USP <645> Water Conductivity is a three-stage test, not a single limit. Almost every article on pharmaceutical water quotes one number — 1.3 µS/cm at 25 °C — and stops there. That figure is only the Stage 1 acceptance value at one temperature. If a sample fails Stage 1 it is not out of specification: it proceeds to Stage 2, and if it fails Stage 2 it proceeds to Stage 3, which evaluates conductivity against measured pH. Water can pass USP <645> at conductivities well above 1.3 µS/cm, and treating 1.3 as a hard limit causes unnecessary investigations and batch holds.
What is USP <645>?
USP General Chapter <645> Water Conductivity defines how conductivity is measured and evaluated for Purified Water and Water for Injection. Conductivity is used as a surrogate for total ionic content: rather than testing for individual ions, the chapter uses a single electrical measurement to bound the total ionic load, which is why it replaced the older wet-chemistry tests for chloride, sulfate, calcium and heavy metals.
Two things about it are routinely misunderstood. First, it is a staged procedure — failing the first test does not fail the water. Second, Stage 1 uses non-temperature-compensated conductivity, which is the opposite of how most process instruments are configured by default.
Why “1.3 µS/cm” is not the specification
1.3 µS/cm is the Stage 1 acceptance value at 25 °C specifically. Stage 1 is a temperature-indexed table: the acceptance value rises with temperature because water’s own self-ionization increases with heat. At 0 °C the Stage 1 value is around 0.6 µS/cm; at 25 °C it is 1.3; at 100 °C it is approximately 3.1.
This matters operationally because hot WFI loops are frequently sampled at temperature. A loop running at 80 °C reading 2.4 µS/cm uncompensated is not automatically out of specification — the Stage 1 value at that temperature is considerably higher than 1.3. Applying the 25 °C number to a hot sample generates false excursions, and we have seen systems investigated and even shut down on exactly that error.
Verify the full Stage 1 table against the current official USP chapter before writing it into a procedure. USP revises these chapters, the table is temperature-indexed in 5 °C steps, and the value used must be the one for the temperature bracket you actually measured in — without interpolating.
The three-stage procedure, stage by stage
Stage 1 — on-line or off-line, non-temperature-compensated
Measure conductivity without temperature compensation and record the temperature at the same moment. Compare the reading against the Stage 1 table value for that temperature, using the next lower temperature bracket rather than interpolating between rows. If the reading is at or below the table value, the water meets the requirement and testing stops. If it is above, proceed to Stage 2.
Because Stage 1 can be performed on-line, this is where most pharmaceutical water is released day to day — a validated in-line conductivity instrument on the distribution loop satisfies the chapter without any laboratory work.
Stage 2 — equilibrate with atmospheric carbon dioxide
Stage 2 exists because dissolved carbon dioxide raises conductivity without indicating any real contamination. Collect a sample, bring it to 25 °C, and stir it in an environment open to the atmosphere until the reading stabilizes — typically five minutes or more — allowing CO2 to reach equilibrium.
The Stage 2 acceptance limit is 2.1 µS/cm at 25 °C. If the equilibrated reading is at or below 2.1, the water meets the requirement. If it exceeds 2.1, proceed to Stage 3.
Stage 3 — conductivity evaluated against pH
Stage 3 recognizes that the conductivity contributed by dissolved CO2 is predictable from pH. Add a saturated potassium chloride solution to the Stage 2 sample to stabilize the pH measurement, measure pH to 0.1 units, and compare the Stage 2 conductivity against the chapter’s pH-indexed limit for that pH value.
If the conductivity is at or below the tabulated limit for the measured pH, the water meets the requirement. If it exceeds it, or if the pH falls outside the range covered by the table, the water fails USP <645> and an investigation is warranted.
The Stage 3 pH-versus-conductivity table must be read from the current official chapter. It is not a formula and should not be reconstructed from secondary sources — the limits are not monotonic across the pH range, and transcription errors here produce false passes.
Stage 1, 2 and 3 compared
| Stage 1 | Stage 2 | Stage 3 | |
|---|---|---|---|
| Where performed | On-line or off-line | Laboratory | Laboratory |
| Temperature compensation | None — uncompensated | Sample brought to 25 °C | 25 °C |
| Acceptance basis | Temperature-indexed table | Single value | pH-indexed table |
| Value at 25 °C | 1.3 µS/cm | 2.1 µS/cm | Varies with measured pH |
| Purpose | Rapid routine release | Remove the CO2 contribution | Account for pH-driven conductivity |
| Additional measurement | Temperature | Stirring to equilibrium | pH to 0.1 units, with KCl addition |
| If it passes | Testing stops — water complies | Testing stops — water complies | Water complies |
| If it fails | Go to Stage 2 | Go to Stage 3 | Water fails <645> |
Compensated vs non-temperature-compensated: the most common instrument error
Most conductivity instruments ship configured to apply automatic temperature compensation, normalizing every reading back to 25 °C. That is the correct behavior for process control and completely wrong for Stage 1 of USP <645>, which requires the raw, uncompensated value read against the temperature table.
The failure mode is subtle and goes both ways. A compensated reading on a hot loop is normalized downward and can mask a genuine excursion; on a cold sample it can be normalized upward and trigger a false one. Any instrument used for <645> Stage 1 release must have compensation disabled or be capable of reporting both values, and the qualification documentation should state explicitly which mode was used.
How is conductivity measured correctly?
- Cell constant. The instrument must use a cell of known, certified constant, and that certification is part of the calibration record required under 21 CFR 211.68.
- Instrument accuracy. The measuring system must be capable of resolving low-conductivity water meaningfully — at 1.3 µS/cm, an instrument with poor resolution at the bottom of its range is not fit for purpose.
- Temperature accuracy. Because Stage 1 is temperature-indexed, the temperature measurement is as critical as the conductivity measurement. An error of a few degrees moves you into the wrong table row.
- Sample handling for Stage 2 and 3. Off-line samples absorb atmospheric CO2 from the moment they are drawn, which is precisely what Stage 2 is designed to normalize — but it also means an off-line Stage 1 reading taken from a sample that has been standing is not equivalent to an in-line reading.
- Location. A single point-of-use reading does not characterize a loop. Sampling points should be defined in the water system validation and cover the worst-case location, typically the point furthest from the return.
How does <645> relate to <643> TOC and microbial limits?
Conductivity and total organic carbon are complementary and neither substitutes for the other. USP <645> bounds the ionic load; USP <643> Total Organic Carbon bounds the organic load at 500 ppb for both Purified Water and Water for Injection. A system can be ionically excellent and organically contaminated, or the reverse.
Microbial quality sits outside both. The action levels given in USP <1231> are 100 CFU/mL for Purified Water and 10 CFU/100 mL for Water for Injection, and WFI additionally carries a bacterial endotoxin limit of 0.25 EU/mL. A complete water monograph compliance position therefore requires conductivity, TOC, bioburden and — for WFI — endotoxin.
What happens when water fails Stage 3?
A genuine Stage 3 failure is a real out-of-specification result and triggers an investigation. In our experience the causes cluster:
- Resin exhaustion in a deionizer or mixed bed, usually visible as a gradual upward trend rather than a step change
- CO2 breakthrough from inadequate degassing upstream of the polishing stage
- Ammonia carryover from a distillation unit, which raises pH and conductivity together
- Sanitant residue after a chemical sanitization with inadequate rinse-down
- Rouge or corrosion products contributing ionic species — particularly in hot loops with a degraded passive layer
- Instrument or calibration fault, which should be excluded before the system is blamed
Trend data usually distinguishes them faster than a single result does, which is why continuous on-line monitoring with alert levels set below the action level is worth far more than a daily grab sample.
Water system failing conductivity, or a loop you cannot get to specification?
Paul Industries designs, builds, remediates and validates USP purified water and WFI systems nationwide — including derouging and re-passivation where corrosion products are driving the ionic load, and the sampling and documentation to demonstrate the fix held.
What a Stage 3 failure sets in motion
No dollar figure attaches to a conductivity result, but a genuine failure carries a cost profile worth knowing before the first excursion. The drivers are these.
- Batch disposition. Water made since the last passing result is suspect, and so is anything it went into. The hold cost depends on how much water the loop delivers between samples.
- The investigation itself: instrument verification, a repeat sample under controlled handling, review of the loop trend, and the written report the deviation system requires.
- Root cause work, usually on the generation train: exhausted EDI or DI capacity, a failed RO element, a carbon bed releasing ions, or a leaking heat exchanger.
- Sanitization and requalification if the investigation opens the loop, which brings the sampling frequency back up for a defined period.
Most apparent failures are handling or instrument errors and never reach this list, which is why the sampling procedure and the instrument records deserve the attention below.
What to require from a conductivity system before it goes into a procedure
The chapter sets instrument requirements; the procedure has to prove they are met and stay met. Ask the supplier of an on-line or laboratory conductivity system for the following, and file it with the method.
- A cell constant certificate for each probe. The chapter requires the constant to be known within a stated tolerance, and the certificate must show the reference solution and the temperature used.
- Meter calibration against traceable precision resistors across the range in use, not a single-point check.
- Temperature sensor calibration, because Stage 1 acceptance is read from a temperature-indexed table and an error in temperature becomes an error in the limit.
- A configuration record showing temperature compensation disabled for Stage 1 readings, signed at commissioning and re-checked after any firmware change.
- The revision of the Stage 1 table the procedure uses, with the official chapter version it was copied from.
- For off-line testing, a sample procedure that states the container material, the fill and cap method, the time allowed before measurement, and the stirring and equilibration steps for Stage 2.
- Stage 3 reagents and the pH meter calibration that goes with them, since the Stage 3 limit is read against measured pH.
- An audit trail for on-line data, so a result can be traced to the probe, the time and the configuration in force.
- Alert and action levels for the loop derived from its own data, set below the Stage 1 value so drift is seen before an excursion.
Verify each tolerance against the current official chapter before writing it into the SOP; the chapter is revised and the values above are described, not quoted, for that reason. See also the USP water specification lookup.
Frequently asked questions about USP <645> conductivity
What is USP 645?
USP General Chapter 645, Water Conductivity, defines how conductivity is measured and evaluated for Purified Water and Water for Injection. It uses a single electrical measurement as a surrogate for total ionic content, replacing the older wet-chemistry tests for chloride, sulfate, calcium and heavy metals. It is a three-stage sequential procedure, not a single pass-fail limit.
Is the USP 645 limit 1.3 microsiemens per centimeter?
No. 1.3 microsiemens per centimeter is the Stage 1 acceptance value at 25 degrees C only. Stage 1 uses a temperature-indexed table whose values rise with temperature, and water that exceeds Stage 1 proceeds to Stage 2 at 2.1 microsiemens per centimeter and then to Stage 3, which evaluates conductivity against measured pH. Water can comply with USP 645 well above 1.3.
What are the three stages of USP 645?
Stage 1 is a non-temperature-compensated measurement, on-line or off-line, compared against a temperature-indexed table. Stage 2 brings a sample to 25 degrees C and stirs it open to the atmosphere until carbon dioxide equilibrates, with an acceptance limit of 2.1 microsiemens per centimeter. Stage 3 adds saturated potassium chloride, measures pH to 0.1 units, and compares the Stage 2 conductivity against a pH-indexed limit.
Should conductivity be temperature compensated for USP 645?
No, not for Stage 1. Stage 1 requires the raw, non-temperature-compensated reading together with the recorded temperature, compared against the table value for that temperature bracket. Most instruments ship with automatic temperature compensation enabled, which is correct for process control and wrong for Stage 1 release testing.
What is the Stage 2 conductivity limit?
The Stage 2 acceptance limit is 2.1 microsiemens per centimeter at 25 degrees C, measured after the sample has been brought to 25 degrees C and stirred in an environment open to the atmosphere until the reading stabilizes, allowing dissolved carbon dioxide to reach equilibrium.
Why does USP 645 have a carbon dioxide equilibration step?
Dissolved atmospheric carbon dioxide forms carbonic acid, which raises conductivity without indicating any real ionic contamination. Stage 2 deliberately equilibrates the sample with atmospheric carbon dioxide so that its contribution is consistent and bounded, rather than variable with how long a sample has been standing.
Can hot WFI fail conductivity just because it is hot?
It can appear to, if the 25 degrees C value is wrongly applied. Water self-ionizes more at higher temperature, so the Stage 1 acceptance value rises with temperature, reaching roughly 3.1 microsiemens per centimeter near 100 degrees C. A hot loop sampled at temperature must be judged against the table value for that temperature, not against 1.3.
What is the difference between USP 645 and USP 643?
USP 645 bounds the ionic content of the water through conductivity. USP 643 bounds the organic content through total organic carbon, with a limit of 500 parts per billion for both Purified Water and Water for Injection. They are complementary; a system can pass one and fail the other.
What are the microbial limits for purified water and WFI?
The action levels given in USP 1231 are 100 CFU per milliliter for Purified Water and 10 CFU per 100 milliliters for Water for Injection. Water for Injection additionally carries a bacterial endotoxin limit of 0.25 endotoxin units per milliliter. These sit outside USP 645 and USP 643 and must be monitored separately.
Can Stage 1 be performed on-line?
Yes, and most pharmaceutical water is released this way. A validated in-line conductivity instrument with temperature measurement on the distribution loop satisfies Stage 1 without laboratory work, provided temperature compensation is disabled or both compensated and uncompensated values are reported.
What causes a USP 645 Stage 3 failure?
Common causes are deionizer or mixed-bed resin exhaustion, carbon dioxide breakthrough from inadequate upstream degassing, ammonia carryover from a still, sanitant residue after inadequate rinse-down, ionic species from rouge and corrosion products in hot loops, and instrument or calibration faults, which should be excluded before the system is investigated.
Do I need to interpolate between temperatures in the Stage 1 table?
No. The Stage 1 table is read at fixed temperature intervals and the value for the next lower temperature bracket is used. Interpolating between rows is not part of the procedure and introduces an unnecessary source of error into a release decision.
What is conductivity actually measuring in pharmaceutical water?
The ability of the water to carry electrical current, which is a function of the dissolved ionic species present. It is a non-specific measure, so it detects ionic contamination without identifying it, which is why it is fast and why it cannot stand alone for every purpose.
Why is carbon dioxide such a problem for water conductivity?
Dissolved carbon dioxide from the air forms carbonic acid, which ionises and raises conductivity even though the water is otherwise pure. That is why Stage 2 deliberately equilibrates the sample with air, so the contribution is accounted for rather than mistaken for contamination.
What is the difference between Stage 1 and Stage 2 in practice?
Stage 1 is performed on the water as it is, typically on-line and without temperature compensation, against a temperature-dependent limit table. Stage 2 is performed off-line on a sample brought to a defined temperature and equilibrated with air, against a single limit.
What does Stage 3 add?
Stage 3 adjusts the sample pH with a defined salt addition and measures conductivity against a pH-dependent limit table, which separates the contribution of carbon dioxide from other ionic contamination. Reaching Stage 3 means the earlier stages could not resolve the question.
Why is temperature compensation not used in Stage 1?
Because the limit table is already temperature-dependent, so applying a compensation algorithm on top of it double-corrects the reading. Instruments left in a compensated mode are a frequent cause of readings that look fine and are not comparable to the standard.
What conductivity instrument requirements does the standard set?
The cell constant has to be known to a stated accuracy, the instrumentation accurate within a stated tolerance, and both traceable to a recognised standard. The instrument qualification is part of the result, which is why an uncalibrated on-line probe cannot support a release decision.
Can hot water in a WFI loop fail Stage 1 simply for being hot?
The Stage 1 limit rises with temperature, so a hot sample is judged against a higher limit rather than the ambient one. Failures at temperature generally come from reading against the wrong row of the table or from a compensated instrument rather than from the heat itself.
What is USP 643 and how does it relate?
USP 643 covers total organic carbon, which measures organic rather than ionic contamination. Conductivity and total organic carbon are complementary, and a water system is monitored against both because each is blind to what the other detects.
What microbial limits apply alongside conductivity?
Purified Water and Water for Injection carry action limits for microbial count, with Water for Injection substantially tighter, and Water for Injection additionally carries a bacterial endotoxin limit. Conductivity says nothing about any of these, which is why the monitoring programme covers all of them.
What sample handling does off-line conductivity testing require?
Minimising contact with air before the measurement, using clean glassware that will not contribute ions, and bringing the sample to the specified temperature. Sample handling error is a more common cause of off-line failures than the water itself.
What typically causes a Stage 3 failure?
Genuine ionic contamination that is not carbon dioxide, such as resin leakage from a deionisation bed, membrane failure allowing feed water through, regenerant carryover, or ingress at a point of use. It points at the treatment train rather than at the sampling.
What cell constant should a conductivity probe have?
One matched to the conductivity range being measured, because a probe optimised for high conductivity loses resolution at the very low values found in purified water. Using a general-purpose probe on high-purity water is a common source of unreliable readings.
Does the same conductivity requirement apply to Purified Water and Water for Injection?
The conductivity requirement is common to both. What separates them is the bacterial endotoxin limit and the tighter microbial expectation on Water for Injection, along with the production methods permitted.