Quick answer
If you are choosing between UV-LED and mercury-lamp technology for a UV-C system, there is no universal winner — the right answer depends on duty cycle, switching pattern, size, and how you value upfront cost versus operating cost.
The single fact that frames every other trade-off: a low-pressure mercury lamp converts roughly 30–35 % of input electrical power into 254 nm UV-C, while commercially available UV-C LEDs typically sit far lower — around 7 % wall-plug efficiency in practice, with the best 280 nm devices reported in the 9–20 % external-quantum-efficiency range (Hull et al., critical review of UV-LEDs, ScienceDirect 2024). So per watt of UV-C delivered, mercury still wins on raw electrical efficiency today.
But efficiency is not the whole picture. LEDs switch instantly, survive unlimited on/off cycles, contain no mercury, are physically tiny, and degrade gracefully instead of failing abruptly. Mercury lamps are cheaper per optical watt, mature, and well-characterised. This guide walks the seven decision factors that actually decide the purchase. For the underlying physics see UV lamp technology; this article is the buyer-facing trade-off layer.
- Quartz-glass envelope transparent to 254 nm
- Electrodes coated tungsten; strike the arc
- Fill gas + mercury argon ignites, Hg vapour emits UV
- Base mount + electrical connection
- Ballast (external) current limiter; sits outside the lamp
- Secondary optics lenses/diffuser; shape the beam
- LED chip (AlGaN) Al content sets the wavelength
- Package (SMD/COB) protects the chip, couples light out
- Carrier plate (AlN) conducts heat away = lifetime
- Driver (integrated) constant current + dimming
Hover a layer to highlight it
Two fundamentally different architectures: the mercury lamp generates UV in a gas discharge and needs an EXTERNAL ballast; the UV-LED module emits from a semiconductor junction and integrates its driver + cooling into the layer stack.
Factor 5 — Switching, dimming and warm-up
This is the factor most often decisive for intermittent or sensor-triggered duty.
A low-pressure mercury lamp needs a warm-up period to reach full UV output and dislikes frequent switching — each cold start erodes the electrodes, so on/off cycling can cut effective service life dramatically. Medium-pressure lamps run hotter still and warm-up/cool-down constraints are stricter.
A UV-C LED reaches full output effectively instantly, can be dimmed by adjusting drive current, and tolerates being switched on and off tens of thousands of times with little measurable degradation penalty (Crystal IS lifetime guidance). For flow- or occupancy-triggered systems — water drawn intermittently, rooms entered occasionally — LED switching behaviour can outweigh its WPE disadvantage entirely. For the driver electronics that enable LED dimming and lamp ballasting see ballasts and drivers.
Factor 6 — Upfront cost, mercury regulation and disposal
Upfront cost: mercury is cheaper per optical watt. Capital cost per watt of UV-C output has been reported at roughly $2/W for low-pressure mercury lamps versus $100–400/W for UV-C LEDs (Hull et al., ScienceDirect 2024). For a large, continuously running system, that gap is hard to close on energy savings alone — which is why mercury still wins many high-throughput municipal and industrial cases.
Mercury and regulation: every mercury lamp contains mercury and is hazardous waste at end of life, requiring controlled disposal. A common misconception is that the Minamata Convention bans UV mercury lamps. It does not: the 2023 COP-5 fluorescent-lighting phase-out explicitly excludes special-purpose UV lamps used for germicidal and curing applications, which retain an exemption (Buildings.com, Minamata fluorescent phase-out; GEW, Update on Mercury Regulation for UV Curing Lamps). So mercury UV lamps remain legal to buy — but the regulatory direction of travel, disposal handling and breakage liability are real reasons some buyers choose mercury-free LED even when the spreadsheet favours mercury.
Factor 7 — Retrofit feasibility
Replacing a mercury lamp in an existing reactor with an LED module is rarely a true drop-in. Because each LED delivers far less power than a lamp (Factor 2) and emits a different beam pattern, an LED conversion of an existing reactor usually needs the array geometry — and often the reactor itself — re-engineered to the water matrix and dose target (Nature Scientific Reports 2023). Some application areas (industrial curing, point-of-use water) now offer engineered LED retrofit kits, but a like-for-like swap that preserves validated dose is the exception, not the rule.
If you are simply replacing a spent lamp with the same technology, that is a different question — see finding the right replacement lamp (coming).
Head-to-head comparison
| Factor | Low-pressure mercury | Amalgam / medium-pressure mercury | UV-C LED |
|---|---|---|---|
| Germicidal wavelength | Fixed 253.7 nm | Amalgam 254 nm (higher output); MP broadband UV-C/B/A | Selectable peak — commonly ~265 / 275 / 285 nm |
| Wall-plug efficiency | ~30–35 % | LP-class for amalgam; MP lower | ~7 % typical; best 280 nm ~4–20 % EQE range |
| Output per device | Tens–hundreds W UV-C | Higher (amalgam/MP) | Fraction of a W to a few W — needs arrays |
| Rated lifetime | ~8,000–10,000 h, degrades from ~9,000 h | Lamp-dependent | L70 drive-dependent (e.g. ~5,000 h hard-driven) |
| Failure mode | Abrupt; warm-up needed | Abrupt; strict warm-up | Gradual fade; instant-on |
| On/off cycling | Erodes electrodes, shortens life | Worse (hotter) | Tolerates tens of thousands of cycles |
| Capital cost (per W UV-C) | ~$2/W | Low-pressure class | ~$100–400/W |
| Mercury content | Yes — hazardous-waste disposal | Yes | None |
| Physical size | Long tube | Long tube | Compact, enables small/point-of-use designs |
All figures: Hull et al., ScienceDirect 2024; Crystal IS; BSC Bulbs; Phoseon. "EQE range" cited because vendors report WPE and EQE inconsistently — confirm which figure a datasheet states.
Decision: when to pick which
Lean mercury (low-pressure / amalgam) when:
- The system runs continuously at high throughput — the WPE advantage compounds and cycling penalties never apply.
- Upfront capital is the binding constraint — the ~$2/W vs ~$100–400/W gap is large.
- You need high UV-C power from a single envelope without designing an array.
Lean medium-pressure mercury when:
- You need broadband UV for photochemistry or advanced oxidation, not just narrowband germicidal dose.
Lean UV-C LED when:
- Duty is intermittent or sensor-triggered — instant-on and unlimited cycling outweigh lower WPE.
- The product is small or point-of-use — appliance integration, compact in-line modules.
- Mercury-free matters for disposal handling, breakage liability or regulatory posture.
- You want to tune the wavelength toward the action-spectrum peak or a specific target.
Get a project-specific comparison rather than a generic rule — see choosing a UV system (coming) and how to read a UV datasheet (coming) so you compare WPE, L70 and dose on a like-for-like basis.
Cross-references
- UV lamp technology — the underlying physics of mercury and LED UV sources
- UV-LED lifetime and degradation — how L70 and drive current interact
- LED area emitters — array geometry and emission patterns
- Wavelengths and action spectra — why 265 nm and 254 nm differ biologically
- Ballasts and drivers — the electronics behind dimming and lamp operation
- UV economics and ROI — modelling total running cost
- Choosing a UV system — picking a system class (coming)
- Finding the right replacement lamp — same-technology lamp replacement (coming)
- How to read a UV datasheet — comparing specs like-for-like (coming)
Sources
- Hull et al., A critical review of ultra-violet light emitting diodes as a one water disinfection technology, ScienceDirect 2024 — WPE, EQE and capital-cost-per-watt figures.
- Improved disinfection performance for 280 nm LEDs over 254 nm low-pressure UV lamps, Nature Scientific Reports 2023 — per-dose efficacy comparison and reactor-design implications.
- US DOE, Operating Lifetime Study of Ultraviolet LEDs, RTI 2022 (PDF) — LED lifetime vs drive current.
- US DOE, Reports Examine the State of Current Commercial UV LEDs — current state and efficiency trajectory of commercial UV LEDs.
- Crystal IS, LED vs Lamp Output Comparison and Understanding UVC LED Lifetime versus Power — per-device output and L70 drive dependence.
- Phoseon, Best UV wavelengths for disinfection — wavelength selection and germicidal mechanism by band.
- Lux Review, How effective are UVC LEDs? — IESNA action spectrum and the 254 nm vs 265 nm misconception.
- BSC Bulbs, How Long Do UV Lights Last? — mercury lamp service-life ranges and cycling sensitivity.
- Buildings.com, Minamata Convention fluorescent phase-out and GEW, Update on Mercury Regulation for UV Curing Lamps — UV-lamp exemption from the mercury phase-out.