Understanding Langmuir probe current-voltage characteristics

Iowa

probristi

cs uure,ainethe

Assoc

plasma, and thus its potential rises and falls to whatever po-tential is necessary to maintain zero net current.

extensive, and new papers appear regularly. My purpose hereis not to discuss the complexities of probe theory, which is

39In a typical plasma, the electrons, because of their smallermass, have significantly higher thermal speeds than the posi-tive ions, even if the electrons and ions are at the same tem-perature. Usually the electrons have a higher temperaturethan the positive ions. Although a plasma is electrically neu-tral, and the electron and ion densities are very nearly equal,a floating probe will tend initially to draw a higher electroncurrent because the electrons reach the probe faster than themore massive ions. Because the net current to the floatingprobe must be zero, the probe floats to a negative potentialrelative to the plasma so that further collection of electrons isretarded and ion collection is enhanced. Thus, the floatingpotential is less than the plasma potential. The plasma poten-

treated in a number of excellent monographs, but to pro-vide a method to help students understand why a Langmuircharacteristic looks the way it does. The difficulty with un-derstanding probe I-V characteristics stems from the fact thatthe electrons and ions are not monoenergetic and often havevery different temperatures. As a result, the probe sometimescollects only ion current, sometimes only electron current,and sometimes both. It is easier to understand and analyzethe full I-V characteristic if the ion and electron current con-tributions are separated.

In Sec. II we discuss the most basic aspects of probetheory needed to calculate the individual electron and ioncurrents, and then construct an ideal probe I-V relation usingRobert L. MerlinoaDepartment of Physics and Astronomy The University ofReceived 26 February 2007; accepted 14 July 2007

I give several simple examples of model Langmuirhelp students learn how to interpret real I-V charactecreate their own Langmuir probe I-V characteristiplasma potential, electron temperature, ion temperatexamples of Langmuir probe I-V characteristics obtanalyzed. A few comments are made advocatingadvanced undergraduate laboratory. 2007 AmericanDOI: 10.1119/1.2772282

I. INTRODUCTION

Plasma physicists use Langmuir probes in low temperatureplasmas1 approximately a few electron volts to measure theplasma density, electron temperature, and the plasma poten-tial. A Langmuir probe consists of a bare wire or metal disk,which is inserted into a plasma and electrically biased withrespect to a reference electrode to collect electron and/orpositive ion currents. Examples of the use of a cylindricalwire probe in a gas discharge tube and a planar disk probein a hot filament discharge plasma are shown in Fig. 1.Probes, initially called sounding electrodes, were first usedin the late 19th and early 20th centuries in an attempt tomeasure the voltage distribution in gas discharges. A gasdischarge Fig 1a is produced in a glass tube of about25 cm diameter and 2040 cm long, which contains metaldisk electrodes anode and cathode at both ends. The tube isfirst evacuated and then refilled with a gas at low pressureabout 1 Torr or less and an electrical discharge ionized gasor plasma is formed by applying a DC voltage of300400 V across the electrodes. A common example of adischarge tube is an ordinary fluorescent light. Probes areinserted at one or more locations along the length of the tube,with the exposed tips protruding into the plasma column. Theearly users of probes naively assumed that the potential ofthe plasma at the location of the probe known as the plasmapotential or space potential and designated as VP could bedetermined by measuring the potential on the probe relativeto one of the electrodes. However, this procedure determinedthe floating potential Vf of the probe which is generally notthe same as the plasma potential. By definition, a probe thatis electrically floating, collects no net current from the1078 Am. J. Phys. 75 12, December 2007 http://aapt.org/ajp, Iowa City, Iowa 52242

e current-voltage I-V characteristics thatcs obtained in a plasma. Students can alsosing a program with the plasma density,and probe area as input parameters. Somed in laboratory plasmas are presented andinclusion of plasma experiments in the

iation of Physics Teachers.

tial is the potential of the plasma with respect to the walls ofthe device at a given location in the plasma. VP is generallya few volts positive with respect to the walls, again becausethe swifter electrons tend to escape to the walls first, leavingthe plasma with a slight excess of positive space charge. Thebulk of the plasma, however, is quasineutralelectron density ion density, and the potential differencebetween the bulk of the plasma and the wall is concentratedin a thin layer or sheath near the wall. The gradient of theplasma potential determines the electric field that is respon-sible for energizing the electrons, which maintain the dis-charge through ionization.

Although physicists knew that Vf and VP were not thesame, they thought that the difference was probably small,and in any case, they had no way of either estimating thedifference or of measuring the actual plasma potential. IrvingLangmuir and Harold Mott-Smith of the General ElectricResearch Laboratory in the 1920s were the first to provide aquantitative understanding of the difference between Vf andVP. They developed an experimental method for determiningthe plasma potential and also showed how it was possible touse the probe now known as a Langmuir probe to deter-mine the plasma density and the electron temperature aswell.2 Langmuirs method consists of obtaining the current-voltage I-V characteristic of the probe as the applied biasvoltage VB, is swept from a negative to a positive potential.

Many students of experimental plasma physics are giventhe task of constructing and implementing a Langmuir probein a plasma. They quickly realize that building the probe andobtaining a I-V characteristic is much easier than extractingaccurate values of the plasma parameters from the data. Theliterature dealing with the theory of the Langmuir probes is1078 2007 American Association of Physics Teachers

collects the ion saturation current Iis. Positive ions continueto be collected by the probe until the bias voltage reaches VP,a set of model plasma parameters. A link is provided to aprogram that allows the user to input the plasma and probeparameters density, temperature, plasma potential, andprobe dimension and plot the resulting Langmuir I-V char-acteristic. Section III provides two examples of real Lang-muir probe I-V characteristics obtained in laboratory plas-mas. I close in Sec. IV with comments advocating theinclusion of plasma experiments in the undergraduate ad-vanced laboratory course.

II. MODEL LANGMUIR PROBE CURRENT-VOLTAGE CHARACTERISTICS

In this section I discuss some of the basic aspects of Lang-muir probe theory that are needed to construct model probeI-V characteristics. Two examples of ideal probe I-V charac-teristics are then given. Finally, a discussion of how the idealcharacteristics must be modified to account for real probeeffects is presented.

A. Ion and electron currents to a Langmuir probe1. The ion current

When the bias voltage VB, on the probe is sufficientlynegative with respect to the plasma potential VP, the probe

Fig. 1. Schematic of basic devices for producing a plasma. a A dischargetube in which a plasma is formed in a low pressure gas 1 Torr byapplying several hundred volts across the cathode and anode. A cylindricalwire probe is inserted into the discharge to measure the properties of theplasma. b Schematic of a multidipole hot filament plasma device with aLangmuir disk probe. The plasma is produced by electron impact ionizationof argon atoms by electrons that are thermionically emitted and acceleratedfrom a hot tungsten W filament. To enhance the ionization efficiency, thewalls of the chamber are lined with rows of permanent magnetic of oppositepolarity. The lower diagram is an end view showing the arrangement ofmagnets. The magnetic field lines are sketched as the dotted curves. In thismagnetic cusp configuration, the bulk plasma is essentially magneticfield-free.1079 Am. J. Phys., Vol. 75, No. 12, December 2007at which point ions begin to be repelled by the probe. ForVBVP, all positive ions are repelled, and the ion current tothe probe vanishes, Ii=0. For a Maxwellian ion distributionat the temperature Ti, the dependence of the ion currentIiVB usually taken to be the negative current on VB isgiven by10

IiVB = Iis expeVP VB/kTi , VB VP, Iis, VB VP,

1

where e is the electrons charge, and k is the Boltzmannconstant. When Ti is comparable to the electron temperatureTe, the ion saturation current, Iis is given by4

Iis =14

enivi,thAprobe, 2

where, ni is the ion density, vi,th=8kTi /mi is the ion ther-mal speed, mi is the ion mass, and Aprobe is the probe collect-ing area. When TeTi,11 the ion saturation current is notdetermined by the ion thermal speed, but rather is given bythe Bohm ion current3,4,12

Iis = IBohm = 0.6enikTemi

Aprobe. 3

The fact that the ion current is determined by the electrontemperature when TeTi is counterintuitive and requiressome explanation. The physical reason for the dependenceIiskTe /mi1/2 has to do with the formation of a sheatharound a negatively biased probe.12,13 If an electrode in aplasma has a potential different from the local plasma poten-tial, the electrons and ions distribute themselves spatiallyaround the electrode in order to limit, or shield, the effect ofthis potential on the bulk plasma. A positively biased elec-trode acquires an electron shielding cloud surrounding it,while a negatively biased electrode acquires a positive spacecharge cloud. For a negatively biased electrode, the charac-teristic shielding distance of the potential disturbance is theelectron Debye length14

De = okTee2ne

1/2. 4In the vicinity of a negatively biased probe, both the

electron and ion densities decrease as the particles ap-proach the probe, but not at the same rate. The electrondensity decreases because electrons are repelled by theprobe. In contrast, the ions are accelerated toward theprobe, and due to the continuity of the current density, theion density decreases. A positive space charge sheath canform only if the ion density exceeds the electron density atthe sheath edge, and for the ion density to decrease moreslowly than the electron density, the ions must approachthe sheath with a speed exceeding the Bohm velocity uB= kTe /mi1/2.13,15 To achieve this speed, the ions must ac-quire an energy corresponding to a potential drop of0.5kTe /e, which occurs over a long distance in theplasma. The factor of 0.6 in Eq. 3 is due to the reductionin the density of the ions in the presheath, which is theregion over which the ions are accelerated up to the Bohmspeed.1079Robert L. Merlino

Table I. Parameters of a typical laboratory plasma used to construct an idealLangmuir probe volt-ampere characteristic.2. The electron currentFor VBVP the probe collects electron saturation current

Ies. For VBVP the electrons are partially repelled by theprobe, and for a Maxwellian electron velocity distribution,the electron current decreases exponentially with decreasingV. For VBVP all electrons are repelled, so that Ie=0. Theelectron current as a function of VB can be expressed as

IeVB = Ies exp eVP VB/kTe , VB VP,Ies, VB VP. 5The electron saturation current Ies is given by

Ies =14

eneve,thAprobe, 6

where ne is the electron density, ve,th8kTe /me is theelectron thermal speed, and me is the electron mass. We seefrom Eqs. 2, 3, and 6 that because ne=ni and memi,the electron saturation current will be much greater thanthe ion saturation current. For example, in an argon plasmafor TeTi we see from Eqs. 3 and 6 that Ies / Iis=mi /me / 0.62=271/1.5=180.

B. Examples of Langmuir probe characteristic1. The ideal Langmuir probe characteristic

For the values in Table I, with Aprobe=2dprobe/22=1.41105 m2 for a planar disk probe that collects fromboth sides, we find from Eqs. 3 and 6, Iis=0.03 mA andIes=5.4 mA, so that Ies / Iis=180. The I-V characteristic cor-responding to these parameters is shown in Fig. 2. The cal-culations of the ion current, Eq. 1, the electron current, Eq.5, and the total current IVB= IeVB+ IiVB for the I-Vcurve in Fig. 2 were performed using the Maple procedurefunction and the standard plotting command.16 Alternately,the data for the I-V curve could be computed and plotted ina spreadsheet program. The heavy solid curve in Fig. 2 is thetotal probe current; the electron current positive and ioncurrent negative are also indicated. The ion current is mag-nified by a factor of 20 in order to see its contribution to thetotal current. Because the electron current is much largerthan the ion current, it is necessary to bias the probe to verynegative voltages to even see the ion current. The total cur-rent is also displayed on a magnified scale 20 to show theprobe bias at which the total current is zero. The probe biasfor which IVB=Vf= Ie+ Ii=0 is the probes floating poten-tial, which occurs at Vf9.5 V. The floating potential can

Parameter Symbol Value Units

Ion species Ar+

Ion mass mi 6.71026 kgElectron density ne 1.01016 m3

Ion density ni 1.01016 m3

Electron temperature Te 2.0 eVIon temperature Ti 0.1 eVPlasma potential VP 1.0 VProbe diameter dprobe 3.0 mm1080 Am. J. Phys., Vol. 75, No. 12, December 2007be calculated from Eqs. 15, as the bias voltage at whichIiVf+ IeVf=0,

Ies expeVf VP/kTe = Iis, 7

or

Vf = VP + kTee

ln0.62me

mi

. 8

If the appropriate parameters are inserted into Eq. 6, wefind that Vf =VP5.2Te, or Vf =9.4 V for kTe /e=2 V, cor-responding to an electron temperature of 2 eV.

The nature of the Langmuir probe I-V characteristic of thetype shown in Fig. 2 is dominated by the fact that the speedof the electrons is considerably higher than that of the posi-tive ions. As a consequence, it is impossible to use the probeto determine the ion temperature, whereas the electron tem-perature can be easily found from the portion of the charac-teristic corresponding to electron repulsion, that is, for VBVP.

2. Probe I-V characteristic for a positive ion (+) /negativeion () plasma with m+=m and T+=T

Consider constructing the I-V characteristic in a plasmaconsisting of positive and negative ions of equal mass andtemperatures, for example. In this case because the thermalspeeds of the positive and negative plasma constituents areidentical, we expect that VP=Vf =0, and I+s= Is. An exampleof such an interesting plasma would be an electron-positronplasma.17 The probe I-V characteristic for this case is shownin Fig. 3, where arbitrary but equal values of the saturationcurrents are used. An important point to take from a consid-eration of the I-V plot in Fig. 3 is that the part of the curvewhere the negative ions are repelled VBVP occurs at10 VVB0 V, and the portion corresponding to positiveion repulsion is 0 VVB10 V. To find the negative ioncurrent the positive ion saturation current must be used as thebaseline and not the I=0 line and vice versa to obtain thepositive ion current. The line extrapolated from the ion cur-

Fig. 2. Ideal Langmuir probe current-voltage characteristic heavy line fora model plasma with the parameters listed in Table I. The individual electronand ion currents that are used to construct the full characteristic are alsoshown. The dotted line is the full probe characteristic magnified by a factorof 20 so that the probe floating potential, Vf the probe voltage where I=0can be easily determined.1080Robert L. Merlino

Sheath expansion occurs for both the ion and electron cur-rents and must be taken into account in the interpretation ofrent should also be used as the baseline to determine theelectron current for the case shown in Fig. 2, although in thatcase the positive ion contribution is negligible, and the I=0line can usually be used to measure Ie. When the plasmacontains a significant fraction of high energy tens of elec-tron volts ionizing primary electrons in addition to the sec-ondary electrons resulting from ionization, it is essential tofirst subtract the ion current from the total current to obtainaccurate values of the electron current.

C. Effect of sheath expansion on probe characteristics

The sharp knee at the plasma potential VP=1.0 V in theI-V characteristic and flat electron and ion saturation currentsshown in Fig. 3 are ideal probe features that are rarely seenin practice. For this reason the I-V characteristic in Fig. 1 isideal. Real Langmuir probe I-V characteristics have roundedknees and saturation currents that increase gradually withincreasing voltage. The lack of saturation is related to thefact that a sheath1315 is formed around the probe, and thissheath expands with increasing bias voltage. Sheaths formaround any electrode in a plasma if the bias voltage differsfrom VP. The formation of a sheath is the plasmas way ofmaintaining charge neutrality in the bulk of the plasma. Anelectrode with a positive bias relative to VP attracts an elec-tron cloud to limit the penetration of the electric field into theplasma to a distance approximately equal to the electron De-bye length, De, defined in Eq. 4. If the plasma density isgreater than 1016 m3 and Te2 eV, then the sheath widthwill be 0.1 mm. In this case the expansion of the sheathwill produce only a negligible increase in current as theprobe bias is increased. For lower plasma densities and smallprobes the sheath expansion produces an increase in the col-lected current because the effective area for particle collec-tion is the sheath area and not the geometric probe area.Another way to think about the sheath expansion effect is torealize that for a finite probe, the collection of plasma par-ticles is limited by fact that some particles that enter thesheath will orbit around the probe and not be collected. Asthe potential on the probe is increased, the minimum impactparameter for which particles are collected increases and thusmore particles will be collected.

Fig. 3. Langmuir probe I-V characteristic for a plasma with positive andnegative ions of equal mass and temperatures. The positive ion and negativeion currents are also shown.1081 Am. J. Phys., Vol. 75, No. 12, December 2007the I-V characteristics. The sheath expansion effect can beincorporated in the ideal probe characteristic so that we canlearn how to deal with it when interpreting real Langmuirprobe characteristics. An illustration of this effect is shown inFig. 4. The parameters used to produce this I-V characteristicwere VP=4 V, Ies=100Iis ,Te=4 eV, and Ti=0.1 eV. Thesheath expansion was modeled as a linear function of thebias voltage with IisVB=0.2VPVB+ Iis for VBVP forthe ions, and IesVB=0.7VBVP+ Ies for VBVP for theelectrons. Figure 4a shows the full I-V characteristic. Realcharacteristics rarely show the sharp knee at the plasma po-tential; rather the knee tends to be rounded as illustrated bythe dotted curve due to the presence of oscillations of theplasma potential15 or averaging in the data acquisition oranalysis process. The rounding of the knee complicates thedetermination of the electron saturation current, but the loca-tion of the knee is made more evident by replotting the cur-rent on a semilog scale, as shown in Fig. 4b. Both Ies andVp can now be easily determined as the coordinates of theintersection of two straight linesone parallel to the curveabove the knee and the other parallel to the sloping part. Theslope of the straight line fit tothe electron current in Fig. 4bis used to determine Te, as Te= V2V1 / lnIe2 / Ie1, where 1and 2 refer to any two points on the line. The electron currentbegins falling off the straight line due to the contribution ofthe ion current. An accurate measurement of the ion currentin this case requires that Ii be obtained for a sufficientlynegative probe bias so that the electron contribution is ex-cluded. The procedure for measuring Iis is shown in Fig.4c. The ion current is plotted on an expanded scale and astraight line is fitted through the points; Iis is taken as thevalue of Ii at Vp. More accurate methods of dealing with thenonsaturation of the ion current are discussed in Refs. 5, 7, 9,and 18.

III. EXAMPLES OF LANGMUIR PROBECHARACTERISTICS FROM LABORATORYPLASMAS

In this section I provide two examples of Langmuir probeI-V characteristics obtained in more realistic laboratory plas-mas. These examples demonstrate how the basic principlespresented in Sec. II are applied in the interpretation of realcharacteristics.

A. Multidipole plasmaA multidipole device1921 is a relatively simple setup for

producing a plasma that can be used for basic plasma physicsexperiments. A schematic diagram of a typical multidipoledevice is shown in Fig. 1b. It is essentially a large about20 l stainless steel soup pot sometimes literally which ispumped down to a base pressure of 106 Torr, and thenfilled with a gas such as argon to a pressure of approximately105103 Torr. The plasma is produced by electron emis-sion from a set of tungsten filaments that are biased to anegative potential of approximately 50 V. The thermioni-cally emitted primary electrons that are accelerated from thefilaments ionize the gas producing the plasma. To enhancethe probability that a primary electron will undergo an ion-izing collision with a neutral atom, the walls of the deviceare lined with permanent magnets in rows of opposite polar-1081Robert L. Merlino

ity creating a magnetic barrier, which inhibits the ionizingelectrons from escaping. Due to the alternating polarity ofthe magnet rows, the magnetic field has substantial strengthonly very close to the walls, so that the main plasma regionis essentially magnetic field free.

A typical Langmuir probe I-V characteristic obtained in

Fig. 4. Model Langmuir probe I-V characteristic including the effect ofsheath expansion, computed with VP=4 V, Te=4 eV, Ti=0.1 eV, andIes / Iis=200. a Total current. The dotted curve depicts the rounding of theknee due to plasma noise or averaging effects. b log IVB versus VB. Theintersection of the horizontal and vertical dotted lines occurs at the coordi-nates VP , Ies. The electron temperature is obtained from the slope of thelinear part of the downward sloping portion of this curve. c Expanded viewof the ion current. The sloping dotted line is a linear fit to the ion current.The ion saturation current is found by extrapolating this line to the plasmapotential.1082 Am. J. Phys., Vol. 75, No. 12, December 2007the multidipole device in the University of Iowas under-graduate Advanced Physics Lab is shown in Fig. 5. Thischaracteristic was taken with a 6 mm diameter planar diskprobe in an argon plasma at a pressure of 0.5 mTorr. The topcurve Fig. 5a is the positive probe current due to electroncollection. As discussed in Sec. II see Fig. 4, the electroncurrent continues to increase slightly with increasing voltageabove the plasma potential due to the sheath expansion ef-fect. The determination of the electron saturation current andplasma potential is facilitated by replotting the current on asemilog scale as shown in Fig. 5b. The break point occursat VP4 V, with Ies= 1005 mA. The slope of the down-ward portion of the line on the semilog plot gives Te1.5 eV. The negative ion current is shown on an ex-panded scale in Fig. 5c. Again we see the sheath expansioneffect as the negative probe voltage increases. The ion satu-ration current is estimated by extrapolating the linear portionof the ion current to the plasma potential, where IiVP= 0.850.05 mA. The floating potential is also found fromFig. 5c as VfI=05 V. The ion and electron densitiescan now be calculated using Eqs. 2 and 4 with Te=1.5 eV. We find that ni= 8.30.51016 m3, and ne= 5.50.551016 m3. Even taking into account the uncer-tainties involved in measuring the saturation currents fromthe plots, there remains a 25% difference between theplasma density obtained from the ion and electron currents.This difference is a typical occurrence with Langmuir probesmeasurements. In a magnetized plasma, the discrepancy inthe densities obtained from the electron and ion saturation

Fig. 5. Langmuir probe I-V characteristic obtained in a multidipole plasmain argon at a pressure of 0.5 mTorr. a Electron current. b log IVB versusVB. The semilog plot of the electron current provides a clear demarcation ofthe plasma potential and electron saturation current. Te is found from theslope of the exponentially decreasing portion. c Expanded scale view ofthe ion current used to find Iis.1082Robert L. Merlino

currents is considerably larger, but expected. The gyroradiusof the electrons is typically much smaller than that of the

ions so that the collection of electrons is affected more thanthe collection of ions. In that case, measurements of theplasma density using the ion saturation current are more re-liable.

It is interesting to calculate the fraction of the neutral ar-gon atoms that are actually ionized in the plasma using themeasured value of the plasma density. This fraction is knownas the percent ionization or ionization fraction. The densityof the neutral argon atoms is na= P /kTg, where Tg is thetemperature of the neutral gas, and P is the neutral gas pres-sure. For P=5104 Torr and Tgas300 K, na=1.651019 m3. With ni=81016 m3, we obtain ni /na=0.005.Thus, only 0.5% of the neutral atoms are ionized.

The fact that the neutral density is roughly 1000 times theplasma density might lead one to wonder about the role ofthese neutral atoms on the plasma and probe measurements.To access the possible effects of collisions of the ions andelectrons with neutrals, we need to estimate a few typicalcollision mean free paths, = na1, where is the crosssection22 for the particular process considered. For ioniza-tion, ionz81020 m2 for 50 eV electrons on argon, soionz75 cm. Thus ionz is on the order of the dimensions ofa typical laboratory plasma device. The relatively long ion-ization mean free path explains, in part, the relatively lowvalue of the ionization percentageelectrons that are ener-getically capable of ionizing atoms are more likely to make itto the wall before ionizing an atom.

The purpose of the magnets on the walls of the multidi-pole device is to reflect the ionizing electrons back into theplasma, thus increasing their chances of having an ionizingcollision. Electrons can also make elastic collisions with neu-tral atoms; a typical cross section in this case is en1020 m2, giving en6 m. For collisions between theions and neutral atoms, the most important process to con-sider is charge exchange, Ar++ArAr+Ar+, in which anargon ion exchanges an electron with an argon atom, result-ing in the production of very slow argon ions and argonatoms with an energy practically equal to the initial energy ofthe ions. The cross section for this process is in51019 m2, giving in12 cm.

We note that for all of the processes considered the meanfree paths are much greater than the probe size and theshielding distance or sheath size, so that even though theneutral density far exceeds the plasma density, the neutralgas atoms produce negligible effects on the probe measure-ments.

B. A positive ion/negative ion plasma in a Q machineFigure 3 is an example of a Langmuir probe I-V charac-

teristic in a plasma in which the positive and negative par-ticles have the same mass. This example might appear to beexotic, but it is not difficult to produce a plasma havingalmost equal numbers of positive and negative ions of com-parable mass. We have produced positive ion/negative ionplasmas also known as electronegative plasmas in a devicecalled a Q machine.23 In a Q machine the plasma is producedby surface ionization, an effect discovered by Langmuir andKingdon in 1923.24 They found that cesium atoms that comeinto contact with a tungsten filament heated to 1200 Kemerge as cesium ions. The reason is that the ionizing poten-tial of cesium is 3.89 eV, and the work function for tungsten1083 Am. J. Phys., Vol. 75, No. 12, December 2007is 4.52 eV. Surface ionization is exploited in a Q machine23by directing an atomic beam of cesium or potassium atomsonto a hot 2000 K tungsten or tantalum plate, usuallyseveral centimeters in diameter. Both positive ions and ther-mionic electrons emerge from the plate forming a nearlyfully ionized plasma that is confined by a strong0.10.5 T longitudinal magnetic field. The relativelygood thermal contact between the plasma and the hot plateresults in a plasma in which both the electrons and positiveions are at roughly the plate temperature, typically 0.2 eV.The Q machine has been used mainly for studying the basicproperties of magnetized plasmas, and in particular plasmawaves. The Q designation refers to the expectation that athermally produced plasma would be quiescent, that is, rela-tively free of low frequency plasma instabilities.

Negative ions are readily formed in a Q machine plasmaby leaking into the vacuum chamber sulfur hexafluoride SF6at a pressure 105104 Torr. Electrons attach to SF6forming SF6

negative ions.25 The cross section for electronattachment to SF6 is energy dependent and peaks in the en-ergy range that coincides closely with that of the Q machineelectrons. Under these circumstances it is possible to produceplasmas in which the ratio of electron density to positive iondensity is ne /n+103. A Langmuir probe I-V characteristicobtained in such a K+ /SF6

m

/m+=3.7 is shown in Fig. 6.Note that the negative ion positive current and positive ionnegative current saturation currents are comparable. Withsuch a characteristic the plasma potential is most easilydetermined as the voltage at which the first derivative ofthe characteristic is a maximum. In this case we see thatVP1 V. The characteristic is roughly symmetric about I=0, with a floating potential VfVP, a result that is to beexpected in a plasma with n+n and nen+. When thenegative ion and positive ion densities are comparable, itmay even be possible to extract both the negative and posi-tive ion temperatures from the Langmuir characteristic.

Fig. 6. Langmuir probe I-V characteristic obtained in a singly ionized po-tassium plasma produced in a Q machine. SF6 gas was introduced into theplasma to form a negative ion plasma by electron attachment. A substantialfraction of the electrons became attached to the heavy SF6 molecules result-ing in a nearly symmetric probe characteristic with I+s Is. The lower curveis the derivative of the probe current, dI /dVB. The plasma potential is thevalue of the VB for which dI /dVB is a maximum.1083Robert L. Merlino

IV. COMMENTS 1T

= necn

1T + nehn 1T . A1

A Langmuir probe I-V characteristic becomes less confus-ing once we are able to see the individual current contribu-tions as well as the total probe current. The procedure forconstructing an I-V characteristic given an appropriate set ofplasma input parameters has been presented. A MAPLE pro-gram that creates the I-V characteristic is available onEPAPS16 and is also available on the authors website.

The inclusion of plasma physics experiments in upperlevel advanced laboratory courses for physics majors canprovide students with much exposure to many important top-ics and methods in experimental physics including basicvacuum techniques, vacuum measurement methods, solder-ing, spot welding, brazing, electronic circuit design and fab-rication, data acquisition methods, curve fitting techniques,and instrument design and construction building a Langmuirprobe. Students also experience using basic concepts in thekinetic theory of gases.

Plasma physics experiments also provide ideal researchtopics for undergraduate thesis projects. For instructors con-templating the inclusion of plasma experiments in advancedlaboratory courses, my suggestion is to start with the basicmultidipole plasma.1921 This device is relatively simple andinexpensive, with the most costly component being thevacuum pumping system. If money is not a concern, it ispossible to purchase fully operational vacuum systems thatare easily adaptable for plasma production. Although it isnow possible to purchase off the shelf Langmuir probe sys-tems, complete with probe and associated electronics, theexperience of constructing probes from scratch is a valuableone that should not be avoided. Building a probe is often thefirst instance in which students are required to use theirhands to create an experimental instrument. Far too oftenstudents are left with the impression that everything neededto perform a measurement can be found at manufacturersweb sites.

ACKNOWLEDGMENT

This work was supported by the U.S. Department ofEnergy Grant DE-FG02-04ER54795.

APPENDIX: SUGGESTED PROBLEMSFOR FURTHER STUDY

The following two problems are intended to extend thebasic probe theory to include some other important effectsoften encountered in using Langmuir probes in realistic plas-mas.

Problem 1. It is not uncommon to find in low pressureplasma discharges that there are two distinct Maxwellian dis-tributions of electronsa cold and hot distribution with tem-peratures Tec and Teh, respectively. Extend the analysis ofSec. II to include a two-temperature electron distribution. Inthis case the electron probe current is written as IeVB= IecVB+ IehVB. Take the respective densities of the coldand hot components to be nec and neh with ne=nec+neh. Tosimplify the analysis, introduce the parameter fehneh /ne asthe fraction of hot electrons, so that nec /ne=1 feh. An inter-esting issue arises as to what value of Te to use in calculatingthe Bohm ion current. It was shown26 that the appropriate Teis the harmonic average of Tec and Teh:1084 Am. J. Phys., Vol. 75, No. 12, December 2007e e ec e eh

After you have produced a Langmuir I-V plot, replot theelectron current as a semilog plot to see more clearly theeffect of the two-temperature electron distribution.

Problem 2. In plasmas produced in hot-filament dis-charges, the effect of the ionizing primary electrons on theprobe I-V trace can be observed, particularly at neutral pres-sures below 104 Torr. Extend the probe analysis to in-clude the presence of these energetic primary electrons,which can be modeled as an isotropic monoenergetic distri-bution. Express the total electron current as IetVB= IeVB+ IepVB, where IeVB is the contribution from the bulk elec-trons, and IepVB is the primary electron contribution, whichfor an isotropic monoenergetic distribution is3

Iep

=Iep

* 14

enepvepAprobe, VB VP,

Iep* 1 2eVP VB

mevep2 , Vp mevep22e VB VP,

0, VB VP mevep22e ,A2

where nep is the density of primary electrons, and vep=2Ep /me is the speed of the primary electrons with energyEp. To produce an I-V plot, assume that the primary electronsare accelerated through a potential drop 5060 V, and thedensity is in the range of 0.0010.1ne.

aElectronic mail: robert-merlino@uiowa.edu1It is common in plasma physics to give temperatures in equivalent energyunits eV. For example, we say that Te=2 eV, which means we are reallygiving kTe converted to electron volts. The actual temperature corre-sponding to 1 eV is 11,600 K.

2I. Langmuir and H. Mott-Smith, The theory of collectors in gaseousdischarges, Phys. Rev. 28, 727763 1926.

3I suggest that Langmuir probe novices start by reading Noah Hershkow-itzs article, How Langmuir probes work, in Plasma Diagnostics, Dis-charge Parameters and Chemistry, edited by O. Auciello and D. L.Flamm Academic, Boston, 1989, Vol. 1, Chap. 3.

4B. E. Cherrington, The use of Langmuir probes for plasma diagnostics:A review, Plasma Chem. Plasma Process. 2, 113140 1982.

5F. F. Chen, Electric Probes, in Plasma Diagnostic Techniques, edited byR. H. Huddlestone and S. L. Leonard Academic, New York, 1965,Chap. 4. A concise summary of Langmuir probe techniques by F. F. Chen,Lecture notes on Langmuir probe diagnostics is available atwww.ee.ucla.edu/~ffchen/Publs/Chen210R.pdf.

6L. Schott, Electrical probes, in Plasma Diagnostics, edited by W.Lochte-Holtgreven North-Holland, Amsterdam, 1968, Chap. 11.

7J. D. Swift and M. J. R. Schwar, Electrical Probes for Plasma Diagnos-tics American Elsevier, New York, 1969.

8I. H. Hutchinson, Principles of Plasma Diagnostics, 2nd ed. CambridgeU.P., Cambridge, 2002, Chap. 3.

9J. G. Laframboise, Theory of spherical and cylindrical Langmuir probesin a collisionless, Maxwellian plasma, Univ. Toronto Aerospace StudiesReport No. 11 1966.

10Reference 3, p. 118.11 It is common in discharge plasmas to have TiTe due to the fact that the

ions are created from neutral atoms at room temperature, while the elec-trons are considerable hotter by a factor of about 100 because they must1084Robert L. Merlino

be energized to ionization energies to maintain the discharge. Energytransfer between the massive ions and light electrons is inefficient, so the

Sheridans model and its verification, Phys. Plasmas 14, 033507-142007.

191085 Am. J. Phys., Vol. 75, No. 12, December 2007ions remain relatively cold.12Reference 3, p. 125; see also the review article by K.-U. Riemann, The

Bohn sheath criterion and sheath formation, J. Phys. D 24, 4935181991, and the recent article by G. D. Severn, A note on the plasmasheath and the Bohm criterion, Am. J. Phys. 75, 9294 2007.

13F. F. Chen, Introduction to Plasma Physics and Controlled Fusion, 2nded. Plenum, New York, 1984, Vol. 1, p. 290.

14Reference 13, p. 8.15M. A. Lieberman and A. J. Lichtenberg, Principles of Plasma Discharges

and Materials Processing, 2nd ed. Wiley, New York, 2005, Chap. 6.16See EPAPS Document No. E-AJPIAS-75-009710 for a MAPLE program

that can be used to produce Langmuir I-V curves. This document can bereached through a direct link in the online articles HTML referencesection or via the EPAPS homepage http://www.aip.org/pubservs/epaps.html. This program can also be accessed fromwww.physics.uiowa.edu/~rmerlino/.

17See, for example, T. Sunn Pedersen, A. H. Boozer, W. Dorland, J. P.Kremer, and R. Schmitt, Prospects for the creation of positron-electronplasmas in a non-neutral stellarator, J. Phys. B 36, 10291039 2003.

18D. Lee and N. Hershkowitz, Ion collection by planar Langmuir probes:

R. Limpaecher and K. R. MacKenzie, Magnetic multipole confinementof large uniform collisionless quiescent plasmas, Rev. Sci. Instrum. 44,726731 1973.

20A. Lang and N. Hershkowitz, Multidipole plasma density, J. Appl.Phys. 49, 47074710 1978.

21R. A. Bosch and R. L. Merlino, Multidipole confinement of argon andpotassium plasmas, Rev. Sci. Instrum. 57, 29402950 1986.

22A good source for cross sections relevant to plasma physics is S. C.Brown, Basic Data of Plasma Physics, 2nd ed. MIT Press, Cambridge,MA, 1967.

23R. W. Motley, Q Machines Academic, New York, 1975.24I. Langmuir and K. H. Kingdon, Thermionic effects caused by alkali

vapors in vacuum tubes, Science 57, 5860 1923.25Bin Song, D. Suszcynsky, N. DAngelo, and R. L. Merlino, Electrostatic

ion-cyclotron waves in a plasma with negative ions, Phys. Fluids B 1,23162318 1989.

26S. B. Song, C. S. Chang, and Duk-In Choi, Effect of two-temperatureelectron distribution on the Bohm sheath criterion, Phys. Rev. E 55,12131216 1997.1085Robert L. Merlino