickly, ultimately nearing zero.
The fates of 3 and 4 in treatment with 2 (4 and 40 nM) were next investigated under the same general conditions with a reaction volume of 500 mL and SPE product extraction prior to analysis by high resolution electrospray ionization (HR-ESI). The HR-ESI mass spectrum (ESI, Fig. S1b ‡) Fig. S1 and S4 ‡).
The ability this chemistry provides to easily achieve a greater than >99% removal of 1a near neutral pH with the generation of a secondary insoluble waste stream which is entirely composed of polymerized BPA promises a very simple technique for removing BPA from near neutral pH waste streams provided other contaminants do not complicate the chemistry.

Degradation of BPA at pH 11
The pH 11 (0.01 M, phosphate) and 25°C, 2/H 2 O 2 oxidations of 1 were studied by HPLC, total organic carbon (TOC) analysis, 1 H NMR spectroscopy, and ion chromatography. In the absence of H 2 O 2 , degradation of 1a over 1 h is negligible at pH 12. At pH 11 in the presence of 2 (1.5 × 10 −7 M) and H 2 O 2 (7.5 × 10 −3 M), the concentration of 1a (1.5 × 10 −4 M or 34 244 μg L −1 ) decreased below the HPLC detection limit (1 × 10 −7 M or 23 μg L −1 ) within 15 min. This represents a 99.9% removal without the generation of a secondary waste stream that requires additional treatment. No oligomeric degradation products were detected.
To further study the rapid and complete degradation of BPA by 2/H 2 O 2 at pH 11, 1 H NMR was employed. The spectrum of the parent BPA in water is simple allowing straightforward monitoring of the catalysed oxidation (Fig. 2). Although 2 is paramagnetic in the resting state, 341 at 150 nM noticeable line broadening of the signals of 1a was not observed. The 1a aliphatic and aromatic resonances reduced quickly and were no longer visible by 7.25 min. Within 2.75 min, the oxidation produced two new singlets at δ 2.24 and 8.46 which were assigned to acetone and formate, respectively. These assignments were confirmed by spiking with authentic samples. Plots of integral intensity versus time (not shown) indicate that the rates of 1a decay and acetone formation are similar though the acetone product only accounts for 25% of the 1a aliphatic signals. After 2.75 min, an AA′BB′ pattern appeared (δ 6.65 and 7.15), as did three higher field (δ 1.62, 1.53, and 1.52) resonances. These signals were no longer distinguishable from the baseline at 15 min. At pH 11, the final 2/H 2 O 2 -1a solution was transparent. As in other TAML degradation studies of phenols and systems with likely phenolic intermediates, 309,318,325,326 ion chromatography (IC) of the final 1b-d reaction solutions confirms the deep oxidation of 1. When 1a (1.5 × 10 −4 M) was subjected to 2/H 2 O 2 (1.5 × 10 −7 M/7 × 10 −3 M) for 15 min at pH 11 (0.01 M carbonate), formate (1.7%) was the major observable product by IC. With the more electron-rich and more reactive 1b (see below for kinetic studies), formate was found at even lower yield (0.7%) under the same conditions. Corresponding The acid-base properties of 1 are pertinent to the interpretation of kinetic data for the processes described above. Therefore, we have studied the effects of pH on the speciation of 1a in the range of 8-12 by UV-vis spectroscopy. The data in Fig. 4 show the pH dependence of the UV-spectra. According to the literature, 1a has a pK a between 9.59 and 11.30. 91 No well-defined isosbestic points were observed in our UV-vis study, suggesting that both phenolic hydroxides may undergo deprotonation in this pH range. The variation in the absorbance at 295 nm (A 295 ) with pH shown in the inset to Fig. 4 could be fit to eqn (1), an analytical form for the dependence of the absorbance on [H + ] which corresponds to the deprotonation sequence AH 2 ⇄ AH − ⇄ A 2− (AH 2 is 1a) in which [1a] t is the total concentration of 1a, K a1 and K a2 are the first and the second dissociation constants for 1a, respectively, and ε AH2 , ε AH , and ε A are the extinction coefficients for the forms AH 2 , AH − , and A 2− , respectively. The solid line in the inset is the calculated dependence of the A 295 on the solution pH using the best-fit values of the parameters of eqn (1). The so-derived pK a1 and pK a2 of 9.4 ± 0.3 and 10.37 ± 0.07, respectively, indicate that at pH 8.5, the neutral form H 2 A is dominant while, at and above pH 11, the dianion A 2− is the dominant form of 1a.
To characterize the relative tendency of substituted phenols 1a-d to undergo 1-electron oxidation, cyclic and differential pulse voltammetry methods have been applied. As was found previously, 342,343 the 1a-d reduction potentials could not be determined in aqueous solutions. However, appropriate data could be obtained in acetonitrile via application of the differential pulse technique. Examples of the voltammograms are shown as the Inset to Fig. 6 and the corresponding reduction potentials of 1a-d are included in Table 1. As expected, the electron-donating methyl substituent lowers the reduction potential of 1b compared to that of 1a, and 1c is the most resistant 1 to oxidation.
Mechanism of pH 8.5 TAML-catalysed 1a oxidation Oligomers such as those reported are commonly produced in enzyme-catalysed polymerizations of phenols 347-349 including 1a. [350][351][352][353][354] The initial step is often proposed to be a one-electron oxidation of the AH 2 or AH − forms of 1a to produce phenolate radicals which can then couple to give higher molecular weight products. [350][351][352][353][354] Our study mirrors the enzymatic results. There is little doubt that the oxidation starts with 1-electron oxidation of either AH 2 or AH − forms of 1a. In the former case, deprotonation of the primary radical-cation affords A and B (Scheme 2), precursors of the coupled C-C and C-O dimers 3 and 4, respectively (Fig. 3).   The initial rates of substrate oxidation (v) are well-modelled by eqn (2), the corresponding rate expression in which [Fe] t is the total initial concentration of all catalyst species. 314 v Obtaining the rate constants k II for oxidation of 1a-d by the initial rates method promised to deliver insight into the nature of the degradation process. As with many artificial peroxidase mimics, 314 the step associated with k I is slow requiring that, when possible, the reaction conditions should be set to ensure that steady-state measurements allow for the determination of k II . For TAML 2 such conditions are favoured by high [H 2 O 2 ] and basic pH around 11, where k I [H 2 O 2 ] ≫ k II [S] usually applies and k −I can be assumed to be negligible. 325,344 Under these conditions, eqn (2) simplifies to v = k II [S][Fe] t . Correspondingly, the initial rate is (i) independent of [H 2 O 2 ] in the range of (0.35-1.40) × 10 −2 M, (ii) proportional to the initial concentration of catalyst, [2], in the range of (0.375-1.50) × 10 −7 M and, (iii) directly proportional to the concentration of substrate, [1a] (Fig. 5).
Similar kinetic measurements were performed for 1c-d and the second-order rate constants k II were determined from the slopes of the linear plots for 1a, 1c and 1d (Fig. 5, Table 1). The oxidation of methyl-substituted 1b was considerably faster than all other cases and the initial rate showed saturation with increasing [1b], suggesting a contribution of the k I pathway to the overall rate. Therefore, the data were fitted to eqn (2) (assuming k −I ∼ 0) and the value of the thus calculated k II appears in Table 1.
We have developed a mathematical tool for modelling TAML catalysis 345 which we have used to examine the comparative behaviour of 15 different catalysts in the oxidation of one substrate under one set of conditions 327 and, more recently, the oxidation of 5 different substrates by 2 different catalysts and 2 different oxidants at two different pHs. 321 We used the k II value for 2/H 2 O 2 oxidation of 1a and the 2 k i value of (7.7 ± 0.3) × 10 −5 s −1 determined in 2/H 2 O 2 oxidation of the azo dye Orange II at pH 11 and 25°C, 346 to estimate the percent removal of 1a by 150 nM 2 treatment. The predicted removal of ca. 100% agrees well with the observed removal of 99.9% providing further validation of the mathematical model.
The 1a-d phenols were characterized electrochemically to examine if rate-limiting electron transfer is the initial event upon encounter of these electron-rich substrates with the powerfully oxidizing TAML reactive intermediate. The negative slope found for the linear relationship between the values of log k II and the reduction potentials E′ (Fig. 6) supports this assignment of the initial step. This correlation and the obser-   vation of acetone as a final 1a degradation product are consistent with the proposed mechanism (Scheme 3).

Tentative mechanism of pH 11 TAML-catalysed 1 oxidation
Since the pK a2 of 1a (10.37 ± 0.07) indicates that the doubly deprotonated A 2− form dominates the speciation of 1 at pH 11, and the k II values in Table 1 correlate with the reduction potentials (Fig. 6), and the observed removal agrees well with the removal predicted by the aforementioned mathematical tool, the initial step of pH 11, 2 catalysed 1 oxidation likely involves rate-limiting electron-transfer from A 2− to the Oxidized TAML species (Scheme 1) to give the primary product 5 (Scheme 3, step i). Oxidation of the second phenolate would give a compound 6 having resonance form 7 (Scheme 3, step ii). 355 Nucleophilic attack of hydroxide and intramolecular electron transfer would give 8 (step iii), an intermediate proposed in degradation of 1a by Sphingomonas sp. Strain TTNP3 via a type II ipso substitution mechanism thought to be enacted by a monooxygenase enzyme. 356 Intermediate 8 can undergo heterolysis to give p-hydroquinone and the compound represented by resonance structures 9 and 10 356 where the aromatization provides an important component of the driving force (step iv). The absence of p-hydroquinone in the 1 H NMR spectra of the reaction mixture (Fig. 2) should not be interpreted as an indication that it is not generated. Molecules like p-hydroquinone are known to undergo fast oxidation by H 2 O 2 under similar reaction conditions without involvement of TAMLs. 325 Since, an intermediate having an AA′BB′ spin system was observed at δ 6.65 and 7.15 (Fig. 2, 7.25 min spectrum) and 4-(2-hydroxypropan-2-yl)phenolate (Scheme 3, 11), 4-( prop-1en-2-yl)phenol (Scheme 4, 12), a tautomer of 10, and p-isopropylphenol, the reduction product of 12, have been observed in the product mixtures of bacterial, 357 enzymatic 6,75,335,350,358 thermal, [276][277][278][279]359 and/or chemical degradations of 1a, 360 including that of TTNP3, 356 and the reaction was performed at basic pH, these compounds were investigated as candidates for the reaction intermediate. 1 H NMR spectra of p-isopropyl-phenol and 12, which was generated from 1a as reported (Scheme 4), 361,362 obtained under the reaction conditions showed AA′BB′ spin systems at δ 6.56 and 7.02 and δ 6.67 and 7.41, respectively, indicating that neither compound accumulated measurably during 2-catalysed oxidation of 1a. 1 H NMR spectra of the crude products of the alkylation of p-hydroxyacetophenone with methylmagnesium bromide 363 showed 11, the presence of which was further confirmed by ESI-MS. However, difficulties were encountered in the isolation of 11, as has been reported. 363 As a result, we tentatively propose the intermediate to be 4-(2-hydroxypropan-2-yl)phenolate (Scheme 3, 11), the 1a analogue of a product observed in 1d oxidation by H 2 O 2 catalysed by 2 immobilized in a layered double hydroxide composite. 310 11 can undergo two successive one electron oxidations, nucleophilic attack of hydroxide and heterolysis with proton transfer (Scheme 3, steps vi, vii, viii and ix, respectively) to give the observed acetone (Fig. 2) and p-hydroquinone which would undergo rapid catalysed or uncatalysed degradation.
The material in the above Experimental section promises that, subject to successful real-world testing, a simple to deploy, technically effective, cheap to install and operate TAML/H 2 O 2 process for removing 1 compounds from water is achievable. The literature covered in the first mini-review establishes ubiquitous 1a occurrences in products and water with broad exposures to humans and wildlife. In the following mini-review, we examine the literature describing the consequences of these exposures to underscore the importance of developing more effective treatment processes that, before deployment, are cleared of low-dose adverse effects.

Mini-review of BPA toxicity
Several derivatives of (4,4′)-dihydroxy-diphenyl methane, which differ in the alkyl substituents of the aliphatic carbon atom, including 1a, are oestrogens as was briefly communicated in 1936, 23 more thoroughly reported in 1938, 364 1940 333 and 1944 334 and reviewed in 1945. 94 Given this history, it is not surprising that global 1a contamination can impact exposed organisms. 1,74,75,128,218,219,365 In a 2008 report to the Canadian Government, authors concluded that, "Bisphenol A is acutely toxic to aquatic organisms and is a known endocrine disruptor" 217 which, in 2010, led the Canadian Government to declare 1a to be toxic and add it to the List of Toxic Substances in Schedule 1 under section 64 of its Environmental Protection Act of 1999. 421 In adult fish of many species, lab tests show that exposure to the levels of 1a found in surface waters (earlier mini-review) causes alteration of gene expression including stimulation of Scheme 3 Tentative mechanism for 2/H 2 O 2 oxidative degradation of 1a at pH 11 consistent with results and working assumptions derived in this work. vitellogenin synthesis (a protein that is a precursor of egg yolk protein and a biomarker for exposure to oestrogens) in males, alteration of reproductive traits including reductions in male sperm quality, and delayed or no ovulation in females. 74,[366][367][368][369][370][371][372] In the embryonic and early life stages of fish development, such exposures have been observed to result in alteration of gene expression, reduction of heart rate, decreased eye pigmentation density, accelerated development, delayed hatching of embryos, testis-ova in males, hyperactivity in larva, and learning deficits in adult males. [373][374][375][376][377][378][379] In addition to these impacts on the exposed fish, lab exposures at typical 1a surface water concentrations can cause effects in offspring not observed in the parental generation. 365 These multigenerational effects can derive from both adult and developmental exposure to EDs and are measured by assessing the impacts of exposure on members of the initial population (F0) and each subsequent generation (Fn) versus unexposed controls (B0-Bn). 77,96,[380][381][382] One such study followed the effects of zebrafish exposure to 0.228 μg L −1 1a on the F0, F1, and F2 generations. 365 In the continuously exposed F1 and F2 generations, decreased male/female sex ratios in the adult population were observed together with lowering of male sperm density and quality including decreased motility and ATP production as well as increased sperm lipid peroxidation. The majority of the F1 adverse effects on sperm quantity and quality were not found in F2 male offspring if, subsequent to the 1a exposure, the F1 generation was incubated for 150 days in water to which 1a was not added, highlighting that these effects may be reduced through reduction of 1a exposure-the decreased sperm ATP production persisted in F2 males. Increases in malformation and mortality at 8 days post fertilization in the F2 offspring of F1 males indicated male-mediated reproductive failures deriving from 1a exposure. In the gonads of F2 males, altered gene expression was observed leading to perturbations of signaling pathways including those regulating mitochondrial biogenesis and testis development. In the larvae of F2 parents, reduction in expression of DNA methyl transferases and the associated transcription factor was observed indicating that continuous 1a exposure can lead to alterations of the epigenome and may result in transgenerational effects such as the observed male specific reproductive failures. 365 A US National Institute for Environmental Health and Safety panel has concluded that the available laboratory rodent studies provide sufficient evidence to be "confident" of human effects including effects on the male reproductive tract arising from adult exposures and effects on the organization of the reproductive tract of males, the brain and metabolism arising from developmental exposures. 92 A US National Toxicology Program panel 383 and the US Food and Drug Administration 79 also concluded that there is "some concern" for effects on the brain, behavior, and prostate gland in foetuses, infants and children arising from exposure to 1a at levels currently observed in the human population (earlier mini-review). However, further studies accounting for contributions to human 1a exposure from dermal absorption and of the toxico-kinetics following dermal absorption need to be performed. 85 We proceed by highlighting laboratory studies that reflect the human effects of 1a exposure to the concentrations found in foetal, child and adult fluids and tissues (0.3-4.4 μg L −1 or 1.3-19 nM). 120 Reviews of the epidemiological studies [384][385][386][387][388] on the effects of human 1a exposure not discussed herein are available. 68,389 Exposure to 1a at and slightly below the concentrations found in foetal, child and adult fluids and tissues can alter cellular development and produce mature cells which are improperly programmed. For example, exposure of HL-60 leukocytes to 1 nM 1a during neutrophilic differentiation induced by 1.25% dimethylsulfoxide and 25 ng mL −1 granulocyte colonystimulating factor results in significant increases in PU.1 transcription factor activity and production of opsonized zymosan (OZ) receptor subunit CD18 and O 2 − stimulating NADPHoxidase components p47phox and p67phox mRNA during differentiation via an oestrogen receptor independent mechanism, as well as differentiated cells with increased CD18 expression on the cell surface and OZ-stimulated O 2 − production suggesting that long-term 1a exposure may significantly affect human immunity. 390 Exposure of day 7 human prostaspheres, prostate stem-progenitor cells, to 1a at human relevant concentrations resulted in rapid membrane-initiated oestrogen signalling with increased levels of p-Akt and p-Erk, downstream targets of membrane ERs, and Erk phosphorylation, which were sustained for at least 60 minutes and returned to baseline by 6 hours, a result similar to that obtained from exposure to the same concentration of E2, an endogenous hormone. 391 Further insights into the effects of 1a exposure on prostate development have been gained through application of an in vivo renal graft model of chimeric human-rat prostate tissues. This model employs mice which express human prostate epithelial stem-progenitor cells that form normal human prostate epithelium, which produces prostate-specific antigen. 392 Treatment of the host mice with testosterone and E2 for 1-4 months to model the elevated E2 levels of later life-staged men increases the transition of the human prostate cells from hyperplasia to prostatic intraepithelial neoplasia and adenocarcinoma demonstrating the role of E2 in cancer of the human prostate epithelium. 391 Oral exposure of the host mice to 1a dosages giving free 1a serum levels of 0.39 and 1.35 μg L −1 , comparable to the internal doses found in human umbilical cord blood and foetal and neonatal serum, for 2 weeks after renal grafting resulted in mice having reduced normal prostate incidences from the control 26% to 11 and 0%, respectively, increased benign lesion incidence from the control 74% to 89 and 100%, respectively, and increased malignant lesion incidence from the control 13% to 36 and 33%, respectively, upon initiation/promotion of hormonal carcinogenesis by administration of testosterone and E2 for 2-4 months. 391 If, to model continuous exposure throughout development, prostaspheres were exposed to 1a in vitro prior to in vivo exposure, incidences of malignant lesions further increased to 45%. These results indicate that exposure of developing human prostate epithelium to doses of 1a relevant to those observed in human development increases its susceptibility to hormonal carcinogenesis. 391 Exposure to 1a at and slightly below the average concentrations found in human blood also alters the functioning of developed cells. For example, exposure of developed human pancreatic β-cells to 1 nM 1a in the absence of glucose rapidly decreased the activity of K ATP channels as effectively as exposure to 8 mM glucose, and in the presence of a stimulatory concentration of glucose (8 mM), exposure of human pancreatic islets of Langerhans to 1 nM 1a enhanced insulin secretion ca. 2 fold demonstrating that 1a exposure may be a risk for the development of type-2 diabetes. 393 Exposure of human breast, subcutaneous and visceral adipose tissue explants and mature adipocytes to 1 and 10 nM 1a suppressed the release of adiponectin, an insulin sensitizer, a result similar to exposure to equimolar concentrations of E2. 83,394 Exposure of human adipose explants to 10 nM 1a stimulated the release of inflammatory cytokines IL-6 and TNFα which promote insulin resistance. Taken together, the adipose tissue and adipocyte results indicate that exposure to 1a at population relevant levels may adversely effect metabolic homeostasis. 83,394 Exposure to 1a at and slightly below the average concentrations found in human blood and serum also alters the behaviour of cancer cells in tumours. For example, exposure of human androgen-dependent prostatic adenocarcinoma cells, a model system for prostate tumours, to 1 nM 1a has been observed to activate AR-T887A leading to androgen independent cellular proliferation like that resulting from exposure to 0.1 nM dihydrotestosterone, an endogenous ligand, though 1a activation may occur indirectly through interaction with ERβ or other proteins. 395 This result indicates that 1a exposure may ease the transition of prostatic adenocarcinomas to androgen independence, 395 thereby challenging treatment. 84 Exposure of both wild-type MCF7 breast cancer cells and an MCF7 subline, MCF7SH, which models the behaviour of long-term oestrogen deprived breast tumor cells, to 10 nM 1a has been observed to stimulate cellular growth, a response similar to, but weaker than that observed upon exposure to 10 nM E2, and this behaviour has been attributed to classical genomic activation of ERα. 191 While the observation of weaker response to 1a than E2 would seem to reflect a general, lower potency of 1a, as would be anticipated to result from the known, weaker binding of 1a to classical oestrogen receptors, this is not always the case. For example, in vitro exposure of MCF-7 cells, which express both ERα and ERβ, MDA-MB-231 cells, which express ERβ only, and SKBR-3 cells, which express neither ERα nor ERβ to 0.1-100 nM 1a in the presence of 2 mM Ca 2+ results in rapid Ca 2+ influxes in all three cell types leading to increases in intracellular calcium concentrations ([Ca 2+ ] i ) comparable to those resulting from exposure to equimolar concentrations of E2. 22 These results implicate 1a in non-genomic signalling pathways, such as membrane ERα (mERα) initiated signalling, and agree with those from in vitro studies employing rat pituitary tumor cell sublines GH3/B6/F10 (F10), which naturally expresses high levels of mERα, and GH3/B6/D9 (D9), which naturally expresses low levels of mERα. In vitro exposure of F10 cells to either 1 pM E2 21 or 1 pM 1a 20 results in rapid and reversible influx of extracellular Ca 2+ leading to nearly equal increases in [Ca 2+ ] i , which results in prolactin secretion and can initiate signaling cascades that lead to changes in cellular protein phosphorylation that alter protein functioning. Exposure of D9 cells to either 1 pM E2 21 or 1 pM 1a 20 does not alter the [Ca 2+ ] i . Thus, 1a can be as potent as E2 in eliciting responses mediated by non-genomic pathways reinforcing that it is inappropriate to label 1a a weak oestrogen. 20,22,393 Similar potency of BPA and EE2 has also been observed in vivo. 396 This may result from the formation of a BPA metabolite that either syngergizes 397 with BPA 396 or is significantly more oestrogenic. 17,396,398 One BPA metabolite produced by human liver S9 fractions in the presence of an NADPH-generating system is 4-methyl-2,4-bis(4-hydroxyphenyl)pent-1-ene (MBP, Fig. 1). 17 MBP has several-to several thousand-fold increased oestrogenicity in cellular assays employing ERα and ERβ. 17 One such assay showed MBP to be as oestrogenic as diethystilbestrol (DES, Fig. 1), 17 prenatal exposure to which is known to increase the risk of breast 42 and vaginal 399 cancers. 105 An in vivo uterotrophic assay using ovariectomized rats has shown MBP to be ca. 500 times more oestrogenic than BPA. 400 The MBP oestrogenicity increase is thought to derive from the increased distance between the phenolic rings which resembles that of benzestrol (Fig. 1), 401 a known oestrogen. 28,402 Studies have also shown MBP to adopt a coplanar conformation like E2 with a similar inter-hydroxyl distance. 17 Therefore, the effects of metabolites like MBP may contribute to the negative environmental 403,404 and human health 405 impacts of BPA, particularly if glucuronidation, the predominant, human, 1a metabolic pathway, 14,15 which gives a monoglucuronide showing almost no oestrogenic activity, 16 does not occur. 17,18 Thus, BPA exposure of foetuses, 68,120,[406][407][408][409][410] which are largely incapable of glucuronidation, 19,68,[411][412][413][414] is particularly concerning and disruption of human development 415 is a potential health impact of BPA and BPA metabolite exposure. 398 This is especially troubling given the known effects of BPA on cellular programming which can increase susceptibility of some organs, including the prostate 391,392,416 and mammary glands 417,418 to the development of cancer. 105,120 Since cytochrome p450 (CYP) operates in the liver fraction S9 metabolism that gives MBP 17,18 and CYP isoforms are expressed in the human foetal liver, 419 in vivo foetal BPA exposure may also result in foetal MBP exposure. 17 Unquestionably, the study of endocrine disruption phenomena associated with BPA metabolites should be a future research priority as there is enough already known to suggest that an under-recognized potent toxicity might be impacting human foetal development.
Along with numerous others, these studies of the effects of exposure to 1a on embryonic, early life stage and adult fish of many species and the zebrafish epigenome at environmentally relevant concentrations and of the effects of exposure to 1a on developing, developed, and cancer cells at concentrations relevant to those observed in the human population reinforce that 1a is an endocrine disruptor with negative environmental and health performances to challenge the on-going viability of many current applications. 12,86,217,420 With these studies providing a useful background we now describe the toxicity assays used to examine the TAML/H 2 O 2 1a treatments described in this work.

Toxicity of the treated 1a solutions
One-hour reaction mixtures (Table 2) were adjusted to neutral pH, quenched with catalase, and tested for acute toxicity using the Microtox assay. 422 Samples obtained at pH 8, where solids form in the dominating oligomerization of this pH region, were filtered before the assay. Although the oligomers are insoluble, we emphasize that these species are important aspects of the environmental profile of pH 7 treatments. A thorough exploration of the endocrine activity of the water insoluble oligomers is beyond the scope of the current studyif precipitation occurs in real-world media this would be an excellent way to isolate waste 1a in a form having diminished potential for long range transport. The results ( Table 2) show that samples treated with H 2 O 2 in the presence of 2 were less toxic than those treated with H 2 O 2 alone. An EC 50 of >100% indicates that the undiluted sample was not toxic enough to affect 50% of the organisms. At pH 8, only 5 molar equivalents of H 2 O 2 were necessary to effect this reduction in toxicity. At pH 12, the H 2 O 2 requirement increased to 50 molar equivalents.
Solution concentrations of 1a can be lowered to nearly zero by polymerizing treatment at pH 8.5 (Fig. 7a). It is prudent to show that these oxidised products, though effectively isolated, do not themselves have oestrogenic activity. This important point was illustrated by a recent study in which it was demonstrated that removal of EE2 from solution through partial oxidation produced soluble products that have similar oestrogenic activity as EE2 itself. 330 Therefore, the oestrogenic activities of the 1a solutions were evaluated by the yeast oestrogen screen (YES) 339 after oxidation with 2 at concentrations of 4, 16, 24 and 40 nM at time 0 and after 10 and 60 minute reaction times. The results are depicted in Fig. 7b. In all cases the oestrogenic activity was reduced, with the higher catalyst levels producing faster and more significant drops in activity. After 60 min, the solutions resulting from the reactions with 2 concentrations of 16, 24 and 40 nM showed almost no residual oestrogenic activity while the solutions resulting from reaction with 4 nM 2 showed approximately 75%. In all cases the residual oestrogenic activity correlated well with the amount of 1a remaining in solution. Therefore, it appears that the product solutions of oxidised 1a have no activity, and the 1a remaining can serve as an indication of the residual oestrogenicity of the treated solution.
To further test for aquatic toxicity of 2/H 2 O 2 treated 1a solutions, starting at 6 hours post fertilization, dechorionated zebrafish embryos were exposed to solutions containing 0.01 to 1% of the treated samples. This was done twice. In the first test, 50 µM 1a was treated with 2/H 2 O 2 (20 nM/5 mM, respectively) at pH 7 (0.01 M, phosphate) and quenched with catalase at 12 h. From the unfiltered, agitated, treated solution, more dilute solutions (0.2-2 µM, based on the initial concentration of 1a) were prepared, and the embryos were exposed to the diluted solutions. In the second test, 80 µM 1a was treated with 2/H 2 O 2 (200 nM/5 mM, respectively) at pH 7 (0.01 M, phosphate) and quenched with catalase at 12 h. From the unfiltered, agitated, treated solution, more dilute solutions (0.08-64 µM, based on the initial concentration of 1a) were prepared, and the embryos were exposed to the diluted solutions. With the treatment conditions chosen, no significant incidences of abnormality among any of the 22 endpoints were observed (ESI, Fig. S5 ‡). However, an insignificant increase in mortality was observed for the treated samples compared to the untreated samples.

Conclusion
In developing Green Chemistry, it is important that chemists come to understand the scope of the challenges posed by everyday-everywhere endocrine disruptors (EDs) to the sustainability of both the chemical enterprise and our complex global civilization. The most troubling such EDs, like BPA, invariably hold their protected positions in the economy because of seductive technical and cost performances that enable large, diverse, profitable markets. For sustainable chemicals, the health, environmental and fairness performances also have to be integral components of the value proposition.
Understanding the negative performances of unsustainable chemicals helps in mapping the properties sustainable chemicals should not have. Key aspects of this understanding include the knowledge of which chemicals are and are not EDs and are and are not capable of eliciting low dose adverse effects by non-endocrine processes, the extent and routes by which the environment and people are exposed to commercial EDs, the environmental and human health consequences of ED exposures, the methods of assessment of endocrine activity 423 including the TiPED, 336 the mechanisms of the low dose adverse effects, the design approaches to attaining new and replacement chemicals free of such effects, 307 and the stewardship methodologies that are currently deployed or might be deployed to better protect health and the environment from commercial EDs. This BPA case study traverses the appropriate multidisciplinary landscape with emphasis on the integration of chemistry and environmental health science in the development of endocrine disruption-free processes to aid the chemical enterprise and society in reducing BPA exposures. Importantly, the litany of unfortunate facts presented about BPA exposures and health and environmental performances is relieved to some extent by the possibility of reduced releases arising from the TAML/H 2 O 2 technology mapped out in the empirical section. This experimental component demonstrates that TAML/ H 2 O 2 provides simple, effective water treatment methodologies, which depending on the pH, either decompose 1a or isolate it in low solubility oligomers. Both processes require only very low concentrations of 2 and H 2 O 2 in further reflection of the remarkable efficiencies of the peroxidase enzymes that are faithfully mimicked by 2 and in marked contrast with the much higher relative iron-and peroxide-requiring Fenton processes. It remains to be established whether the current laboratory studies project to real world scenarios. These may include treatment of 1a-contaminated landfill leachates and paper plant processing solutions where the concentrations are similar to those employed in this study. In such scenarios, TAML/H 2 O 2 would present an enzyme-mimicking method which in the case of 2 is comprised exclusively of biochemically common elements and has passed multiple TiPED assays that, in contrast with existing real world processes, avoids generation of 1a-contaminated sludges and associated subsequent releases to soil, that does not generate a 1a-contaminated adsorbent which must be replaced or regenerated at elevated temperature, that does not generate chlorinated forms of 1a, that does not generate a concentrated retentate, and that is remarkably simple to deploy using very low and cheap chemical inputs with all the positive potential consequences thereof for capital and operating expenses.
Finally, in order to avoid the habit or perception of greenwashing, a realistic perspective is essential to the integrity of green chemistry. We view the sustainability challenges posed by BPA as enormous-the experimental work presented could evolve into a solution for some of these problems but is, by no means, a general quick fix. BPA markets large and small are expanding rapidly, especially as the industry has learned how to produce even more effective replacements for glass and metal products. Huge new markets are developing such as those of plastic glass houses, and even houses, and automobile body parts that are comprised primarily of BPA. In this build-up, BPA's unfortunate health and environmental performances continue to be given short shrift. Continuation of the present BPA expansion trends without limits, technical corrections and more aggressive stewardship advances of multiple kinds will menace society with an ever increasing oestrogenization of the entire ecosphere.