Kelvin Probe & Air Photoemission Systems Knowledge Database
The Kelvin probe is a non-contact, non-destructive measurement device used to investigate properties of materials. It is based on a vibrating capacitor and measures the work function difference or, for non-metals, the surface potential, between a conducting specimen and a vibrating tip. The work function is an extremely sensitive indicator of surface condition and is affected by adsorbed or evaporated layers, surface reconstruction, surface charging, oxide layer imperfections, surface and bulk contamination etc.
This webpage provides information on the science behind the Kelvin probe and the air photoemission systems, ideas for applications, examples of measurements and some comments on current apparatus available. All have been compiled by Professor Iain D. Baikie, who has extensive experience in these techniques.
If you did not find what you were looking for, please contact us directly. We welcome any comments and questions related to the Kelvin Probe, Air Photoemission Systems or this site.
The Kelvin Probe is an extremely sensitive analytical tool measuring changes in contact potential difference (cpd) between a reference material and a sample to less than 0.001 V.
The essence of the cpd is the difference in fermi-levels (in the simple case of a metal this is the energy of the most energetic electron within the outer electron band with respect to the vacuum level, Evac).
If we assume that the reference material fermi-level is unchanged during the measurement (in practical situations this may have to be determined via control experiments), then the changes in contact potential difference of the Kelvin 'Junction' can be wholly ascribed to changes occurring at the sample surface.
Depending upon the direction of your study you may find this term expressed in different ways: contact potential, fermi-level, work function, surface potential, corrosion potential, surface dipole, etc. I would be delighted to learn of other expressions to extend this list.
The Kelvin method was first postulated by the renowned Scottish scientist Sir William Thomson, later to be known as Lord Kelvin, in 1898 when he presented a public lecture to the British Institution on the 'contact electrification of metals'.
Over a century later the method he proposed is at the forefront of materials research and start-of-the-art equipment development. In essence the method is simplicity itself: take two conducting materials, allow them to come into electrical contact, then sense any flow of charge from one material to the other.
Kelvin used two large metal plates and a gold leaf electroscope to demonstrate this surface charging effect: he showed that a potential is generated between the surfaces of two conductors when they are brought into electrical contact.
This experiment forms the basis of the Kelvin Probe which has been developed into a highly sensitive tool for analysing the surface properties of materials. The range of materials it can be applied to is constantly increasing: metals, alloys, semiconductors and even insulators, yes insulators!
The air photoemission systems are the latest addition to our large surface analysis range. These systems measure the absolute work function of a material by photoemission in air, with no vacuum required. With an excitation range of 3.3 eV to over 7 eV, our system is capable of measuring the absolute work function of metals. The addition of a surface photovoltage (SPV) and surface photovoltage spectroscopy (SPS) source, enables the full bands of semiconductors to be measured in one system. Each APS system comes with our gold-standard Kelvin probe system, capable of contact potential difference measurements backed up by the APS absolute measurement.
The work function (wf) is the least amount of energy required to remove an electron from the surface of a conducting material, to a point just outside the metal with zero kinetic energy.
As the electron has to move through the surface region, it’s energy is influenced by the optical, electric and mechanical characteristics of the region.
Hence, wf is an extremely sensitive indicator of surface condition and is affected by absorbed or evaporated layers, surface reconstruction, surface charging, oxide layer imperfections, surface and bulk contamination, etc.
When light is incident on a material such as a metal or a semiconductor, the photons may have enough energy to liberate electrons from the surface, a process known as the Photoelectric Effect. Photons having insufficient energy will not liberate electrons, while photons of just enough energy will liberate a few electrons; photons of much more energy than the work function will liberate a lot of electrons.
The energy required for electrons to escape the material is termed the work function. By varying the energy of the incoming light, the absolute work function can be established. Based on Fowler’s analysis of photoemission, the square root (cube root for semiconductors) of the photoelectron yield is plotted on a graph versus the incident photon energy.
Lord Kelvin was born William Thomson in Belfast in 1824. Son of a mathematics professor, he was educated in Scotland at the University of Glasgow (entering when he was 10 years old) and in Cambridge, where he graduated in 1845. He also spent a year working with Pierre and Marie Curie and Henri Becquerel in Paris. Thomson set-up the first dedicated physics laboratory in the United Kingdom at the University of Glasgow in 1846. For his impressive and wide ranging efforts to science, including notable contributions to thermodynamics, electromagnetic theory, communication (first transatlantic cable link) and of course contact potentials, he was knighted in 1866 and received the title Baron Kelvin of Largs in 1892. The day after Kelvin formally retired from Glasgow University he enrolled as a student.
Lord Kelvin died in 1907 and was appropriately buried next to Sir Issac Newton in Westminster Abbey. He published 661 papers, took out 70 patents and had more initials after his name than any other man in the British Commonwealth!
The picture below shows a statue of Lord Kelvin that was erected in Glasgow to commemorate his achievements. Kelvin was one of the most gifted scientists of his generation. He was, by all accounts, a talented orator and was sufficiently confident in his experimental methods to incorporate public demonstration of his methods. It is perhaps fitting to reproduce a sketch of his original experimental arrangement which incorporated large plates of copper and zinc (which have work functions of 4.6 eV and 4.7 eV respectively when cleaned) mounted on insulating shafts. I am indebted to Ing. Kees van der Werf, of the University of Twente, in the Netherlands, a skilled scientist and artist, for the reproduction.
Kelvin detected the flow of charge, upon making electrical contact, using either the gold leaf electroscope as shown here, or a more sensitive quadrant electrometer. A further development, using a home-made Daniell s cell, allowed nulling of the field between the plates. Remembering that this method pre-dated the discover of the electron by J. J. Thomson (no relation to William) in 1905, Kelvin explained the effect of leaf separation using the terms resinous electricity equating to the flow of positive charge and glasseous electricity for negative charge..
A diligent reporter of the physical world, Kelvin described an additional phenomenon he observed upon introducing a substance he had recently received from Henri Becquerel derived, from uranium ore pitchblende between the plates. He remarked that the electroscope immediately discharged indicating the presence of a conducting pathway between the plates. This last result is not surprising since the radiation given off by the decay of radium from the zincblende would ionize the air. This was not especially good news for the health of his audience and perhaps indicates why, at many notable British institutions, the senior members of staff tend to sit at the back of the auditorium especially for experimental demonstrations!
Kelvin was 73 at the time of this demonstration and died of influenza 10 years later. Herni Bequerel was awarded the Nobel Prize in 1920 together with Marie and Pierre Currie on the discovery and isolation of radium from uranium ore. Marie Curie however suffered radiation burns in her study of radium.
The Kelvin probe is a non-contact, non-destructive vibrating capacitor device used to measure the work function difference, or for non-metals, the surface potential, between a conducting specimen and a vibrating tip. Although not as well known as some other surface analysis techniques, the Kelvin probe has undergone a dramatic renaissance over the last few years. Advances in hardware design and signal processing technology have improved the resolution of the instrument and have facilitated use in vacuum.
Improvements have also been made to the spatial resolution of the technique, whilst new designs can map surface properties with resolution in the 50 nm range. Several spectroscopic variants have also been developed for the analysis of semiconductor surfaces and thin films. The Kelvin probe is a non-invasive technique, yet it is extremely sensitive to changes in the top-most atomic layers, such as those caused by deposition, absorption, corrosion and atomic displacement. In some cases it can detect less than one-thousandth of an absorbed layer.
The Kelvin probe is a non-contact, non-destructive vibrating capacitor device used to measure the work function (wf) of conducting materials or surface potential (sp) of semiconductor or insulating surfaces. The wf of a surface is typically defined by the topmost 1-3 layers of atoms or molecules, so the Kelvin probe is one of the most sensitive surface analysis techniques available. KP Technology systems offer very high wf resolution of 1-3 meV, currently the highest achieved by any commercial device.
The Kelvin probe does not actually touch the surface; rather an electrical contact is made to another part of the sample or sample holder. The probe tip is typically 0.2 - 2.0 mm away from the sample and it measures the traditional work function , i.e. that found in literature tables. Other techniques, using very sharp tips some 10 s of nanometers away from the sample, measure very reduced and distorted work functions due to the close separation of the tip and the sample.
TThe physical form of a Kelvin probe is a head unit containing a voice coil driving system and integral amplifier suspended above a sample. The vibrating tip and the sample form a capacitor, having ideal, or parallel-plate, geometry. As the tip vibrates electric charge is pushed around the external detection circuit. By careful control of the tip potential and automatic capture and analysis of the resulting waveform, both the potential across the capacitor and the capacitor spacing can be calculated to very high resolution. In scanning form the tip is steered across the sample surface using a high resolution 3-axis translator. The spatial resolution of the tip is approximately the tip diameter; we typically use tips of 2 mm and 0.05 mm as standard in air, and anything from a sharp tip to 10 mm diameter in ultrahigh vacuum. The resulting 3D surface work function images contain information about surface structures, surface composition, thin films, defects, etc. In time-varying mode artifacts such as oxidation (corrosion) and defect relaxation can be observed.
For semiconductor surfaces, both organic (polymer) and inorganic (Si, Ge, CdS, etc), the Kelvin Probe is the only way to directly measure the Fermi-level. Changes in Fermi-level, caused by illumination with white or monochromatic light, result in energy band shifts which can be used to characterize interface and bulk defect states. These techniques are termed surface photovoltage (SPV) and surface photovoltage spectroscopy (SPS) for which KP Technology can supply both software and accessories for our systems.
The traditional Kelvin probe actually produces the work function difference between the tip and sample. Typically the tip is calibrated against a reference surface, such as gold. However KP Technology is the only company to offer absolute Kelvin probes, which combine the Kelvin method and Einstein photoelectric effect to produce absolute work function values (in eV).
KP Technology has developed dedicated head units for ambient, controlled atmosphere, relative humidity and ultrahigh vacuum environments. All of our systems share the off-null, height regulation (ONHR) method invented by Prof. Baikie and consequently produce stable high signal levels, repeatable and high- resolution measurements. The rapid growth of the company since 2000 means that our systems are found in leading materials research laboratories worldwide. We pride ourselves on our rapid post-sales support and our ability to assist technical and data interpretation queries.
The properties of many materials are governed by their density of states (DOS) near the Fermi level. Our air photoemission system (APS) is capable of probing these DOS by differentiating the detected photoelectron yields by the incident photon energy.
The DOS measurement can thus be compared to molecular orbital calculations for the material under investigation. DOS data collected with the APS in air is shown for copper with the data for all measured samples consistent with literature.
All KP Technology systems are built upon the Baikie System . This design is unique to KP Technology and allows for features unsurpassed by any other company. We offer the highest work function/surface potential resolution, and the highest signal to noise ratio thanks to our voice coil driver which allows for high rejection of driver talk Our off null signal detection, means our probes have vastly improved resolution, and our height regulation features allow for stable, reliable and repeatable results. Full digital control of all the probe s parameters, along with customer support recognised by our Queen s Award , makes our Kelvin probes a delight to use.
When two materials with different work functions are brought together, electrons in the material with the lower work function flow to the one with the higher work function. If these materials are made into a parallel plate capacitor, then equal and opposite surface charges will form. The voltage developed over this capacitor is called the contact potential and measuring it is done by applying an external backing potential to the capacitor, until the surface charges disappear., At this point the backing potential will equal the contact potential.
The traditional Kelvin probe method consists of a flat circular electrode (termed the reference electrode) suspended above and parallel to a stationary electrode (the specimen), thus creating a simple capacitor. In 1932 William Zisman, of Harvard University, introduced a new method to measure the contact potential. He mounted a vibrating reference surface, or tip, just above a sample electrode. Here the output voltage varied periodically as the tip vibrated, with the peak-to-peak voltage dependent on the difference between the contact potential and the external voltage.
This technique led to the development of systems that automatically track shifts in the contact potential due to changes in the work function of the sample.
A major asset of this method is that the surfaces do not need to touch each other. It also requires only very weak electrical fields, which are not likely to influence the electrical or chemical structure of the material. Several ingenious mechanisms have been used to achieve the required variation in spacing between the tip and sample. For vacuum applications piezoelectric and voice-coil drivers are most convenient. A typical probe design for ultrahigh vacuum is made from stainless steel, including the suspension system that controls the tip movement.
The tip vibrates with an amplitude of 0.1 to 1mm at a frequency of 30-300 Hz with it’s mean position kept constant to within 50 nm. A complete scanning Kelvin probe system includes a digital oscillator to drive the tip movement, a tip actuator, a signal amplifier and a scan controller. A computer with a data acquisition system is used to control the instrument and to capture measured data.
The wf difference between two surfaces can be found by measuring the flow of charge when the two conducting materials are connected (Figure B). However, this produces a once only measurement, as the surfaces become charged, and the charge must dissipate before another measurement can be made. By using a vibrating probe, a varying capacitance is produced. This is given by: C=Q/V=(epsilon.A)/t (Equation 1) Where C is the capacitance, Q is the Charge, V is the Potential, Epsilon 0 is the permittivity of the dielectric (in an air probe the dielectric is air). A is the surface area of the capacitor and d is the separation between the plates. Therefore, as the separation d increases the capacitance C decreases. As the charge remains constant, the voltage V must increase.
As the probe oscillates above the sample the voltage change is recorded. The peak-to-peak output voltage Vptp, is given by: Vptp = (Delta V - Vb)R.C0.w.e.sin(w.t + f) (Equation 2) Where Delta V represents the voltage difference between probe tip and sample, Vb is the externally applied emf used to balance (or null) the circuit, R is the I/V converter feedback resistance, C0 is the mean Kelvin Probe capacitance, Omega the angular frequency of vibration and Theta the phase angle. Epsilon is the modulation index (d1/d0) where d0 is the average distance between the sample and the probe tip and d1 is the amplitude of oscillation of the probe.
The backing potential Vb is set to a range of potentials and a plot of Vptp verses Vb is made. As can be seen from Equation 2 this is a straight line. When Vptp = 0 the work function or contact potential of the surface is equal and opposite to Vb. Traditional phase sensitive detection methods utilise a LIA (lock in amplifier) to detect the null output condition. However, these systems have the disadvantage that at the balanced point, the signal to noise (S/N) ratio reaches a minimum, with the noise creating an offset voltage. A better method is to set Vb to a range of voltages and then extrapolate to the null signal condition, thus working to a high S/N ratio. Figures A to C (below) show various electron energy diagrams for two conducting specimens, 1 and 2.
Figure A Electron energy level diagram for two conducting specimens, where 1 and 2 are the work functions of the materials, and 1 and 2 represent their fermi levels.
Figure B If an external electrical contact is made between the two electrodes their fermi levels equalise and the resulting flow of charge (in direction indicated) produces a potential gradient, termed the contact potential Vc, between the plates. The two surfaces become equally and oppositely charged.
Figure C Inclusion of a variable "backing potential" Vb in the external circuit, permits biasing of one electrode with respect to the other. At the unique point where the (average) electric field between the plates vanishes, a null output signal results.
A variety of ingenious mechanisms have been used to generate the required spacing modulation including rotation, piezoelectric and voice-coil drivers- even piano wire has been attempted! In terms of vacuum applications piezoelectric and voice coil drivers remain the easiest to implement. The majority of the probe construction is stainless steel including the suspension system which consists of three laser-cut diaphragm springs. The vibration frequency (30 - 300 Hz), and amplitude of oscillation (0.01 - 1.00 mm) and probe mean spacing (± 50 nm) are controlled by a digital oscillator located in the host PC.
The voice-coil driving system is deliberately well removed (250 mm) from the tip ensuring that no traces actuator signal can be detected on the input signal, i.e. driver talk-over. The stainless steel tip customarily has a diameter of 2-3 mm and can be gold plated if desired. The suspension system can be easily adapted to multiple tip geometry allowing both high and low spatial resolution on one head stage. Alternative piezoelectric actuator designs such as Besocke s are customarily located in close proximity (10-20) mm from the sample. My experience of such drivers indicates that capacitive coupling in the form of talk-over noise can pose considerable problems particularly if a LIA based self-nulling detection system is used. In such cases it can be difficult, in the vacuum environment, to distinguish real changes in Vc from stray capacity effects- a clean Kelvin signal is essential for high resolution measurements.
The figures below show a 300 x 400 µm^2 section of an operational transistor imaged using a 0.1 mm diameter Pt tip. The bias potential changes are clearly visible through the native oxide layer above a single p-n junction, together with the associated topographic image.
One major problem associated with the null method is obviously the low signal to noise level at balance. Together with co-workers at Twente University in the Netherlands I developed an off-null approach. Here Vb is derived from a computer steered digital-to-analogue converter which is set to a range of potentials around the balance point. At each value Vptp is determined: a plot of Vptrp versus Vb is thus a straight line, the intersection with the Vb axis producing Vc. In this fashion changes in Vc can be determined to sub-mV resolution.
An advantage of this off-null method is that measurements are performed on high signal levels and the balance point is determined via extrapolation. A further benefit is that the gradient of the Vptp versus Vb line is very sensitive to changes in mean spacing and can be used to maintain a constant spacing either during an experiment or throughout a scan. In this fashion the Kelvin probe can map out surface potential changes simultaneously with sub-micron scale sample topography.
The mounting flange for the non-scanning probe is DN40 (CF2.75 inch OD) with CF64, CF100 used for our SKP systems. The UHV head stage features a voice coil actuator and user defined flange-to-sample spacing. Please note that our standard tip sizes are 4-8mm diameter and you would achieve 1-3 meV resolution with such tips.
The standard head unit is designed for normal geometry, i.e.; the insertion port is at 90^o to the sample plane. The quoted voltage resolution for a 4-8mm diameter tip is 1-3 meV: this is the highest resolution of any commercially available Kelvin probe on the market and sub mV resolution is achievable with the larger tip sizes. Note the actual work function resolution will depend upon the electrical and mechanical noise in the vicinity of the measurement cell.
Some customers monitor the work function change as a function of system pressure or substrate temperature. Accordingly the system software package allows for these other parameters to be recorded using an additional input channel and plotted against the change in work function. This is a digital system such that the Kelvin Probe frequency, amplitude of oscillation, mean spacing, backing potential and data acquisition parameters can be set by the user.
Further to this is an off-null system; i.e. the work function is calculated using two or more signal peak-to-peak measurements performed at backing potentials on either side of balance. Thus the actual Kelvin probe signal is visible at all times during measurement. In contrast, null-based detection systems require a phase sensitive detector and the Kelvin probe signal is nulled throughout.
AAmbient Kelvin probes (KP, SKP and APS in Kelvin probe mode) use a non-invasive, non-contact vibrating capacitor technique, which measures to mV resolution, the voltage between a vibrating micro-electrode and a conducting or semiconducting sample. The surface potential maps contain information on the sample s work function (which can be used to indicate chemical composition) but more importantly how the surface responds to treatment: such as deposited thin films, oxidation, illumination. Applications include the fundamental study of corrosion, surface passivation, low and high work function surfaces for ion and electron emission, fundamental surface adsorption studies, semiconductor oxide quality, etc. This system has also been utilised to provide surface charge imaging of non-conductors such as ptfe, platinium-pfte mixtures and charges deposited on top of silicon oxide. The vibrating tip (amplitude of vibration can be digitally changed from a few microns to 1-2 mm) moves in a plane-parallel fashion with respect to the sample. The mean tip-to-sample spacing is usually in the 0.3 - 1.0 mm range, although measurement can be made with larger tips with an off set of 2-5 mm.
The tip actuator is voice-coil driven, powered from a computer controller digital oscillator with a 0.2 - 0.3 Hz frequency resolution. Some of our smaller tips (diameter < 200 um) are shielded using a thin guard electrode. An automatic measurement procedure controls the (x,y,z) micro-translator (0.4 um resolution), amplitude and frequency of oscillation, tip and shield bias.
We utilise an off-null measurement procedure featuring a tracking routine to maintain an average tip to sample spacing to within approximately 1 micron, to avoid spurious work function changes associated with spacing changes. The voltage gradient above most samples is considerable and without the tracking routine measurements, made at different heights will, often not correlate well. The general-purpose system also includes a peeltier stage to allow sample heating and cooling and an optical illumination source to allow surface photovoltage and surface photovoltage spectroscopy measurements.
The Kelvin probe provides the relative work function between the tip and sample. The work function is highly sensitive to surface conditions: surface processing, roughness, chemical composition, charge and contamination will produce measurable changes in metal or semiconductor work functions.
The figure below shows how equilibrium is attained between two surfaces when they are placed in external electrical contact.
The surface Fermi-level of the semiconductor shifts during illumination. In this case the semiconductor is a p-n junction and we obtain the open circuit voltage. However as the Kelvin probe measures this via the front surface we also measure (in series) the surface trapping too. The figure below explains this.
This figure shows the effect the semiconductor-oxide interface plays in understanding surface potential changes measured with the Kelvin probe. The oxide and/or semiconductor-oxide interface charges up during illumination. This alters the threshold voltage as measured by the probe. These changes can be quite large: up to 0.5 V. The semiconductor interface can be very sensitive to even low levels of illumination of even a few lux.
We are currently using our Scanning Kelvin probes with tips of diameter 0.15-0.5 mm to scan the front surface of semiconductor solar cell materials. The resulting surface potential topographies provide information on the quality of the front electrical contact; the electrical characteristics of the buried p-n junction; the surface interface traps and the effects of varying light energy and intensity.
Such multi-crystalline cells are not highly efficient and this study produces information on the surface traps, recombination centers and degradation of cell performance with temperature. Via an appropriate biasing technique, the Kelvin probe can selectively provide information on the surface and/or the bulk phenomena. This data can be used to optimise solar cell efficiency. The diagrams show the front electrical contact composed of silver bus-bar (coloured red) and associated fingers.
These contacts are screen printed on top of a thin layer (60 nm) of either SiO2 or Si3N4 passivation. Avrid van der Heide, ECN Petten, from the Netherlands has kindly supplied these samples. We observe that the electrical contact between the silver layer and the underlying, highly-doped emitter layer (100 nm thick) is excellent, indicating that the drive-in diffusion of silver through the passivation was successful. Further SKP scans and point measurements performed under dark and white light illumination, termed the surface photovoltage (see the SPV section on this website) clearly show that the charge trapping lifetimes in the interface between the two types of surface passivation is completely different.
Further we can show, via a non-contact surface potential measurement that the open circuit voltage degrades with temperature. In this fashion we can directly measure the changes in sub-surface quasi Fermi-level as a function of light intensity and thus provide information on how the cell responds to different light intensities.
This is a Kelvin probe scan of a SiO2 passivated multicrystalline solar cell showing bus bar, silver fingers and semiconductor regions.
This scan is of a Si3N4 passivated solar cell. This is a higher performance cell than the SiO2 coated one due to the gettering effect of hydrogen on bulk traps. The silicon surface is much smoother and the open circuit voltage (measured with the Kelvin probe) is much larger.
An example of an in-situ work function topography of a clean Si(111) 7x7 single crystal is shown here. The work function is mostly flat with a depression in one area which is in the vicinity of the sample contact. Using optical pyrometry we can determine that this region of the sample was 50 ºC cooler and subsequent AESanalysis indicates a modest build-up of carbon contamination (< 10%).
The Kelvin probe has been used to study, in detail, the oxidation of Si surfaces, serving as a reminder of the sensitivity of the method. As the induced surface dipole at monolayer (ML) coverage is quite large- upwards of 1 V-, sub-mV work function accuracy infers resolutions of approximately 1/1000 ML in coverage. STM and AES can also be used to follow this reaction however STM topographies quickly become confused at relatively low coverage, e.g. > 0.1 ML and it is very difficult to obtain accurate chemical information below 0.3 ML. The KP data however is continuous throughout this region and offers additional information regarding atomic re-arrangements and density of surface states, etc.
We have employed similar measurements on polycrystalline metals such as Re and W in order to follow their high temperature (800 - 1400 oC) oxidation characteristics. The resulting high work function surface, i.e. >7 eV, will be used as positive ion sources.
Although the work function of an element is not unique, the Kelvin probe can nevertheless be used in chemical analysis. The figure below shows a Scanning Kelvin probe image of a polycrystalline aluminum sample upon which a very thin layer (few nm thick) of gold has been applied, except in the region of the Al mask. The gold layer is invisible to the naked eye, however the aluminum-gold interface is clearly visible as the work function difference between these two metals is approximately 1 V.
When a group of atoms or molecules are brought together to form a solid the highest occupied energy level, or Fermi-level is termed the work function. The work function is continuous across the interior of the solid, however at the surface the electron energy is influenced by the exact state of the surface, e.g. type, orientation and direction of the outer atoms and molecules. Thus different crystallographic orientations of the same solid may have different work functions. The work function of a surface can be modified by adsorbed layers, for instance in corrosion or oxidation of an iron surface, or by modification of composition or controlled contamination as in the case of polymer or inorganic semiconductors.
When two or more materials are brought together the Fermi-levels equalize by a flow of electrons from the lower work function to the high. Detecting these electrons, in essence, is the way all Kelvin probe systems work.
The work function is often described in introductory textbooks as 'the energy required to remove an electron from a surface atom to infinity or equivalently the vacuum level'. Although this description appears easy, in surface analysis one has to ask further questions such as which type of electron, where is it taken from and where does it end up?
For example as an electron is removed a short distance 'd' from a conducting solid there is a so-called attractive image force (due to an imaginary positive charge a distance 'd' inside the solid). The force on the electron diminishes as 'd' gets larger and by approximately 30-50 nm it is exceedingly small. However some work function detection systems actually operate within this region thus the electron is not removed to infinity. This effect, coupled with the local electric fields in the vicinity of the tip, distort the work function data recorded.
In all KP Technology systems the Kelvin probe tip always operates at sufficiently large distances to produce the true work function . Furthermore the mean spacing of the tip electrode above the sample is tightly controlled producing high- stability, repeatable measurements, allowing automatic system set-up.
The standard relative humidity chamber creates a local, 80 litre, environment whereby the relative humidity (RH) can be adjusted within 1%. We also have an optional sample heater working from room temperature to 90˚C. Both can be automatically controlled.
The RHC can also act as a light enclosure with light source either integral or piped in via semi flexible fibre optic tube. The light sources can be optionally controlled in terms of intensity and wavelength for Surface Photovoltage Spectroscopy. We have both DC and AC SPV modes. For metals the RHC would be used to measure the corrosion potential (changes in electrochemical potential) under standard condition.
[For example measuring specific metal samples versus varying RH levels both on cleaned surfaces and surfaces covered with a thin chemical layer.] For some insulating surfaces (or dielectric) RH control is necessary to dissipate surface charge.
For some solar cells the surface and interface trapping constants are very sensitive to cell temperature, i.e. Voc and solar cell fill factor is affected. For some organic semiconductors control of RH during measurement is required. We have also encountered other forensic projects where RH control is required.
Lastly the chambers are sufficiently large (and the bottom is an optical table with mounting holes) to permit mounting of other equipment and easy mounting/dismounting of samples.
For semiconductor surfaces, both organic (polymer) and inorganic (Si, Ge, CdS, etc), the Kelvin probe is the only way to directly measure the Fermi-level. Changes in Fermi-level, caused by illumination with white or monochromatic light, results in energy band shifts which can be used to characterize interface and bulk defect states. These techniques are termed surface photovoltage (SPV) and surface photovoltage spectroscopy (SPS), for which KP Technology can supply both software and accessories for our systems.
We offer several forms of SPV, essentially they are the same across our product range, however the physical implementation may differ, for example in ambient and ultra high vacuum environments. Firstly there is the simple programmed DC-SPV (SPV010) of a light source being switched on and off, synchronised with surface potential (SP) measurements to produce delta (SPV) = SP (Illuminated) -SP(Dark). One can also manually switch the light source on and off and watch the SP transient (approx 1000 measurements/min). Most devices have decay times in the seconds to minutes range. Results over longer periods are not well known as probably the only way of measuring the SP on longer timeshares is with the Kelvin probe.
The light source can be either high intensity LEDs (which work very well) or QTH (Quartz Tungsten Halogen) lamp. The QTH lamp is useful both by itself and for a source of wavelength specific SPS.
The next SPV system SPV020 is of most interest to general customers, in this case all the features of SPV010 apply but in addition the light source intensity is controllable, thus a plot of SPV versus light flux is produced. This can be useful for scoping out the whole response: in the cells I have measured to date both Voc is variable (say from 250 to 450 mV) as is the light sensitivity curve. There is certainly interesting data there and if the light source output was independently calibrated perhaps a rough guide to air mass would result.
We can offer the above systems with or without a light enclosure, secondly we can offer a variant of SPV020 with an optical chopper resulting in AC-SPV.
Lastly we offer either motorised linear variable filter (SPS030) or motorised monochromator (SPS040) systems to probe the specific wavelength dependency of SPV from 400-700 and 400-800 nm respectively. The LVF produces perhaps a higher light intensity than the monochromator system, however the pass band is obviously somewhat larger. We can optionally additionally populate the system with infra-red LEDS to cover higher wavelengths than can be easily obtained with either filter or monochromator systems.
The base, system and SKP packages do not offer automatic SPV however obviously one can implement one's own light source and switch it on and off to measure SP. Having used both ways in the past we know that it is much more efficient researching photosensitive behaviour with SPV/SPS using the automated systems. It allows the possibility to specify the experimental protocol so that a non-specialist can undertake the measurements and they will be comparable. Our surface Photovoltage Spectroscopy modules are the perfect all-in-one solution for in-depth studies of light sensitive materials such as organic semiconductors, solar cells or light sensitive dyes.
The modules offer a comprehensive range of measurement modes including DC and AC surface photovoltage studies utilising the built-in optical chopper. Total digital control of all parameters including light intensity and wavelength (400-700nm or 400-1000nm) gives the opportunity to investigate the quality of samples, characterise interfaces and bulk defect states.
The monitor/camera arrangement can be configured to watch the probe position on the sample or in higher magnification the separation of the tip and sample. On some surfaces this can be difficult to see by eye. Also with our smaller tips, e.g. 50 microns diameter, camera positioning is necessary. Note that this is used for initial tip position and as a general visual check to see where the tip is currently on the sample, for instance on several of our in-house systems the PC is separate for the optical table or desktop supporting the KP/SKP/RHC-KP.
AAfter initial macro-positioning beside the equipment, reverting to computer control of tip to sample spacing is far more accurate. There is an option to keep sample to tip spacing constant to within 1 micron during scanning and this is done by monitoring the Kelvin probe signal and automatically adjusting the sample height to the vibrating tip. Automatic approach to sample is now available in most software packages.
In non-motorised systems it is possible to use a DC offset applied to the voice coil. However the motorised systems give 25 mm of height control and the DC offset only about 0.1 - 0.3 mm.
In our probes samples typically can be moved vertically over 10cm with respect to the tip. As most samples are quite flat, during a scan, only a short range is used.
The sample can be retracted for sample changing. Both the manual and motorised translators mean that, with minimum fuss a wide range of sample sizes can be accommodated.
We estimate that a 2mm tip provides sufficient spatial resolution for most of our users to get started. However we also offer a 50 micron tip and in this case the scan size is typical 500-3000 microns per side.
The Kelvin method is a capacitive method and thus the probe averages the WF underneath the tip. The spatial resolution is essentially the probe diameter. Newton’s fringing field terms do apply (to broaden the effective area) however in general this is not an issue with macroscopic tips. We can achieved 2 micron resolution and on binary systems (2 work function involved) hasve applied 2D deconvolution to determine the precise location of an interface with 1/20th of the tip diameter. Though we don't claim to be able to do both at once!
Higher spatial resolution is feasible however as we are approaching the limits of the measurement environment and operator capabilities, we have elected to supply 2 mm tips as standard and 50 micron tips for anyone who wants higher resolution. We also offer an in-line tip holder that allows the client to screw in any diameter tip they require. We use this system form our larger SPV tips: 5, 10 and 20 mm diameter (semi-transparent).
In summary the tip diameter is the lateral spatial resolution, the length of the tip is not so important, obviously if it is too long then the tip may oscillate in the lateral direction and this would degrade resolution.
A simple Faraday cage will shields the equipment from electric fields due to the operator (the chair the chair they are sitting on), equipment on the desk and in the vicinity. Note an operator can be at a static potential of 2000-3000V after arising from a plastic chair. If the first thing they contact is the Faraday cage there is no problem, however if they touch a dielectric sample they can charge it up, also if they touch the tip amplifier they will damage it, if not destroy it.
Moving up a scale to the optical or relative humidity enclosures will keep sample dust free, their mass tends to minimise mechanical disturbances (from the environment) and isolate the experiment from scientists (and visitors) passing through a lab generating air currents and noise. In a general purpose busy laboratory we cannot stress enough the importance of a suitable enclosure.
It is correct that the Kelvin method is a relative one , however if the tip is calibrated against another 'reference' surface then the tip work function is known too. However this value is obviously only as good as the value taken for the reference surface. Note that the air photoemission system is NOT a relative technique, but absolute and the following does not apply to it.
Our ambient KP Technology Kelvin probe systems are shipped with a 'reference' sample composed of two metals, Aluminum (Al) and Gold (Au). The objective of this sample is for you to measure on the gold surface, and taking the gold reference as 5.10 eV absolute, calculate the tip work function. The objective of the Al surface is as the work function of Al is about 1V less than Au, measurement on this surface tells you whether the tip is below or above Au (5.10eV). Our experience is that the tip is between 50 and 200 mV below gold. The range depends upon tip handling during mounting and tip treatment in operation (i.e. history of tip to sample collisions).
For instance if the tip to gold contact potential difference (CPD) is 200 mV and the tip to aluminum CPD is -800 mV then the tip work function is 200 mV below gold, i.e. approximately 4.90 eV. The aluminum work function is then 4.90 - 0.80 eV, i.e. approximately 4.10 eV. The stated value for aluminum of about 4.2 - 4.3 eV and the reason your Al surface is initially lower, is that we mirror polish the Al, (the surface changes and the polishing media tend to lower the bare Al surface work function slightly).
You will probably find that the Al surface slowly oxidises (over a period of months and years). This will cause the mirror finish to become tarnished and the work function to rise to 4.5 - 4.6 eV. However the gold surface, if held pristine, will maintain its work function. One must remember when using gold in this way that it is sensitive to humidity and if the humidity in your laboratory changes the measured CPD will too, however generally these changes are quite small 10-30 mV for the typical humidity changes in a laboratory, some of which are climate controlled in any case. When using the reference sample one must also remember that the gold surface is fragile and can be damaged by tip contact, fingerprints etc. We recommend that you consider changing the reference sample every 3-6 month (depending upon its use and condition). It cannot be mechanically or ultrasonically cleaned without damaging the gold. Note - it should be possible to extend the life of your sample with careful handling and storage in a separate enclosed box.
Replacement reference samples can be purchased from the KP Technology online store.
If you find that the data you get for your reference sample on gold is -200 mV and for Al is +800 mV then there are two possibilities (i) you have applied the backing potential to the sample rather than the tip, or (ii) you have changed the trigger position of the incoming waveform so that it now triggers at a point 180 degrees advanced in time than when we set the default parameter. Both of these operations would result in the CPD data being multiplied by -1, neither will however change the magnitude of the CPD.
Here is another important point to consider. As a PhD student Prof. Baikie found that the CPD between two surfaces changes with probe parameter (amplitude of oscillation, mean spacing between tip and sample, Vb applied to tip or sample). These results are published in both his thesis and in the journal "Review of Scientific Instruments". They showed the importance of using exactly the same set-up to compare CPD measurements on Al, Au and other samples. In his system this involved keeping the gradient approximately constant (for example 300 with an allowed error of 20). If you do not maintain this consistency (either manually or automatically and using the same parameter file) then, in general, the CPD data will differ each time you take a new measurement. It is this property of accurate, repeatable data, that distinguishes KP Technology equipment from other vendors.
In experiments of polar liquids (where we scanned across the liquid surface above the gold - aluminum interface). we found that for some liquids the interface (with 1V built-in surface work function change) was invisible, in other cases the interface was still visible but the potential changes was reduced. In these cases the liquid layer was not less than 2mm thick.Never-the-less the liquid build up still has some practical implications:
1. If evaporation occurs this may destabilise the liquid surface level. Use of our dedicated relative humidity chamber makes this less of an issue.
2. If you pre-treat the surface to reduce surface tension then the water may flow. You may wish to consider making a trough or reservoir in a sample to accommodate the fluid.
In order to determine the water thickness we suggest that you first make a measurement (at gradient 300) on the bare surface and note the Z motor displacement. Then back the probe away automatically, add the salt solution and re-insert the probe towards the liquid surface. If you now record the Z displacement at gradient 300 you know the liquid layer thickness.
If you are using the SKP for these measurements a lot, it may be worth considering an upgrade to a larger diameter tip and our new in-line tip. We can accommodate any diameters (though most commonly 0.05 to 20 mm). We can also make the tip with holes in it to allows air penetration. The larger tips will result in a larger stand-off height and thus minimal disruption to the liquid surface.
'Our in house research has give us the opportunity to work with polycrystalline Si thin film for a number of years. The key point for us was understanding that the long term decay states at the dielectric interfaces played an important role in determining Voc and thus the fill factor and energy extraction. A surprising aside was to discover how the cells responded (detrimentally) to temperature change and for this point we have been very careful in interpretation, claimed specifications determined at 20°C as in strong sunlight, the surface temperature of the cell can be considerably higher and cell efficiency actually starts to degrade.
From the beginning of our research we were aware of SPV mapping on semiconductors using pulsed light through a semitransparent electrode. The objective of this experimentation was to determine chemical contamination of crystalline Si wafer via changes in de-trapping lifetime and involved high speed data acquisition (>1MHz). However such systems simply cannot measure slow states (this is clearly the domain of the Kelvin probe), e.g. at a speed of 1000 measurements per minute (referred to as DC-SPV), and it is within these slow states that much of the effect of Voc is determined. These SPV transients are easy to record and product surface potential changes of between 250 and 450 mV (depending on semiconductor and cell structure) make them easy to measure.
Our recent experiences have been with a series of solar cells and dye sensitive films. Our work has resulted in the commercial release of our surface photovoltage spectroscopy equipment based upon a motorised linear variable filter (high intensity, 400-700nm, broad pass band) and a motorised monochromator (low intensity, 400-800nm, narrow pass band). We currently use the SPS equipment in both DC-SPV and AC-SPV modes.
We have used a 50 micron diameter tip in our SKP5050 system to seek obvious macroscopic (optical) grain boundaries in oxide and nitride coated mc-Si and have recorded approx 70mV surface potential changes across the grain boundaries. Further SEM/X-ray crystallographic studies of micro-crystalline metal alloy surfaces indicate a complex grain boundary structure and although we have not yet found a way to resolve these surface potential changes laterally, the Kelvin probe, electrically, via charge trapping and de-trapping is another matter. There is a high probability that charge trapping and release can be observed, however it would involve a separate technique to establish a relationship between electrical and physical/chemical characteristics.
'Clearly you are dealing with organic semiconductors. At a conference in Riga in 2006 there was a discussion of the electrical behaviour of traditional and organic semiconductors and in many ways they appear to be the same. For a number of years here at KP Technology we have been consulting for several large OLED manufacturers many of whom purchased our equipment. The principle research was into energy barrier height minimisation at the cathode, anode and HIL. However things are moving on and as solar cells/detector devices are being considered a better understanding of interface defects is also required (as has been the case with traditional materials)
Starting from the beginning: the Kelvin probe is the only method that gives the position of the semiconductor Fermi-level. Photoemission is of course sensitive to the topmost populated states of the valence band, inverse PE respectively for conduction. However with hole based conduction in organics PE data is a little unclear. In the best case (with electronic states) one cannot distinguish a PE threshold better than 30-50 meV. In fact photoemission data can have several turn-on points indicating several photoelectric thresholds and thus state distribution in the energies near the valence band maxima, so the PE data (and corresponding work functions) reported are sometimes averages and recent data suggests 0.25-0.3 eV lower than that predicted by the Kelvin (capacitance) method. In this picture HOMO is not perhaps an ‘edge’, but a series of energy states.
Typically the way to get information with the Kelvin method on semiconductors is to make a measurement in darkness, then illuminate the semiconductor. If the illumination is sufficiently intense the surface energy bands will flatten as the photon flux is sufficient to induce intrinsic characteristics with ni=pi. The surface band-bending is thus obtained from the shift in Fermi level upon illumination. As the Fermi level is known from the dark measurement and the band-bending, the doping gives the bulk Fermi level, the band-gap is known (or can be determined from another measurement) so in theory everything that is necessary to draw the energy band diagram including HOMO level.
'For a 50 micron diameter tip the closest approach of the tip to the sample is within 5-10 microns, for a 2 mm tip a few hundred microns and for a 10-20 mm tip then say 2-3 mm. This can be automatically maintained for the duration of a scan by the software.
The Kelvin probe is sensitive to the top 3 atomic layers and, depending on the work function change between the substrate and the film, can see sub mono-layer coverage. The air photoemission system can 'see' considerably deeper than the Kelvin probe.
Professor Baikie's experience lies mostly with oxidation, and in his experience with oxygen adsorption on clean silicon surface claimed 1/1000 of a monolayer. It is the most sensitive technique available, certainly more sensitive in some cases than Auger Electron Spectroscopy (AES) and much more so than XPS (X-ray Photoelectron Spectroscopy).
When additional parameters are required the KP technique can be used as a supplement to these other techniques. The spectroscopies give a "finger print" of the sample not just one parameter. However take for example a UHV application - The UHV-KP is very good for scoping out changes in the surface during an experiment and then suggesting where to look for critical changes in behaviour. If you used UHV scanning probes such as AFM or SMT this could take weeks or months. With KP possibly 1 hour.
'Null based measurements are prone to error and noise. KP Technology systems have an "off-null" algorithm whereby the Kelvin probe signal is measured at two or more values of Vb, (for instance at high signal levels). The balance point can then be obtained by extrapolation.'