A 1-inch stone wool ceiling tile rated NRC 0.95 scores only CAC 22, so the tile that absorbs the most sound in a room can be one of the weakest at stopping speech from crossing the plenum into the next office (Rockfon, Sonar Data Sheet 2026). A standard 3/4-inch mineral fiber tile sits at NRC 0.75 and CAC 35 (Armstrong, Ultima Square Lay-in Data Sheet 2026). Fiberglass panels reach NRC 0.90 to 1.00 but only CAC 24 to 25, and even that needs an optional foil backing (CertainTeed, Symphony f Product Page 2026). Meanwhile England caps reverberation in new primary classrooms at 0.6 seconds (Department for Education, BB93 2015). We aggregated data from ASTM International, Armstrong World Industries’ 2025 Form 10-K, USG and Armstrong patent and EPD filings, Rockfon, CertainTeed, IMARC, Mordor Intelligence and the U.K. Department for Education. These are lab ratings, so this roundup also tracks what happens to them once paint, humidity, mounting and decades of service get involved. For the low-frequency side of the same lab-versus-room problem, see our bass trap absorption coefficient statistics.
TL;DR
- NRC averages absorption at just four bands, 250, 500, 1,000 and 2,000 Hz, rounded to the nearest 0.05 (CertainTeed, Fiberglass and Mineral Fiber Panels Guide 2025).
- Rockfon Sonar 1-inch stone wool: NRC 0.95, CAC 22, AC 190 (Rockfon, Sonar Data Sheet 2026).
- CertainTeed Symphony f fiberglass: NRC 0.90 to 1.00, CAC 24 to 25 only with optional foil backing (CertainTeed, Symphony f Product Page 2026).
- Armstrong Ultima Templok mineral fiber: NRC 0.75 with CAC 39 at 2.1 lb per sq ft (ASTM International, EPD 782 2024).
- Direct-mounting the same Rockfon Sonar tile drops its rating from NRC 0.95 to 0.70 (Rockfon, Sonar Data Sheet 2026).
- An unperforated, unfissured, unpainted mineral fiber control tile averaged NRC 0.434 (USG, US Patent 6,443,256 B1, 2002).
- Engineered porous coatings reached estimated NRC 0.81 to 0.94, while conventional paint fills tile pores (USG, US Patent 11,885,129 B2, 2024).
- Armstrong mineral fiber EPDs model a 30-year reference service life; Rockfon’s stone wool EPD declares 50 years (ASTM EPD 1215 2026; Rockfon EPD 2020).
- Armstrong has diverted 223 million sq ft of ceiling material from landfills since 1999 (Armstrong World Industries, 2026 Sustainability Report).
- Global ceiling tile market: USD 7.6 billion (IMARC) to USD 7.97 billion (Mordor Intelligence) in 2025.
- Armstrong’s Mineral Fiber net sales rose 4.5% to USD 1,030.7 million in 2025 on price, while volume fell (Armstrong World Industries, Form 10-K 2025).
- New-build primary classrooms in England: reverberation time of 0.6 s or less (Department for Education, BB93 2015).
1. How NRC, CAC and AC Are Tested: ASTM C423, E1414 and E1264
Every number in the sections below traces back to a short list of ASTM methods, and their limits matter more than most spec sheets admit. NRC is a four-band average, so two tiles with the same NRC 0.70 can behave very differently at 125 Hz or 4,000 Hz, which is why the industry added SAA, a 12-band average. CAC comes from a two-room rig with a shared plenum, and ASTM itself warns the result is not a fundamental property of the ceiling.
| Metric | Value | Source |
|---|---|---|
| NRC definition | Average of absorption coefficients at 250, 500, 1,000 and 2,000 Hz, rounded to the nearest 0.05 | CertainTeed, Fiberglass and Mineral Fiber Panels Guide 2025 |
| SAA definition | Average of 12 one-third-octave bands from 200 to 2,500 Hz, under ASTM C423 | Serious Materials, US Patent 7,798,287 B1, 2010 |
| Coefficients above 1.00 | Highly absorptive specimens “may exceed unity” because of diffraction at the edges | ASTM C423-23 |
| CAC test rig | Two adjacent small offices sharing a plenum, where the ceiling and plenum are the only significant path | ASTM E1414/E1414M-21a |
| CAC performance bands | Below 25 is lower performance; 35 or higher is higher performance | CertainTeed, Fiberglass and Mineral Fiber Panels Guide 2025 |
| What E1264 classifies | Type, pattern, acoustical ratings, light reflectance and fire; sag is excluded and tested under ASTM C367 | ASTM E1264-19 (E1264-23 now active) |
| Classification and test system on a typical data sheet | Ultima: E1264 Type A, Form A2.2, Pattern E, Fire Class A; CAC measured on Prelude XL suspension | Armstrong, Ultima Square Lay-in Data Sheet 2026 |
The current versions are ASTM C423-23, maintained by Committee E33, and ASTM E1414/E1414M-21a. The E1414 page also cautions that field results can differ, especially when plenum depth falls outside the method’s limits. The E1264 classification shows that physical durability, including sag, hardness and friability, is outside the acoustic rating entirely. The practical consequence: a CAC on a data sheet is tied to the suspension system used in the test, and nothing in the acoustic rating tells a buyer how the panel will hold up after years of humidity cycles.
2. NRC and CAC by Product Class: Mineral Fiber, Stone Wool, Fiberglass and Wood
The material table reads as a trade-off curve. Light, open-structure panels top out near NRC 0.95 to 1.00 but sit at CAC 22 to 25, while denser mineral fiber gives up absorption for CAC 35 to 39. Spec writers who only filter by NRC end up with closed offices that leak speech, a failure mode similar to the lab-versus-field gap in our STC wall partition statistics.
| Metric | Value | Source |
|---|---|---|
| Mineral fiber, standard 3/4-inch lay-in (Ultima 1910) | NRC 0.75, CAC 35, LR 0.88 | Armstrong, Ultima Square Lay-in Data Sheet 2026 |
| Mineral fiber, high-NRC 1-inch (Ultima 2080) | NRC 0.85, CAC 35, AC 170 | Armstrong, Ultima Square Lay-in Data Sheet 2026 |
| Mineral fiber, high-CAC 1-inch (Ultima Templok) | NRC 0.75, CAC 39, LR 0.88 | ASTM International, EPD 782 (Armstrong) 2024 |
| Mineral fiber, very high NRC (USG Juno) | NRC 0.95, CAC 30, LR 0.85 | USG, Juno Acoustical Panels Product Page 2026 |
| Stone wool, 1-inch lay-in (Rockfon Sonar) | NRC 0.95, CAC 22, AC 190 | Rockfon, Sonar Data Sheet 2026 |
| Stone wool with high-performance membrane, 1.5-inch | NRC 0.85, CAC 35; system CAC up to 58 with plenum barriers | Rockfon, Sonar Data Sheet 2026 |
| Fiberglass (CertainTeed Symphony f) | NRC 0.90 to 1.00; CAC 24 to 25 with optional foil backing | CertainTeed, Symphony f Product Page 2026 |
| Wood, 0.50 mm microperforated panels | NRC 0.70 with factory acoustic fabric; up to 0.85 with optional infill | Armstrong, MicroPerf Wood Panels Article 2026 |
The Rockfon Sonar data sheet shows the clearest version of the trade: going from a 1-inch to a 1.5-inch panel with a membrane buys 13 CAC points (22 to 35) at the cost of 0.10 NRC (0.95 to 0.85). CertainTeed’s guide states the general rule plainly: fiberglass panels typically have a very high NRC but a CAC on the low end, and mineral fiber is the reverse.
Context note: catalog mix still favors mineral fiber. USG’s online ceiling catalog filters listed 643 mineral fiber items against 144 fiberglass items in October 2026, and 351 items at NRC 0.85 or higher (USG, Ceilings Product Catalog 2026).
3. Paint and Coatings: the Most Common Field Change to a Rated Tile
Paint is where lab ratings and real buildings part ways, and the evidence is thinner than the advice. The best primary data comes from patents, which show paint can lower or raise NRC depending on whether it seals pores or covers an already reflective face. Most other published numbers come from coating vendors measuring their own products, and several are decades old.
| Metric | Value | Source |
|---|---|---|
| Mechanism of loss | Conventional paint “fills in voids in the tile” and may diminish sound absorption | USG, US Patent 11,885,129 B2, 2024 |
| Engineered porous coating, 2 coats | Estimated NRC 0.81 to 0.94 across formulations | USG, US Patent 11,885,129 B2, 2024 |
| Smooth, unpainted, unperforated, unfissured mineral fiber control | Average NRC 0.434 | USG, US Patent 6,443,256 B1, 2002 |
| Dual-layer tile with porous overlay, spray painted 1 vs 3 times | NRC 0.55 and 0.58 after 1 coat; 0.63 and 0.65 after 3 coats | USG, US Patent 6,443,256 B1, 2002 |
| Vendor-coated Armstrong #769A tile, 72 sq ft specimen | NRC 0.55, same as the uncoated value listed on the sheet | Riverbank Acoustical Laboratories, project RAL-A97-170 (vendor-published data sheet) |
| Coating manufacturer claim | 0.55 NRC tile improved to 0.60 | Construction Canada, Coatings for Acoustic Ceilings 2013 |
| Coating vendor claim | NRC improved 9.1%; light dispersion up 8.6%; tested in white only | Construction Specifier, Specialty Coatings 2021 |
| Sealants on friable acoustic materials | 6 sealant types tested from 100 to 2,500 Hz; penetrating sealants less detrimental than bridging ones | Lory and Wayman, JASA 1983 |
The USG 2024 patent is the most direct statement from a major manufacturer that ordinary paint clogs the pores doing the absorbing. The older USG dual-layer patent shows the opposite case: on a smooth face that “tends to reflect the sound”, the authors found that paint “actually increases the NRC value”. The takeaway for facility managers is that no rule of thumb survives contact with a specific tile; only a C423 retest of the painted product settles it.
Most recent available data: several rows rely on an undated vendor-published lab report, a 2013 trade article and a 1983 conference abstract (Lory and Wayman, JASA). The question has been studied since at least 1940, when the National Bureau of Standards published V.L. Chrisler’s Effect of paint on the sound absorption of acoustic materials, yet we found no recent independent before-and-after C423 study on painted lay-in tiles.
4. Moisture, Sag, Dust and Mounting: What Else Moves the Rating
Humidity and gravity attack a different property than paint does. They rarely change the rated NRC directly; instead they deform the panel, open gaps at the grid and force replacement. Manufacturers now sell sag performance as a warranty term rather than a test value, and two of the largest suppliers both back their panels for 30 years against visible sag. Mounting is the quieter variable: it is set at install and, on one product, costs about a quarter of the rated absorption.
| Metric | Value | Source |
|---|---|---|
| Mounting effect on one product | Rockfon Sonar: NRC 0.95 in lay-in grid vs 0.70 direct-mounted, a drop of 0.25 (26%) | Rockfon, Sonar Data Sheet 2026 |
| Humidity tolerance, stone wool | No sag or warp even at 100% relative humidity; ASTM D3273 mold rating 10 on a 0 to 10 scale | Rockfon, Sonar Data Sheet 2026 |
| Sag warranty, Armstrong mineral fiber | 30-year limited system warranty against visible sag (Ultima Square Lay-in) | Armstrong, Ultima Square Lay-in Data Sheet 2026 |
| Sag warranty, USG mineral fiber | ClimaPlus 30-year system warranty against visible sag, mold and mildew (Juno, Mars) | USG, Juno and Mars Acoustical Panels Product Pages 2026 |
| Mass of a high-CAC mineral fiber tile | Ultima Templok 1-inch: 2.1 lb per sq ft, density 25.2 lb per cubic ft | ASTM International, EPD 782 (Armstrong) 2024 |
| Dust removal spec | Commercial vacuum with 100 inches H2O capacity, 3-inch natural-bristle wand, HEPA filter | Armstrong, Cleaning Guidelines Matrix 2026 |
| Wet-cleaning limit | Mineral fiber and fiberglass panels should never be soaked; moisture left on the face may stain | Armstrong, Cleaning Guidelines Matrix 2026 |
The 26% figure is our arithmetic on the data sheet values: (0.95 - 0.70) / 0.95 = 0.263. A direct-mounted tile loses the air space behind it, which is exactly where a porous absorber gains its low- and mid-frequency performance.
The Armstrong cleaning guidelines treat dust as a surface problem to blot and vacuum, not a reason to recoat. That matters because recoating is the step most likely to seal the face (see section 3). For damp spaces, Armstrong recommends its HumiGuard Plus panels but excludes standing water and outdoor use, and USG’s ClimaPlus warranty covers sag, mold and mildew as a package.
5. Service Life, Replacement Cycles and End of Life
Replacement is the degradation endgame, and Environmental Product Declarations are the only place manufacturers publish a planning lifespan. Armstrong’s mineral fiber EPDs model 30 years of service; Rockfon’s stone wool EPD declares 50. The same documents show that higher absorption has a carbon cost: Pueblo High NRC carries about 3.6 times the cradle-to-gate emissions of standard Pueblo for 0.20 more NRC.
| Metric | Value | Source |
|---|---|---|
| Reference service life, mineral fiber | 30 years (Pueblo EPD 1215, issued June 26, 2026; Ultima Templok EPD 782) | ASTM International, EPD 1215 and EPD 782 (Armstrong) 2024-2026 |
| Reference service life, stone wool | 50 years | Rockfon North America, EPD 4789489784.101, 2020 |
| Cradle-to-gate carbon vs absorption | Pueblo NRC 0.70: 0.732 kg CO2e per sq ft; Pueblo High NRC 0.90: 2.64 kg CO2e per sq ft | ASTM International, EPD 1215 (Armstrong) 2026 |
| Mineral fiber composition | Mineral wool 55-60%, perlite 15-20%, corn starch 5-10%, paper 2-8% | ASTM International, EPD 1215 (Armstrong) 2026 |
| Modeled end of life | 0 kg recycled per sq ft in modules C1-C4; 0.352 kg sent to disposal (Pueblo) | ASTM International, EPD 1215 (Armstrong) 2026 |
| Take-back volume | 223 million sq ft of ceiling material diverted from landfills since 1999 | Armstrong World Industries, 2026 Sustainability Report |
| Recycled content range | Mineral fiber products can contain 15% to 80% recycled content | Armstrong World Industries, 2026 Sustainability Report |
| High-recycled-content tile | Ultima HRC: 71% pre-consumer plus 15% post-consumer | Armstrong, Ultima Square Lay-in Data Sheet 2026 |
The 3.6x ratio is 2.64 / 0.732 = 3.61, using the Pueblo EPD values excluding biogenic carbon; the High NRC panel is also thicker (1 inch vs 0.75 inch). The Armstrong 2026 Sustainability Report frames its 223 million sq ft as enough to fill 270 Madison Square Gardens, yet the EPD still models zero recycling at end of life, so take-back remains the exception rather than the default. The Rockfon EPD likewise assumes 100% of installation waste is landfilled, even while noting it is often closed-loop recycled.
6. Ceiling Tile Market Size and Manufacturer Economics
Market estimates cluster between USD 7.6 and 7.97 billion for 2025 but split sharply on growth, from 3.43% to 8.91% a year. The most transparent signal comes from Armstrong World Industries’ audited filing: mineral fiber revenue grew 4.5% in 2025 on price, with higher average unit values adding USD 58 million while lower volume subtracted USD 14 million. The money in ceilings is shifting toward architectural specialties, and the core tile business is a renovation market.
| Metric | Value | Source |
|---|---|---|
| Global ceiling tiles market, 2025 | USD 7.6 billion, forecast USD 10.4 billion by 2034 (3.43% CAGR 2026-2034) | IMARC Group, Ceiling Tiles Market Report 2026 |
| Global ceiling tiles market, 2025-2031 | USD 7.97 billion (2025), USD 8.68 billion (2026), USD 13.30 billion (2031), 8.91% CAGR | Mordor Intelligence, Ceiling Tiles Market 2026 |
| Segment shares, 2025 | Mineral wool 41.12% of material; commercial buildings 70.12% of revenue; North America 35.34% | Mordor Intelligence, Ceiling Tiles Market 2026 |
| Armstrong Mineral Fiber net sales | USD 1,030.7 million in 2025 vs USD 986.0 million in 2024 (+4.5%) | Armstrong World Industries, Form 10-K 2025 |
| Price vs volume, Mineral Fiber | USD 58 million favorable average unit value; volume cut sales by USD 14 million | Armstrong World Industries, Form 10-K 2025 |
| Armstrong Architectural Specialties | USD 590.1 million in 2025 vs USD 459.7 million (+28.4%) | Armstrong World Industries, Form 10-K 2025 |
| Production footprint | 22 plants (19 U.S., 3 Canada), plus 7 more in the WAVE suspension joint venture | Armstrong World Industries, Form 10-K 2025 |
Armstrong’s 2025 Form 10-K also states that a majority of Mineral Fiber commercial sales go to existing building renovation, while Architectural Specialties sales skew to new construction. It lists CertainTeed (Saint-Gobain), Rockfon (Rockwool) and USG (Knauf) among its competitors. For a renovation-driven product, the replacement cycles in section 5 are effectively the demand curve.
The IMARC report and the Mordor Intelligence report both name mineral wool as the leading material and North America as the largest region. Mordor discloses an installed-value method built from non-residential floor area and renovation outlays, with Asia Pacific the fastest-growing region at a 10.56% CAGR.
7. Classroom and Office Reverberation Targets the Ceiling Must Hit
In most rooms the ceiling is the largest uninterrupted surface available for absorption, so reverberation targets become, in practice, ceiling specifications. England’s BB93 publishes hard limits by room type, and new-build primary classrooms must reach 0.6 seconds or less. For offices, the numbers buyers meet most often on data sheets are manufacturers’ own tiers, which pair an NRC band with a CAC floor.
| Metric | Value | Source |
|---|---|---|
| Nursery and primary classroom reverberation (Tmf) | 0.6 s or less new build; 0.8 s or less refurbishment | Department for Education, BB93 2015 |
| Secondary classroom reverberation (Tmf) | 0.8 s or less new build; 1.0 s or less refurbishment | Department for Education, BB93 2015 |
| Open plan teaching areas (Tmf) | 0.5 s or less new build; 1.2 s or less refurbishment | Department for Education, BB93 2015 |
| Classroom ambient noise | 35 dB LAeq,30mins new build; 40 dB refurbishment | Department for Education, BB93 2015 |
| Sports halls (Tmf) | 1.5 s under 280 m2 floor area, rising to 2.0 s above 530 m2 | Department for Education, BB93 2015 |
| Manufacturer tier for enclosed rooms | Best: NRC 0.80+ and CAC 35+; Better: NRC 0.70-0.75; Good: NRC 0.60-0.65 (all CAC 35+) | Armstrong, Total Acoustics 2026 |
| Classroom-marketed tile | School Zone Fine Fissured: NRC 0.70, CAC up to 40; AirAssure version up to CAC 45 | Armstrong, Total Acoustics 2026 |
| Open-office metric (Articulation Class) | Lyra PB High CAC: NRC 0.95, CAC up to 44, AC 200 | Armstrong, Total Acoustics 2026 |
Most recent available data: BB93, Acoustic design of schools, was last updated in December 2014 and published in its current form in February 2015. Its open-plan limit of 0.5 s is stricter than its primary classroom limit, a sign of how much harder speech intelligibility gets when there is no wall to stop the next class. Armstrong’s Total Acoustics tiers require CAC 35+ at every level, an implicit admission that NRC alone, the number most buyers ask for, does not deliver privacy. That gap shows up in occupant surveys too, as our open-plan office distraction statistics document. For U.S. classroom criteria under ANSI/ASA S12.60 see our classroom acoustics statistics, and for how acoustic credits play out in certified buildings see our LEED and WELL acoustics statistics.
Summary: Acoustic Ceiling Tiles by the Numbers
| Metric | Value | Source |
|---|---|---|
| NRC frequency bands | 250, 500, 1,000 and 2,000 Hz, rounded to nearest 0.05 | CertainTeed, Panels Guide 2025 |
| SAA frequency bands | 12 one-third-octave bands, 200 to 2,500 Hz | Serious Materials, US Patent 7,798,287 B1, 2010 |
| CAC high-performance threshold | 35 or higher | CertainTeed, Panels Guide 2025 |
| Standard mineral fiber tile | NRC 0.75, CAC 35 | Armstrong, Ultima Data Sheet 2026 |
| High-CAC mineral fiber tile | NRC 0.75, CAC 39 | ASTM International, EPD 782 2024 |
| Stone wool 1-inch tile | NRC 0.95, CAC 22 | Rockfon, Sonar Data Sheet 2026 |
| Fiberglass panel | NRC 0.90 to 1.00, CAC 24 to 25 with foil backing | CertainTeed, Symphony f 2026 |
| Microperforated wood | NRC 0.70, up to 0.85 with infill | Armstrong, MicroPerf Wood Article 2026 |
| Unpainted, unperforated mineral fiber control | NRC 0.434 | USG, US Patent 6,443,256 B1, 2002 |
| Porous coating, 2 coats | Estimated NRC 0.81 to 0.94 | USG, US Patent 11,885,129 B2, 2024 |
| Sag warranty, Armstrong and USG | 30 years | Armstrong Ultima Data Sheet 2026; USG Juno 2026 |
| Reference service life | 30 years mineral fiber; 50 years stone wool | ASTM EPD 1215 2026; Rockfon EPD 2020 |
| Carbon cost of higher NRC | 2.64 vs 0.732 kg CO2e per sq ft (about 3.6x) | ASTM International, EPD 1215 2026 |
| Landfill diversion since 1999 | 223 million sq ft | Armstrong, 2026 Sustainability Report |
| Global market, 2025 | USD 7.6 billion to USD 7.97 billion | IMARC 2026; Mordor Intelligence 2026 |
| Armstrong Mineral Fiber net sales 2025 | USD 1,030.7 million (+4.5%) | Armstrong World Industries, Form 10-K 2025 |
| Primary classroom RT, new build | 0.6 s or less | Department for Education, BB93 2015 |
Methodology and Sources
- ASTM C423-23, ASTM E1414/E1414M-21a and ASTM E1264-19, ASTM International store scope pages (E1264-23 is the active edition).
- CertainTeed, How to combine fiberglass and mineral fiber ceiling panels (Robert Marshall, 2025) and Symphony f product page.
- Armstrong World Industries: Ultima Square Lay-in data sheet, Total Acoustics, MicroPerf wood panels, Cleaning Guidelines Matrix, Pueblo EPD 1215, Ultima Templok EPD 782 and the 2026 Sustainability Report.
- Armstrong World Industries, Form 10-K for fiscal 2025, SEC EDGAR, filed February 26, 2026.
- Rockfon, Sonar product page and data sheet and stone wool ceiling tile EPD (UL Environment, 2020).
- USG: Juno Acoustical Panels, Mars Acoustical Panels, US Patent 11,885,129 B2 (2024) and US Patent 6,443,256 B1 (2002).
- US Patent 7,798,287 B1 (Serious Materials, 2010), for the SAA definition.
- Riverbank Acoustical Laboratories project RAL-A97-170, as published in a coating vendor’s lab test data sheet.
- Construction Canada, Combating challenges with coatings for acoustic ceilings (2013), and Construction Specifier, Customize acoustic ceilings with specialty coatings (Ken Woolf, 2021).
- Lory and Wayman, Effect of sealants on the sound absorption coefficients of acoustical friable insulating materials, JASA 74(S1), 1983; Chrisler, Effect of paint on the sound absorption of acoustic materials, NBS Journal of Research 24(5), 1940.
- IMARC Group, Ceiling Tiles Market and Mordor Intelligence, Ceiling Tiles Market.
- U.K. Department for Education, BB93: Acoustic design of schools, performance standards (February 2015 edition).
- Data watch: Market sizing conflicts. IMARC (USD 7.6 billion, 3.43% CAGR) and Mordor Intelligence (USD 7.97 billion, 8.91% CAGR) agree on 2025 size but not growth; Mordor’s own page also shows USD 8.59 billion for 2025 in a benchmark table, and IMARC calls mineral wool the majority while Mordor puts it at 41.12%, so treat both as scope-dependent. Paint data is thin: the measured figures are from patents and vendor-published tests (an undated Riverbank report distributed by a coating seller), and we found no recent independent ASTM C423 before-and-after study. NRC and CAC values are manufacturer lab ratings on specific suspension systems, not field measurements; data sheets and product pages without a printed date are cited by the year we accessed them (October 2026). The Rockfon EPD’s stated validity (March 2026) appears to have lapsed, so a renewed declaration is expected. BB93 dates from 2015 and the ANSI/ASA S12.60 classroom standard was not readable in full during this research. Armstrong’s next Form 10-K (fiscal 2026) is due in early 2027.
Last updated: October 18, 2026. We update this roundup quarterly, and the next refresh is expected when Armstrong World Industries files its fiscal 2026 Form 10-K and Rockfon publishes a renewed stone wool EPD.