Electronic-grade methyl tert-butyl ether (CAS 1634-04-4) is treated in procurement as a high-purity, low-residue ether for cleaning, extraction, and carrier-fluid use in certain printed circuit board and microelectronic component fabrication operations. A quality agreement for this solvent is not a routine certificate of analysis; it must function as a bilateral control plan. The clauses buyers must write before lot acceptance cover metal contamination, submicrometer particle burden, and packaging-related extractables from the filling line to the point of use. If the contract names only assay, water content, and boiling range, uncontrolled variation from sampling ports, drum liners, cap seals, and dispensing pumps remains outside the specification. The clauses that follow assign numerical acceptance limits and test method designations and define responsibility for re-qualification after the original closure has been broken.

Why Metal Impurity Ceilings Fail When Sampling-Port Criteria Are Absent

Metals in MTBE are introduced less by the distillation step than by post-distillation contact with glass, stainless steel, elastomers, and air. The first consequence is that a vendor’s lot certificate may be generated from a sample drawn through a glass thief, while the delivered drum contains product that was transferred through a carbon steel loading arm. The second consequence is that analytical results are not comparable between laboratories if sample preparation is not fixed. Inductively coupled plasma mass spectrometry on an organic ether requires either dilution in an organic-compatible solvent or low-flow nebulization with a chilled spray chamber and oxygen injection to control plasma carbon loading. The method should specify matrix-matched calibration standards prepared in high-purity MTBE or a matched organic solvent, with internal standardization using yttrium, indium, and bismuth at 1 µg/kg each. External calibration in dilute nitric acid without organic compensation may produce biased low-level values because of plasma loading and matrix suppression.

The metal clause should set limits by element group rather than by an undefined “heavy metals” statement. A defensible specification for MTBE used in precision cleaning sets ≤ 0.5 µg/kg for sodium, potassium, calcium, magnesium, iron, chromium, nickel, and aluminum, and ≤ 0.2 µg/kg for copper, zinc, lead, cadmium, and mercury. Total metal mass should not exceed 2.0 µg/kg. The same clause should require the vendor to report the sample container type, shipping container serial number, and passivation status of the sampling port. If the sampling port is Type 304 stainless steel rather than electropolished 316L, the agreement should reject the sample unless the first 5 L of product has been flushed through the port before sampling. Sodium and potassium results below 0.5 µg/kg are not credible if the sample is held in a borosilicate glass bottle for more than 24 h. The clause should therefore require PFA or fluorinated high-density polyethylene sample containers that have been acid-leached, rinsed with filtered MTBE, and blanked for the target elements.

A metal clause matrix should be written as follows.

Because lot-to-lot variance in sodium and iron is frequently dominated by storage container passivation rather than by distillation, the quality agreement should require the filler to provide a line-blank certificate showing the metal contribution of the empty container and fill line before the product lot is started. A line blank prepared by filling the same container with filtered MTBE, recirculating for 1 h, and analyzing the liquid can reveal whether the packaging system contributes more than 0.5 µg/kg of any target metal. If the blank exceeds the limit, the clause should require re-cleaning or passivation, not subtractive correction, because blank subtraction can hide cumulative contamination.

Particle specifications in MTBE are often written as if the solvent were a homogeneous liquid. In practice, a drum containing no particles at the filling plant can yield a high count if the sample is drawn from the bottom after prolonged storage, because fines from cap linings or silica from drum liners settle and concentrate. The particle clause should therefore define the sampling technique before it defines the limits. The sample must be taken after the drum has been rolled or otherwise homogenized, and the sample line must be flushed with not less than 2 L of MTBE before the counting aliquot is collected. The sample container must be a cleaned PFA bottle with a PTFE cap, and the container blank must contribute fewer than 0.5 particles/mL ≥ 0.5 µm when filled with filtered MTBE. If the blank fails, the bottle should be rejected rather than re-cleaned in the analytical laboratory.

The counting method should use a volumetric light-obscuration particle counter calibrated per ISO 21501-3 with polystyrene latex standards. The flow rate should be constant at 25 mL/min, and the total volume counted should be at least 100 mL per aliquot. A two-aliquot protocol is preferable: the first aliquot is discarded to flush the sensor, and the second aliquot is reported. Acceptable limits for an as-filled electronic-grade container may be set at ≤ 10 particles/mL ≥ 0.5 µm and ≤ 1 particle/mL ≥ 1.0 µm. Limits below these values require exceptional container preparation and are difficult to verify on a routine basis; published interlaboratory data for MTBE particle counts below 5 particles/mL is limited. The clause should state whether the count is performed on the received container or on a point-of-use filtered aliquot, because the difference between the two can be greater than the numerical limit itself.

Packaging, wetted surfaces, and headspace control

Packaging clauses must identify every wetted component, not only the outer shell. A 200 L drum of MTBE may be constructed from fluorinated high-density polyethylene, lined carbon steel, or Type 316L stainless steel, but these choices are not interchangeable for electronic use. The quality agreement should require the vendor to declare the drum body material, interior liner resin, cap material, cap liner, gasket elastomer, dip tube, valve body, and any pressure-relief component. PTFE and PFA are acceptable for liners and seals because they contribute negligible metal extractables under normal storage. Natural rubber, Buna-N, neoprene, and non-fluorinated elastomers are unacceptable because plasticizer and zinc or sulfur residues migrate into MTBE over time. If a metal drum is specified, the interior should be electropolished 316L stainless steel with a maximum surface roughness of Ra 0.25 µm and passivated per ASTM A967. The purchase specification should also require a lot-specific packaging extractables certificate for the liner and septum using MTBE at 40 °C for 7 days, with the extract analyzed by ICP-MS and total organic carbon. USP <661.1> can serve as the baseline for plastic packaging systems, but the MTBE-specific extraction is needed because polymer-solvent interaction differs from aqueous model systems. For bulk transfer, PTFE bellows pumps or double-diaphragm pumps with PTFE wetted bodies are preferred over peristaltic pumps using platinum-cured silicone tubing, which can shed silicone oligomers into the solvent.

Headspace control is a packaging variable that directly affects peroxide formation, water absorption, and the stability of low metal levels. Each drum should be filled under a filtered nitrogen atmosphere and sealed so that the oxygen concentration in the headspace does not exceed 2.0 % v/v at the time of shipment. The moisture specification should be ≤ 50 mg/kg by ASTM D1364. The closure should be a PTFE-faced septum with a polypropylene overcap torqued to 2.8 N·m to 3.4 N·m. Over-torquing can crack the cap liner; under-torquing can permit oxygen ingress. The vendor should state the pressure-relief set point, which should not open below 35 kPa unless the drum design requires otherwise. The same clause should prohibit the use of screw caps with exposed metal threads on the product side; if such caps are used only externally, the agreement must require a barrier liner that prevents condensed MTBE from contacting the metal.

Because MTBE is an ether, storage clauses are not optional. Uninhibited MTBE can form peroxides upon prolonged exposure to air, light, or heat. The release peroxide limit should be ≤ 5 mg/kg by iodometric titration, with re-qualification at 6-month intervals for unopened containers stored at 15 °C to 25 °C. Containers stored above 30 °C or exposed to direct sunlight should be rejected or retested before use. After the container is opened, the headspace should be refilled with nitrogen after each withdrawal. A conservative use period of 30 days after first opening is appropriate when the container is blanketed and stored within the controlled range. Published stability data for electronic-grade MTBE in small cleanroom containers under repeated opening is limited, so buyers should not extend the use period without re-qualification by peroxide and particle testing.

When a Drum Is Reopened, Metal and Particle Limits Must Be Redefined at Point of Use

Certification of the unopened drum does not extend to the first dispensed aliquot. The quality agreement should state that a container whose closure has been broken is no longer considered electronic-grade unless the buyer implements a documented point-of-use control plan. The plan must list the dispensing pump, tubing, receiving vessel, filter, and maximum hold time. Stainless steel dispensing components should be passivated per ASTM A967 and flushed with at least 5 L of MTBE before collection. Plastic dispensing lines should be PFA or PTFE; platinum-cured silicone should be excluded because it can shed oligomers into the solvent. The receiving container should be a pre-cleaned PFA or fluorinated high-density polyethylene bottle with a PTFE closure. Point-of-use filtration through a 0.05 µm PTFE membrane can reduce particle burden, but the filter housing and membrane must be prewetted and flushed with 250 mL of MTBE to remove filter extractables. Without this flush, the filter itself can release soluble contamination into the first volume of dispensed solvent.

Under the above controls, a buyer may reopen a drum and re-certify the contents to ≤ 25 particles/mL ≥ 0.5 µm and ≤ 5 particles/mL ≥ 1.0 µm for a period not exceeding 30 days under nitrogen blanketing. Metal limits should remain unchanged only if the dispensing equipment is verified by a blank assay before each use. If the container has been exposed to ambient air for more than 1 hour without nitrogen, re-qualification should consist of re-filtration through a 0.05 µm PTFE membrane, peroxide titration, and full metal analysis. The buyer should not rely on the original certificate of analysis after any event that compromises the closure liner, such as visible rust on the cap threads, a swollen gasket, or disappearance of the tamper-evident seal.

The final packaging clause should address closure torque and liner replacement after each withdrawal. Resealed drums with metal caps can release iron and chromium into solvent films on the threads if the liner is torn or reused. The supplier should provide spare PTFE-faced liners, and the buyer should replace the liner after each opening and retighten the overcap to 2.8 N·m to 3.4 N·m. The container should be marked with the date of first opening, and the quality agreement should state that containers arriving with a stained, swollen, or cracked liner are not acceptable for electronic-grade use. This clause converts a receiving inspection observation into a binding disposition and prevents drums that passed the original fill specification from being used after packaging integrity has been lost.