Bioreactor Sensors and Automation

Purpose and scope

Bioreactor instrumentation provides the measurements required to observe biological conditions, execute automated control, detect abnormal operation, and document process performance. Reliable control depends on the complete measurement chain—from the process sensor through signal transmission, controller configuration, operator display, alarm handling, and electronic record.

This article addresses instrumentation and automation used with stainless-steel and single-use bioreactors and fermenters. It covers sensor selection, installation, measurement performance, calibration, control loops, recipes, alarms, data handling, failure detection, maintenance, and lifecycle control.

The mechanical and process architecture is addressed in Bioreactor and Fermenter Design, Functionality, and Process Control. Detailed verification is addressed in Bioreactor Qualification and Lifecycle Control. Full computerized-system validation and Part 11 assessment remain separate but connected activities.

Lifecycle position and system boundary

Instrumentation and automation requirements should be established during bioreactor design rather than added after mechanical selection. The intended process determines:

Parameters that must be measured
Required measurement ranges
Accuracy and resolution
Response-time requirements
Control-loop functions
Alarm and interlock requirements
Data-recording frequency
Recipe and phase logic
Calibration strategy
Redundancy or alternate measurements
Failure-response requirements
Electronic-record and data-retention needs

The system boundary may include:

Process sensors and probes
Sensor housings and process connections
Transmitters and signal converters
Mass-flow controllers
Pumps and variable-frequency drives
Valve actuators and position feedback
Programmable logic controller or distributed control system
Human-machine interface
Recipe-management functions
Alarm and event handling
Local and remote data storage
Historian or batch-reporting interface
Network, time-service and external-system interfaces

A boundary drawing should distinguish bioreactor equipment functions from shared automation infrastructure, facility utilities, manufacturing systems and separately managed computerized systems.

Measurement and control architecture

A bioreactor measurement does not originate at the HMI. It passes through a chain of physical and electronic elements:

The process condition acts on the sensing element.
The sensor produces a physical or electrical response.
A transmitter or signal conditioner converts that response.
The signal is transferred to a controller input.
Scaling and engineering units are applied.
Control logic compares the value with the applicable setpoint.
The controller manipulates an output.
The value, output and system state are displayed and recorded.

Each element can introduce error, delay, noise, loss of resolution or incorrect configuration.

The illustration below presents this relationship from the process sensor through the control and operator interfaces.

Bioreactor sensors connected through transmitters and a PLC to the HMI, control outputs, alarms, and electronic records — Bioreactor measurement architecture connecting process sensors and transmitters with the controller, HMI, control outputs, alarms and electronic process records.

Instrument criticality and measurement requirements

Not every bioreactor measurement has the same GMP significance. Criticality depends on intended use rather than instrument type.

A temperature sensor used for product control can have a different impact from a temperature indicator used only for maintenance troubleshooting. A pressure transmitter used to protect the vessel may perform a safety function, while another pressure signal may support a GMP process record.

Criticality assessment should consider:

Relationship to a critical process parameter
Potential influence on product quality
Use in automated control
Use for batch acceptance or release
Role in contamination or sterile-boundary control
Role in an alarm, interlock or protective function
Availability of an independent measurement
Detectability of failure
Consequence of incorrect, delayed or missing data

Measurement requirements should define more than the nominal operating setpoint.

Requirement	Design consideration
Range	Must include normal, challenge, cleaning, sterilization and credible excursion conditions
Accuracy	Must be suitable relative to the process tolerance
Resolution	Must allow meaningful display, control and recording
Repeatability	Must support consistent measurement under equivalent conditions
Response time	Must be fast enough for the process dynamics and control objective
Drift	Must remain acceptable throughout the intended operating period
Environmental resistance	Must address temperature, pressure, moisture, vibration and cleaning exposure
Calibration	Must be achievable using appropriate standards and procedures
Data use	Must identify whether the value controls, alarms, calculates or records GMP information

Using an instrument with a range far beyond the actual process can reduce useful resolution or measurement capability. Selection should reflect the complete operating and maintenance envelope without unnecessarily sacrificing performance in the normal process range.

pH measurement

Glass-electrode pH probes are widely used in reusable bioreactors. A typical assembly includes:

pH-sensitive glass membrane
Internal reference system
Reference junction
Electrolyte
Temperature-compensation element
Sensor body and sanitary process connection
Cable or digital transmitter

Probe performance can be affected by:

Coating or fouling
Reference-junction degradation
Electrolyte depletion
Cracked glass
Electrical noise
Temperature-compensation errors
Repeated SIP exposure
Long operating duration
Improper storage
Process-medium composition

The illustration below shows the principal internal features responsible for pH measurement and the locations susceptible to degradation.

Cross-section of a sterilizable bioreactor pH probe showing the glass membrane, reference junction, electrolyte, and temperature sensor — A sterilizable pH probe uses a glass measurement membrane and reference system to determine process pH. Membrane condition, reference stability, temperature compensation and repeated processing exposure affect measurement reliability.

Calibration and in-process verification

A common strategy includes:

Two-point calibration before use
Use of appropriate, traceable buffer standards
Verification that the calibration slope and offset are acceptable
Confirmation of temperature compensation
Post-process or post-use verification
Assessment of drift between pre-use and post-use checks

The specific strategy depends on process duration, probe technology, system design and whether the probe can be removed or checked during operation.

A post-use discrepancy does not automatically establish when the measurement became inaccurate. The assessment should consider the error magnitude, direction, process trends, redundant indications, laboratory results, control activity and last documented acceptable condition.

Dissolved-oxygen measurement

Dissolved oxygen may be measured using electrochemical or optical technology.

Electrochemical sensors

Electrochemical probes consume a small amount of oxygen as part of the measurement mechanism. Performance depends on:

Membrane condition
Electrolyte condition
Polarization
Flow past the sensing surface
Temperature compensation
Cathode and anode condition
Calibration stability

Maintenance can include membrane and electrolyte replacement.

Optical sensors

Optical sensors use an oxygen-sensitive luminescent material. Oxygen affects the measured light response, allowing dissolved-oxygen concentration to be calculated.

Potential advantages include reduced oxygen consumption and lower routine maintenance. However, optical components can age, drift or be affected by repeated sterilization and process exposure.

The following illustration compares the two measurement principles.

Comparison of electrochemical and optical dissolved-oxygen probes used for bioreactor process measurement — Electrochemical and optical dissolved-oxygen probes use different sensing principles and therefore have different maintenance, response, drift and sterilization-resistance characteristics.

Calibration conditions

Dissolved-oxygen calibration should define:

Zero reference
Span reference
Temperature
Pressure
Gas composition
Agitation and flow conditions
Stabilization time
Acceptance criteria

A nominal “100%” calibration depends on the defined gas, pressure, temperature and liquid conditions. Changes in atmospheric pressure, vessel pressure or gas composition can affect the relationship between oxygen concentration and percent saturation.

Temperature measurement

Resistance temperature detectors are commonly used because they provide appropriate accuracy and stability across typical bioreactor operating ranges. Thermocouples may be used where their characteristics better suit the application.

Design considerations include:

Measurement range
Required accuracy
Sensor class
Number of sensing elements
Immersion depth
Thermowell design
Response time
Installation location
SIP temperature exposure
Transmitter and controller scaling

Large bioreactors may contain multiple temperature sensors for control, indication, verification or sterilization monitoring. Their functions should be distinguished. Agreement criteria should reflect expected sensor accuracy, location and process uniformity.

The process-control sensor should not be assumed to demonstrate complete vessel temperature distribution during qualification or SIP validation.

Pressure measurement

Pressure measurements may support:

Vessel-pressure control
Exhaust backpressure control
Filter-differential pressure
Single-use assembly protection
SIP pressure monitoring
Leak or pressure-hold testing
Alarm and relief-system coordination

Pressure transmitters must be compatible with the applicable process fluids, cleaning agents and SIP conditions. Installation should avoid condensate pockets, gas traps, blocked impulse paths and orientations that impair drainage.

The control range should provide useful measurement resolution while accommodating credible operating and excursion conditions.

Gas-flow measurement and control

Mass-flow controllers can regulate air, oxygen, nitrogen and carbon dioxide supplied to the vessel. Their performance affects:

Dissolved-oxygen control
pH control
Gas-transfer capability
Carbon dioxide removal
Overlay conditions
Foam behavior
Process calculations

Requirements should address:

Gas identity
Minimum and maximum flow
Accuracy and repeatability
Turndown
Inlet-pressure requirements
Gas correction factors
Leak-tightness
Response time
Communication protocol
Failure position
Calibration for the intended gas

Substitution of one gas, controller or calibration basis for another requires technical evaluation. A controller calibrated for one gas may not report the correct mass flow for a different gas without appropriate configuration or correction.

Facility supply and distribution requirements are addressed in Process Gas Systems for GMP Manufacturing.

Weight, level and feed measurement

Bioreactor systems may use load cells, platform scales, balance feedback, pressure-derived level or other methods to determine vessel contents and material additions. Load-cell performance can be affected by:

Rigid piping connections
Hose tension
Vessel supports
Thermal expansion
Agitator vibration
Uneven load distribution
Zeroing practices
Environmental disturbance
Addition-system contact with the vessel

Feed delivery may be verified by:

Pump calibration
Source-container weight loss
Receiving-vessel weight gain
Flow measurement
Time and pump-speed calculation
Material reconciliation

The selected method should be capable of detecting delivery errors relevant to the process. Pump speed or running time alone may not demonstrate actual delivery when tubing wear, occlusion, loss of prime or blockage is possible.

Foam and process-state detection

Foam probes commonly detect contact between foam and a conductive or capacitive sensing element. Their performance can be affected by:

Probe coating
Liquid splashing
Condensate
Installation height
Sensitivity settings
Process conductivity
Electrical grounding

Automated antifoam addition should include appropriate limits because excessive antifoam can affect oxygen transfer and downstream processing.

Other process-state measurements may include:

Capacitance-based viable biomass
Optical density
Raman or near-infrared measurements
Conductivity
Redox potential
Glucose, lactate or metabolite analyzers
Off-gas oxygen and carbon dioxide

These technologies require an intended-use-specific assessment of measurement reliability, calibration or model maintenance, sampling representativeness, interfaces and data use.

Single-use sensors

Single-use bioreactors can use disposable:

Optical pH and dissolved-oxygen sensor patches
Pressure sensors
Temperature elements
Conductivity sensors
Flow sensors
Foam or level detectors

The complete measurement function may include a disposable sensing element and reusable reader or transmitter. Controls should address:

Correct sensor type and range
Supplier and lot traceability
Irradiation effects
Shelf life and storage
Reader compatibility
Installation and alignment
Calibration or factory calibration
Single-point adjustment, where applicable
Connection to the control system
Replacement or failure response

Factory calibration does not eliminate the need to establish suitability for the intended measurement and process conditions.

Sensor installation and sanitary integration

Calibration cannot compensate for an inappropriate installation. Probe location and orientation affect whether the sensor measures representative process conditions. Installation design should consider:

Required immersion depth
Working-volume range
Impeller and sparger location
Vessel circulation pattern
Gas entrainment
Surface turbulence
Addition points
Potential stagnant regions
Mechanical vibration
Accessibility
Drainability
Hygienic connection design
SIP steam access and condensate removal

The illustration below compares representative placement within the mixed process volume with locations susceptible to gas pockets or stagnant conditions.

Comparison of proper bioreactor sensor placement in representative mixed zones with improper placement near the liquid surface, bubble plume, and vessel wall — Properly installed bioreactor sensors are fully immersed in representative mixed-liquid regions and positioned away from direct bubble interference, impeller sweep, vessel walls, and stagnant zones.

Sensor location should be justified by vessel geometry and process behavior rather than mechanical convenience. A stable signal can still be incorrect if it represents a localized condition rather than the bulk process.

Signal transmission and configuration

Sensor signals may be analog, digital or network-based. The measurement chain can include:

Sensor
Preamplifier
Transmitter
Analog-to-digital conversion
Input module
Controller scaling
Signal filtering
Display formatting
Historian processing

Configuration elements requiring control include:

Input type
Engineering units
Lower and upper range
Linearization
Damping or filtering
Temperature compensation
Gas correction factor
Calibration coefficients
Decimal precision
Alarm limits
Bad-signal behavior

Incorrect scaling can produce a believable but inaccurate displayed value. Verification must therefore challenge the complete measurement chain, not only the sensor or local transmitter.

Electrical design should address shielding, grounding, isolation, cable routing, moisture protection and electromagnetic interference. Noise filtering should not be so aggressive that it conceals a meaningful process change or creates unacceptable response delay.

Calibration and metrology control

Calibration establishes the relationship between the instrument indication and an appropriate reference over the required measurement range. The strategy should be based on intended use, criticality, drift history, process tolerance and manufacturer information.

The following illustration presents the controlled lifecycle from scheduled calibration through assessment and return to service.

Bioreactor sensor calibration lifecycle showing as-found verification, acceptance decision, out-of-tolerance assessment, adjustment, as-left verification, documentation, and return to service — The calibration lifecycle uses as-found results to determine whether an instrument remains acceptable. Out-of-tolerance instruments require impact assessment, adjustment or repair, and acceptable as-left verification before documentation and return to service.

Calibration procedure

The calibration procedure should define:

Instrument identification
Calibration range and points
Reference standards
Environmental or process conditions
Stabilization requirements
Permitted adjustments
Acceptance limits
As-found and as-left recording
Calculation and rounding conventions
Failure and escalation requirements
Labeling and return-to-service controls

Acceptance limits should be appropriate relative to the process requirement. A calibration tolerance equal to the complete allowable process variation may leave no allowance for other sources of measurement uncertainty or process variation.

As-found and as-left results

The as-found result shows instrument condition before adjustment. It is essential for determining whether previously generated data may have been affected.

The as-left result demonstrates instrument condition after adjustment or calibration work. Recording only the as-left result removes evidence needed to evaluate historical performance.

An out-of-tolerance condition requires assessment of:

Error magnitude and direction
Potentially affected period
GMP uses of the measurement
Process values recorded during that period
Control actions driven by the signal
Available independent or redundant evidence
Potential product or batch impact
Corrective action and recurrence risk

Program-level requirements are addressed in GMP Calibration Program and Metrology Control.

Sterilization, cleaning and sensor survivability

Reusable probes can be exposed to:

Steam-in-place cycles
Clean-in-place chemicals
Elevated temperature
Pressure cycling
Rapid temperature changes
Process media
Storage solutions
Repeated insertion and removal

These exposures can cause:

pH reference-junction deterioration
Glass-membrane aging
DO membrane damage
Optical-patch degradation
Seal or O-ring deterioration
Cable damage
Moisture intrusion
Changed response time
Increased drift

The sensor specification and maintenance strategy should reflect actual exposure. A probe described as sterilizable may still have a finite validated or manufacturer-supported life.

CIP and SIP interfaces should be evaluated with the applicable Clean-in-Place system qualification and Steam-in-Place Utility Systems rather than treated only as sensor environmental conditions.

Response time and measurement dynamics

A sensor can remain accurate at equilibrium while responding too slowly for effective process control.

Dynamic performance depends on:

Sensor technology
Protective membrane or coating
Thermowell construction
Process flow at the sensing surface
Signal filtering
transmitter damping
controller scan time
network update frequency
historian collection interval

Response requirements should be related to the process. A rapid addition or gas-control loop may require faster measurement than a slowly changing parameter used only for trending.

Changes in response time can be an early indication of fouling, membrane aging, poor flow or signal-processing changes.

Drift trending and predictive control

Calibration history can reveal progressive degradation before a complete failure occurs. Trending may use:

Calibration offset
Calibration slope
Zero and span error
Response time
Number of SIP cycles
Operating hours
Probe age
Failure and replacement history

The illustration below shows gradual drift approaching an established limit.

Sensor calibration drift increasing over operating time or repeated sterilization cycles toward warning and acceptance limits — Trending calibration deviation against operating time or SIP cycles can identify progressive sensor drift before the acceptance limit is exceeded and support risk-based replacement or calibration-interval decisions.

A warning threshold may support investigation or planned replacement before an out-of-tolerance condition occurs. It should not be confused with the approved calibration acceptance limit.

Calibration intervals may be shortened or extended when supported by adequate history, risk assessment and controlled change. Interval changes should not be based only on scheduling convenience.

Control loops and cascade control

A control loop connects:

Measured process variable
Setpoint
Control algorithm
Manipulated output
Physical process response

Proportional-integral-derivative control is commonly used, but appropriate tuning depends on process dynamics. Poor tuning can produce:

Sustained oscillation
Overshoot
Slow recovery
Excessive actuator movement
Alternating acid and base addition
Abrupt cascade transitions
Apparent control that masks equipment-capacity limitations

Loop performance should be evaluated across the approved operating range and relevant process phases.

Dissolved-oxygen cascade

A dissolved-oxygen cascade may manipulate:

Agitation
Airflow
Oxygen enrichment
Backpressure

The sequence should define:

Order of outputs
Transition points
Output limits
Ramp rates
Interaction with foam control
Shear restrictions
Failure behavior
Manual-mode restrictions

If all outputs reach their limits while dissolved oxygen remains below setpoint, the automation may be operating correctly while the physical system lacks sufficient oxygen-transfer capability.

pH control

pH control may manipulate:

Acid addition
Base addition
Carbon dioxide
Gas stripping
Feed composition

The control strategy should account for mixing delay and local concentration effects. Dosing limits, deadbands and totalized additions can help prevent uncontrolled correction following a faulty or slowly responding measurement.

Temperature control

Temperature control may manipulate heating and cooling valves, temperature-control-unit setpoints or circulation functions. Tuning must reflect process volume, jacket response, metabolic heat generation and utility capacity.

Recipes, phases and parameter management

Automation may organize operation into controlled phases such as:

Preparation
Leak testing
CIP
SIP
Media charge
Temperature conditioning
Inoculation
Cultivation
Feeding
Induction
Perfusion
Harvest
Shutdown

Each phase should define:

Entry conditions
Required equipment state
Setpoints and limits
Active control loops
Permitted operator actions
Transition criteria
Hold conditions
Alarm behavior
Abort and recovery logic

Critical parameters should be protected from unauthorized or undocumented modification. The system should distinguish approved recipe values, temporary operator adjustments, engineering configuration and maintenance settings.

Alarms, interlocks and failure response

Alarms notify personnel that defined conditions require attention. Interlocks prevent or terminate an action when required conditions are not satisfied. They are related but not interchangeable.

The strategy should address:

Alarm priority
Setpoint and delay
Deadband
Latching
Acknowledgement
Suppression or shelving
Required response
Escalation
Event recording
Return-to-normal behavior

Relevant failure conditions include:

Failure	Detection or secondary evidence	Possible automated response
Sensor signal loss	Bad-quality flag, open circuit or communication alarm	Hold output, transfer control or enter defined safe mode
Frozen signal	No response despite known process change	Alarm or sensor-disagreement detection
Gradual drift	Calibration trend, redundant sensor or material balance	Investigation, recalibration or replacement
Excessive noise	Rate-of-change or variability monitoring	Alarm, alternate measurement or controlled filtering
Mass-flow-controller failure	Command-versus-feedback disagreement	Isolate affected gas or transfer control
Feed-pump failure	Weight, flow or delivered-mass discrepancy	Stop phase, alarm or controlled recovery
Valve-position failure	Command-versus-position disagreement	Prevent sequence transition
PLC or communication failure	Watchdog or communication alarm	Defined fail-safe equipment state
Historian failure	Interface or data-collection alarm	Local buffering or controlled operating restriction
Loss of power	System-status and equipment-state detection	UPS-supported control, controlled shutdown or recovery procedure

Fail-safe behavior must consider the biological process. Abruptly closing every valve or stopping every output may protect equipment but destroy the batch or create another hazard. Safe-state design must evaluate equipment, personnel, containment and product consequences.

Manual mode and overrides

Manual control may be necessary for setup, maintenance, troubleshooting or controlled recovery. It can also bypass normal automation protections.

Controls should define:

Authorized users
Permitted manual functions
Status indication
Alarm behavior
Interlock behavior
Time limits
Reason documentation
Event or audit-trail recording
Review and return to automatic operation

A manual override should not silently disable a GMP control or remove evidence of the intervention.

Electronic records and data integrity

Bioreactor automation can create or maintain records used for:

Batch execution
Process trending
Deviation assessment
Process validation
Batch review or release
Calibration
Maintenance
Recipe management
Alarm and event history

Relevant controls may include:

Unique user identification
Role-based access
Audit trails
Accurate timestamps
Time synchronization
Record protection
Backup and recovery
Data retention
Review of changes
Controlled report generation
Secure interfaces

The applicability of 21 CFR Part 11 depends on whether electronic records or signatures are used to meet FDA record requirements. Detailed assessment is addressed in 21 CFR Part 11 Compliance and Checklist.

Data should remain attributable, legible, contemporaneous, original, accurate, complete, consistent, enduring and available. Application of these principles is addressed in ALCOA+ Principles and Implementation.

Under 21 CFR 211.68, automatic, mechanical and electronic equipment used in drug manufacturing must be routinely calibrated, inspected or checked under a written program designed to assure proper performance.

Interfaces and time synchronization

Bioreactor controls may interface with:

Data historians
Manufacturing execution systems
Electronic batch records
Laboratory systems
Building-management systems
Utility monitoring systems
Recipe servers
Network time services
Remote alarm-notification platforms

Interface requirements should define:

Source and destination
Data elements
Units and precision
Transfer frequency
Data ownership
Error detection
Retry or buffering
Duplicate prevention
Time basis
Reconciliation
Recovery following interruption

The sending system, receiving system and interface can each function individually while transferred data remain incomplete or incorrect. Interface verification must therefore confirm end-to-end transfer and error handling.

Instrument and automation qualification boundary

Equipment qualification should verify GMP-relevant instrumentation and automation functions such as:

Installed sensor and transmitter identity
Range and scaling
Calibration status
Control-loop behavior
Recipe and phase execution
Alarm and interlock operation
Valve sequencing
Failure and recovery response
Data display and recording
Applicable interfaces
Configuration baseline

A broader computerized-system lifecycle may also address:

Supplier assessment
Software and configuration specifications
Requirements traceability
User and access management
Audit trails
Electronic signatures
Backup and recovery
Cybersecurity
Infrastructure
Periodic access and audit-trail review
Software change control

The two scopes must be coordinated. Mechanical qualification should not claim that the entire computerized system has been validated merely because several HMI screens and control loops were tested. The validation strategy is addressed in Computerized Systems Validation Planning.

Maintenance and lifecycle control

Lifecycle control should include:

Scheduled calibration
Preventive maintenance
Probe inspection and replacement
Review of calibration drift
Review of recurring alarms
Control-loop performance review
Backup and recovery verification
Configuration management
Spare-part and obsolescence control
Supplier notices
Change control
Periodic assessment
Risk-based requalification

Relevant changes include:

Sensor make, model or technology
Sensor location or insertion depth
Range or scaling
Calibration procedure or interval
Alarm or interlock setting
Control-loop tuning
Cascade order
Recipe parameter
PLC logic
HMI configuration
Firmware
Interface
Historian or reporting function
Network or infrastructure component

Replacement described as “like-for-like” still requires confirmation that specifications, firmware, configuration, materials, performance and installation remain equivalent.

General reliability controls are addressed in Preventive Maintenance and Equipment Reliability. Instrument and maintenance modifications should follow Change Impact on Calibration and Maintenance.

Periodic review

Periodic review should integrate:

Calibration completion and out-of-tolerance history
Drift and adjustment trends
Sensor failures and replacement frequency
Control-loop performance
Recurring or nuisance alarms
Overrides and manual operation
Automation deviations
Recipe and configuration changes
Access and audit-trail concerns
Data losses or interface failures
Backup and recovery results
Maintenance effectiveness
Obsolescence and supplier support
Open corrective actions
Requalification status

The conclusion should determine whether the existing calibration, maintenance, monitoring and qualification controls remain adequate or require revision.

Documentation

Controlled documentation should include, as applicable:

Instrument index
Instrument specifications
Criticality assessment
Process and instrumentation diagrams
Sensor-location drawings
Input/output list
Range and scaling records
Alarm and interlock schedule
Control narrative
Recipe and phase specifications
Calibration procedures and records
Configuration baseline
Software and firmware versions
Interface specifications
User and role definitions
Backup and recovery procedures
Maintenance instructions
Qualification records
Change history
Periodic-review conclusions

Documentation must connect each important process measurement and automated function with its requirement, design configuration, verification method and lifecycle control.

Conclusion

Bioreactor control depends on more than selecting accurate probes. Measurement reliability results from correct sensor technology, representative installation, controlled signal processing, appropriate calibration, adequate response, sound control-loop design, effective alarms, trustworthy electronic records and disciplined lifecycle management.

Instrumentation and automation must be designed as part of the biological processing system. The objective is to ensure that process conditions are measured correctly, manipulated predictably, recorded reliably and maintained within the approved operating strategy throughout the equipment lifecycle.