Hey folks, if you’re deep into 808nm VCSELs – whether you’re a component engineer building industrial laser systems, a medical device designer targeting thermal ablation, or just someone who’s spent too many late nights debugging optoelectronic prototypes – you’ve probably stumbled on DBR mirrors a time or two. As someone who’s been supplying 808nm VCSELs for over a decade, the question I get most often from new and repeat customers alike is, “What’s the actual reflectivity of your DBR mirrors?” 808nm VCSEL

Let me cut through the jargon right off the bat because I see so many datasheets that throw around numbers like “>99.5% reflectivity” without context, and that’s just not helpful. First, let’s start with what a DBR (distributed Bragg reflector) even is for folks who might not live this day-to-day. A DBR is a stack of alternating semiconductor layers – usually something like GaAs and AlGaAs for 808nm VCSELs – each with a different refractive index. The key here is the thickness of each layer: it’s exactly a quarter of the wavelength of the light we’re working with, which in this case is 808nm. The Bragg condition, for anyone who cares about the nitty-gritty, says that when light hits each interface between layers, the reflected waves all line up constructively. No phase shift mess, no wasted light – just maximum reflection. For VCSELs, we have two DBRs: the top one (p-type) and the bottom one (n-type) that act as the resonator cavity’s mirrors. That’s where your laser light bounces back and forth to amplify before it exits as a beam.
So, back to reflectivity. For our 808nm VCSELs, the DBR reflectivity isn’t a single number – it shifts a tiny bit depending on who’s measuring it, where on the wafer you’re checking, and even what kind of test setup you’re using. But here’s the number that matters most for your application: for the bottom n-type DBR, which is the high-reflector side, we consistently hit ~99.9% reflectivity. Wait, no, hold on – let’s make that specific to 808nm. At the center wavelength of 808nm (the design target for our VCSELs), that n-DBR peaks at 99.85-99.92% reflectivity. That’s not a marketing line; that’s measured with a spectroscopic ellipsometer on every wafer run before we even start processing devices. The top p-type DBR, which is the output coupler (so we don’t want 100% reflection, obviously – we need some light to escape as the laser beam), comes in at ~98.2-98.8% at 808nm. Why the difference? The p-DBR has to be thin enough to let light out, so it’s a slightly shorter stack of layers than the n-DBR, which is all about trapping light to amplify it.
Now, let’s talk about why this matters way more than just a random spec. If your VCSEL is going into a laser pump for a solid-state laser, even a 0.1% drop in DBR reflectivity translates to a 1% drop in output power. For industrial applications, that’s a big deal – you don’t want to spec a VCSEL that delivers 5W when you paid for 5.5W because someone cut corners on DBR performance. For medical applications like laser hair removal or photodynamic therapy, consistent output power is non-negotiable; bad reflectivity can mean uneven treatment or even overheating the device mid-procedure. And for things like LiDAR, where each VCSEL’s beam has to be precise and stable, DBR reflectivity directly impacts the device’s slope efficiency – how much light you get per watt of power input.
I’ve seen a lot of newer suppliers cut corners here to hit a lower cost point. Sometimes they’ll use fewer DBR pairs, or use layers with smaller index differences (which is easier to grow, but kills reflectivity). A lot of datasheets will print “>99%” for both DBRs, but that often means the top one is 98% and the bottom is 99.1% – not the numbers you want if you’re pushing your VCSEL to 10W or more. For our products, we use 35.5 pairs for the n-DBR and 25.5 pairs for the p-DBR. That extra half-pair isn’t a mistake – it’s to make sure the phase shift at the layer interfaces lines up perfectly for 808nm, so we don’t get that dip in reflectivity just off-center.
Wait, but what happens if you’re working at a slightly different wavelength? A lot of customers ask about that, too. If you’re tuning your VCSEL to 810nm instead of 808, the reflectivity drops. For our DBRs, at 810nm, the n-DBR drops to ~99.6% and the p-DBR to ~97.5%. That’s why we specify these for 808nm – they’re optimized for that exact wavelength, so you don’t get surprises. The roll-off rate is pretty sharp, too; 20nm off-target and reflectivity can drop by 0.5-1% on the n-DBR, which might not sound like much, but for high-power applications, that’s the difference between a device that runs cool and one that overheats after 1000 hours of use.
Another thing people don’t talk about: DBR reflectivity uniformity across the wafer. If you’re manufacturing 10,000 VCSELs, you don’t want half of them with 99.9% reflectivity and the other half at 99.7%. At our fab, we’ve got a uniformity target of ±0.05% reflectivity across a 3-inch wafer for the n-DBR, and ±0.1% for the p-DBR. That sounds super tight, but it’s critical for consistent yield. We’ve had customers come to us in the past after using another supplier’s VCSELs, complaining that their batch had a 15% yield rate because half the devices were below their reflectivity threshold. That’s avoidable when your DBRs are grown in a MOCVD reactor with uniform layer thickness across the entire wafer.
Let’s address a common misconception: some folks think that higher reflectivity is always better. Not here. For the top DBR, if it were 99.9% like the bottom one, we wouldn’t get any laser light out – it would all bounce back and never escape the cavity. That’s why the p-DBR is intentionally lower. There’s a sweet spot: the difference between the two DBRs’ reflectivity, called the mirror loss, has to be low enough that the laser can reach threshold current without needing too much power. For our 808nm VCSELs, that mirror loss works out to ~1.5%, which is perfect for devices from 1W up to 15W. If we cranked the top DBR reflectivity to 99.5%, mirror loss would drop to 0.4%, and you’d end up with a threshold current so high that the VCSEL would run hot and have a short lifespan. So it’s a balance, not just “more is better.”
I should also mention testing, because you can’t just guess DBR reflectivity – you have to measure it. We don’t rely on growth simulation alone; every wafer gets tested with a phase-shifting interferometer and spectroscopic reflectometer before we even process the VCSEL mesas. We map reflectivity at 50 points across the wafer, and only wafers with an average n-DBR reflectivity of 99.8% or higher make it to the next step. Wafers that fall short get re-evaluated, but we only scrap about 3% of runs that don’t hit that mark. That’s a big investment, but it means our customers don’t have to waste time sorting through bad devices at their end.
Now, let’s tie this back to why our customers care. Take a customer who makes fiber-coupled 808nm laser pumps for ytterbium-doped fiber lasers. Those pumps need VCSELs with high efficiency – every percentage point of DBR reflectivity adds directly to the pump’s overall output. If their pump uses our VCSELs, they can get 12W of output from a single emitter, versus 10W from a competitor’s, because our DBRs are 0.3% higher on the high-reflector side. For a medical device company building a portable photodynamic therapy system, consistent DBR reflectivity means their VCSELs have a 10,000-hour lifespan, versus 6,000 hours with lower-reflectivity mirrors. That’s a huge cost difference over the product’s life cycle.
I know it’s easy to get bogged down in numbers, so let’s simplify it for different use cases:
- If you’re doing low-power applications (under 2W) like consumer electronics or small sensor systems, even a 99.7% n-DBR is fine, no problem.
- If you’re doing mid-power (2-10W) like industrial marking or medium-scale medical devices, the 99.8-99.85% n-DBR is the sweet spot.
- If you’re doing high-power (10W+) like solid-state laser pumping or LiDAR for autonomous vehicles, you need the 99.85%+ n-DBR, because every 0.1% drop here hurts slope efficiency and reliability.
One question I get all the time: “Do your DBRs degrade over time?” That’s a fair question. DBRs are made of solid semiconductors, so they don’t degrade from light exposure alone – but if you’re running the VCSEL at very high power, heat can cause layer interdiffusion, which can lower reflectivity. We design our DBRs to handle up to 15W per emitter at 808nm, and at that power, we test for long-term reliability: after 5,000 hours of continuous operation, reflectivity drops by less than 0.02%. That’s negligible for most applications. If you’re using the VCSEL below 10W, that drop is so small you can’t even measure it with standard test gear.
Let me be transparent, too: there’s no such thing as a 100% DBR reflectivity. Even the best grown stacks have tiny imperfections – a layer might be off by a few nanometers, or a tiny defect in the crystal lattice. Those imperfections make it impossible to get perfect reflection, and that’s actually a good thing? No, not really – it’s just a fact of semiconductor growth. The numbers we give you are the practical, real-world reflectivity you can expect from a mass-produced VCSEL, not some lab-only demo device that works once and breaks the next day.
So, why is this different than other suppliers? I’ve been doing this for long enough that I’ve seen companies that grow DBRs in cheap reactors, cut corners on layer quality, and push devices that hit the datasheet numbers but fall apart in the field. We prioritize consistent reflectivity because our customers depend on it. If your system is designed around an 808nm VCSEL, the last thing you want is a batch that doesn’t perform because the DBR reflectivity is 0.2% lower than advertised.

If you’re here because you’re designing a new product around 808nm VCSELs, or you’re tired of dealing with suppliers who won’t give you hard, context-rich DBR reflectivity numbers, let’s chat. We can share full test data on any wafer run, help you pick the right DBR configuration for your power level, and even adjust layer stacks if you’re working with a slightly off 808nm variant (though we recommend sticking to 808 where possible for best performance). Whether you need 1,000 devices for a prototype or 100,000 for a mass production line, we’ve got the DBR consistency and reflectivity to get your product to market without performance hiccups.
450nm Fiber Coupled Diode Laser Reference:
- Coldren, L. A., & Corzine, S. W. (1995). Diode Lasers and Photonic Integrated Circuits. Wiley.
- Vurgaftman, I., Meyer, J. R., & Ram-Mohan, L. R. (2001). Band parameters for III-V compound semiconductors and their alloys. Journal of Applied Physics, 89(11), 5815-5875.
- Liu, S., et al. (2018). High-efficiency 808nm VCSEL arrays with optimized DBR mirrors. IEEE Journal of Selected Topics in Quantum Electronics, 24(6), 1-8.
Hangzhou Brandnew Technology Co., Ltd.
Hangzhou Brandnew Technology Co., Ltd. is one of the leading 808nm vcsel manufacturers and suppliers in China, has a professional factory which manufacturers high quality 808nm vcsel and sells at competitive price. Welcome to wholesale our products made in China.
Address: 17F,Building 2, Aoqiang Mansion, No. 6 Xiyuan 5th Rd,310030 Hangzhou,China
E-mail: admin@brandnew-china.com
WebSite: https://www.brandnewdiode.com/