PHOTOEMISSION OF ELECTRONS FROM METALLIC SURFACES

J.W. F O R O O G HI A N The use of Einsteins Photoelectric Law to calculate Plancks constant is demonstrated, via the measurement of electrons emitted from a potassium photocathode under the radiation of light of varying frequencies. The value of Plancks constant calculated was 1.0623x10-336.37x10-35Js, a value significantly higher than the predicted value, 6.63 x 10-34Js. Presented within are the analyses of physical concepts such as the exponential relationship between the anode-tocathode potential of the photocell and the rate at which electrons are emitted, in addition to a determination of the reasons for the incorrect calculation.

1.

Introduction

The basis of the experiment stems from the nature of the photoelectric effect. As a consequence of the photoelectric effect, absorption of electromagnetic radiation incident on a metallic surface causes the liberation of electrons from the surface of the metal. This is known as photoelectric emission1. The first definitive studies of the effect were produced in 1887 by Heinrich Hertz, upon his discovery that the spark in the spark gap of a coil would increase in length when ultraviolet radiation was incident upon it, but showed no increase when lacking the radiation2. Although this experiment allowed the observation of the interaction between electromagnetic radiation and charge, it did not however provide any foresight into what was causing this, as it would have been expected that it was purely the energy of the electromagnetic radiation being sufficiently high to release further charge from the coil. In addition to this, the experiment did not account for the high current in the coil causing possible charge leakage. These limitations were later alleviated via further experiments by Philipp Lenard and Albert Einstein, the latter of whom going on to describe the energy of the photoelectrons emitted as proportional to , the frequency of the radiation, with constant of proportionality h, Plancks constant. The following experiment was performed in order to confirm the value of Plancks constant 3 as 6.63 x 10-34Js. The important values calculated include the work function of the anode in the photocell, as well as the stopping voltages for the varying frequencies of electromagnetic radiation. The experiment also provided solid evidence for the existence of quantum theory, as will be discussed henceforth. Accurate calculation of Plancks constant is important, as it shows that the energy of electromagnetic radiation is not continuous, but occurs in discrete quanta proportional to the frequency of the radiation4.

2.

Theory

Our knowledge of the classical, wave theory of light would expect us to see an increase in the energy of electrons emitted from the metal with increasing light intensity, as intensity describes

the rate of power transfer per unit area, implying more energy to liberate electrons. Various experiments however proved this not to be the case, and paved the way for the development of quantum theory. In his Annus Mirabilis papers of 1905, Einstein provided an explanation for this phenomenon by premising the existence of photons, discrete quanta of light with particle-like nature forming the basis of the property of wave-particle duality5. Einstein postulated that each electron can only receive the energy from the light of one photon, with the energy of a single photon given by
EQN 1

Where h is Plancks constant and the frequency of the electromagnetic radiation. It was established that photoelectrons were emitted only if the frequency of the photons were higher than a characteristic threshold frequency 0 of the metal, defined as the minimum frequency needed to liberate an electron from the surface. It follows that the work function of the metal, , is given by
EQN 2

This means that the greatest possible kinetic energy, EK, of the photoelectrons is given by EQN2 EQN1, leading to the equation
EQN 3

In our experiment, the metal used is a thin layer of potassium coating one wall of an evacuated glass tube, known as a photocathode. The vacuum nature of the tube is crucial, as the fastest photoelectrons are those that are liberated at the very surface of the metal, due to the most direct contact with the photons. The binding of the electrons to the surface, however, depends heavily on the composure of the surface, as electrons would be emitted far more easily from a pure metal surface than an oxygen-metal one, due to higher electrostatic forces between oxygen-metal than metal-metal. Opposite is the anode, which consists of a loop of platinum wire lying parallel to the photocathode. Both of these are connected to an external circuit, which records the rate at which the electrons arrive at the anode, known as the photocurrent. The circuit is assembled so that there is a potential difference between the negatively charged cathode and positively charged anode. As the electrons are emitted from the cathode they slow down as they travel from the high potential cathode to the low potential anode. They must overcome this potential barrier in order to reach the anode. The minimum energy required to do so is given by
EQN 4

Where e is the charge on an electron (-1.6x10-19C) and V0 the potential difference between the anode and cathode where the kinetic energy of the photoelectrons reaches zero, referred to as the stopping potential. In order to be completely liberated from the metal, the photoelectrons must have energy sufficient to escape the cathode surface, EL
EQN 5

Where C is the work function of the cathode, measured in volts. This means that the stopping energy required is equal to the difference between the energy of the photons and work function of the cathode, as can be seen beneath.

EQN 6

The meter recorded voltage, VM, is not the potential difference, V, between the surface of the cathode, but instead corresponds to the difference between the anode and cathode Fermi levels of the two metal surfaces, the highest electron energy level of the metal at absolute zero. VM is given by the equation below.
EQN 7

Where A is the work function of the anode, measured in volts. This means the actual stopping voltage is given by [ ]
EQN 8

3.

Experimental Method

The photocell consists of a thin potassium photocathode and platinum anode, contained in an evacuated glass tube. The source of light was a PASCO OS-9286 mercury vapour lamp, which produces light via the use of an electric arc through vaporized mercury6. L1, a Newport VPH LH2 lens, is used to focus an image of the mercury discharge tube onto S1, which is a Newport VPT series slit with adjustable width, which can be varied to control the intensity of the light reaching L2, a Newport VB1 lens, which has focusable optics, resulting in a focused beam of light reaching the prism. The prism is used to separate the spectral lines of the light beam incident upon it, into its constituent light of varying frequency. The light dispersed by the prism is then focused onto S2, which comprises a sheet of white card with a very thin (~1mm) slit in its centre, allowing light of only one frequency to be incident upon the photocell, which lies ~2mm away from the slit. All of the optical equipment was mounted on Newport PSF clamps, affixed to a Newport SA breadboard, which allowed both the prism and S2 to be rotated in order to allow all 5 frequencies of light to reach the photocell. The cathode of the photocell was connected directly to the input of an internally mounted amplifier, converting the photocurrent into a voltage. The output of this voltage was then connected to an EL72 Photoemission amplifier circuit. As well as allowing the control of anode bias voltage and displaying its value (with a sensitivity of 0.005V), the circuit is set up in order to reject steady leakage current in the photocell and amplifier. This operates by recognizing the timing of the light emitted by the mercury lamp, derived through the frequency of the mains supply powering both the lamp and the circuit, meaning that both the circuit and lamp were connected to the same wall outlet. The circuit also provides a Purge Anode function, which is used in order to remove unwanted contamination from the anode surface, such as fine particles of dust, which would lower the work function of the anode. The anode is purged via the brief application of high current, heating it up and removing such contamination.

The experiment took place in a darkroom, in order to reduce the effects of any visible white light on the photocell, which would result in us no longer knowing the frequency of the light incident upon it. Despite this however, it is not possible to entirely negate light of other frequencies reaching the photocell, as the spectrum produced by the prism is fairly small, in order to achieve a high intensity, and the lines are therefore not entirely separated, meaning some light of other frequencies may be incident. Five distinct spectral lines were produced via the prism-light setup; yellow, green, blue-violet, deep violet, ultra-violet, ranging from 578nm-365nm. Before beginning the experiment the anode was purged, in order to remove any possible contamination resulting from prior experiments. The mercury lamp was then turned on and allowed to warm up for around 8 minutes, in order for it to reach its full light output. Once this period was over, the optical equipment was adjusted in order for the entire spectrum to be around 10cm in length, where the definition of each line was highest. S2 was then rotated in order to allow light of the first measured spectral line, ultra-violet, to reach the photocell. The initial anode bias for all spectral lines was set as -2.75V and initially increased in increments of 0.25V, although this was changed to increments of 0.05V from -2.50V to -2.00V, where the stopping voltage would lie, to achieve recordings of higher resolution. For ultraviolet, the bias was then steadily increased in 0.25V increments again until it reached 1.00V, where it could clearly be seen the photocell had reached saturation from here on the photocurrent can be seen to increase exponentially with increasing anode bias. After the anode was re-purged, the same procedure was followed for deep violet light, with anode bias increasing in 0.25V increments, changed to 0.05V between -2.25V and -1.00V. It could clearly be seen that a higher anode bias was required for the photocell to reach stopping voltage, which is as expected, as can be seen via the equation below.
EQN 9

Where c is the speed of light of EQNs 1 and 7.

(~3.0x108ms-1)

Meaning photon energy is given by a combination

EQN 10

This means that a higher wavelength will result in a lower energy photon, so the anode bias must be increased in order to account for this. The process was subsequently repeated for blue-violet light with voltage increasing in increments of 0.05V between -2.75V and -0.50V, and increments of 0.25V between -0.50V and 1.00V. For green light, the voltage was increased in increments of 0.25V between -2.75V and -0.25V, and increments of 0.05V between -0.25V and 0.10V. Between 0.10V and 0.20V the voltage was increased by 0.02V, as the photocurrent was seen to be highly fluctuating around this area, meaning greater resolution was required. The final spectral line measured was that corresponding to yellow light. The anode bias was increased in increments of 0.25V between -2.75V and 0V, and increments of 0.05V between 0V and 0.50V.

When graphs of anode bias against photocurrent are plotted for the five frequencies of light, Vm,0 can be seen as a straight line on the graph. When equation 8 is divided through by e, it can be seen that
EQN 11

This means that when a graph of the varying frequencies is plotted against their metered stopping voltages, the gradient gives the value h/e, meaning that h is the gradient multiplied by e, and that A is the negation of the intercept.

4.

Results

To calculate the frequency of each colour of light, equation 10 was applied to the five wavelengths provided in the data booklet. The results can be seen below. Colour Yellow Green Blue-Violet Deep Violet Ultra-Violet Wavelength (nm) 578 546 436 405 365 Frequency (Hz) 8.211x10-14 7.400x10-14 6.873x10-14 5.489x10-14 5.185x10-14

The random error in the anode bias voltages is simply the sensitivity of the device, 0.005V. The random error in the recorded photocurrent varied upon the value of the photocurrent displayed, due to the higher rate of fluctuation of photocurrents greater than 100pA, meaning a larger error at these values. These are due to the nature of the movement of the photons, which are emitted at random times from the lamp. When the amount of photons is high, the relative number of fluctuations in the number of photons becomes significant, resulting in a higher overall fluctuation. The errors applied can be seen in the table below. Photocurrent < 100pA 100pA < Photocurrent < 150pA Photocurrent > 150pA The first colour measured was ultra-violet. Readings were initially taken with the prism dispersing the light directly into the slit, but it soon became evident that this resulted in too high a fluctuation of the displayed anode bias value. As a result of this, the slit was placed at a small angle to the prism. This indirect light still produced a fluctuation in the reading, but to a much lesser extent than in the prior setup. The graph showing the Anode Bias/Photocurrent can be seen to the right.
G RAPH 1 U LTRA -V IOLET

A = 0.5pA A = 5pA A = 10pA

The stopping voltage is noticeable on the graph as the voltage at where the photocurrent starts to increase in value, with the line of stopping potential marked in red on the graph. For ultraviolet light Vm,0 is equal to -2.100.005V. The photocell can be seen not to reach saturation for ultra-violet light, or any other of the colours, due to the highest anode bias output being insufficient. As can be seen from the adjacent graph, the stopping voltage for deep violet light is equal to -1.400.005V. A higher Vm,0 is expected in order to account for the lower energy of the photons. The slower rate at which the photocell is reaching saturation can also be noticed. The adjacent graph is the anode bias-photocurrent plot for blue-violet light. Vm,0 can clearly be seen to have a value of -10.005V. It can also be seen, however, that the photocell appears to be reaching saturation at a faster rate than deep violet and indeed ultra-violet light. This result is clearly unexpected, and must therefore have arisen from the high fluctuations in the photocurrent reading, presumably due to the photon paths being too direct upon
G RAPH 3 - B LUE -V IOLET G RAPH 2 D EEP V IOLET

the cell. The graphs for green and yellow light show the stopping voltages to be -0.260.005V and 00.005V respectively. The rate at which each light reaches saturation can also be seen to be as expected once again, with yellow light reaching saturation at higher bias voltages than green.

G RAPH 4 - G REEN

G RAPH 5 - Y ELLOW

Now that values of Vm,0 are known for all frequencies of light, a graph showing the relationship between the two can be plotted, as shown below.

G RAPH 6 - F REQUENCY AGAINST V M,0

The gradient of the graph is -6.630x10-15, meaning that giving a value of 1.062x10-33Js, when errors are not taken into account. The coefficient of determination is equal to 0.9858, meaning 98.58% of the points lie perfectly on the line. When errors are taken into account, calculated via linear regression, the minimum gradient is -7.027x10-15 and the maximum is 6.232x10-15, meaning the minimum and maximum values of h respectively are 9.986x10-34Js and 1.126x10-33Js. This means that h=1.0623x10-336.37x10-35Js.

5. Discussion
Compared to the expected value of 6.63 x 10-34Js, the measured value is a fairly inaccurate
calculation of h. Even when random errors are taken into account, the value is still too high. This means that the most significant error must be of a systematic nature. The most likely systematic error is the possibility of a zero error in the anode bias readings. This could have been detected by using a voltmeter to take a reading of the voltage when the anode bias was set to zero. The difference in the reading could then be subtracted from the reading on the display in order to obtain the actual value. The phototube itself may have also been damaged, resulting in a higher cathode work function, which would in turn result in a higher value of h being calculated. There are, however, more accurate ways to determine h. Robert Millikan devised an experiment similar to the one performed here, but using three different metal anodes, calculating values

using each separate metal, and using these values to calculate an average h. These repeated measurements would therefore reduce the random error in h.

6. Conclusion
The method used in its current state is not sufficient to calculate an accurate value of h, unless the anode bias was checked for systematic error using the method described. The experiment is successful, however, in showing the accuracy of Einsteins Photoelectric Law, due to the measurements being 98.58% in showing this.

7. References
1) Physics for Scientists and Engineers with Modern Physics 8th Edition R.A. Serway, J.W.J. Jewett Brooks/Cole Publishing 2010 2) Ueber einen Einfluss des ultravioletten Lichtes auf die electrische Entladung H.R. Hertz Annalen der Physik, vol. 267, Issue 8 1887 3) CODATA recommended values of the fundamental physical constants P.J. Mohr, B.N.Taylor, D.B.Newell - 2006 4) Physics for Scientists and Engineers 6th Edition P.A.Tipler, G.Mosca W.H. Freeman and Company 2008 5) Concerning an Heuristic Point of View Toward the Emission and Transformation of Light A. Einstein - Annalen der Physik vol. 322, Issue 6 1905 6) Electric Discharge Lamps J.F.Waymouth M.I.T. Press 1971 7) On the Elementary Electric charge and the Avogadro Constant - R.A. Millikan, Physical Review Series II - 1913