Repairing a smartphone or replacing it: what lifecycle research can tell us

Research question

What does accessible published research tell us about the greenhouse-gas implications of extending a smartphone's life through repair, compared with replacing it?

The question concerns climate impacts. Price, safety, data security, accessibility and whether the device remains suitable still matter to the practical decision. A carbon comparison cannot settle those issues on its own.

How this review was prepared

We used a targeted, AI-assisted web search on 20 October 2026. Search combinations included “smartphone lifetime repair life cycle assessment”, “Fairphone 4 life cycle assessment repair”, “smartphone reuse rebound effects”, and exact paper titles found through references. We opened publisher, institutional and manufacturer full texts.

We selected sources that explain relevant boundaries or examine lifetime extension, component replacement, footprint variation or reuse behaviour. Six research/report sources are discussed, with an IPCC glossary for terminology. Manufacturer reports provide model-specific evidence and are identified as such. We excluded marketing summaries from the evidence table when the underlying report was accessible.

Selection was purposive and English-language, rather than an exhaustive database search. There was no preregistered protocol, independent duplicate screening or formal risk-of-bias scoring. We therefore call this a narrative literature review. Different models and functional units were not statistically combined.

CO₂ and CO₂e mean different things

CO₂ describes carbon dioxide. CO₂e expresses different greenhouse gases on a common climate-impact scale, using a stated metric and time horizon. A common approach uses 100-year global warming potentials. A result reported in kilograms of CO₂e is not necessarily that mass of carbon dioxide alone.1

The studies below generally address lifecycle greenhouse-gas footprints. Their assumptions and indicator versions should travel with any numerical comparison.

Finding 1: compare useful service, not just two devices

Cordella and colleagues modelled different replacement cycles while keeping the reference service period constant. Moving from a two-year to a three-year replacement cycle reduced the modelled footprint by 23–30%, depending on whether repairs were needed. Their baseline and sensitivity cases have defined boundaries; this range is not a savings guarantee for every phone.2

Our interpretation is that an extra year matters when it supplies service that would otherwise require another purchase. The comparison needs a defined period and realistic replacement behaviour.

Ask what happens with and without the repair, using equivalent communication needs. This is a proposed decision framework, rather than another numerical study finding.

Finding 2: a longer life can raise the total while lowering the yearly average

Fraunhofer IZM's Fairphone 4 report gives lifecycle totals of 43.0 kg CO₂e for a three-year scenario and 49.9 kg CO₂e for five years with one battery replacement. The longer scenario includes additional charging and replacement-battery burdens. The study also examines module-repair routes.3

Dividing those reported totals by their respective assumed durations gives approximately 14.3 and 10.0 kg CO₂e per year. That is our simple arithmetic, not a new LCA or a pooled result. The totals rise because the second scenario supplies more years of use; the annual average falls.

This illustrates the difference between a total and a rate. Neither number alone establishes savings from repairing a different model.

Finding 3: production matters, but published footprints are not repair credits

Ercan and colleagues estimated 57 kg CO₂e for a Sony Z5 and accessories over three years, excluding network usage. Their inventory identifies integrated-circuit production as an important contributor.4

Apple's June 2025 report gives 56 kg CO₂e for the iPhone 15 128GB and attributes approximately 80% of that footprint to production. It uses a three-year assumption for first-owner power use and includes in-box items.5 These are product-specific illustrations, not a ranking: different models, years, inventories and electricity assumptions are involved.

Our interpretation is that avoiding premature replacement can preserve the usefulness of manufacturing already carried out. However, subtracting a manufacturer's complete footprint from every repair would credit activities that may still occur, use an unsuitable replacement baseline, and ignore the repair's own impacts.

A defensible estimate would explain which future burdens change because of the repair. If sufficient information is unavailable, a qualitative conclusion is more honest than a precise-looking “CO₂ saved” badge.

Finding 4: the repair and the wider market both need attention

For a particular job, we propose documenting the parts, route and expected further service before discussing avoided emissions. A generic credit would conceal the distinction between a small successful intervention and extensive work with uncertain remaining life.

Makov and Font Vivanco show a separate limitation: in their US reuse model, used-phone purchases and spending changes affect the environmental outcome. A second-hand sale does not necessarily replace a new purchase one-for-one; it may be an additional device or displace another used phone.7

This does not show that repair is ineffective. It shows that technical comparisons and market-wide claims answer different questions. Extending the life of one's main phone and increasing the total number of devices owned are not automatically equivalent interventions.

How strong is the evidence?

Clément and colleagues found that smartphone LCA results vary with modelling choices, including production inventories and electricity mixes. Their analysis makes cross-report comparisons especially sensitive to methodological alignment.6

The selected evidence supports a conditional direction: useful lifetime extension can reduce the footprint of an equivalent period of smartphone service. It does not establish one universal percentage for UK users or eServUK repairs.

Important uncertainties remain:

• Transferability: device-specific and older inventories may not represent another handset or today's supply chain.
• Actual remaining life: a modelled additional year is an assumption until service is delivered.
• Comparable functionality: an unsuitable or unsupported device may not provide the service assumed in the comparison.
• Evidence independence: manufacturer reports offer detailed product data alongside a commercial interest; open reporting helps scrutiny but does not remove that interest.
• Wider impacts: climate indicators do not fully represent resource depletion, toxicity or social conditions.

These are reasons to state the boundaries of a conclusion, rather than to treat every estimate as equally uncertain.

What this means for a repair decision

Our practical interpretation is to ask four questions before making an environmental claim:

1. Will the repaired phone meet the user's needs for a realistic further period?
2. What components, transport and other work does the proposed repair require?
3. Which purchase, if any, would the repair actually postpone or avoid?
4. Does the evidence use comparable service, geography and lifecycle boundaries?

Safety, appropriate software support and data preparation should be considered alongside those questions. This article provides no instruction to open a device or work on a battery.

The evidence gives a sound reason to investigate useful repair before unnecessary replacement. It also sets a standard for communication: explain the comparison, disclose the assumptions and keep product-level examples separate from verified savings in a particular job. eServUK has not measured an individual repair's carbon saving in this review.

Disclosures

This article was prepared using AI-assisted research and editorial synthesis. eServUK has a commercial interest in repair services. The article has not undergone external academic peer review. Source reports include manufacturer-funded or manufacturer-specific data, identified in the evidence map and text.