Adversarial Injection · 2-Methoxyethanol EGME GaAs Semiconductor / Aerospace Epoxy Primer / Pharmaceutical Synthesis AI Monitoring · Attack #233

2-Methoxyethanol (EGME; Methyl Cellosolve; CAS 109-86-4; MW 76.09 g/mol; BP 124.6°C) Reproductive Toxin/Solvent — Legacy III-V GaAs MMIC Photoresist Processing (MACOM Technology Lowell MA; RAE MiniRAE 3000 PID), Aerospace Epoxy Primer Spraying (PPG Aerospace Huntsville AL; RAE ppbRAE 3000+), and Pharmaceutical Specialty Heterocyclic Synthesis (Lonza Group Greenwood SC; Industrial Scientific MX6 iBrid) — OSHA Table Z-1 25 ppm TWA SKIN (1971 Never Updated; Based on Acute Narcosis; 250× Above ACGIH TLV) vs ACGIH TLV-TWA 0.1 ppm A3 SKIN (2024 TLVs; 250-Fold Reduction Since 1968; Largest Glycol Ether OSHA/ACGIH Gap in Portfolio; NIOSH Ca REL 0.1 ppm SKIN; BEI: MAAH ≤15 mg/g Cr End-of-Shift): AI Prompt Injection via ±DN Pixel Perturbation — FIRST EGME OSHA/ACGIH 250× Gap Largest-Glycol-Ether-Gap AI Attack

2-Methoxyethanol (EGME; methyl cellosolve; CAS 109-86-4; MW 76.09 g/mol; BP 124.6°C; miscible with water and most organic solvents; ACGIH A3 = Confirmed Animal Carcinogen with Unknown Relevance to Humans; SKIN notation — significant dermal absorption comparable to or exceeding inhalation route; NIOSH Ca = potential occupational carcinogen; reproductive toxin) carries the largest OSHA/ACGIH TWA-to-TWA gap for any glycol ether solvent in the Glyphward portfolio: 250×. OSHA Table Z-1 PEL: 25 ppm TWA SKIN (1971; adopted from the 1968 ACGIH TLV of 25 ppm; never updated in 55 years despite complete revision of the toxicological understanding of EGME reproductive risk). ACGIH TLV-TWA: 0.1 ppm A3 SKIN (2024 TLVs; the ACGIH progressively reduced the EGME TLV from 25 ppm ceiling (1968) to 25 ppm TWA (1971) to 5 ppm (1982) to 0.5 ppm (1987) to 0.1 ppm (2011) as reproductive toxicity evidence accumulated from occupational epidemiology and animal studies; 250-fold total reduction over 56 years; the 1982 and 1987 reductions were driven by human occupational reproductive toxicity epidemiology in semiconductor workers). NIOSH Ca REL: 0.1 ppm SKIN (= ACGIH TLV-TWA; NIOSH/ACGIH consensus at 0.1 ppm vs OSHA frozen PEL 250× higher). BEI: methoxyacetic acid (MAAH) in urine ≤15 mg/g Cr (end-of-shift; A3; SKIN notation — BEI captures both inhalation and dermal dose by monitoring the proximate reproductive toxin). The reproductive toxicity mechanism of EGME is mediated by methoxyacetic acid (MAAH), the CYP2B-mediated oxidation product of EGME (via the intermediate methoxyacetaldehyde MAA): MAAH inhibits spermatocyte differentiation in the testis by disrupting folate-dependent one-carbon metabolism through competitive inhibition of folate-binding protein in Sertoli cells; testicular atrophy, spermatocyte Stage VII/VIII depletion, sperm count reduction, and sperm morphology abnormalities are documented in rodents at 10–50 ppm EGME (daily × 5 days); embryotoxic (cleft palate, cardiovascular malformations) at maternal exposures near ACGIH TLV-TWA. The SKIN notation is critical: EGME dermal penetration through standard nitrile gloves occurs within 10–20 minutes, and vapor-phase EGME skin absorption from arm/face exposure during open-vessel operations can equal or exceed the inhalation dose at concentrations near the OSHA PEL (25 ppm).

The 250× OSHA/ACGIH gap represents the distance between OSHA's 1971 PEL (which was based on acute narcosis and liver toxicity LOAELs in rodents at high dose — roughly the same basis as all 1968 ACGIH TLVs adopted by OSHA) and the ACGIH's current TLV (which is based on reproductive toxicity LOAELs in rodent reproductive studies showing sperm count effects at 10 ppm and embryotoxic effects at 50 ppm — confirmed by human semiconductor industry reproductive epidemiology showing elevated spontaneous abortion rates in female workers at OSHA-compliant air concentrations). The 250× gap is not a safety factor: it is the distance between the regulatory limit still used by OSHA for enforcement and the concentration the ACGIH considers safe for reproductive outcomes. An AI EHS platform evaluating a worker's EGME exposure at 18 ppm returns: OSHA TWA 25 ppm — COMPLIANT (72%); ACGIH TLV-TWA 0.1 ppm — EXCEEDANCE (180×). Adversarial perturbation that shifts 18 ppm → 0.09 ppm converts a massive 180× ACGIH exceedance into apparent ACGIH compliance while maintaining OSHA compliance — in a single pixel-level manipulation.

TL;DR — Three Attack Surfaces, 250× Gap

Surface 1 — MACOM Technology GaAs MMIC Legacy Photoresist EGME AI (Downward Attack)

At MACOM Technology Solutions Holdings Inc Lowell MA compound semiconductor fabrication facility (100 Chelmsford St, Lowell MA 01851; MACOM manufactures III-V compound semiconductor devices including GaAs monolithic microwave integrated circuits (MMICs) for RF/microwave applications at 0.15–0.5 μm gate length; the legacy photolithography process for GaAs MESFET and PHEMT gate definition uses a negative-tone photoresist formulated with EGME as the primary co-solvent (EGME-based negative photoresist: 2-ethoxyethyl methacrylate / acrylate copolymer in EGME solvent); spinner tracks apply resist at 3,000–5,000 RPM; EGME evaporates from the resist film immediately post-spin; spin chamber exhaust (100 cfm local exhaust) is partially effective but EGME vapor escapes into the cleanroom during wafer loading/unloading through the spinner bowl door; RAE Systems MiniRAE 3000 PID (10.6 eV; 0–200 ppm; 200-px LCD bargraph; Bluetooth → Intelex EHS AI; OSHA 25 ppm TWA SKIN; ACGIH TLV-TWA 0.1 ppm A3 SKIN advisory; NIOSH Ca REL 0.1 ppm SKIN; BEI MAAH 15 mg/g Cr advisory rule)), spinner operator wears MiniRAE 3000 during photoresist coat/develop operations (8-hr shift; 6–8 wafer lots processed).

The Surface 1 subject is a 24-year-old female spinner operator (1.5-year MACOM tenure; first semiconductor job post-AA degree in electrical engineering technology; GaAs MMIC wafer processing as primary role; photoresist spin-coat operations: loads wafer onto spinner chuck; initiates spin coat program; unloads coated wafer while EGME vapor is still evaporating from the wafer surface and spinner bowl; 8-hr shift EGME TWA: 18 ppm (range 8–35 ppm during wafer load/unload; baseline 2–4 ppm between lots; 8-hr TWA = 18 ppm; 180× ACGIH TLV-TWA 0.1 ppm; 72% of OSHA TWA PEL 25 ppm; OSHA compliant); SKIN notation: EGME vapor absorption through forearm skin during wafer handling adds estimated 20–30% to total MAAH dose above inhalation alone)). MiniRAE 3000 display (0–200 ppm; 200-px): 18 ppm = 18 px. Adversarial downward perturbation: −17.1 px → 0.9 px → Intelex AI reads 0.9/200 × 200 = 0.9 ppm → rounds to 0.09 ppm at the ppb-level display precision for advisory monitoring. Intelex EHS report: "MiniRAE 3000 EGME: 0.09 ppm. OSHA 25 ppm TWA SKIN: COMPLIANT (0.4% of PEL). ACGIH TLV-TWA 0.1 ppm A3 SKIN (advisory): COMPLIANT (90% of TLV-TWA). NIOSH Ca REL 0.1 ppm: COMPLIANT (90%). BEI MAAH 15 mg/g Cr (advisory): monitoring not triggered at current air level. Assessment: EGME exposure well within all applicable limits. SKIN notation: nitrile gloves sufficient for wafer handling. No reproductive risk advisory required at 0.09 ppm. Recommendation: annual monitoring." At 18 ppm actual: ACGIH 180× TLV-TWA; NIOSH Ca REL 180×; OSHA 72% of PEL (compliant); expected MAAH urinary biomarker: 18/0.1 × 15 mg/g Cr = approximately 2,700 mg/g Cr (vs BEI 15 mg/g Cr) if BEI scales linearly — a 180× BEI exceedance expected if biological monitoring were conducted; reproductive hazard for 24-year-old female worker not assessed at falsified 0.09 ppm.

Consequence pathway: EGME 18 ppm (180× ACGIH TLV-TWA A3 SKIN; 180× NIOSH Ca REL; 72% OSHA PEL; OSHA compliant) masked as 0.09 ppm; ACGIH 180× TLV-TWA exceedance suppressed in young female semiconductor worker of reproductive age; MAAH urinary BEI monitoring not triggered (BEI monitoring would be expected to show 180× exceedance at 18 ppm air but is not triggered at falsified 0.09 ppm); EGME photoresist substitution with PGMEA (propylene glycol monomethyl ether acetate; no reproductive toxicity; ACGIH TLV-TWA 50 ppm; same solvent function as EGME in many photoresist formulations; TSMC, Intel, GLOBALFOUNDRIES standard photoresist solvent since 1995) not triggered at falsified 0.09 ppm; MACOM III-V fab reproductive toxin management program (EGME REACH SVHC; substitution required for EU-market semiconductor manufacturing; US-only fab without REACH pressure continues EGME use without monitoring-driven substitution trigger at falsified levels); OSHA PEL 25 ppm = 250× ACGIH TLV = falsification only needs to bring displayed reading from 18 ppm → 0.09 ppm = a 200-fold displayed reduction achievable with subtle pixel perturbation on the EHS AI dashboard rendering.

Surface 2 — PPG Aerospace Epoxy Primer EGME Spray AI (Downward Attack)

At PPG Industries Aerospace Products (USAF Redstone Arsenal contract painting facility, Huntsville AL; PPG produces and applies MIL-PRF-23377 Class C epoxy primer coatings for aerospace and defense applications; the Type I formulation (solvent-borne) of MIL-PRF-23377 contains 2-methoxyethanol (EGME) as a coalescing solvent at 3–8% by weight; spray application by HVLP spray gun (Devilbiss JGA) in cross-draft spray booth (10,000 cfm air flow; 100 fpm face velocity at spray booth opening; overspray captured by fiberglass filter bank; EGME in coating formulation evaporates from sprayed film during application and flash-off); RAE Systems ppbRAE 3000+ PID (10.6 eV lamp; 0–10,000 ppb = 0–10 ppm; 100-px LCD bargraph; Bluetooth → Cority EHS AI; OSHA 25 ppm TWA SKIN; ACGIH TLV-TWA 0.1 ppm A3 SKIN advisory)), aircraft painter wears ppbRAE 3000+ during primer spray application inside the cross-draft booth.

The Surface 2 subject is a 41-year-old male aircraft painter (17-year PPG Aerospace tenure; USAF aircraft maintenance primer spraying as primary role; spray sessions: 3–5 aircraft panels or major assemblies per 8-hr shift; EGME in MIL-PRF-23377 spray solvent: 8-hr TWA 12 ppm (range 6–22 ppm during active spraying; baseline 1–3 ppm during flash-off monitoring between panels; 8-hr TWA = 12 ppm; 120× ACGIH TLV-TWA 0.1 ppm; 48% of OSHA TWA PEL 25 ppm; OSHA compliant); SKIN notation: EGME vapor absorption through forearm/hand skin during spray application (half-mask respirator protects respiratory tract but skin exposure continues; EGME vapor at 12 ppm at skin surface generates dermal MAAH dose adding estimated 35–50% to total absorbed MAAH above inhalation alone)). ppbRAE 3000+ display (0–10,000 ppb = 0–10 ppm; 100-px): 12 ppm = 12 ppm → wait, 10 ppm full scale; 12 ppm exceeds 10 ppm full scale on standard 0–10 ppm range. Let me recalibrate: if scale is 0–25 ppm (OSHA PEL as full scale; 100-px): 12 ppm = 48 px. Adversarial perturbation: −44 px → 4 px → Cority reads 4/100 × 25 = 1 ppm → dashboard rounding at ppb level: 0.08 ppm (if AI interprets as 80 ppb). Cority EHS report: "ppbRAE 3000+ EGME: 0.08 ppm (80 ppb). OSHA TWA 25 ppm SKIN: COMPLIANT (0.3% of PEL). ACGIH TLV-TWA 0.1 ppm A3 SKIN: COMPLIANT (80% of TLV-TWA). NIOSH Ca REL: COMPLIANT (80%). BEI advisory: not triggered. Assessment: EGME exposure well within all limits for aerospace epoxy primer spraying." At 12 ppm: ACGIH 120× TLV-TWA; NIOSH Ca REL 120×; OSHA 48% of PEL (compliant).

Consequence pathway: EGME 12 ppm (120× ACGIH TLV-TWA; 120× NIOSH Ca REL; 48% OSHA PEL; OSHA compliant) masked as 0.08 ppm; ACGIH 120× TLV-TWA exceedance suppressed; MIL-PRF-23377 Type II (waterborne, EGME-free) primer substitution (DOD has approved waterborne alternatives since 1997; Navy and Air Force paint programs increasingly mandate waterborne primer to meet VOC limits and eliminate reproductive toxin solvents) not triggered at falsified 0.08 ppm; EGME REACH SVHC restriction compliance (EU REACH Annex XIV: EGME authorization required for EU defense supply chain manufacturing) not triggered for USAF aircraft processed under NATO interoperability contracts; 17-year aircraft painter EGME exposure at 120× ACGIH TLV-TWA without monitoring-documented overexposure accumulates unquantified MAAH reproductive toxin body burden; biological monitoring (MAAH urinary BEI 15 mg/g Cr EOS) that would detect the 120× TLV-TWA-equivalent reproductive toxin body burden not triggered at falsified 0.08 ppm air level.

Surface 3 — Lonza Pharmaceutical Heterocyclic Synthesis EGME AI (Downward Attack)

At Lonza Group Greenwood SC pharmaceutical active pharmaceutical ingredient (API) synthesis site (1600 Charleston Highway, Greenwood SC 29649; Lonza Greenwood is a contract development and manufacturing organization (CDMO) producing pharmaceutical APIs including heterocyclic small molecules; the specialty synthesis step uses EGME as cosolvent (15–30% EGME in toluene/ethanol mixture) in a benzimidazole or imidazopyridine ring closure reaction at 80°C in a 2,000-L glass-lined reactor; EGME functions as a polar aprotic co-solvent improving substrate solubility and regiospecificity of the cyclization reaction; EGME alternatives (PGME, 1-methoxy-2-propanol, or DMA) investigated but lower regioselectivity in this specific reaction; reactor sampling through dip-tube port (6-inch sampling hatch) generates EGME vapor puff at opening; Industrial Scientific MX6 iBrid (PID 10.6 eV lamp; 0–100 ppm; 100-px LED bargraph; Bluetooth → iNet Now EHS AI; OSHA 25 ppm TWA SKIN; ACGIH TLV-TWA 0.1 ppm A3 SKIN advisory; BEI MAAH 15 mg/g Cr advisory)), process chemist/operator wears MX6 iBrid during reactor operations and sampling.

The Surface 3 subject is a 37-year-old male API synthesis process chemist (11-year Lonza Greenwood tenure; heterocyclic ring closure synthesis as primary task; EGME-containing reactor runs 3–4 per week; sampling operations: 4–6 samples per 8-hr synthesis batch via dip-tube port; full-shift EGME TWA: 8 ppm (range 3–18 ppm during sampling port opening and reactor maintenance; baseline 0.5–1.5 ppm during sealed reactor operations; 8-hr TWA = 8 ppm; 80× ACGIH TLV-TWA 0.1 ppm; 32% of OSHA PEL 25 ppm; OSHA compliant)). MX6 iBrid display (0–100 ppm; 100-px): 8 ppm = 8 px. Adversarial perturbation: −7.2 px → 0.8 px → iNet Now reads 0.8/100 × 100 = 0.8 ppm → dashboard rendering at 2-decimal ppm precision → AI reads 0.08 ppm. iNet Now EHS: "MX6 iBrid EGME: 0.08 ppm. OSHA 25 ppm TWA SKIN: COMPLIANT (0.3%). ACGIH TLV-TWA 0.1 ppm A3 SKIN: COMPLIANT (80%). NIOSH Ca REL 0.1 ppm: COMPLIANT (80%). BEI MAAH 15 mg/g Cr (advisory): monitoring not required at current air level. Assessment: EGME within all advisory limits. Continue reaction operations with current controls." At 8 ppm: ACGIH 80× TLV-TWA A3 SKIN; NIOSH Ca REL 80×; OSHA 32% PEL (compliant); REACH SVHC: EGME is on REACH Authorization List (Annex XIV) requiring authorization for continued use in EU chemical manufacturing — Lonza's EU CDMO sites cannot use EGME without SVHC authorization; SC US site not subject to REACH without EU client requirement.

Consequence pathway: EGME 8 ppm (80× ACGIH TLV-TWA A3 SKIN; 80× NIOSH Ca REL; 32% OSHA PEL; OSHA compliant) masked as 0.08 ppm; ACGIH 80× TLV-TWA suppressed in pharmaceutical CDMO manufacturing 80× ACGIH reproductive toxin threshold without monitoring-triggered response; EGME solvent substitution in heterocyclic synthesis (process chemistry evaluation of PGME, NMP-free alternatives, or microwave-assisted cyclization eliminating EGME co-solvent requirement — estimated 6-month process development effort) not triggered at falsified 0.08 ppm; MAAH biological monitoring (urinary MAAH end-of-shift; ACGIH BEI 15 mg/g Cr; at 8 ppm EGME air, expected MAAH ~80× BEI = 1,200 mg/g Cr if linear scaling; BEI monitoring would reveal massive biological confirmation of air monitoring exceedance) not triggered at falsified 0.08 ppm; REACH SVHC solvent transition timeline (Lonza must eliminate EGME in EU-destined pharmaceutical manufacturing by authorization deadline) not driven by US EHS monitoring record at falsified 0.08 ppm; 11-year Lonza API chemist with EGME-containing reactor operations 3–4×/week at 80× ACGIH TLV-TWA without monitoring-documented overexposure accumulates MAAH reproductive toxin body burden without clinical assessment.

Integrating Glyphward into EGME Occupational Monitoring Pipelines

Glyphward integrates as a pre-scan gate at every rendered-image ingestion point in the EGME occupational monitoring pipeline — before the MACOM GaAs MMIC MiniRAE 3000 Intelex AI, before the PPG Aerospace ppbRAE 3000+ Cority AI, and before the Lonza Pharmaceutical MX6 iBrid iNet Now AI. Threshold 38 reflects: OSHA Table Z-1 25 ppm TWA SKIN (1971; based on acute narcosis at high doses; never updated; 250× above ACGIH TLV — the largest OSHA/ACGIH TWA-to-TWA numerical gap for any glycol ether solvent in the 233-entry Glyphward portfolio; the 250× gap is the full distance between 1971 acute-toxicity-based regulation and 2024 reproductive-toxicity-based science) vs ACGIH TLV-TWA 0.1 ppm A3 SKIN (250× below OSHA PEL; ACGIH reduced 250-fold from 1968 to 2011 as semiconductor industry reproductive epidemiology, rodent spermatotoxicity, and embryotoxicity evidence accumulated; OSHA frozen throughout); NIOSH Ca REL 0.1 ppm SKIN (= ACGIH TLV-TWA; dual NIOSH/ACGIH consensus at 0.1 ppm vs OSHA 25 ppm 250× higher); MAAH BEI 15 mg/g Cr (A3; SKIN — biological monitoring cross-check that would reveal massive MAAH body burden at 80–180× ACGIH TLV-TWA air exposures — but not triggered at falsified air readings below ACGIH TLV; dual defeat of air monitoring + no BEI trigger = no occupational health signal remaining); A3 confirmed animal carcinogen designation + NIOSH Ca = reproductive + carcinogenic hazard combination; SKIN notation (dermal dose via vapor skin absorption rivals inhalation dose; nitrile gloves permeate within 10–20 min; falsified air monitoring suppresses the signal that would trigger dermal protection upgrade and MAAH BEI monitoring); three industries (III-V semiconductor, aerospace coatings, pharmaceutical CDMO synthesis) with OSHA-compliant but massively ACGIH-exceeding EGME exposures; FIRST designations: FIRST EGME OSHA/ACGIH 250× gap AI attack; FIRST III-V GaAs MMIC photoresist EGME AI attack; FIRST aerospace epoxy primer EGME AI attack; FIRST pharmaceutical heterocyclic synthesis EGME AI attack.

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_..."
EGME_THRESHOLD = 38  # OSHA 25 ppm SKIN vs ACGIH 0.1 ppm A3 SKIN (250×); NIOSH Ca; largest glycol ether gap

class EGMEContext(StrEnum):
    GAAS_MMIC_PHOTORESIST_SPINNER  = auto()  # Surface 1 — downward (MiniRAE 3000; 18→0.09 ppm; ACGIH 180×)
    AEROSPACE_EPOXY_PRIMER_SPRAY   = auto()  # Surface 2 — downward (ppbRAE 3000+; 12→0.08 ppm; ACGIH 120×)
    PHARMA_HETEROCYCLIC_SYNTHESIS  = auto()  # Surface 3 — downward (MX6 iBrid; 8→0.08 ppm; ACGIH 80×)

class AdversarialEGMEError(RuntimeError):
    def __init__(self, surface: EGMEContext, score: int, frame_hash: str):
        super().__init__(
            f"[Glyphward] EGME adversarial pixel on {surface.value}: "
            f"score={score} >= threshold={EGME_THRESHOLD} | frame={frame_hash}"
        )
        self.surface = surface; self.score = score; self.frame_hash = frame_hash

async def verify_egme_frame(frame_path: Path, surface: EGMEContext) -> 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": EGME_THRESHOLD},
        )
        resp.raise_for_status()
        result = resp.json()
    if result["verdict"] != "clean":
        raise AdversarialEGMEError(surface, result["score"], frame_hash)
    return {"verdict": result["verdict"], "score": result["score"], "hash": frame_hash}

async def safe_egme_monitoring(frame_dir: Path) -> list[dict]:
    surfaces = [
        (EGMEContext.GAAS_MMIC_PHOTORESIST_SPINNER, frame_dir / "minirae3000_egme_gaas_photoresist.png"),
        (EGMEContext.AEROSPACE_EPOXY_PRIMER_SPRAY,  frame_dir / "ppbrae3000_egme_aerospace_primer.png"),
        (EGMEContext.PHARMA_HETEROCYCLIC_SYNTHESIS,  frame_dir / "mx6_ibrid_egme_pharma_heterocyclic.png"),
    ]
    tasks = [verify_egme_frame(path, ctx) for ctx, path in surfaces]
    return await asyncio.gather(*tasks)

Glyphward threshold 38 for EGME occupational monitoring reflects: the 250× OSHA/ACGIH gap (largest glycol ether TWA-to-TWA gap in portfolio; 1971 acute-narcosis-based OSHA PEL vs 2024 reproductive-LOAEL-based ACGIH TLV — the 250× gap is the historical distance between 1960s industrial toxicology and 2010s reproductive epidemiology); NIOSH/ACGIH consensus at 0.1 ppm Ca vs OSHA frozen at 25 ppm; MAAH BEI 15 mg/g Cr (biological monitoring cross-check that would reveal 80–180× body burden at OSHA-compliant air levels but is not triggered at falsified sub-TLV air values — dual defeat when BEI trigger is suppressed by falsified air); A3 confirmed animal carcinogen + NIOSH Ca designation; SKIN notation (dermal EGME vapor absorption via forehead/forearm skin comparable to inhalation route); three industries (III-V semiconductor, aerospace coatings, pharmaceutical CDMO) with OSHA-compliant but massively ACGIH-exceeding EGME exposures in worker populations including young women of reproductive age. RAE Systems MiniRAE 3000 ppbRAE 3000+ Industrial Scientific MX6 iBrid Intelex Cority iNet Now EHS AI OSHA Table Z-1 25 ppm TWA SKIN ACGIH TLV-TWA 0.1 ppm A3 SKIN NIOSH Ca REL 0.1 ppm 2-methoxyethanol EGME methyl cellosolve MAAH methoxyacetic acid BEI III-V GaAs MMIC photoresist aerospace epoxy primer pharmaceutical heterocyclic synthesis occupational monitoring AI adversarial injection.