Back to Blog
solar technology8 min read

NREL Solar Cell Efficiency Chart 2026: Records and Timeline

Read the dated NREL/NLR efficiency chart correctly: research cells, concentrators, tandem categories, record timelines and commercial module procurement.

Keyur Rakholiya

Written by

Keyur Rakholiya

CEO & Co-Founder · SurgePV

Rainer Neumann

Edited by

Rainer Neumann

Editorial contributor · SurgePV

Published ·Updated

Answer

On the NLR (formerly NREL) Best Research-Cell Efficiency chart, revision 17 July 2026, the top result is 47.6%. It comes from a four-junction concentrator cell under 665 suns (light 665 times as strong as normal sunlight). Perovskite/silicon tandem cells reach 35.2% and silicon heterojunction cells 28.1%. Fraunhofer ISE puts commercial modules at around 24%.

A laboratory headline is useful only when its category and source stay attached. This guide lists the current record in each chart category and explains the research timeline. It also shows where a research record stops and a buying decision starts.

A junction is the part of a cell where light is turned into electric current. A tandem cell stacks more than one junction to use more of the sunlight.

Chart read on 28 October 2026. The chart PDF is marked Rev. 07-17-2026, and the NLR page shows “Last Updated Aug. 14, 2026”. The linked file can change, so keep its revision with any figure you quote. A manufacturer’s newer announcement and an older chart entry are different evidence records.

Current Records by Category

The right edge of the NLR Best Research-Cell Efficiency chart prints the current best value for each category. The table copies those values from revision 07-17-2026. The category names come from the chart legend. Where the chart labels a group on that data point, the group follows the category name. III-V means cells made from elements in groups III and V of the periodic table, such as gallium arsenide (GaAs).

Chart category Current best
III-V multijunction, 2-, 3- or 4-junction (concentrator), Fraunhofer ISE 4-J at 665x 47.6%
III-V multijunction, 3 or more junctions (non-concentrator), NREL 6-J 39.5%
Hybrid tandem: III-V/Si (Fraunhofer ISE/AMOLF) 36.1%
Hybrid tandem: perovskite/Si (LONGi) 35.2%
III-V multijunction, 2-junction (non-concentrator) 32.9%
Single-junction GaAs, concentrator 30.8%
Emerging PV: perovskite tandem cells 30.1%
Single-junction GaAs, thin-film crystal 29.1%
Crystalline Si: heterostructure (HIT) 28.1%
Emerging PV: perovskite cells 28.0%
Single-junction GaAs, single crystal 27.8%
Crystalline Si: single crystal, concentrator 27.6%
Crystalline Si: single crystal, non-concentrator 26.7%
Hybrid tandem: perovskite/CIGS 26.7%
Thin-film: CIGS 23.6%
Hybrid tandem: perovskite/organic 23.4%
Thin-film: CIGS (concentrator) 23.3%
Crystalline Si: multicrystalline 23.3%
Thin-film: CdTe 23.1%
Crystalline Si: thin-film crystal 21.2%
Emerging PV: organic cells 19.4%
Emerging PV: quantum dot cells 19.1%
Emerging PV: CZTSSe cells 16.6%
Emerging PV: organic tandem cells 14.2%
Thin-film: amorphous Si:H (stabilized) 14.0%
Emerging PV: dye-sensitized cells 13.0%

Two cautions on reading the table. First, the values are read from the printed chart. So confirm against NLR’s chart page before you cite one in a report. Second, LONGi’s July 2026 announcement of 35.5% for a perovskite/silicon cell is newer than this chart revision. So the chart still shows 35.2% for that category. Both numbers are given here with their sources.

How to read the NREL chart

  • The horizontal axis is the year of the measurement. The vertical axis is cell efficiency in percent.
  • Colour marks the technology family: purple for III-V multijunction and GaAs, blue for crystalline silicon, green for thin-film, orange-red for emerging PV and brown for hybrid tandems.
  • The marker shape marks the category. A dot inside a marker means a concentrator cell, which uses lenses or mirrors to focus sunlight.
  • The percentage at the right edge is the category’s current best. Labels along each line name the group that set earlier records.

Start With the Official NREL/NLR Chart

The Best Research-Cell Efficiency Chart, widely known as the NREL efficiency chart, is now hosted by the National Laboratory of the Rockies (NLR). Its technology lines cover research-cell results from 1976 onward. When you cite a record, use the dated chart PDF, not an unversioned search snippet.

The explanatory notes treat two kinds of tests separately. One is global-spectrum flat-plate testing. The other is direct-spectrum concentrator testing. Global-spectrum testing uses direct plus scattered sunlight on a flat panel. Direct-spectrum testing uses only direct sunlight. The notes use a reference temperature of 25°C and total or aperture cell area. Entries rest on measurements by recognized independent laboratories. Those are research measurement conditions, not outdoor energy guarantees.

Which date are you reading?

The official data-table guide lists several separate fields. They are measurement date, first publication reference, efficiency, uncertainty, area and area type. A revised efficiency can also differ from an earlier reported value when reference spectra change. Keep those fields together.

Record:

  • Chart or table revision and access date.
  • Device technology and number of junctions.
  • Light and concentration conditions.
  • Efficiency, uncertainty and the area measured.
  • Who made the cell, who tested it, and where it was published.

Do not call a result “new in 2026” simply because it appears in a 2026 chart. Do not use a producer’s publication date in place of an independently measured date that you have not retrieved.

A Source-Checked Timeline of Selected Results

These are selected milestones and evidence updates. They are not an exhaustive history or one interchangeable ranking. Dates identify the cited publication or chart revision. The last column keeps the qualification visible.

Evidence date Result Device and source Interpretation
April 2020 paper 47.1% at 143 suns NREL six-junction III-V research cell Concentrated direct-spectrum result
April 2020 paper 39.2% at one sun A global-spectrum variation of that research structure Separate illumination and device variant
May 2022 institute report 47.6% at 665 suns Fraunhofer ISE four-junction cell Concentrator result, not a rooftop module
February 2026 institute report 36.1% cell result discussed Fraunhofer ISE III-V/silicon research Cell result achieved earlier; not a newly dated perovskite/silicon result
17 February 2026 press release 31.3% module (218 cm²) Fraunhofer ISE III-V/silicon module Small research module built from adapted III-V/silicon cells of the 36.1% type, not a production panel
17 February 2026 press release 34.2% module (833 cm²) Fraunhofer ISE III-V/germanium module Described by Fraunhofer as the most efficient solar module in the world
17 July 2026 chart revision 35.2% LONGi perovskite/silicon, as listed on the NLR chart Chart entry for the hybrid tandem category
July 2026 producer announcement 35.5% LONGi crystalline-silicon/perovskite tandem Producer reports ESTI certification; newer than the chart revision

The 2020 peer-reviewed paper record identifies both the 47.1% concentrated result and the 39.2% one-sun variation. Neither is the 47.6% record.

Fraunhofer’s 2022 result report states four junctions and 665 suns. Its September 2026 CalLab retrospective still names the 47.6% cell as the highest result. It places its confirmation in 2022.

Keep tandem categories separate

Fraunhofer’s 17 February 2026 tandem-module press release explicitly names the earlier 36.1% cell as III-V/silicon. It is not evidence for a 36.1% perovskite/silicon cell.

LONGi’s July 2026 research announcement reports 35.5%. It says a 35.2% two-terminal result was included in the 68th Solar Cell Efficiency Tables. These are dated research claims. Cell records are measured on small research devices. Commercial panels get their own rating, called module efficiency.

Why a Single Percentage Is Not a Fair Comparison

Concentrated versus one-sun testing

Concentrating sunlight changes the irradiance at the tested cell. Irradiance is the sunlight power reaching a surface. The 47.6% result comes with a concentration factor. So you cannot treat it as the efficiency to expect from a normal flat-plate array in ordinary one-sun conditions. One sun is standard sunlight strength, 1,000 W/m². Check the illumination definition before you compare it with another record.

Single junction versus tandem

The Shockley–Queisser framework describes the efficiency limit of a single-junction device under defined assumptions. A tandem uses multiple junctions to respond to different parts of the spectrum. Exceeding a quoted single-junction limit does not violate physics. The device no longer has the same architecture.

A numerical limit always comes with assumptions. These cover its bandgap, spectrum, temperature and recombination. Bandgap is the minimum light energy a material absorbs. Recombination is when electric charges are lost before they become current. So 33.7% is not a ceiling for every device. For silicon, Fraunhofer ISE states that conventional cells cannot exceed 29.4%. The best silicon cell on the chart is 28.1%.

Cell versus module

A cell result and a complete module rating have different boundaries. Several things affect the assembled result. These are inactive module area, encapsulation (the protective layers around the cells), optical effects, interconnection and cell mismatch. Fraunhofer ISE’s own numbers show the size of the step. Its III-V/silicon cell is 36.1%. The 218 cm² module built from that cell design is 31.3%. That is a drop of 36.1 − 31.3 = 4.8 percentage points.

That is one research example, not a rule for converting a cell record into a module rating. Heat matters too: see solar temperature derating.

A cell-to-module power ratio and a module-area efficiency are two different measurements. State the numerator and denominator before you compare a ratio above or below 100%. Neither is annual system yield. The cell efficiency glossary entry gives the basic definition.

A Clear Module-Area Calculation

Module efficiency is output power divided by the sunlight power on the stated area (incident irradiance times area). Consider a hypothetical 500 W module with an area of 2.0 m², tested at 1,000 W/m²:

500 ÷ (1,000 × 2.0) = 0.25 = 25%.

This is an illustrative rated-power calculation, not a product. The 500 W is the rating at 1,000 W/m². Real output varies with conditions, spectral response, temperature, shading and electrical operation.

Now assume a design can genuinely fit 50 m² of module area. This excludes paths, setbacks and other unused roof space. Under the same rating basis:

  • 20% efficient modules represent 10 kWp DC.
  • 25% efficient modules represent 12.5 kWp DC.
  • The difference is 2.5 kWp. That is five percentage points of efficiency, or 25% relative additional nameplate capacity (rated capacity).

Check the area arithmetic: 50 m² × 1,000 W/m² × 0.20 = 10,000 W, and 50 × 1,000 × 0.25 = 12,500 W. Different module sizes may not pack equally well. The inverter, structure or connection may also limit the larger array. A real panel-layout design process tests those limits.

Test a candidate module in a layout

SurgePV supports solar array layout and energy-yield modeling. A guided demo shows how to compare module options on your roof.

Book a Free Demo

Commercial Modules Versus Research Records

Fraunhofer ISE’s February 2026 release says commercial PV modules on sale now are around 24% efficient. The best silicon cell on the chart is 28.1%. So a typical commercial module sits roughly four percentage points below the best silicon research cell (28.1 − 24 = 4.1). The multijunction records of 35.2% to 47.6% use other materials. The 47.6% record also uses concentrated light.

A research record tells you nothing about these points. Get them from the product documents and the supplier:

  • Outdoor stability and service life.
  • The production-module warranty and its exclusions.
  • Regional stock, delivered price and factory ramp dates.
  • Roof load, mounting approvals and electrical fit.
  • The temperature coefficient and rear-side contribution of a specific model.
  • Annual kWh, bill savings and financial return for your site.

For products you can buy, the module-efficiency comparison uses manufacturer specifications for each model. It does not use research-cell values. The perovskite commercial-use guide covers how mature the technology is and how long it lasts.

Translate Research Into Procurement Checks

Keep the useful installer question from the original timeline: does the proposed module solve this project’s actual constraint?

  1. Find the bottleneck. Available layout, structural limits, grid connection capacity, budget or demand profile may limit the project.
  2. Get the exact product document. Record model, region, revision, dimensions, nameplate power and test basis.
  3. Check the fit. Compare inverter voltage and current, mounting needs, installation instructions and any certification that applies.
  4. Judge performance on its own. Use the model’s actual temperature and warranty information. Do not assign one technology-wide value.
  5. Confirm supply and contract scope. A prototype result or an announced manufacturing plan is not a binding delivery commitment.
  6. Model the accepted layout. Use stated weather, shade and electrical assumptions for energy output. Use the customer’s tariffs and load for savings.

Use solar design software to evaluate the actual project configuration. Bring the candidate module and its datasheet to the demo.

Keep the Record Register Current

Keep a small dated register of the research figures relevant to your work. When a new entry appears, check what kind it is. It may be a different category. It may be an independently confirmed measurement. It may be a revised spectrum basis, or a producer announcement that still needs fuller documentation.

Keep research records separate from the approved product list. Move a module into that list only after its documents, supply and project checks pass. The research timeline is context, not a schedule. It does not say when tandems reach the market or what they will cost.

Frequently Asked Questions

What is the highest solar cell efficiency record in 2026?

The NLR chart (revision 17 July 2026) lists 47.6% as the highest research-cell result. It is a four-junction III-V concentrator cell from Fraunhofer ISE, measured under 665-sun concentration and achieved in 2022. The best non-concentrator multijunction cell on the chart is 39.5%. The best perovskite/silicon tandem is 35.2%. Compare within one category, not across them.

What does the NREL efficiency chart show?

It plots the best confirmed research-cell efficiency for each technology from 1976 onward. Each colour is a technology family, and each marker is a category. The percentage at the right edge is the current best for that category. The chart is now hosted by the National Laboratory of the Rockies (NLR). The PDF is dated by revision.

What is the best silicon solar cell efficiency on the NREL chart?

On revision 17 July 2026, the silicon categories top out at 28.1% for heterojunction (HIT) cells. They are 27.6% for single-crystal concentrator cells and 26.7% for single-crystal non-concentrator cells. Fraunhofer ISE states that conventional silicon cells cannot exceed 29.4%.

What is the difference between cell efficiency and module efficiency?

Cell efficiency is measured on one tested cell under stated conditions. Module efficiency is the complete module’s output compared with its area and the sunlight on it (irradiance). So inactive area, optics and interconnection losses count. Fraunhofer ISE’s III-V/silicon cell is 36.1% and its module built from those cells is 31.3%. That is a drop of 4.8 percentage points.

What efficiency can I expect from a 2026 commercial solar panel?

Fraunhofer ISE says commercial modules are around 24% efficient. Use the efficiency and dimensions on the exact model’s current datasheet. Also compare layout fit, temperature behavior, warranted performance and project energy output. Do not look at nameplate efficiency alone.

Do tandem cells exceed the single-junction efficiency limit?

Yes. A tandem stacks junctions that respond to different parts of the spectrum. So a limit calculated for one junction does not apply to it. State the architecture, spectrum and concentration whenever you quote a tandem percentage.

Will perovskite solar panels replace silicon?

A laboratory result cannot answer that. Perovskite can be a single-junction cell (28.0% on the chart). It can also be part of a tandem with silicon (35.2%). Whether people buy it depends on durability, manufacturing, certification, price, supply and warranty for the specific module.

Why do solar cell records matter to installers?

They show where the technology is heading and help you ask better questions about new products. Stock, price, lifetime performance and fit come from the model’s datasheet, warranty and supplier. They do not come from a lab record.

Where this fits

This article is part of SurgePV's Solar Technology hub, which works through the topic from first principles to the decisions a project team actually has to make.

About the Contributors

Author
Keyur Rakholiya
Keyur Rakholiya

CEO & Co-Founder · SurgePV

Keyur Rakholiya is identified by SurgePV as its CEO and a company co-founder. His SurgePV author page lists only role information that can be tied to the public profile below; credentials, project totals, testing claims, media appearances, and speaking engagements are not asserted without retained evidence.

Editor
Rainer Neumann
Rainer Neumann

Editorial contributor · SurgePV

Rainer Neumann is credited as an editorial contributor on SurgePV content. This profile does not assert engineering credentials, project totals, software-testing experience, education, speaking engagements, or media citations because independent verification evidence is not retained in the publication record.

Get Solar Design Tips in Your Inbox

Join 2,000+ solar professionals. One email per week - no spam.

No spam · Unsubscribe anytime

Book Free Demo

Choose which optional technologies SurgePV may use. Essential storage remains active for security and requested features.