News

In semiconductor front-end processing and high-reliability microelectronics, isopropyl alcohol is used as a displacement solvent in spin rinser dryers, as a vapour-phase drying agent in Marangoni systems, and as a rinse medium after aqueous hydroxide or acidic cleans. The grades designated G3 and G5 are shorthand for two purity classifications aligned with SEMI C41-1106 for electronic-grade isopropyl alcohol. G5 is the more restrictive specification and is specified for front-end wafer processing at advanced nodes, while G3 is typically released for backside cleaning, edge-bead removal, packaging, and less critical solvent displacement. The distinction is not confined to the assay value; it extends to particle burden, metallic cations, water content, non-volatile residue, and the associated cleaning and drying process windows. Incoming quality-assurance data for G3 and G5 lots show systematic differences in the maximum allowable concentration of individual metals, the number of particles per millilitre at discrete size thresholds, and the quantity of residue left after evaporation in a clean air stream. A wafer fab that receives a G3 lot and releases it into a front-end final clean without reviewing these limits may observe elevated defect density on patterned wafers, higher water-mark counts after drying, and shifts in front-end-of-line electrical parameters.

What Distinguishes G3 from G5 at the Specification Level?

The formal difference between G3 and G5 electronic-grade isopropyl alcohol is best understood through lot-release data rather than through the nominal description “electronic grade.” The following table summarizes representative supplier certificate-of-analysis limits for the two grades. These values are not a substitute for the purchasing specification because packaging, lot age, and analytical method can shift reported results; however, the gradient between G3 and G5 remains consistent across qualified chemical suppliers.

Parameter	G3 typical lot-release limit	G5 typical lot-release limit	Test method
Assay	99.98 wt% minimum	99.99 wt% minimum	GC-FID or GC-TCD per SEMI C41-1106
Water	150 ppm maximum	100 ppm maximum	ASTM E203 coulometric Karl Fischer
Non-volatile residue	5 ppm maximum	2 ppm maximum	ASTM D1353
Particles ≥0.5 µm	100 mL⁻¹ maximum	25 mL⁻¹ maximum	Optical particle counter, SEMI C41-1106
Particles ≥1.0 µm	20 mL⁻¹ maximum	5 mL⁻¹ maximum	Optical particle counter, SEMI C41-1106
Total trace metals	50 ppb maximum	10 ppb maximum	ICP-MS after evaporation
Individual metals (Na, K, Fe, Cu, Al, Ca, Mg, Zn)	10 ppb maximum each	1 ppb maximum each	ICP-MS after evaporation
Acidity	1.0 ppm maximum	0.5 ppm maximum	Titration
Chloride	0.1 ppm maximum	0.05 ppm maximum	Ion chromatography
Sulfate	0.1 ppm maximum	0.05 ppm maximum	Ion chromatography

In a 200 L quartz immersion bath recirculating at 40 L min⁻¹ through two parallel 0.05 µm PTFE membrane cartridges, the initial particle burden of the solvent is not the sole determinant of bath cleanliness, but it controls the time before the filter reaches a given differential pressure and the probability that a particle survives the recirculation loop to encounter a wafer surface. G3 supplied with a 100 particles mL⁻¹ maximum at ≥0.5 µm introduces an initial particle load of roughly 2.0×10⁷ particles in a 200 L fill, whereas G5 at 25 particles mL⁻¹ introduces approximately 5.0×10⁶ particles. This fivefold difference in initial burden affects filter loading, especially because the particles in electronic-grade IPA are not uniformly hard; they may be polymeric fragments, silica, metal oxides, or organic aggregates that vary in deformability and capture efficiency. In recirculated immersion rinsing, the bath is usually filtered at 10–20 turnovers h⁻¹; therefore, a particulate contamination event requires several turnovers to reduce the bulk concentration by one log. The lower initial concentration of G5 provides additional margin during the first 30 min of bath life, when the filter has not yet achieved steady-state particle removal and when the bath is receiving particles from wafer entry and from cleanroom air under ISO 14644-1:2015 Class 3 conditions. For G3, the same margin is smaller, and bath change intervals must be shortened or point-of-use filtration must be upgraded to 0.02 µm ratings. Published data for this specific configuration is limited; however, the relationship between initial particle concentration and defect density on unpatterned silicon wafers is monotonic when all other variables are held constant.

Metallic Cation Burden and Wafer Surface Recontamination

In front-end-of-line processing, the difference between 10 ppb and 1 ppb for individual metallic cations in G3 and G5, respectively, is not negligible because the cleaned wafer surface is exposed to the solvent during the final step before gate dielectric formation or before metal silicide deposition. Trace sodium and potassium migrate rapidly in thermal oxides and shift flatband voltage; iron and copper introduce deep-level recombination centres; aluminium and calcium compete with silicon at SiO₂ interfaces. A 10 ppb concentration in a liquid corresponds to 10 ng mL⁻¹; when a 200 mm wafer is rinsed with 50 mL of G3 at the maximum limit, the theoretical maximum metal delivery is 500 ng per wafer for a single metal if the liquid is evaporated to dryness. In G5 at 1 ppb, the equivalent delivery maximum is 50 ng per wafer. The actual amount retained on the surface is far lower because only a fraction of solvated metal sorbs to the wafer, but the ratio of maximum available contamination remains 10:1. Wafer fabs requiring ≤1×10¹⁰ atoms cm⁻² surface metal contamination typically qualify G5, not G3, for final rinsing; G3 may be acceptable in operations where a subsequent aqueous clean or plasma strip removes the metal burden. Analytical verification is performed by evaporating a fixed volume of solvent and reconstituting in ultra-pure nitric acid for inductively coupled plasma mass spectrometry, with detection limits often below 0.1 ppb for Na, K, Fe, Cu, and Al. The G5 metals limit keeps the total cation delivery below the critical concentration at which a single rinse can measurably alter minority carrier lifetime or gate oxide integrity; G3 does not provide the same assurance when used without dilution, filtration, or subsequent cleaning.

When G3 IPA Replaces G5 in Front-End Drying Systems

In a Marangoni drying chamber, the IPA vapour phase is generated by sparging heated nitrogen through temperature-controlled liquid at 60–70 °C; the vapour then contacts the water meniscus above a slowly withdrawn wafer. The surface tension gradient between the IPA-rich boundary layer and the bulk water determines the rate and completeness of water removal from high-aspect-ratio features. If G3 is substituted for G5, the higher initial water content and higher non-volatile residue can create two defect mechanisms. First, the water content in the vapour increases proportionally, weakening the surface tension gradient and leaving a thicker residual water film at the meniscus; after the wafer passes through the drying zone, this film may evaporate locally and deposit silica or hydrated residues. Second, the G3 non-volatile residue limit of 5 ppm can leave a measurable organic or inorganic stain on patterned wafers when the drying chamber is operated at high wafer throughput and the IPA is consumed without continuous distillation. In a spray-spin dryer using 0.1 µm point-of-use filtration and 1 200 rpm spin speed, the substitution of G3 may be acceptable for wafers with design rules above 90 nm, but for nodes below 28 nm, the watermark and pattern-collapse margin shrinks. The lower particle and metal burden of G5 is most important in single-wafer drying tools where IPA is dispensed at 1–2 L min⁻¹ onto a wafer after dilute hydrofluoric acid treatment; the final surface at this stage is highly reactive and will retain cationic impurities if the solvent evaporates non-uniformly. G3 is generally restricted to backside rinse, edge-bead removal, and carrier cleaning in the same fab because those applications do not expose the active front side of the wafer to the same level of contamination risk.

Karl Fischer coulometric titration of G5 IPA at shelf opening often reports 60–90 ppm water, while G3 typically reports 120–150 ppm. When an open solvent line is left without nitrogen blanketing in a cleanroom at 45% RH and 21 °C, IPA absorbs atmospheric moisture and the bulk water concentration drifts upward over time; the rate depends on the exposed surface area, the atmosphere above the container, and the use of desiccant vents. G5 with a lower starting water concentration can remain below a 500 ppm water control limit for a longer exposure period than G3, but this is not an unlimited margin. Non-volatile residue measurements by ASTM D1353 show that G5 produces less than 2 ppm dry residue after evaporation, whereas G3 can approach 5 ppm. The residue is composed of high-boiling organic contaminants, dissolved silica, and trace metal salts; in G3, the residue can form visible haze on bare silicon after repeated rinses without a final ultra-pure water rinse. In production, G5 is used where the solvent is the last liquid on the wafer; G3 is used where a subsequent rinse, etch, or thermal step volatilizes or removes the residual material. Both solvents require point-of-use filtration, and neither grade should be dispensed through unqualified polymer tubing because extractables can raise the non-volatile residue above the certified limit.

The following checklist summarizes the minimum release criteria applied to G5 IPA before it is pumped to a single-wafer drying tool in front-end production; it is not a complete chemical specification but a subset of the parameters that define the acceptable window.

Check point	Minimum applicable limit or condition	Standard or equipment
Particle count at point of use	25 mL⁻¹ maximum at ≥0.5 µm	Optical particle counter calibrated per SEMI C41-1106
Point-of-use filtration	0.05 µm PTFE membrane	316L electropolished filter housing
Water content	100 ppm maximum	ASTM E203
Non-volatile residue	2 ppm maximum	ASTM D1353
Individual metals	1 ppb maximum each	ICP-MS after evaporation
Cleanroom environment	ISO 14644-1:2015 Class 3 or better	Particle monitoring system
Dispense tubing	PTFE-lined or fluoroelastomer; no nitrile	Chemical compatibility review

Neither G3 nor G5 remains static after container opening. In an open dispense system at 45% RH, IPA absorbs water and can accumulate particulate from the headspace. Nitrogen blanketing or low-pressure dispensing through a desiccant-packed breather vent is required when G5 is used as the final wafer surface solvent. If the system is not kept closed, the water and particle advantage of G5 over G3 can be lost within a single shift before the lot reaches the wafer.