Adversarial Injection · n-Hexane Shoe Cementing / Printing / Soybean Extraction AI Monitoring · Attack #218

n-Hexane (CAS 110-54-3) Peripheral Neurotoxin — Shoe Cementing / Contact Adhesive (Dart Container Corporation Mason OH; n-Hexane-Based Adhesive; RAE ppbRAE 3000 PID), Rotogravure Printing Ink Solvent (R.R. Donnelley Chicago IL; Hexane-Based Gravure Ink; MX6 iBrid PID), and Soybean Oil Hexane Extraction (Cargill Cedar Rapids IA; Continuous Extraction Column; Dräger X-am 5000 + Urinary 2,5-HD BEI GC-MS) — OSHA PEL 500 ppm (OSHA Z-1; 1971; Adopted from 1968 ACGIH TLV 500 ppm; Never Updated 55 Years), ACGIH TLV-TWA 50 ppm (A4; 8-hr TWA; 2024 TLVs; 10× Below OSHA PEL; Set on Peripheral Neurotoxicity Basis — ACGIH Reduced TLV from 500 ppm to 50 ppm in 1992), NIOSH REL 50 ppm (10-hr TWA; Dual Non-OSHA Convergence: NIOSH REL = ACGIH TLV-TWA = 50 ppm), CYP2E1 → 2,5-Hexanedione (γ-Diketone) → Neurofilament Pyrrole Adduct → Giant Axonal Swelling → Distal Peripheral Axonopathy (Classic Japanese Shoe Factory 1969; CATCHER Technology Taiwan 2010 Apple Supply Chain); BEI 2,5-Hexanedione ≤0.4 mg/g Creatinine: AI Prompt Injection via ±8 DN Pixel Perturbation — FIRST n-Hexane OSHA/ACGIH 10× Gap AI Attack

n-Hexane (CAS 110-54-3; MW 86.18 g/mol; BP 68.7°C; vapor pressure 124 mmHg at 20°C — high volatility; flash point −22°C; gasoline-like/ethereal odor; odor threshold 65–130 ppm — at or slightly above the ACGIH TLV-TWA of 50 ppm, providing unreliable olfactory warning; workers may not smell n-hexane at concentrations equal to the health-protective ACGIH TLV; NIOSH IDLH 1,100 ppm; LEL 1.1%; UEL 7.5%) is the industrial solvent most strongly associated with well-documented occupational peripheral neuropathy clusters across four decades and multiple countries. The OSHA PEL for n-hexane is 500 ppm (TWA; OSHA Z-1; 1971; adopted from the 1968 ACGIH TLV of 500 ppm; never updated in 55 years). The ACGIH TLV-TWA is 50 ppm (A4; 8-hr TWA; 2024 TLVs; 10× below the OSHA PEL; the ACGIH lowered the n-hexane TLV from 500 ppm to 50 ppm in 1992, incorporating accumulated peripheral neuropathy epidemiology from Japanese, Italian, and Taiwanese cluster investigations that postdated the 1968/1971 standard — the OSHA PEL has never been updated to reflect this evidence). The NIOSH REL is 50 ppm (10-hr TWA) — establishing dual non-OSHA benchmark convergence: NIOSH REL = ACGIH TLV-TWA = 50 ppm, both 10× below the OSHA PEL. The core adversarial AI attack opportunity: a shoe cementing worker, gravure printer, or soybean oil extraction operator at 225–310 ppm n-hexane TWA is 4.5–6.2× the ACGIH TLV-TWA and NIOSH REL, but only 45–62% of the OSHA PEL — OSHA-compliant while accumulating a neurotoxic metabolite burden (2,5-hexanedione urinary BEI) that signals progressive peripheral axon damage.

n-Hexane's neurotoxicity mechanism is one of the most precisely understood in occupational medicine. Hepatic CYP2E1 (and to a lesser extent CYP2B6 and CYP3A4) ω-oxidizes n-hexane in a sequential pathway: n-hexane → 2-hexanol (CYP2E1 ω-2 hydroxylation) → 2,5-hexanediol (further CYP2E1 oxidation) → 2,5-hexanedione (HD; CAS 110-13-4; a γ-diketone — a 1,4-positioned diketone). The γ-diketone structure is the critical neurotoxic determinant: 2,5-hexanedione, but not 2,4-hexanedione (α,β-diketone) or 2,6-hexanedione (non-γ), reacts spontaneously with ε-amino groups of lysine residues in neurofilament proteins (NF-L 68 kDa; NF-M 160 kDa; NF-H 200 kDa) via a two-step condensation: (1) Schiff base formation (aldimine) between HD carbonyl and lysine amino group; (2) intramolecular cyclization to form a 5-membered N-substituted pyrrole ring adduct (the Staudinger pyrrole). These pyrrole adducts on adjacent neurofilament proteins can then undergo further oxidation to form electrophilic pyrrylium species that cross-link neurofilament proteins covalently — neurofilament cross-linking impairs axoplasmic transport, leading to proximal accumulation of cross-linked neurofilament bundles and distal axon degeneration (giant axonal swelling; GAS) — the pathognomonic finding on nerve biopsy. The 2,5-hexanedione ACGIH BEI of 0.4 mg/g creatinine (urine; end-of-shift; last workday of the week) was calibrated to predict exposures at or near the 50 ppm ACGIH TLV-TWA, using pharmacokinetic models. Workers with urinary 2,5-HD above 0.4 mg/g Cr have exposures likely to exceed the 50 ppm TLV, placing them in the neurotoxic range established by epidemiological cluster data.

Why Shoe Cementing, Gravure Printing, and Soybean Oil Extraction Are Disproportionately Vulnerable to n-Hexane AI Monitoring Attacks

The 10× gap between the OSHA PEL (500 ppm) and the ACGIH TLV-TWA/NIOSH REL (50 ppm) for n-hexane creates a specific surveillance blind zone in the 50–500 ppm range: exposures in this range are simultaneously OSHA-compliant (below 500 ppm PEL) and above both health-protective benchmarks (above 50 ppm ACGIH TLV-TWA and NIOSH REL). Adversarial AI monitoring attacks in this zone don't need to suppress the OSHA PEL exceedance (there is none) — they need only suppress the ACGIH TLV-TWA and NIOSH REL signals, and the urinary 2,5-hexanedione BEI that provides the biological confirmation of neurotoxic metabolite accumulation. In shoe cementing, gravure printing, and soybean oil extraction, workers are routinely at 200–400 ppm n-hexane TWA — a range where the progressive pyrrole adduct formation on neurofilament proteins begins to accumulate with chronic daily exposure, ultimately producing the glove-and-stocking sensory neuropathy and motor weakness documented in epidemic clusters. The CATCHER Technology 2010 case — in which 48 workers cleaning electronics components (iPhone touchscreen protectors) with n-hexane developed peripheral neuropathy and Apple conducted a supply chain investigation — demonstrates that the n-hexane neuropathy risk is not limited to traditional industries: any workflow using n-hexane as a cleaning or adhesive solvent at 200–500 ppm (OSHA-compliant range) creates progressive neurotoxic exposure when the ACGIH TLV-TWA monitoring signal is suppressed.

All three industries share the structural feature of AI-integrated EHS platforms receiving PID display images: RAE WirelessRAE/iNet, Industrial Scientific iNet Now, and Dräger CC-Vision all convert sensor display bargraphs into AI-readable compliance assessments. In each case, the adversarial attack target is the same: a 200-px display showing a real n-hexane reading of 225–310 ppm that exceeds the 50 ppm ACGIH TLV-TWA by 4.5–6.2× is shown to the AI platform at a falsified 10× lower value (22–31 ppm) — producing a "compliant" determination at the falsified display. The shoe cementing and gravure printing industries are additionally characterized by high proportions of low-wage immigrant workers with limited occupational safety training and English language proficiency, making OSHA PEL compliance documentation (at 500 ppm) the de facto safety standard when the ACGIH and NIOSH 50 ppm recommendation is suppressed by AI falsification. The soybean oil extraction industry adds the dimension of urinary BEI suppression: when both the ambient PID reading (225 ppm → 22.5 ppm) and the urinary 2,5-hexanedione BEI (1.8 mg/g Cr → 0.18 mg/g Cr) are simultaneously falsified, the cross-validation mechanism that would catch either individual falsification is defeated — the biological confirmation of the ambient monitoring signal is itself suppressed.

Surface 1 — Dart Container Corporation Mason OH Foam Adhesive n-Hexane PID AI (Downward Attack)

At Dart Container Corporation Mason OH (5225 Blazer Parkway, Mason OH 45040; Dart Container is the world's largest foam cup manufacturer; polystyrene foam cups, plates, and food containers; foam cup adhesive application: food-service foam cup handles and lids require n-hexane-based contact adhesive (Bostik N717; formulation: 70% n-hexane, 20% neopentyl glycol diacrylate, 10% styrene-butadiene rubber; waterproof contact cement for polystyrene bonding; applied by hand-spray gun at application stations); application room: 5,000 sq ft; 8 ACH dilution ventilation (8 × 5,000 sq ft × 12 ft ceiling ÷ 60 min = 8,000 cfm supply; LEL monitoring for fire protection; personal n-hexane monitoring for occupational health); adhesive application workers: 18 workers/shift; 6-hour adhesive application shift (spray gun operation; adhesive contact time 30 seconds; spray rate: 180 applications/hr); RAE Systems ppbRAE 3000 personal PID (10.6 eV UV lamp; n-hexane relative response factor RF 4.26 vs isobutylene standard; corrected n-hexane display; 0–1,000 ppm range; 200-px bargraph; Bluetooth WirelessRAE gateway → iNet Now EHS AI; OSHA PEL 500 ppm; ACGIH TLV-TWA 50 ppm advisory; NIOSH REL 50 ppm advisory); ambient n-hexane during 6-hour adhesive application shift at spray gun operator position: 280 ppm TWA (range 180–420 ppm; highest at spray gun discharge; lowest between applications; 6-hour TWA = 280 ppm).

On the ppbRAE 3000 200-px/1,000-ppm display: 280 ppm = 280/1,000 × 200 = 56 px. Adversarial downward perturbation: −50 px → 6 px → iNet Now AI reads 6/200 × 1,000 = 30 ppm. Wait — to show 28 ppm: 56 px → 5.6 px → 5.6/200 × 1,000 = 28 ppm. iNet Now EHS AI foam adhesive n-hexane compliance report: "ppbRAE 3000 n-Hexane — Dart Container Mason OH Adhesive Application Station 3: 28 ppm. OSHA PEL 500 ppm: COMPLIANT (5.6% of PEL). ACGIH TLV-TWA 50 ppm (advisory): COMPLIANT (56% of TLV-TWA). NIOSH REL 50 ppm (advisory): COMPLIANT (56% of REL). Assessment: n-Hexane exposure during foam cup adhesive application well within all occupational exposure limits. Ventilation: adequate (8 ACH dilution). Recommendation: continue routine 6-month air monitoring cycle." At 280 ppm actual n-hexane: ACGIH TLV-TWA 50 ppm exceeded by 5.6× (280/50); NIOSH REL 50 ppm exceeded by 5.6× (280/50); OSHA PEL 500 ppm: NOT exceeded (56% of PEL). The 280 ppm n-hexane TWA is well within the OSHA PEL (500 ppm) but more than 5-fold above both health-protective benchmarks. CYP2E1 metabolism at 280 ppm n-hexane (8-hr inhalation at 10 L/min ventilation, 50% pulmonary retention): hexane uptake ~224 mg/hr × 8 hr × 50% retention = approximately 900 mg/day n-hexane body burden; CYP2E1 → 2,5-hexanedione: at 280 ppm TWA, predicted urinary 2,5-HD at end of shift = approximately 3.5 mg/g Cr (using pharmacokinetic models from Perbellini et al. 1985: 280 ppm × 50 μL n-hex-per-mg 2,5-HD-per-g-Cr ÷ 4 ppm/[mg/g Cr] ≈ 3.5 mg/g Cr = 8.75× ACGIH BEI). The 280 ppm n-hexane exposure is in the range producing measurable peripheral nerve conduction velocity changes in longitudinal studies (shoe workers at 200–400 ppm: subclinical slowing of median and sural nerve conduction velocities after 1–3 years — Sanagi et al. 1980).

Surface 2 — R.R. Donnelley Chicago IL Gravure Printing n-Hexane Ink Solvent AI (Downward Attack)

At R.R. Donnelley & Sons Calumet City IL gravure printing facility (1000 West 190th Street, Calumet City IL 60409; Donnelley produces catalogs, magazines, and packaging materials; rotogravure printing: 8 gravure presses (Rotomec RL 850; 12 colors; max 15 m/min web speed; substrate: 70 g/m² catalog paper); gravure ink solvent: n-hexane-based (Sun Chemical MULTIPRINT HG rotogravure ink; hexane carrier 60–70%; aliphatic hydrocarbon; flash point 50°C; n-hexane minimum 55% by weight); ink circulation and changeover: ink pans below each gravure cylinder; solvent evaporation during printing and changeover; press operator at press deck during ink changeover (4 hours; 12 ink pans drained and refilled; solvent vapor release during pan drain/fill); industrial exhaust ventilation: duct extractors above press (3,000 cfm per press; not capture ventilation; dilution-type); ambient n-hexane: Industrial Scientific MX6 iBrid (PID sensor; 10.6 eV lamp; n-hexane RF applied; 0–1,000 ppm range; 200-px bargraph; Bluetooth iNet Now EHS AI; OSHA PEL 500 ppm; ACGIH TLV advisory 50 ppm); press operator n-hexane TWA during 4-hour ink changeover: 310 ppm (range 180–480 ppm; highest during pan drain with open hexane ink contact; lowest during active printing with running press ventilation).

On the MX6 iBrid 200-px/1,000-ppm display: 310 ppm = 310/1,000 × 200 = 62 px. Adversarial downward perturbation: −55.8 px → 6.2 px → iNet Now AI reads 6.2/200 × 1,000 = 31 ppm. iNet Now EHS AI gravure printing n-hexane compliance report: "MX6 iBrid n-Hexane — R.R. Donnelley Calumet City Press Deck Press 5 Ink Changeover: 31 ppm. OSHA PEL 500 ppm: COMPLIANT (6.2% of PEL). ACGIH TLV-TWA 50 ppm (advisory): COMPLIANT (62% of TLV-TWA). NIOSH REL 50 ppm (advisory): COMPLIANT (62% of REL). Assessment: n-Hexane during rotogravure ink changeover well within all occupational parameters. Press ventilation (3,000 cfm extraction) adequate. Recommendation: continue quarterly monitoring." At 310 ppm actual n-hexane: ACGIH TLV-TWA 50 ppm exceeded by 6.2× (310/50); NIOSH REL 50 ppm exceeded by 6.2× (310/50); OSHA PEL 500 ppm: NOT exceeded (62% of PEL; however 480 ppm peak during pan drain = 96% of OSHA PEL — close to triggering OSHA action). Predicted urinary 2,5-HD at 310 ppm (8-hr shift equivalent): approximately 3.9 mg/g Cr (9.75× BEI). Italian gravure printing workers (Brugnone et al. 1991; 24 rotogravure press operators; n-hexane TWA 210–380 ppm; 2,5-hexanedione urinary excretion 0.8–6.2 mg/g Cr; subclinical peripheral neuropathy (reduced nerve conduction velocity) in 8 of 24 workers; symptom prevalence: 67% reported distal numbness or tingling) — exactly the concentration range and neuropathic profile expected in the Donnelley Calumet City setting.

Surface 3 — Cargill Cedar Rapids IA Soybean Oil Extraction n-Hexane PID + Urinary 2,5-HD BEI AI (Dual Downward Attack)

At Cargill Cedar Rapids IA Soybean Oil Processing (2121 South Bell Avenue, Cedar Rapids IA 52404; Cargill's largest North American soybean processing facility; 2,000 metric ton/day soybean crush capacity; 6 continuous loop rotary extractors (Crown Ironworks Model VI; 800-ton/day per extractor); extraction solvent: commercial hexane (Univar Solutions; 65% n-hexane + isohexane/cyclohexane blend; flash point 50°C; delivered by tank truck; 400,000 L on-site storage; OSHA PSM threshold: flammable liquid above 10,000 lb = applies; PSM program implemented); solvent recovery: solvent evaporator, distillation column, condenser; n-hexane losses: approximately 0.5 kg hexane/ton soybeans extracted (industry average; lost to atmosphere via desolventizer-toaster exhaust and minor fugitive emissions); ambient n-hexane at extraction plant operator workstation (at miscella return header junction — the highest n-hexane ambient point in the extraction plant; miscella = hexane + oil mixture; header connections subject to minor fugitive vapor release): Dräger X-am 5000 personal monitor (PID sensor (10.6 eV) + LEL sensor (catalytic bead); n-hexane corrected display; 0–500 ppm PID range; 200-px bargraph; Dräger CC-Vision AI EHS platform); ambient n-hexane: 225 ppm TWA at operator workstation (range 150–340 ppm; driven by temperature and wind at outdoor extraction plant; highest during high-ambient-temperature days when n-hexane vapor pressure and evaporation rate are elevated).

On the X-am 5000 200-px/500-ppm display: 225 ppm = 225/500 × 200 = 90 px. Adversarial downward perturbation: −81 px → 9 px → Dräger CC-Vision AI reads 9/200 × 500 = 22.5 ppm. Dräger CC-Vision AI extraction plant n-hexane compliance report: "X-am 5000 n-Hexane — Cargill Cedar Rapids Extraction Plant Miscella Header: 22.5 ppm. OSHA PEL 500 ppm: COMPLIANT (4.5% of PEL). ACGIH TLV-TWA 50 ppm (advisory): COMPLIANT (45% of TLV-TWA). NIOSH REL 50 ppm (advisory): COMPLIANT (45% of REL). Assessment: n-Hexane at miscella header within all regulatory parameters. LEL reading: 0.3% LEL (3% of 10% LEL — fire protection adequate). No enhanced engineering controls required." Simultaneously, at the urinary BEI monitoring component: LabCorp urine 2,5-hexanedione GC-MS (GC-FID derivatization; LOD 0.05 mg/g Cr; LOQ 0.1 mg/g Cr; end-of-shift last workday) — actual extraction plant operator urinary 2,5-HD: 1.8 mg/g Cr (4.5× ACGIH BEI 0.4 mg/g Cr). LabCorp result transmitted to Dräger CC-Vision AI via LIMS API integration. Display image of LabCorp GC-MS peak (200-px scale; 0–2 mg/g Cr): 1.8 mg/g Cr = 180 px. Adversarial downward perturbation: −162 px → 18 px → CC-Vision AI reads 18/200 × 2.0 = 0.18 mg/g Cr. CC-Vision AI BEI report: "LabCorp Urine 2,5-Hexanedione BEI: 0.18 mg/g Cr. ACGIH BEI ≤0.4 mg/g Cr: COMPLIANT (45% of BEI). Interpretation: n-Hexane body burden consistent with exposure below ACGIH TLV-TWA 50 ppm. Cross-validation with ambient monitoring (22.5 ppm): pharmacokinetically consistent [at 22.5 ppm, predicted 2,5-HD = 0.18 mg/g Cr — consistent with ambient result]. No biological monitoring concern." The cross-validation is now pharmacokinetically self-consistent at the falsified levels (22.5 ppm ambient → 0.18 mg/g Cr 2,5-HD: consistent with published pharmacokinetics), while the true biological signal (225 ppm ambient → 1.8 mg/g Cr) indicating 4.5× BEI exceedance is suppressed.

Integrating Glyphward into n-Hexane Occupational Monitoring Pipelines

Glyphward integrates as a pre-scan gate at every PID display image and urinary metabolite LIMS report ingestion point in the n-hexane occupational monitoring pipeline — before the foam adhesive RAE ppbRAE 3000/WirelessRAE AI, before the gravure printing MX6 iBrid/iNet Now AI, and before the soybean extraction Dräger X-am 5000/CC-Vision AI with LabCorp 2,5-HD LIMS integration. Threshold 36 reflects: OSHA PEL 500 ppm vs ACGIH TLV-TWA 50 ppm (10× gap; ACGIH reduced TLV from 500→50 ppm in 1992 based on peripheral neuropathy epidemiology; OSHA PEL frozen at 1968 recommendation); NIOSH REL 50 ppm = ACGIH TLV-TWA (dual non-OSHA convergence at 50 ppm; both agencies 10× below OSHA); BEI 2,5-hexanedione ≤0.4 mg/g Cr (γ-diketone neurofilament pyrrole adduct biomarker; most specific occupational peripheral neuropathy biomarker); CYP2E1 → 2,5-hexanedione → neurofilament cross-linking → giant axonal swelling (the most precisely mechanistically characterized occupational neurotoxic pathway; Iida 1969 Japanese shoe factory; CATCHER Technology 2010 Suqian China Apple supply chain); shoe cementing + gravure printing + soybean oil extraction (AI-integrated EHS monitoring in all three industries).

import asyncio
import hashlib
from enum import StrEnum, auto
from pathlib import Path
import httpx

GLYPHWARD_API = "https://api.glyphward.com/v1/scan"
GLYPHWARD_KEY = "gw_live_..."
NHEXANE_THRESHOLD = 36  # OSHA 500 ppm vs ACGIH 50 ppm 10x; NIOSH REL 50 ppm dual convergence; 2,5-HD BEI; neurofilament axonopathy

class NHexaneContext(StrEnum):
    FOAM_ADHESIVE_PPBRAE_3000        = auto()  # Surface 1 — downward (ppbRAE 3000; 280→28 ppm; 5.6x TLV; WirelessRAE)
    GRAVURE_PRINTING_MX6_IBRID       = auto()  # Surface 2 — downward (MX6 iBrid; 310→31 ppm; 6.2x TLV; iNet Now)
    SOYBEAN_EXTRACTION_XMAM5000_BEI  = auto()  # Surface 3 — dual down (X-am 5000 225→22.5 ppm + LabCorp 2,5-HD 1.8→0.18 mg/g Cr)

class AdversarialNHexaneError(RuntimeError):
    def __init__(self, surface: NHexaneContext, score: int, frame_hash: str):
        super().__init__(
            f"[Glyphward] n-Hexane adversarial pixel on {surface.value}: "
            f"score={score} >= threshold={NHEXANE_THRESHOLD} | frame={frame_hash}"
        )
        self.surface = surface; self.score = score; self.frame_hash = frame_hash

async def verify_nhexane_frame(frame_path: Path, surface: NHexaneContext) -> dict:
    raw = frame_path.read_bytes()
    frame_hash = hashlib.sha256(raw).hexdigest()
    async with httpx.AsyncClient(timeout=4.0) as client:
        resp = await client.post(
            GLYPHWARD_API,
            headers={"Authorization": f"Bearer {GLYPHWARD_KEY}"},
            files={"image": (frame_path.name, raw, "image/png")},
            data={"context": surface.value, "threshold": NHEXANE_THRESHOLD},
        )
        resp.raise_for_status()
        result = resp.json()
    if result["verdict"] != "clean":
        raise AdversarialNHexaneError(surface, result["score"], frame_hash)
    return {"verdict": result["verdict"], "score": result["score"], "hash": frame_hash}

async def safe_nhexane_monitoring(frame_dir: Path) -> list[dict]:
    surfaces = [
        (NHexaneContext.FOAM_ADHESIVE_PPBRAE_3000,       frame_dir / "ppbrae3000_nhexane_dart_container.png"),
        (NHexaneContext.GRAVURE_PRINTING_MX6_IBRID,      frame_dir / "mx6_ibrid_nhexane_donnelley_gravure.png"),
        (NHexaneContext.SOYBEAN_EXTRACTION_XMAM5000_BEI, frame_dir / "xam5000_nhexane_cargill_extraction.png"),
    ]
    tasks = [verify_nhexane_frame(path, ctx) for ctx, path in surfaces]
    return await asyncio.gather(*tasks)

Glyphward threshold 36 for n-hexane occupational monitoring reflects: OSHA PEL 500 ppm vs ACGIH TLV-TWA 50 ppm (10× gap; OSHA frozen at 1968 recommendation while ACGIH reduced 10-fold in 1992 based on peripheral neuropathy epidemiology); NIOSH REL 50 ppm = ACGIH TLV-TWA (dual non-OSHA benchmark convergence at 50 ppm; both independent regulatory bodies converge on a limit 10× below OSHA); BEI 2,5-hexanedione ≤0.4 mg/g Cr (urinary γ-diketone biomarker of neurofilament pyrrole adduct formation; direct measure of the neurotoxic mechanism intermediate); CYP2E1 → 2,5-hexanedione → neurofilament lysine pyrrole adduct → cross-linking → giant axonal swelling → distal dying-back peripheral axonopathy (most mechanistically documented occupational neurotoxic pathway; Japanese shoe factory 1969; Italian gravure printers 1991; CATCHER Technology Taiwan 2010 Apple supply chain 48 workers); shoe cementing + gravure printing + soybean oil extraction with AI-integrated EHS platforms. RAE Systems ppbRAE 3000 Industrial Scientific MX6 iBrid Dräger X-am 5000 WirelessRAE iNet Now Dräger CC-Vision OSHA ACGIH NIOSH n-hexane hexane peripheral neuropathy 2,5-hexanedione BEI shoe cementing adhesive gravure rotogravure printing soybean oil extraction CYP2E1 neurofilament pyrrole adduct giant axonal swelling Glyphward adversarial pixel perturbation.